Clinical Review

An 8-month-old boy with a history of hypotonia, developmental delay, and seizure disorder refractory to multiple anticonvulsant medications was brought to the ED with a two-week history of intermittent fever and poor oral intake. His current medications included sodium bromide (185 mg bid, orally) for his seizure disorder.

On physical examination, the boy appeared small for his age, with diffuse hypotonia and diminished reflexes. He was able to track with his eyes but was otherwise unresponsive. No rash was present. Results of initial laboratory studies were sodium, 144 mEq/L; potassium, 4.8 mEq/L; chloride, 179 mEq/L; bicarbonate, 21 mEq/L; blood urea nitrogen, 6 mg/dL; creatinine, 0.1 mg/dL; and glucose, 63 mg/dL. His anion gap (AG) was −56.

WHAT DOES THE ANION GAP REPRESENT?
The AG is a valuable clinical calculation derived from the measured extracellular electrolytes and provides an index of acid-base status.¹ Due to the necessity of electroneutrality, the sum of positive charges (cations) in the extracellular fluid must be balanced exactly with the sum of negative charges (anions). However, to routinely measure all of the cations and anions in the serum would be time-consuming and is also unnecessary. Because most clinical laboratories commonly only measure one relevant cation (sodium) and two anions (chloride and bicarbonate), the positive and negative sums are not completely balanced. The AG therefore refers to this difference (ie, AG = Na – [Cl + HCO₃]).

Of course, electroneutrality exists in vivo and is accomplished by the presence of unmeasured anions (UA; eg, lactate and phosphate) and unmeasured cations (UC; eg, potassium and calcium) not accounted for in the AG (ie, AG = UA – UC). In other words, the sum of measured plus unmeasured anions must equal the sum of the measured plus unmeasured cations.

WHAT CAUSES A LOW OR NEGATIVE ANION GAP?
While most health care providers are well versed in the clinical significance of an elevated AG (eg, MUDPILES [methanol, uremia, diabetic ketoacidosis, propylene glycol or phenformin, iron or isoniazid, lactate, ethylene glycol, salicylates]), the meaning of a low or negative AG is underappreciated. There are several scenarios that could potentially yield a low or negative AG, including decreased concentration of UA, increased concentrations of nonsodium cations (UC), and overestimation of serum chloride.

Decreased concentration of unmeasured anions. This most commonly occurs by two mechanisms: dilution of the extracellular fluid or hypoalbuminemia. The addition of water to the extracellular fluid will cause a proportionate dilution of all the measured electrolytes. Since the concentration of measured cations is higher than that of the measured anions, there is a small and relatively insignificant decrease in the AG.

Alternatively, hypoalbuminemia results in a low AG due to the change in UA; albumin is negatively charged. At physiologic pH, the overwhelming majority of serum proteins are anionic and counter-balanced by the positive charge of sodium. Albumin, the most abundant serum protein, accounts for approximately 75% of the normal AG. Hypoalbuminemic states, such as cirrhosis or nephrotic syndrome, can therefore cause low AG due to the retention of chloride to replace the lost negative charge. The albumin concentration can be corrected to calculate the AG.²

Nonsodium cations. There are a number of clinical conditions that result in the retention of nonsodium cations. For example, the excess positively charged paraproteins associated with IgG myeloma raise the UC concentration, resulting in a low AG. Similarly, elevations of unmeasured cationic electrolytes, such as calcium and magnesium, may also result in a lower AG. Significant changes in AG, though, are caused only by profound (and often life-threatening) hypercalcemia or hypermagnesemia.

Overestimation of serum chloride. Overestimation of serum chloride most commonly occurs in the clinical scenario of bromide exposure. In normal physiologic conditions, chloride is the only halide present in the extracellular fluid. With intake of brominated products, chloride may be partially replaced by bromide. As there is greater renal tubular avidity for bromide, chronic ingestion of bromide results in a gradual rise in serum bromide concentrations with a proportional fall in chloride.

However, and more importantly, bromide interferes with a number of laboratory techniques measuring chloride concentrations, resulting in a spuriously elevated chloride, or pseudohyperchloremia. Because the measured sodium and bicarbonate concentrations will remain unchanged, this falsely elevated chloride measurement will result in a negative AG.

Continue for causes of the falsely elevated chloride >>

WHAT CAUSES THE FALSELY ELEVATED CHLORIDE?
All of the current laboratory techniques for measurement of serum chloride concentration can potentially result in a falsely elevated value. However, the degree of pseudohyperchloremia will depend on the specific assay used for measurement. The ion-selective electrode method used by many common laboratory analyzers appears to have the greatest interference on chloride measurement in the presence of bromide. This is simply due to the molecular similarity of bromide and chloride.

Conversely, the coulometry method, often used as a reference standard, has the least interference of current laboratory methods.³ This is because coulometry does not completely rely on molecular structure to measure concentration; rather, it measures the amount of energy produced or consumed in an electrolysis reaction. Iodide, another halide compound, has also been described as a cause of pseudohyperchloremia, whereas fluoride does not seem to have significant interference.⁴

HOW ARE PATIENTS EXPOSED TO BROMIDE SALTS?
Bromide salts, specifically sodium bromide, are infrequently used to treat seizure disorders but are generally reserved for patients with epilepsy refractory to other, less toxic anticonvulsant medications. During the era when bromide salts were more commonly used to treat epilepsy, bromide intoxication, or bromism, was frequently observed.

Bromism may manifest as a constellation of nonspecific neurologic and psychiatric symptoms. These most commonly include headache, weakness, agitation, confusion, and hallucinations. In more severe cases of bromism, stupor and coma may occur.^3,5

Although bromide salts are no longer commonly prescribed, a number of products still contain brominated ingredients. Symptoms of bromide intoxication can occur with chronic use of a cough syrup containing dextromethorphan hydrobromide, as well as the brominated vegetable oils found in some soft drinks.⁵

Continue for how bromism is treated >>

HOW IS BROMISM TREATED?
The treatment of bromism involves preventing further exposure to bromide and promoting bromide excretion. Bromide has a long half-life (10 to 12 days), but in patients with normal renal function, it is possible to reduce this half-life to approximately three days with hydration and diuresis with sodium chloride.³

Alternatively, in patients with impaired renal function or severe intoxication, hemodialysis has been used effectively.⁵

CASE CONCLUSION
The patient was admitted for observation and treated with IV sodium chloride. After consultation with his neurologist, he was discharged home in the care of his parents, who were advised to continue him on sodium bromide (185 mg bid, orally) since his seizures were refractory to other anticonvulsant medications.

REFERENCES
1. Emmett M, Narins RG. Clinical use of the anion gap. Medicine (Baltimore). 1977;56(1):38-54.
2. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26(11):1807-1810.
3. Vasuyattakul S, Lertpattanasuwan N, Vareesangthip K, et al. A negative anion gap as a clue to diagnose bromide intoxication. Nephron. 1995;69(3):311-313.
4. Yamamoto K, Kobayashi H, Kobayashi T, Murakami S. False hyperchloremia in bromism. J Anesth. 1991;5(1):88-91.
5. Ng YY, Lin WL, Chen TW. Spurious hyperchloremia and decreased anion gap in a patient with dextromethorphan bromide. Am J Nephrol. 1992;12(4):268-270.

An 8-month-old boy with a history of hypotonia, developmental delay, and seizure disorder refractory to multiple anticonvulsant medications was brought to the ED with a two-week history of intermittent fever and poor oral intake. His current medications included sodium bromide (185 mg bid, orally) for his seizure disorder.

On physical examination, the boy appeared small for his age, with diffuse hypotonia and diminished reflexes. He was able to track with his eyes but was otherwise unresponsive. No rash was present. Results of initial laboratory studies were sodium, 144 mEq/L; potassium, 4.8 mEq/L; chloride, 179 mEq/L; bicarbonate, 21 mEq/L; blood urea nitrogen, 6 mg/dL; creatinine, 0.1 mg/dL; and glucose, 63 mg/dL. His anion gap (AG) was −56.

WHAT DOES THE ANION GAP REPRESENT?
The AG is a valuable clinical calculation derived from the measured extracellular electrolytes and provides an index of acid-base status.¹ Due to the necessity of electroneutrality, the sum of positive charges (cations) in the extracellular fluid must be balanced exactly with the sum of negative charges (anions). However, to routinely measure all of the cations and anions in the serum would be time-consuming and is also unnecessary. Because most clinical laboratories commonly only measure one relevant cation (sodium) and two anions (chloride and bicarbonate), the positive and negative sums are not completely balanced. The AG therefore refers to this difference (ie, AG = Na – [Cl + HCO₃]).

Of course, electroneutrality exists in vivo and is accomplished by the presence of unmeasured anions (UA; eg, lactate and phosphate) and unmeasured cations (UC; eg, potassium and calcium) not accounted for in the AG (ie, AG = UA – UC). In other words, the sum of measured plus unmeasured anions must equal the sum of the measured plus unmeasured cations.

WHAT CAUSES A LOW OR NEGATIVE ANION GAP?
While most health care providers are well versed in the clinical significance of an elevated AG (eg, MUDPILES [methanol, uremia, diabetic ketoacidosis, propylene glycol or phenformin, iron or isoniazid, lactate, ethylene glycol, salicylates]), the meaning of a low or negative AG is underappreciated. There are several scenarios that could potentially yield a low or negative AG, including decreased concentration of UA, increased concentrations of nonsodium cations (UC), and overestimation of serum chloride.

Decreased concentration of unmeasured anions. This most commonly occurs by two mechanisms: dilution of the extracellular fluid or hypoalbuminemia. The addition of water to the extracellular fluid will cause a proportionate dilution of all the measured electrolytes. Since the concentration of measured cations is higher than that of the measured anions, there is a small and relatively insignificant decrease in the AG.

Alternatively, hypoalbuminemia results in a low AG due to the change in UA; albumin is negatively charged. At physiologic pH, the overwhelming majority of serum proteins are anionic and counter-balanced by the positive charge of sodium. Albumin, the most abundant serum protein, accounts for approximately 75% of the normal AG. Hypoalbuminemic states, such as cirrhosis or nephrotic syndrome, can therefore cause low AG due to the retention of chloride to replace the lost negative charge. The albumin concentration can be corrected to calculate the AG.²

Nonsodium cations. There are a number of clinical conditions that result in the retention of nonsodium cations. For example, the excess positively charged paraproteins associated with IgG myeloma raise the UC concentration, resulting in a low AG. Similarly, elevations of unmeasured cationic electrolytes, such as calcium and magnesium, may also result in a lower AG. Significant changes in AG, though, are caused only by profound (and often life-threatening) hypercalcemia or hypermagnesemia.

Overestimation of serum chloride. Overestimation of serum chloride most commonly occurs in the clinical scenario of bromide exposure. In normal physiologic conditions, chloride is the only halide present in the extracellular fluid. With intake of brominated products, chloride may be partially replaced by bromide. As there is greater renal tubular avidity for bromide, chronic ingestion of bromide results in a gradual rise in serum bromide concentrations with a proportional fall in chloride.

However, and more importantly, bromide interferes with a number of laboratory techniques measuring chloride concentrations, resulting in a spuriously elevated chloride, or pseudohyperchloremia. Because the measured sodium and bicarbonate concentrations will remain unchanged, this falsely elevated chloride measurement will result in a negative AG.

Continue for causes of the falsely elevated chloride >>

WHAT CAUSES THE FALSELY ELEVATED CHLORIDE?
All of the current laboratory techniques for measurement of serum chloride concentration can potentially result in a falsely elevated value. However, the degree of pseudohyperchloremia will depend on the specific assay used for measurement. The ion-selective electrode method used by many common laboratory analyzers appears to have the greatest interference on chloride measurement in the presence of bromide. This is simply due to the molecular similarity of bromide and chloride.

Conversely, the coulometry method, often used as a reference standard, has the least interference of current laboratory methods.³ This is because coulometry does not completely rely on molecular structure to measure concentration; rather, it measures the amount of energy produced or consumed in an electrolysis reaction. Iodide, another halide compound, has also been described as a cause of pseudohyperchloremia, whereas fluoride does not seem to have significant interference.⁴

HOW ARE PATIENTS EXPOSED TO BROMIDE SALTS?
Bromide salts, specifically sodium bromide, are infrequently used to treat seizure disorders but are generally reserved for patients with epilepsy refractory to other, less toxic anticonvulsant medications. During the era when bromide salts were more commonly used to treat epilepsy, bromide intoxication, or bromism, was frequently observed.

Bromism may manifest as a constellation of nonspecific neurologic and psychiatric symptoms. These most commonly include headache, weakness, agitation, confusion, and hallucinations. In more severe cases of bromism, stupor and coma may occur.^3,5

Although bromide salts are no longer commonly prescribed, a number of products still contain brominated ingredients. Symptoms of bromide intoxication can occur with chronic use of a cough syrup containing dextromethorphan hydrobromide, as well as the brominated vegetable oils found in some soft drinks.⁵

Continue for how bromism is treated >>

HOW IS BROMISM TREATED?
The treatment of bromism involves preventing further exposure to bromide and promoting bromide excretion. Bromide has a long half-life (10 to 12 days), but in patients with normal renal function, it is possible to reduce this half-life to approximately three days with hydration and diuresis with sodium chloride.³

Alternatively, in patients with impaired renal function or severe intoxication, hemodialysis has been used effectively.⁵

CASE CONCLUSION
The patient was admitted for observation and treated with IV sodium chloride. After consultation with his neurologist, he was discharged home in the care of his parents, who were advised to continue him on sodium bromide (185 mg bid, orally) since his seizures were refractory to other anticonvulsant medications.

REFERENCES
1. Emmett M, Narins RG. Clinical use of the anion gap. Medicine (Baltimore). 1977;56(1):38-54.
2. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26(11):1807-1810.
3. Vasuyattakul S, Lertpattanasuwan N, Vareesangthip K, et al. A negative anion gap as a clue to diagnose bromide intoxication. Nephron. 1995;69(3):311-313.
4. Yamamoto K, Kobayashi H, Kobayashi T, Murakami S. False hyperchloremia in bromism. J Anesth. 1991;5(1):88-91.
5. Ng YY, Lin WL, Chen TW. Spurious hyperchloremia and decreased anion gap in a patient with dextromethorphan bromide. Am J Nephrol. 1992;12(4):268-270.

An 8-month-old boy with a history of hypotonia, developmental delay, and seizure disorder refractory to multiple anticonvulsant medications was brought to the ED with a two-week history of intermittent fever and poor oral intake. His current medications included sodium bromide (185 mg bid, orally) for his seizure disorder.

On physical examination, the boy appeared small for his age, with diffuse hypotonia and diminished reflexes. He was able to track with his eyes but was otherwise unresponsive. No rash was present. Results of initial laboratory studies were sodium, 144 mEq/L; potassium, 4.8 mEq/L; chloride, 179 mEq/L; bicarbonate, 21 mEq/L; blood urea nitrogen, 6 mg/dL; creatinine, 0.1 mg/dL; and glucose, 63 mg/dL. His anion gap (AG) was −56.

WHAT DOES THE ANION GAP REPRESENT?
The AG is a valuable clinical calculation derived from the measured extracellular electrolytes and provides an index of acid-base status.¹ Due to the necessity of electroneutrality, the sum of positive charges (cations) in the extracellular fluid must be balanced exactly with the sum of negative charges (anions). However, to routinely measure all of the cations and anions in the serum would be time-consuming and is also unnecessary. Because most clinical laboratories commonly only measure one relevant cation (sodium) and two anions (chloride and bicarbonate), the positive and negative sums are not completely balanced. The AG therefore refers to this difference (ie, AG = Na – [Cl + HCO₃]).

Of course, electroneutrality exists in vivo and is accomplished by the presence of unmeasured anions (UA; eg, lactate and phosphate) and unmeasured cations (UC; eg, potassium and calcium) not accounted for in the AG (ie, AG = UA – UC). In other words, the sum of measured plus unmeasured anions must equal the sum of the measured plus unmeasured cations.

WHAT CAUSES A LOW OR NEGATIVE ANION GAP?
While most health care providers are well versed in the clinical significance of an elevated AG (eg, MUDPILES [methanol, uremia, diabetic ketoacidosis, propylene glycol or phenformin, iron or isoniazid, lactate, ethylene glycol, salicylates]), the meaning of a low or negative AG is underappreciated. There are several scenarios that could potentially yield a low or negative AG, including decreased concentration of UA, increased concentrations of nonsodium cations (UC), and overestimation of serum chloride.

Decreased concentration of unmeasured anions. This most commonly occurs by two mechanisms: dilution of the extracellular fluid or hypoalbuminemia. The addition of water to the extracellular fluid will cause a proportionate dilution of all the measured electrolytes. Since the concentration of measured cations is higher than that of the measured anions, there is a small and relatively insignificant decrease in the AG.

Alternatively, hypoalbuminemia results in a low AG due to the change in UA; albumin is negatively charged. At physiologic pH, the overwhelming majority of serum proteins are anionic and counter-balanced by the positive charge of sodium. Albumin, the most abundant serum protein, accounts for approximately 75% of the normal AG. Hypoalbuminemic states, such as cirrhosis or nephrotic syndrome, can therefore cause low AG due to the retention of chloride to replace the lost negative charge. The albumin concentration can be corrected to calculate the AG.²

Nonsodium cations. There are a number of clinical conditions that result in the retention of nonsodium cations. For example, the excess positively charged paraproteins associated with IgG myeloma raise the UC concentration, resulting in a low AG. Similarly, elevations of unmeasured cationic electrolytes, such as calcium and magnesium, may also result in a lower AG. Significant changes in AG, though, are caused only by profound (and often life-threatening) hypercalcemia or hypermagnesemia.

Overestimation of serum chloride. Overestimation of serum chloride most commonly occurs in the clinical scenario of bromide exposure. In normal physiologic conditions, chloride is the only halide present in the extracellular fluid. With intake of brominated products, chloride may be partially replaced by bromide. As there is greater renal tubular avidity for bromide, chronic ingestion of bromide results in a gradual rise in serum bromide concentrations with a proportional fall in chloride.

However, and more importantly, bromide interferes with a number of laboratory techniques measuring chloride concentrations, resulting in a spuriously elevated chloride, or pseudohyperchloremia. Because the measured sodium and bicarbonate concentrations will remain unchanged, this falsely elevated chloride measurement will result in a negative AG.

Continue for causes of the falsely elevated chloride >>

WHAT CAUSES THE FALSELY ELEVATED CHLORIDE?
All of the current laboratory techniques for measurement of serum chloride concentration can potentially result in a falsely elevated value. However, the degree of pseudohyperchloremia will depend on the specific assay used for measurement. The ion-selective electrode method used by many common laboratory analyzers appears to have the greatest interference on chloride measurement in the presence of bromide. This is simply due to the molecular similarity of bromide and chloride.

Conversely, the coulometry method, often used as a reference standard, has the least interference of current laboratory methods.³ This is because coulometry does not completely rely on molecular structure to measure concentration; rather, it measures the amount of energy produced or consumed in an electrolysis reaction. Iodide, another halide compound, has also been described as a cause of pseudohyperchloremia, whereas fluoride does not seem to have significant interference.⁴

HOW ARE PATIENTS EXPOSED TO BROMIDE SALTS?
Bromide salts, specifically sodium bromide, are infrequently used to treat seizure disorders but are generally reserved for patients with epilepsy refractory to other, less toxic anticonvulsant medications. During the era when bromide salts were more commonly used to treat epilepsy, bromide intoxication, or bromism, was frequently observed.

Bromism may manifest as a constellation of nonspecific neurologic and psychiatric symptoms. These most commonly include headache, weakness, agitation, confusion, and hallucinations. In more severe cases of bromism, stupor and coma may occur.^3,5

Although bromide salts are no longer commonly prescribed, a number of products still contain brominated ingredients. Symptoms of bromide intoxication can occur with chronic use of a cough syrup containing dextromethorphan hydrobromide, as well as the brominated vegetable oils found in some soft drinks.⁵

Continue for how bromism is treated >>

HOW IS BROMISM TREATED?
The treatment of bromism involves preventing further exposure to bromide and promoting bromide excretion. Bromide has a long half-life (10 to 12 days), but in patients with normal renal function, it is possible to reduce this half-life to approximately three days with hydration and diuresis with sodium chloride.³

Alternatively, in patients with impaired renal function or severe intoxication, hemodialysis has been used effectively.⁵

CASE CONCLUSION
The patient was admitted for observation and treated with IV sodium chloride. After consultation with his neurologist, he was discharged home in the care of his parents, who were advised to continue him on sodium bromide (185 mg bid, orally) since his seizures were refractory to other anticonvulsant medications.

REFERENCES
1. Emmett M, Narins RG. Clinical use of the anion gap. Medicine (Baltimore). 1977;56(1):38-54.
2. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26(11):1807-1810.
3. Vasuyattakul S, Lertpattanasuwan N, Vareesangthip K, et al. A negative anion gap as a clue to diagnose bromide intoxication. Nephron. 1995;69(3):311-313.
4. Yamamoto K, Kobayashi H, Kobayashi T, Murakami S. False hyperchloremia in bromism. J Anesth. 1991;5(1):88-91.
5. Ng YY, Lin WL, Chen TW. Spurious hyperchloremia and decreased anion gap in a patient with dextromethorphan bromide. Am J Nephrol. 1992;12(4):268-270.

There’s a cyclical lament in obstetrics, and it goes something like this: Forceps are waning and are going to fade away completely if something isn’t done about it. This lament resounds every few decades, as a look at the literature confirms:

In this, our latest cycle of lament, 4 or 5 papers have suggested that forceps in general and Kielland forceps in particular ought not be abandoned because outcomes are better than those suggested by the older literature. With the cesarean delivery rate hovering at about 31% in the United States, perhaps it is time to revisit the issue.

This Update is not intended to be a comprehensive review of the literature. Rather, it offers a snapshot of articles published within the past year—articles that highlight some new features of a very old debate:

Forceps and vacuum device placement is “suboptimal” in almost 30% of operative vaginal deliveries
Ramphul M, Kennelly MM, Burke G, Murphy DJ. Risk factors and morbidity associated with suboptimal instrument placement at instrumental delivery: observational study nested within the Instrumental Delivery & Ultrasound randomised controlled trial ISRCTN 72230496. BJOG. 2015;122(4):558–563.

Rouse DJ. Instrument placement is sub-optimal in three of ten attempted operative vaginal deliveries. BJOG. 2015;122(4):564.

Over the years, many clinicians have argued that we don’t do enough forceps deliveries to maintain our own competence with the procedure, let alone teach residents how to perform it. This observational study nested in a randomized clinical trial is intriguing because Ramphul and colleagues looked for objective evidence of clinicians’ skill at the vacuum and forceps. Specifically, they looked for evidence that the forceps or vacuum was malpositioned during attempts at operative vaginal delivery. In the process, they nicely documented the absolute rate of malpositioning of the forceps and vacuum, finding that it is much higher than expected, even in an institution that performs a lot of operative vaginal deliveries.

Details of the trial
A cohort of 478 nulliparous women at term (≥37 weeks) underwent instrumental delivery at 2 university-affiliated maternity hospitals in Ireland. Ramphul and colleagues documented fetal head position prior to application of the instrument and at delivery. The midwife or neonatologist attending each delivery examined the neonate after birth and recorded the markings of the instrument on the infant’s head to determine whether instrument placement had been optimal.

Instrument placement was considered optimal when the vacuum cup included the flexion point (3 cm anterior to the posterior fontanelle) and the posterior fontanelle, with central placement. For forceps, instrument placement was considered optimal when the blades were positioned bilaterally and symmetrically over the malar bones. Two main types of forceps were used in this study—direct-traction Neville Barnes forceps (n = 138) and rotational Kielland forceps (n = 13)—and the rates of optimal and suboptimal placement were similar between them. 

Each case was labeled as “optimal” or “suboptimal” by 2 investigators, with a third observer arbitrating when the 2 investigators differed in opinion.

Instrument placement was clearly documented in 478 deliveries, 138 of which (28.8%) involved suboptimal placement. There was a lower rate of induction of labor among deliveries with suboptimal placement (42.8% vs 53.2%; odds ratio [OR], 0.66; 95% confidence interval [CI], 0.44–0.98; P = .038). There were no differences between the optimal and suboptimal groups in terms of duration of labor, use of oxytocin, and analgesia. In addition, the seniority of obstetricians performing operative vaginal delivery was similar between groups.

Fetal malposition was more common in the suboptimal group (58.7% vs 37.4%; OR, 2.44; 95% CI, 1.62–3.66; P<.0001). Midcavity station also was more common in the suboptimal group (82.6% vs 73.8%; OR, 1.68; 95% CI, 1.02–2.78; P = .042).

Maternal and neonatal outcomes
Postpartum hemorrhage was more common in the suboptimal placement group (24.6% vs 14.4%; OR, 1.94; 95% CI, 1.19–3.17; P = .008), as was prolonged hospitalization (26.8% vs 14.7%; OR, 2.13; 95% CI, 1.31–3.44; P = .02).

In addition, the incidence of neonatal trauma was higher in the group with suboptimal placement (15.9% vs 3.9%; OR, 4.64; 95% CI, 2.25–9.58; P<.0001) and included such effects as Erb’s palsy, fracture, retinal hemorrhage, cephalhematoma, and cerebral hemorrhage.

After adjustment for potential confounding factors, including induction of labor, seniority of the obstetrician, fetal malposition, caput above +1, midcavity station, regional analgesia, and the instrument used, the association remained significant between suboptimal placement and prolonged hospitalization (adjusted OR, 2.28; 95% CI, 1.30–4.02) and neonatal trauma (adjusted OR, 4.25; 95% CI, 1.85–9.72).

Dwindling statistics for operative vaginal delivery
In an editorial accompanying the study by Ramphul and colleagues, Dwight J. Rouse, MD,points to the waning of instrumental vaginal delivery in many parts of the world, most notably the United States, where, in 2012, only 2.8% of live births involved use of a vacuum device and only 0.6% involved the forceps.⁵

“When the rate of cesarean delivery is 10 times the combined rate of vaginal vacuum and forceps delivery (as it is in the USA), it is fair to argue that operative vaginal delivery is underutilized,” Dr. Rouse writes. “So kudos to Ramphul et al for providing insight into how we might continue to perform operative vaginal delivery safely.”

What this EVIDENCE means for practice
The study by Ramphul and colleagues very clearly confirms that correct placement of the vacuum device or forceps is key to safety.

Should we continue forceps education using the apprenticeship model of training?
Kyser KL, Lu X, Santillan D, et al. Forceps delivery volumes in teaching and nonteaching hospitals: Are volumes sufficient for physicians to acquire and maintain competence? Acad Med. 2014;89(1):71–76.

Ericsson KA. Necessity is the mother of invention: ­video recording firsthand perspectives of critical medical procedures to make simulated training more effective. Acad Med. 2014;89(1):17–20.

Kyser and colleagues have provided the best current snapshot of the opportunity for teaching instrumental vaginal delivery in the United States. They conducted a retrospective cohort study using new state inpatient data from 9 states in diverse geographic locations to capture experience at large and small teaching hospitals, as well as nonteaching institutions. They demonstrated that the opportunity for hands-on experience with these difficult and technically demanding deliveries is extremely limited and probably insufficient for all practicing physicians to maintain their skills if we continue to rely on traditional ways of teaching.

Details of the study
Using State Inpatient Data from 9 states, Kyser and colleagues identified all women hospitalized for childbirth in 2008. Of 1,344,305 deliveries in 835 hospitals, the final cohort included 624,000 operative deliveries—424,224 cesarean deliveries, 174,036 vacuum extractions, 6,158 forceps deliveries, and 19,582 deliveries that required more than 1 method. Of the 835 hospitals in this study, 68 were major teaching hospitals, 130 were minor teaching facilities, and 637 were nonteaching institutions.

The mean annual volumes for cesarean delivery for major teaching, minor teaching, and nonteaching hospitals were 969.8, 757.8, and 406.9, respectively (P<.0001).

The mean annual volumes for vacuum delivery were 301.0, 304.2, and 190.4, respectively (P<.0001).

The mean annual volumes for forceps delivery were 25.2, 15.3, and 8.9, respectively (P<.0001).

Three hundred twenty hospitals (38.3% of all hospitals) failed to perform a single forceps delivery in 2008, including 11 major teaching hospitals (16.2% of major teaching hospitals), 30 minor teaching hospitals (23.1% of minor teaching hospitals), and 279 nonteaching hospitals (43.8% of nonteaching hospitals) (P<.0001).

We need to rethink the apprenticeship model
In a commentary accompanying the study by Kyser and colleagues, K. Anders ­Ericsson, PhD, revisits the “see one, do one, teach one” model that has long characterized medical education. “Both the limitations on learning opportunities available in the clinics and the restrictions on resident work hours have created a real problem for the traditional apprenticeship model for training doctors,” he writes.

Ericsson notes that other specialists, such as concert musicians, chess players, and professional athletes do not learn using an apprenticeship model. For example, chess players do not play game after game of chess to become expert. And when a game is concluded, usually after several hours have passed, they are unlikely to be aware of the specific moves that lost or won them the game (unless an observer points them out). That is why, when training, chess players focus on particular aspects of the game (often identified by a mentor) as being crucial to improve their overall performance.

In today’s chess-learning environment, Ericsson notes, the computer plays a key role and can provide accurate feedback on each move the player executes. Computer chess programs have evolved to the point that they “are far superior in skill to any human chess player. Most important, computers can provide more accurate feedback on each chess move and are available at any time for practice,” writes Ericsson.

The same is true in sports. A tennis player does not practice by playing an endless series of games—though an ability to win a game is the ultimate goal. Rather, the athlete focuses on aspects of the game—the serve, for example—that can make the difference between winning or losing. Ericsson also notes that most musicians, dancers, and athletes “spend most of their time training by themselves to get ready to exhibit their skills for the first time in front of a large audience.”

These approaches are a better model for improving performance than the apprenticeship model, Ericsson argues. In medicine, one alternative might be the video recording of medical procedures in the clinic from multiple points of view—so that later viewers get both the “big picture” and a close-up view from the point of technical performance. After the recording is digitized and stored on a server, it can serve as valuable teaching for an unlimited number of residents.

Simulator training offers another venue for education, as it makes possible the isolation of difficult aspects of a procedure, which can then be repeated by the trainee as many times as necessary. In the future, it should be possible to link video recordings directly to simulators “so trainees could focus on particular aspects of the procedures and be required to respond to prompts with recordable actions,” Ericsson writes.

What this EVIDENCE means for practice
Given the extremely limited opportunities for observing forceps deliveries in the United States, it is time for us to explore new avenues for teaching other than the traditional apprenticeship model.

Is ultrasound evidence of levator muscle “avulsion” a real anatomic entity?
Dietz HP. Forceps: toward obsolescence or revival? Acta Obstet Gynecol Scand. 2015;94(4):347–351.

Da Silva AS, Digesu GA, Dell’Utri C, Fritsch H, ­Piffarotti P, Khullar V. Do ultrasound findings of levator ani “avulsion” correlate with anatomical findings? A multicenter cadaveric study [published online ahead of print May 15, 2015]. Neurourol Urodyn. doi:10.1002 /nau.22781.

Dietz takes a new tack in the debate over cesarean versus forceps by pointing to a recently highlighted abnormality in women who deliver by forceps: levator ani muscle avulsion, or LMA—traumatic disconnection of the levator ani from the pelvic sidewall. It has long been known that forceps deliveries can increase the risk of obstetric anal sphincter injuries (OASIS). Dietz contends that OASIS occurs at a rate as high as 40% to 60% after forceps delivery. He also notes, with some consternation, that the American ­College of Obstetricians and Gynecologists and the Society for Maternal-Fetal Medicine have advocated forceps as a way of reducing the high cesarean delivery rate.

When a parturient has been pushing for an extended period of time and there is a positional abnormality of the fetus, such as persistent occiput posterior position, cesarean delivery is often favored as a way of protecting the rectal sphincter. Dietz argues that cesarean delivery also protects against LMA, which “has only recently been recognized as a major etiological factor in pelvic floor dysfunction.” Dietz then presents a list of studies that have produced ultrasound findings of LMA in a high percentage of women undergoing forceps delivery—percentages on the order of 10% to 40%.

Enter Da Silva and colleagues, who argue that “the only true place to visualize the 3D structure of the human body, [and] thus validate imaging findings, [is] on cadaveric or live tissue dissections.” They undertook a cadaveric study to validate—or not—some of the findings of LMA summarized by Dietz.

Details of the study
The pubovisceral muscle (PVM) anatomy of 30 female cadavers was analyzed via 3D translabial ultrasonography to confirm LMA. The cadavers were then dissected to assess the finding anatomically. Da Silva and colleagues found LMA on imaging in 11 (36.7%) cadavers. LMA was unilateral in 10 (33.3%) cadavers and bilateral in 1 (3.3%). However, no LMA was found at dissection.

When an additional 39 cadavers were dissected, no LMA was identified.

On ultrasound, LMA is strongly associated with a narrower PVM insertion depth (mean of 4.79 mm vs 6.32 mm; P = .001). Da Silva and colleagues concluded that “there is a clear difference between anatomical and ultrasonographic findings. The imaged appearance of an ‘avulsion’ does not represent a true anatomical ‘avulsion’ as confirmed on dissection.”

What this EVIDENCE means for practice
Before we prematurely adopt ultrasound evidence of LMA as a significant morbidity, we need to learn more about its true etiology, pathophysiology, and epidemiology. We don’t yet know enough to say that it’s such a bad injury, when imaged via ultrasound, that it warrants cesarean delivery to avoid it.

When deciding between cesarean and forceps, keep the risks of second-stage cesarean in mind
Pergialiotis V, Vlachos DG, Rodolakis A, Haidopoulos D, Thomakos N, Vlachos GD. First versus second stage C/S maternal and neonatal morbidity: a systematic review and meta-analysis. Eur J Obstet Gynecol Reprod Biol. 2014;175:15–24.

This expert systematic review and meta-analysis summarizes the morbidity of second-stage cesarean delivery. When an obstetrician has a patient who is arrested at persistent occiput posterior position, say, and is trying to decide on cesarean delivery versus Kielland’s rotation or other forceps delivery, it is necessary to balance the risks and benefits of the 2 options. And as all clinicians are aware, when cesarean delivery is performed late in labor and the patient has been pushing for a prolonged period of time in the second stage—cesarean can be a challenging procedure. Moreover, these late cesareans are associated with much greater risks than cesarean deliveries performed earlier in labor.

Details of the review
Pergialiotis and colleagues selected 10 studies comparing maternal and neonatal morbidity and mortality between cesarean delivery at full dilatation and cesarean delivery prior to full dilatation. These studies involved 23,104 women with a singleton fetus who underwent cesarean delivery in the first (n = 18,160) or second (n = 4,944) stage of labor.

They found that second-stage cesarean was associated with a higher rate of maternal death (OR, 7.96; 95% CI, 1.61–39.39), a higher rate of maternal admission to the intensive care unit (OR, 7.41; 95% CI, 2.47–22.5), and a higher maternal transfusion rate (OR, 2.60; 95% CI, 1.49–2.54).

The rate of neonatal death also was higher among second-stage cesareans (OR, 5.20; 95% CI, 2.49–10.85), as was admission to the neonatal intensive care unit (OR, 1.63; 95% CI, 0.91–2.91), and the 5-minute Apgar score was more likely to be less than 7 (OR, 2.77; 95% CI, 1.02–7.50).

According to the authors, this study is the “first systematic review and meta-­analysis that investigates the impact of the stage of labor on maternal and neonatal outcomes among women delivering by cesarean section.” The findings demonstrate with authority that second-stage cesareans can be a risky undertaking.

What this EVIDENCE means for practice
Cesareans performed late in the second stage of labor are distinct from those performed in the first stage, carrying much higher risks, especially for the mother. When deciding whether to proceed with cesarean, vacuum, or forceps, the added risk of second-stage cesarean is an important aspect of both the consent conversation and clinical decision making.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

There’s a cyclical lament in obstetrics, and it goes something like this: Forceps are waning and are going to fade away completely if something isn’t done about it. This lament resounds every few decades, as a look at the literature confirms:

In this, our latest cycle of lament, 4 or 5 papers have suggested that forceps in general and Kielland forceps in particular ought not be abandoned because outcomes are better than those suggested by the older literature. With the cesarean delivery rate hovering at about 31% in the United States, perhaps it is time to revisit the issue.

This Update is not intended to be a comprehensive review of the literature. Rather, it offers a snapshot of articles published within the past year—articles that highlight some new features of a very old debate:

Forceps and vacuum device placement is “suboptimal” in almost 30% of operative vaginal deliveries
Ramphul M, Kennelly MM, Burke G, Murphy DJ. Risk factors and morbidity associated with suboptimal instrument placement at instrumental delivery: observational study nested within the Instrumental Delivery & Ultrasound randomised controlled trial ISRCTN 72230496. BJOG. 2015;122(4):558–563.

Rouse DJ. Instrument placement is sub-optimal in three of ten attempted operative vaginal deliveries. BJOG. 2015;122(4):564.

Over the years, many clinicians have argued that we don’t do enough forceps deliveries to maintain our own competence with the procedure, let alone teach residents how to perform it. This observational study nested in a randomized clinical trial is intriguing because Ramphul and colleagues looked for objective evidence of clinicians’ skill at the vacuum and forceps. Specifically, they looked for evidence that the forceps or vacuum was malpositioned during attempts at operative vaginal delivery. In the process, they nicely documented the absolute rate of malpositioning of the forceps and vacuum, finding that it is much higher than expected, even in an institution that performs a lot of operative vaginal deliveries.

Details of the trial
A cohort of 478 nulliparous women at term (≥37 weeks) underwent instrumental delivery at 2 university-affiliated maternity hospitals in Ireland. Ramphul and colleagues documented fetal head position prior to application of the instrument and at delivery. The midwife or neonatologist attending each delivery examined the neonate after birth and recorded the markings of the instrument on the infant’s head to determine whether instrument placement had been optimal.

Instrument placement was considered optimal when the vacuum cup included the flexion point (3 cm anterior to the posterior fontanelle) and the posterior fontanelle, with central placement. For forceps, instrument placement was considered optimal when the blades were positioned bilaterally and symmetrically over the malar bones. Two main types of forceps were used in this study—direct-traction Neville Barnes forceps (n = 138) and rotational Kielland forceps (n = 13)—and the rates of optimal and suboptimal placement were similar between them. 

Each case was labeled as “optimal” or “suboptimal” by 2 investigators, with a third observer arbitrating when the 2 investigators differed in opinion.

Instrument placement was clearly documented in 478 deliveries, 138 of which (28.8%) involved suboptimal placement. There was a lower rate of induction of labor among deliveries with suboptimal placement (42.8% vs 53.2%; odds ratio [OR], 0.66; 95% confidence interval [CI], 0.44–0.98; P = .038). There were no differences between the optimal and suboptimal groups in terms of duration of labor, use of oxytocin, and analgesia. In addition, the seniority of obstetricians performing operative vaginal delivery was similar between groups.

Fetal malposition was more common in the suboptimal group (58.7% vs 37.4%; OR, 2.44; 95% CI, 1.62–3.66; P<.0001). Midcavity station also was more common in the suboptimal group (82.6% vs 73.8%; OR, 1.68; 95% CI, 1.02–2.78; P = .042).

Maternal and neonatal outcomes
Postpartum hemorrhage was more common in the suboptimal placement group (24.6% vs 14.4%; OR, 1.94; 95% CI, 1.19–3.17; P = .008), as was prolonged hospitalization (26.8% vs 14.7%; OR, 2.13; 95% CI, 1.31–3.44; P = .02).

In addition, the incidence of neonatal trauma was higher in the group with suboptimal placement (15.9% vs 3.9%; OR, 4.64; 95% CI, 2.25–9.58; P<.0001) and included such effects as Erb’s palsy, fracture, retinal hemorrhage, cephalhematoma, and cerebral hemorrhage.

After adjustment for potential confounding factors, including induction of labor, seniority of the obstetrician, fetal malposition, caput above +1, midcavity station, regional analgesia, and the instrument used, the association remained significant between suboptimal placement and prolonged hospitalization (adjusted OR, 2.28; 95% CI, 1.30–4.02) and neonatal trauma (adjusted OR, 4.25; 95% CI, 1.85–9.72).

Dwindling statistics for operative vaginal delivery
In an editorial accompanying the study by Ramphul and colleagues, Dwight J. Rouse, MD,points to the waning of instrumental vaginal delivery in many parts of the world, most notably the United States, where, in 2012, only 2.8% of live births involved use of a vacuum device and only 0.6% involved the forceps.⁵

“When the rate of cesarean delivery is 10 times the combined rate of vaginal vacuum and forceps delivery (as it is in the USA), it is fair to argue that operative vaginal delivery is underutilized,” Dr. Rouse writes. “So kudos to Ramphul et al for providing insight into how we might continue to perform operative vaginal delivery safely.”

What this EVIDENCE means for practice
The study by Ramphul and colleagues very clearly confirms that correct placement of the vacuum device or forceps is key to safety.

Should we continue forceps education using the apprenticeship model of training?
Kyser KL, Lu X, Santillan D, et al. Forceps delivery volumes in teaching and nonteaching hospitals: Are volumes sufficient for physicians to acquire and maintain competence? Acad Med. 2014;89(1):71–76.

Ericsson KA. Necessity is the mother of invention: ­video recording firsthand perspectives of critical medical procedures to make simulated training more effective. Acad Med. 2014;89(1):17–20.

Kyser and colleagues have provided the best current snapshot of the opportunity for teaching instrumental vaginal delivery in the United States. They conducted a retrospective cohort study using new state inpatient data from 9 states in diverse geographic locations to capture experience at large and small teaching hospitals, as well as nonteaching institutions. They demonstrated that the opportunity for hands-on experience with these difficult and technically demanding deliveries is extremely limited and probably insufficient for all practicing physicians to maintain their skills if we continue to rely on traditional ways of teaching.

Details of the study
Using State Inpatient Data from 9 states, Kyser and colleagues identified all women hospitalized for childbirth in 2008. Of 1,344,305 deliveries in 835 hospitals, the final cohort included 624,000 operative deliveries—424,224 cesarean deliveries, 174,036 vacuum extractions, 6,158 forceps deliveries, and 19,582 deliveries that required more than 1 method. Of the 835 hospitals in this study, 68 were major teaching hospitals, 130 were minor teaching facilities, and 637 were nonteaching institutions.

The mean annual volumes for cesarean delivery for major teaching, minor teaching, and nonteaching hospitals were 969.8, 757.8, and 406.9, respectively (P<.0001).

The mean annual volumes for vacuum delivery were 301.0, 304.2, and 190.4, respectively (P<.0001).

The mean annual volumes for forceps delivery were 25.2, 15.3, and 8.9, respectively (P<.0001).

Three hundred twenty hospitals (38.3% of all hospitals) failed to perform a single forceps delivery in 2008, including 11 major teaching hospitals (16.2% of major teaching hospitals), 30 minor teaching hospitals (23.1% of minor teaching hospitals), and 279 nonteaching hospitals (43.8% of nonteaching hospitals) (P<.0001).

We need to rethink the apprenticeship model
In a commentary accompanying the study by Kyser and colleagues, K. Anders ­Ericsson, PhD, revisits the “see one, do one, teach one” model that has long characterized medical education. “Both the limitations on learning opportunities available in the clinics and the restrictions on resident work hours have created a real problem for the traditional apprenticeship model for training doctors,” he writes.

Ericsson notes that other specialists, such as concert musicians, chess players, and professional athletes do not learn using an apprenticeship model. For example, chess players do not play game after game of chess to become expert. And when a game is concluded, usually after several hours have passed, they are unlikely to be aware of the specific moves that lost or won them the game (unless an observer points them out). That is why, when training, chess players focus on particular aspects of the game (often identified by a mentor) as being crucial to improve their overall performance.

In today’s chess-learning environment, Ericsson notes, the computer plays a key role and can provide accurate feedback on each move the player executes. Computer chess programs have evolved to the point that they “are far superior in skill to any human chess player. Most important, computers can provide more accurate feedback on each chess move and are available at any time for practice,” writes Ericsson.

The same is true in sports. A tennis player does not practice by playing an endless series of games—though an ability to win a game is the ultimate goal. Rather, the athlete focuses on aspects of the game—the serve, for example—that can make the difference between winning or losing. Ericsson also notes that most musicians, dancers, and athletes “spend most of their time training by themselves to get ready to exhibit their skills for the first time in front of a large audience.”

These approaches are a better model for improving performance than the apprenticeship model, Ericsson argues. In medicine, one alternative might be the video recording of medical procedures in the clinic from multiple points of view—so that later viewers get both the “big picture” and a close-up view from the point of technical performance. After the recording is digitized and stored on a server, it can serve as valuable teaching for an unlimited number of residents.

Simulator training offers another venue for education, as it makes possible the isolation of difficult aspects of a procedure, which can then be repeated by the trainee as many times as necessary. In the future, it should be possible to link video recordings directly to simulators “so trainees could focus on particular aspects of the procedures and be required to respond to prompts with recordable actions,” Ericsson writes.

What this EVIDENCE means for practice
Given the extremely limited opportunities for observing forceps deliveries in the United States, it is time for us to explore new avenues for teaching other than the traditional apprenticeship model.

Is ultrasound evidence of levator muscle “avulsion” a real anatomic entity?
Dietz HP. Forceps: toward obsolescence or revival? Acta Obstet Gynecol Scand. 2015;94(4):347–351.

Da Silva AS, Digesu GA, Dell’Utri C, Fritsch H, ­Piffarotti P, Khullar V. Do ultrasound findings of levator ani “avulsion” correlate with anatomical findings? A multicenter cadaveric study [published online ahead of print May 15, 2015]. Neurourol Urodyn. doi:10.1002 /nau.22781.

Dietz takes a new tack in the debate over cesarean versus forceps by pointing to a recently highlighted abnormality in women who deliver by forceps: levator ani muscle avulsion, or LMA—traumatic disconnection of the levator ani from the pelvic sidewall. It has long been known that forceps deliveries can increase the risk of obstetric anal sphincter injuries (OASIS). Dietz contends that OASIS occurs at a rate as high as 40% to 60% after forceps delivery. He also notes, with some consternation, that the American ­College of Obstetricians and Gynecologists and the Society for Maternal-Fetal Medicine have advocated forceps as a way of reducing the high cesarean delivery rate.

When a parturient has been pushing for an extended period of time and there is a positional abnormality of the fetus, such as persistent occiput posterior position, cesarean delivery is often favored as a way of protecting the rectal sphincter. Dietz argues that cesarean delivery also protects against LMA, which “has only recently been recognized as a major etiological factor in pelvic floor dysfunction.” Dietz then presents a list of studies that have produced ultrasound findings of LMA in a high percentage of women undergoing forceps delivery—percentages on the order of 10% to 40%.

Enter Da Silva and colleagues, who argue that “the only true place to visualize the 3D structure of the human body, [and] thus validate imaging findings, [is] on cadaveric or live tissue dissections.” They undertook a cadaveric study to validate—or not—some of the findings of LMA summarized by Dietz.

Details of the study
The pubovisceral muscle (PVM) anatomy of 30 female cadavers was analyzed via 3D translabial ultrasonography to confirm LMA. The cadavers were then dissected to assess the finding anatomically. Da Silva and colleagues found LMA on imaging in 11 (36.7%) cadavers. LMA was unilateral in 10 (33.3%) cadavers and bilateral in 1 (3.3%). However, no LMA was found at dissection.

When an additional 39 cadavers were dissected, no LMA was identified.

On ultrasound, LMA is strongly associated with a narrower PVM insertion depth (mean of 4.79 mm vs 6.32 mm; P = .001). Da Silva and colleagues concluded that “there is a clear difference between anatomical and ultrasonographic findings. The imaged appearance of an ‘avulsion’ does not represent a true anatomical ‘avulsion’ as confirmed on dissection.”

What this EVIDENCE means for practice
Before we prematurely adopt ultrasound evidence of LMA as a significant morbidity, we need to learn more about its true etiology, pathophysiology, and epidemiology. We don’t yet know enough to say that it’s such a bad injury, when imaged via ultrasound, that it warrants cesarean delivery to avoid it.

When deciding between cesarean and forceps, keep the risks of second-stage cesarean in mind
Pergialiotis V, Vlachos DG, Rodolakis A, Haidopoulos D, Thomakos N, Vlachos GD. First versus second stage C/S maternal and neonatal morbidity: a systematic review and meta-analysis. Eur J Obstet Gynecol Reprod Biol. 2014;175:15–24.

This expert systematic review and meta-analysis summarizes the morbidity of second-stage cesarean delivery. When an obstetrician has a patient who is arrested at persistent occiput posterior position, say, and is trying to decide on cesarean delivery versus Kielland’s rotation or other forceps delivery, it is necessary to balance the risks and benefits of the 2 options. And as all clinicians are aware, when cesarean delivery is performed late in labor and the patient has been pushing for a prolonged period of time in the second stage—cesarean can be a challenging procedure. Moreover, these late cesareans are associated with much greater risks than cesarean deliveries performed earlier in labor.

Details of the review
Pergialiotis and colleagues selected 10 studies comparing maternal and neonatal morbidity and mortality between cesarean delivery at full dilatation and cesarean delivery prior to full dilatation. These studies involved 23,104 women with a singleton fetus who underwent cesarean delivery in the first (n = 18,160) or second (n = 4,944) stage of labor.

They found that second-stage cesarean was associated with a higher rate of maternal death (OR, 7.96; 95% CI, 1.61–39.39), a higher rate of maternal admission to the intensive care unit (OR, 7.41; 95% CI, 2.47–22.5), and a higher maternal transfusion rate (OR, 2.60; 95% CI, 1.49–2.54).

The rate of neonatal death also was higher among second-stage cesareans (OR, 5.20; 95% CI, 2.49–10.85), as was admission to the neonatal intensive care unit (OR, 1.63; 95% CI, 0.91–2.91), and the 5-minute Apgar score was more likely to be less than 7 (OR, 2.77; 95% CI, 1.02–7.50).

According to the authors, this study is the “first systematic review and meta-­analysis that investigates the impact of the stage of labor on maternal and neonatal outcomes among women delivering by cesarean section.” The findings demonstrate with authority that second-stage cesareans can be a risky undertaking.

What this EVIDENCE means for practice
Cesareans performed late in the second stage of labor are distinct from those performed in the first stage, carrying much higher risks, especially for the mother. When deciding whether to proceed with cesarean, vacuum, or forceps, the added risk of second-stage cesarean is an important aspect of both the consent conversation and clinical decision making.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

There’s a cyclical lament in obstetrics, and it goes something like this: Forceps are waning and are going to fade away completely if something isn’t done about it. This lament resounds every few decades, as a look at the literature confirms:

In this, our latest cycle of lament, 4 or 5 papers have suggested that forceps in general and Kielland forceps in particular ought not be abandoned because outcomes are better than those suggested by the older literature. With the cesarean delivery rate hovering at about 31% in the United States, perhaps it is time to revisit the issue.

This Update is not intended to be a comprehensive review of the literature. Rather, it offers a snapshot of articles published within the past year—articles that highlight some new features of a very old debate:

Forceps and vacuum device placement is “suboptimal” in almost 30% of operative vaginal deliveries
Ramphul M, Kennelly MM, Burke G, Murphy DJ. Risk factors and morbidity associated with suboptimal instrument placement at instrumental delivery: observational study nested within the Instrumental Delivery & Ultrasound randomised controlled trial ISRCTN 72230496. BJOG. 2015;122(4):558–563.

Rouse DJ. Instrument placement is sub-optimal in three of ten attempted operative vaginal deliveries. BJOG. 2015;122(4):564.

Over the years, many clinicians have argued that we don’t do enough forceps deliveries to maintain our own competence with the procedure, let alone teach residents how to perform it. This observational study nested in a randomized clinical trial is intriguing because Ramphul and colleagues looked for objective evidence of clinicians’ skill at the vacuum and forceps. Specifically, they looked for evidence that the forceps or vacuum was malpositioned during attempts at operative vaginal delivery. In the process, they nicely documented the absolute rate of malpositioning of the forceps and vacuum, finding that it is much higher than expected, even in an institution that performs a lot of operative vaginal deliveries.

Details of the trial
A cohort of 478 nulliparous women at term (≥37 weeks) underwent instrumental delivery at 2 university-affiliated maternity hospitals in Ireland. Ramphul and colleagues documented fetal head position prior to application of the instrument and at delivery. The midwife or neonatologist attending each delivery examined the neonate after birth and recorded the markings of the instrument on the infant’s head to determine whether instrument placement had been optimal.

Instrument placement was considered optimal when the vacuum cup included the flexion point (3 cm anterior to the posterior fontanelle) and the posterior fontanelle, with central placement. For forceps, instrument placement was considered optimal when the blades were positioned bilaterally and symmetrically over the malar bones. Two main types of forceps were used in this study—direct-traction Neville Barnes forceps (n = 138) and rotational Kielland forceps (n = 13)—and the rates of optimal and suboptimal placement were similar between them. 

Each case was labeled as “optimal” or “suboptimal” by 2 investigators, with a third observer arbitrating when the 2 investigators differed in opinion.

Instrument placement was clearly documented in 478 deliveries, 138 of which (28.8%) involved suboptimal placement. There was a lower rate of induction of labor among deliveries with suboptimal placement (42.8% vs 53.2%; odds ratio [OR], 0.66; 95% confidence interval [CI], 0.44–0.98; P = .038). There were no differences between the optimal and suboptimal groups in terms of duration of labor, use of oxytocin, and analgesia. In addition, the seniority of obstetricians performing operative vaginal delivery was similar between groups.

Fetal malposition was more common in the suboptimal group (58.7% vs 37.4%; OR, 2.44; 95% CI, 1.62–3.66; P<.0001). Midcavity station also was more common in the suboptimal group (82.6% vs 73.8%; OR, 1.68; 95% CI, 1.02–2.78; P = .042).

Maternal and neonatal outcomes
Postpartum hemorrhage was more common in the suboptimal placement group (24.6% vs 14.4%; OR, 1.94; 95% CI, 1.19–3.17; P = .008), as was prolonged hospitalization (26.8% vs 14.7%; OR, 2.13; 95% CI, 1.31–3.44; P = .02).

In addition, the incidence of neonatal trauma was higher in the group with suboptimal placement (15.9% vs 3.9%; OR, 4.64; 95% CI, 2.25–9.58; P<.0001) and included such effects as Erb’s palsy, fracture, retinal hemorrhage, cephalhematoma, and cerebral hemorrhage.

After adjustment for potential confounding factors, including induction of labor, seniority of the obstetrician, fetal malposition, caput above +1, midcavity station, regional analgesia, and the instrument used, the association remained significant between suboptimal placement and prolonged hospitalization (adjusted OR, 2.28; 95% CI, 1.30–4.02) and neonatal trauma (adjusted OR, 4.25; 95% CI, 1.85–9.72).

Dwindling statistics for operative vaginal delivery
In an editorial accompanying the study by Ramphul and colleagues, Dwight J. Rouse, MD,points to the waning of instrumental vaginal delivery in many parts of the world, most notably the United States, where, in 2012, only 2.8% of live births involved use of a vacuum device and only 0.6% involved the forceps.⁵

“When the rate of cesarean delivery is 10 times the combined rate of vaginal vacuum and forceps delivery (as it is in the USA), it is fair to argue that operative vaginal delivery is underutilized,” Dr. Rouse writes. “So kudos to Ramphul et al for providing insight into how we might continue to perform operative vaginal delivery safely.”

What this EVIDENCE means for practice
The study by Ramphul and colleagues very clearly confirms that correct placement of the vacuum device or forceps is key to safety.

Should we continue forceps education using the apprenticeship model of training?
Kyser KL, Lu X, Santillan D, et al. Forceps delivery volumes in teaching and nonteaching hospitals: Are volumes sufficient for physicians to acquire and maintain competence? Acad Med. 2014;89(1):71–76.

Ericsson KA. Necessity is the mother of invention: ­video recording firsthand perspectives of critical medical procedures to make simulated training more effective. Acad Med. 2014;89(1):17–20.

Kyser and colleagues have provided the best current snapshot of the opportunity for teaching instrumental vaginal delivery in the United States. They conducted a retrospective cohort study using new state inpatient data from 9 states in diverse geographic locations to capture experience at large and small teaching hospitals, as well as nonteaching institutions. They demonstrated that the opportunity for hands-on experience with these difficult and technically demanding deliveries is extremely limited and probably insufficient for all practicing physicians to maintain their skills if we continue to rely on traditional ways of teaching.

Details of the study
Using State Inpatient Data from 9 states, Kyser and colleagues identified all women hospitalized for childbirth in 2008. Of 1,344,305 deliveries in 835 hospitals, the final cohort included 624,000 operative deliveries—424,224 cesarean deliveries, 174,036 vacuum extractions, 6,158 forceps deliveries, and 19,582 deliveries that required more than 1 method. Of the 835 hospitals in this study, 68 were major teaching hospitals, 130 were minor teaching facilities, and 637 were nonteaching institutions.

The mean annual volumes for cesarean delivery for major teaching, minor teaching, and nonteaching hospitals were 969.8, 757.8, and 406.9, respectively (P<.0001).

The mean annual volumes for vacuum delivery were 301.0, 304.2, and 190.4, respectively (P<.0001).

The mean annual volumes for forceps delivery were 25.2, 15.3, and 8.9, respectively (P<.0001).

Three hundred twenty hospitals (38.3% of all hospitals) failed to perform a single forceps delivery in 2008, including 11 major teaching hospitals (16.2% of major teaching hospitals), 30 minor teaching hospitals (23.1% of minor teaching hospitals), and 279 nonteaching hospitals (43.8% of nonteaching hospitals) (P<.0001).

We need to rethink the apprenticeship model
In a commentary accompanying the study by Kyser and colleagues, K. Anders ­Ericsson, PhD, revisits the “see one, do one, teach one” model that has long characterized medical education. “Both the limitations on learning opportunities available in the clinics and the restrictions on resident work hours have created a real problem for the traditional apprenticeship model for training doctors,” he writes.

Ericsson notes that other specialists, such as concert musicians, chess players, and professional athletes do not learn using an apprenticeship model. For example, chess players do not play game after game of chess to become expert. And when a game is concluded, usually after several hours have passed, they are unlikely to be aware of the specific moves that lost or won them the game (unless an observer points them out). That is why, when training, chess players focus on particular aspects of the game (often identified by a mentor) as being crucial to improve their overall performance.

In today’s chess-learning environment, Ericsson notes, the computer plays a key role and can provide accurate feedback on each move the player executes. Computer chess programs have evolved to the point that they “are far superior in skill to any human chess player. Most important, computers can provide more accurate feedback on each chess move and are available at any time for practice,” writes Ericsson.

The same is true in sports. A tennis player does not practice by playing an endless series of games—though an ability to win a game is the ultimate goal. Rather, the athlete focuses on aspects of the game—the serve, for example—that can make the difference between winning or losing. Ericsson also notes that most musicians, dancers, and athletes “spend most of their time training by themselves to get ready to exhibit their skills for the first time in front of a large audience.”

These approaches are a better model for improving performance than the apprenticeship model, Ericsson argues. In medicine, one alternative might be the video recording of medical procedures in the clinic from multiple points of view—so that later viewers get both the “big picture” and a close-up view from the point of technical performance. After the recording is digitized and stored on a server, it can serve as valuable teaching for an unlimited number of residents.

Simulator training offers another venue for education, as it makes possible the isolation of difficult aspects of a procedure, which can then be repeated by the trainee as many times as necessary. In the future, it should be possible to link video recordings directly to simulators “so trainees could focus on particular aspects of the procedures and be required to respond to prompts with recordable actions,” Ericsson writes.

What this EVIDENCE means for practice
Given the extremely limited opportunities for observing forceps deliveries in the United States, it is time for us to explore new avenues for teaching other than the traditional apprenticeship model.

Is ultrasound evidence of levator muscle “avulsion” a real anatomic entity?
Dietz HP. Forceps: toward obsolescence or revival? Acta Obstet Gynecol Scand. 2015;94(4):347–351.

Da Silva AS, Digesu GA, Dell’Utri C, Fritsch H, ­Piffarotti P, Khullar V. Do ultrasound findings of levator ani “avulsion” correlate with anatomical findings? A multicenter cadaveric study [published online ahead of print May 15, 2015]. Neurourol Urodyn. doi:10.1002 /nau.22781.

Dietz takes a new tack in the debate over cesarean versus forceps by pointing to a recently highlighted abnormality in women who deliver by forceps: levator ani muscle avulsion, or LMA—traumatic disconnection of the levator ani from the pelvic sidewall. It has long been known that forceps deliveries can increase the risk of obstetric anal sphincter injuries (OASIS). Dietz contends that OASIS occurs at a rate as high as 40% to 60% after forceps delivery. He also notes, with some consternation, that the American ­College of Obstetricians and Gynecologists and the Society for Maternal-Fetal Medicine have advocated forceps as a way of reducing the high cesarean delivery rate.

When a parturient has been pushing for an extended period of time and there is a positional abnormality of the fetus, such as persistent occiput posterior position, cesarean delivery is often favored as a way of protecting the rectal sphincter. Dietz argues that cesarean delivery also protects against LMA, which “has only recently been recognized as a major etiological factor in pelvic floor dysfunction.” Dietz then presents a list of studies that have produced ultrasound findings of LMA in a high percentage of women undergoing forceps delivery—percentages on the order of 10% to 40%.

Enter Da Silva and colleagues, who argue that “the only true place to visualize the 3D structure of the human body, [and] thus validate imaging findings, [is] on cadaveric or live tissue dissections.” They undertook a cadaveric study to validate—or not—some of the findings of LMA summarized by Dietz.

Details of the study
The pubovisceral muscle (PVM) anatomy of 30 female cadavers was analyzed via 3D translabial ultrasonography to confirm LMA. The cadavers were then dissected to assess the finding anatomically. Da Silva and colleagues found LMA on imaging in 11 (36.7%) cadavers. LMA was unilateral in 10 (33.3%) cadavers and bilateral in 1 (3.3%). However, no LMA was found at dissection.

When an additional 39 cadavers were dissected, no LMA was identified.

On ultrasound, LMA is strongly associated with a narrower PVM insertion depth (mean of 4.79 mm vs 6.32 mm; P = .001). Da Silva and colleagues concluded that “there is a clear difference between anatomical and ultrasonographic findings. The imaged appearance of an ‘avulsion’ does not represent a true anatomical ‘avulsion’ as confirmed on dissection.”

What this EVIDENCE means for practice
Before we prematurely adopt ultrasound evidence of LMA as a significant morbidity, we need to learn more about its true etiology, pathophysiology, and epidemiology. We don’t yet know enough to say that it’s such a bad injury, when imaged via ultrasound, that it warrants cesarean delivery to avoid it.

When deciding between cesarean and forceps, keep the risks of second-stage cesarean in mind
Pergialiotis V, Vlachos DG, Rodolakis A, Haidopoulos D, Thomakos N, Vlachos GD. First versus second stage C/S maternal and neonatal morbidity: a systematic review and meta-analysis. Eur J Obstet Gynecol Reprod Biol. 2014;175:15–24.

This expert systematic review and meta-analysis summarizes the morbidity of second-stage cesarean delivery. When an obstetrician has a patient who is arrested at persistent occiput posterior position, say, and is trying to decide on cesarean delivery versus Kielland’s rotation or other forceps delivery, it is necessary to balance the risks and benefits of the 2 options. And as all clinicians are aware, when cesarean delivery is performed late in labor and the patient has been pushing for a prolonged period of time in the second stage—cesarean can be a challenging procedure. Moreover, these late cesareans are associated with much greater risks than cesarean deliveries performed earlier in labor.

Details of the review
Pergialiotis and colleagues selected 10 studies comparing maternal and neonatal morbidity and mortality between cesarean delivery at full dilatation and cesarean delivery prior to full dilatation. These studies involved 23,104 women with a singleton fetus who underwent cesarean delivery in the first (n = 18,160) or second (n = 4,944) stage of labor.

They found that second-stage cesarean was associated with a higher rate of maternal death (OR, 7.96; 95% CI, 1.61–39.39), a higher rate of maternal admission to the intensive care unit (OR, 7.41; 95% CI, 2.47–22.5), and a higher maternal transfusion rate (OR, 2.60; 95% CI, 1.49–2.54).

The rate of neonatal death also was higher among second-stage cesareans (OR, 5.20; 95% CI, 2.49–10.85), as was admission to the neonatal intensive care unit (OR, 1.63; 95% CI, 0.91–2.91), and the 5-minute Apgar score was more likely to be less than 7 (OR, 2.77; 95% CI, 1.02–7.50).

According to the authors, this study is the “first systematic review and meta-­analysis that investigates the impact of the stage of labor on maternal and neonatal outcomes among women delivering by cesarean section.” The findings demonstrate with authority that second-stage cesareans can be a risky undertaking.

What this EVIDENCE means for practice
Cesareans performed late in the second stage of labor are distinct from those performed in the first stage, carrying much higher risks, especially for the mother. When deciding whether to proceed with cesarean, vacuum, or forceps, the added risk of second-stage cesarean is an important aspect of both the consent conversation and clinical decision making.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Pelvic masses frequently are the reason for the medical evaluation of young women and girls. Regardless of what prompted the work-up that led to the mass’ discovery, the patient inevitably will be sent to a gynecologist for further evaluation, and such a practitioner should be involved whenever there is suspicion for a mass involving the reproductive tract.

While it does not happen often, it is possible that a mass diagnosed as ovarian, based on imaging, is then determined to be of another organ system at the time of surgery. Most frequently, this occurs with ruptured appendicitis, as the presence of an appendiceal abscess can mimic a complex ovarian mass or tubo-ovarian abscess (TOA).¹

The full differential diagnosis of non-gynecologic pelvic masses is extensive and includes mesenteric duplication cysts, presacral masses, pelvic kidney, peritoneal inclusion cysts, and urachal cysts (TABLE). It can be difficult to distinguish pathology as gynecologic or nongynecologic even if a thorough work-up is performed.

In this review, we offer several cases involving varying presentations of pelvic masses related to the reproductive tract.

Case 1: Severe pelvic pain in an 18-year-old
An 18-year-old adolescent presents to your office reporting worsening pelvic pain over the past 3 days. The pain is severe in the left lower quadrant. She reports a foul discharge and thinks she has a fever but hasn’t checked her temperature. She says she has been sexually active in the past few months with 2 different male partners. She has not been using condoms consistently and hasn’t been tested for sexually transmitted infections. Physical examination reveals a mucopurulent discharge at the cervix and copious white blood cells noted on wet mount. Bimanual examination reveals cervical motion tenderness and tenderness over the left adnexa. Ultrasound reveals a mass in the left adnexa with debris and internal echoes.
Diagnosis: Tubo-ovarian abscess.
Treatment: Admission to the hospital for intravenous antibiotic therapy.

It is important to note that TOAs can be seen in patients who have not been sexually active, as well as in cases related not to an ascending infection but rather to a history of pelvic surgery or complex structural anomaly.² The majority of the time a TOA is a result of pelvic inflammatory disease (PID). Often, patients with uncomplicated PID can be treated on an outpatient basis if they meet strict criteria, but patients with a TOA need to be treated as an inpatient due to the severity of this infection.

Clinical pearl. If a patient has an IUD in place, close clinical follow up is critical to determine response to therapy. The Centers for Disease Control and Prevention’s sexually transmitted infection treatment guidelines state that removal of the IUD is not mandatory, but if the patient is not responding to treatment removal ultimately may be necessary. The IUD should be removed if there is no improvement in the patient’s symptoms with antibiotic therapy, there is no decrease in size of the TOA with antibiotic therapy, or if there is no positive test of cure after treatment for the TOA is completed.

If a patient has progressive abdominal pain and other findings consistent with infectious etiology, consider that a ruptured appendix could have a very similar appearance to a TOA. Computed tomography can be a useful tool to aid in firm diagnosis in cases in which gastroenterologic entities must be ruled out, but ultimately the gold standard of diagnosis for both of these procedures is diagnostic laparoscopy. Diagnostic surgery can be performed if the patient does not respond to medical therapy. In an effort to avoid surgical intervention, interventional radiology may be an option to drain the TOA. If this is performed, it is useful to repeat the ultrasound to confirm resolution prior to removal of the drain.

Case 2: Acute-onset severe pelvic pain in a young adolescent
A 12-year-old girl presents to the emergency department with acute-onset right lower quadrant pain. She states that about 2 hours ago she was playing in the yard and suddenly doubled over with pain. She also has had nausea and vomiting since that time.

She is in obvious distress and is resting in the fetal position. Examination reveals normal vital signs and tenderness to palpation over the right lower pelvic quadrant. There is no palpable abdominal mass. Genital examination reveals Tanner stage 4 external genitalia with normal introitus and patent, intact, annular hymen.

An abdominal ultrasound reveals a normal appendix, and pelvic ultrasonography reveals that the right ovary is enlarged (volume = 25 mL). The left ovary shows no obvious mass and a volume of 8 mL.

She is taken to the operating room, where you perform diagnostic laparoscopy.
Diagnosis: Adnexal torsion (FIGURE 1).
Treatment: Surgical detorsion with or without cystectomy.

Ultrasonography certainly can be useful in determining the size of an adnexal mass. An adnexal volume of less than 20 mL is strong evidence against adnexal torsion in an adolescent. This information, in addition to the remainder of the clinical picture, can be used to determine if surgery is immediately necessary or can be delayed.³

Several studies have attempted to draw a link between size of an ovarian mass and risk of malignancy. Unfortunately, such attempts have been unsuccessful, especially for large and small masses.⁴ In addition, many studies have explored the use of Doppler technology to confirm a diagnosis of torsion found on sonography. Studies have shown, however, that diminished or absent Doppler flow is not a reliable finding and that ovarian blood flow can be preserved in cases of surgically confirmed adnexal torsion.⁵

Torsion ultimately is a clinical diagnosis, and medical history and physical examination are critical in the decision-making process. The decision to go to the operating room for further evaluation never should be made based on ultrasound findings alone, as not all ovarian torsions result in a mass greater than 20 mL.

Clinical pearl. In the setting of a known adnexal cyst, it is important to impress upon patients and their parents the warning signs of torsion and the need to proceed directly to the emergency center if acute pelvic pain occurs.

Historically, adnexal torsion is correlated with oophorectomy, but recent studies indicate that ovarian function can be preserved in the majority of cases with detorsion and cystectomy alone.^6,7 In cases in which no cyst is present, detorsion is therapeutic.

In addition, studies have shown that the appearance of the ovary is not indicative of damage to the ovary. Regardless of “necrotic” appearance, the adnexa should be preserved.^8,9 Ovarian function after detorsion also has been assessed in a case series that showed normal follicular development on ultrasonography in more than 90% of patients after detorsion. In this group, 6 of 102 patients with torsion eventually underwent in vitro fertilization and, in all 6 patients, oocytes retrieved from the ischemic ovary were fertilized.¹⁰

Case 3: 15-year-old girl with nontender, palpable abdominal mass
A 15-year-old adolescent comes to your office for a well woman exam and to establish gynecologic care. Abdominal examination reveals a mass below the umbilicus that is nontender to palpation. The remainder of the examination is normal, and the patient is sent for ultrasonography. Results reveal a complex-appearing mass in the left ovary that measures 8 cm x 7 cm x 8 cm. The report states that the mass shows areas of fat and calcifications. There are no other abnormalities noted in the abdomen or pelvis.
Diagnosis: Dermoid cyst.
Treatment: Surgical intervention versus expectant management.

Germ cell ovarian tumors are a diverse category of tumors that include both benign and malignant disease. The ovarian teratoma (“dermoid”) is the most common and perhaps best-known example of a benign ovarian germ cell tumor (FIGURE 2). While the incidence in the general population is unknown, dermoids account for approximately 65% of adnexal masses in pediatric patients presenting for treatment.¹¹ Most of the time, patients with ovarian dermoids will be asymptomatic; however, pain is the most common presenting symptom.

The spectrum of sonographic features includes:

The notation of internal vascularity is concerning for malignancy.¹²

Fortunately, malignant ovarian germ cell tumors are much less common than benign lesions. Of these, the most common pediatric ovarian germ cell tumor is dysgerminoma (FIGURE 3).¹³

Clinical pearl. A unique consideration in patients with dysgerminoma or choriocarcinoma is the possible diagnosis of XY gonadal dysgenesis, or Swyer syndrome.^14,15 This may be seen in young girls with female external genitalia and primary amenorrhea. Depending on the level of concern, obtaining a karyotype also may be beneficial.¹³

Sex cord−stromal tumors also are relatively common in the pediatric population.¹⁶ Of these, granulosa cell tumors are the most common and account for 2% to 5% of ovarian malignancies regardless of age at diagnosis. Juvenile-type granulosa ovarian cancers occur mainly in premenarchal girls and comprise roughly 5% of all granulosa cell tumors.¹⁷ The presenting problem usually is precocious puberty. Therefore, in any situation in which a prepubertal girl is developing too early and peripheral precocious puberty is suspected, sonography should be obtained to rule out a hormone-producing ovarian mass. Tumor markers most helpful in this situation include estradiol, testosterone, and inhibin B.¹⁷

When an adolescent is diagnosed with ovarian cancer, it is ideal to perform a fertility-sparing procedure whenever it is reasonable.¹⁸ While dermoid cysts can look concerning on sonography because of their heterogeneous appearance, the vast majority can be safely and effectively resected without oophorectomy in order to preserve fertility, as in most cases they are benign. Nonetheless, cystectomy does have a small, theoretic risk of cyst rupture, with the potential for pelvic peritonitis from dermoid content spillage.¹⁹ In the vast majority of cases in which a benign adnexal mass is identified, ovarian cystectomy is appropriate and oophorectomy is not indicated.²⁰

Another very rare presentation of mature cystic teratoma can include acute neurologic decline in cases of paraneoplastic anti-N-methyl-D-aspartate receptor encephalitis. Frequently, these teratomas will be small in size and discovered only incidentally during the work-up of a patient with altered mental status. Resection is indicated as soon as possible to stop the neurologic decline.²¹

Case 4: Cyclic pelvic pain and a pelvic mass in a 16-year-old
A 16-year-old adolescent presents to your office for evaluation of cyclic pelvic pain. She states that menarche occurred at age 12 years and menses have been irregular ever since, occurring every few months and associated with significant pain with the onset of bleeding.

Physical examination reveals Tanner stage 4 breasts and Tanner stage 4 external genitalia. The introitus is normal with a visible intact, annular hymen. A mildly tender palpable mass at the level of the umbilicus is noted. The patient consents to having a Q-tip placed in the vagina, which reveals a bulge in the left vaginal wall that is nontender and fluctuant. Pelvic ultrasonography reveals uterine didelphys and an obstructed left hemivagina. A renal ultrasound reveals an absent left kidney.
Diagnosis: OHVIRA (obstructed hemivagina and ipsilateral renal anomaly) syndrome.
Treatment: Surgical resection of the vaginal septum.

Masses that appear complex on imaging can be deceiving, as they also can be related to obstructive reproductive tract anomalies. “Pelvic mass” is often the working diagnosis in cases of imperforate hymen, vaginal atresia, cervical agenesis, and uterine didelphys with obstructed hemivagina. This underscores the importance of taking an accurate menstrual history as well as performing a thorough physical examination. Usually this does not require an internal vaginal examination if the patient is unable to tolerate one, but a rectal examination can provide similar information in a patient presenting with a “pelvic mass” who will consent to this portion of the exam.

Clinical pearl. If a patient is not comfortable consenting to a rectal exam, a lubricated Q-tip can be used to palpate the vagina to minimize patient discomfort.

Before performing surgery…
Vaginal surgery can be corrective in the majority of these cases; however, magnetic resonance imaging is the gold standard for diagnosis and should be performed prior to surgical planning to further characterize the anomaly.²² Because Müllerian anomalies are associated with renal malformation such as absent kidney, pelvic kidney, collecting system duplication, or ectopic ureteral insertion around 40% of the time, imaging studies to assess for these structures is important prior to surgical intervention.²³ If the patient is symptomatic and surgery cannot be performed immediately in a safe manner, she may require admission for pain control and placement of a Foley catheter (if the mass is obstructing urinary flow) until surgery can be performed safely.

A comprehensive review of Müllerian anomalies is beyond the scope of this article; it is important to note that these clinical scenarios are always unique and treatment should be individualized.

Conclusion
There are many sources of pelvic masses in children, adolescents, and young women; not all sources will be gynecologic. To avoid unnecessary surgical intervention, it is important to obtain as much information as possible from the patient’s history, physical examination, and laboratory and imaging studies.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Pelvic masses frequently are the reason for the medical evaluation of young women and girls. Regardless of what prompted the work-up that led to the mass’ discovery, the patient inevitably will be sent to a gynecologist for further evaluation, and such a practitioner should be involved whenever there is suspicion for a mass involving the reproductive tract.

While it does not happen often, it is possible that a mass diagnosed as ovarian, based on imaging, is then determined to be of another organ system at the time of surgery. Most frequently, this occurs with ruptured appendicitis, as the presence of an appendiceal abscess can mimic a complex ovarian mass or tubo-ovarian abscess (TOA).¹

The full differential diagnosis of non-gynecologic pelvic masses is extensive and includes mesenteric duplication cysts, presacral masses, pelvic kidney, peritoneal inclusion cysts, and urachal cysts (TABLE). It can be difficult to distinguish pathology as gynecologic or nongynecologic even if a thorough work-up is performed.

In this review, we offer several cases involving varying presentations of pelvic masses related to the reproductive tract.

Case 1: Severe pelvic pain in an 18-year-old
An 18-year-old adolescent presents to your office reporting worsening pelvic pain over the past 3 days. The pain is severe in the left lower quadrant. She reports a foul discharge and thinks she has a fever but hasn’t checked her temperature. She says she has been sexually active in the past few months with 2 different male partners. She has not been using condoms consistently and hasn’t been tested for sexually transmitted infections. Physical examination reveals a mucopurulent discharge at the cervix and copious white blood cells noted on wet mount. Bimanual examination reveals cervical motion tenderness and tenderness over the left adnexa. Ultrasound reveals a mass in the left adnexa with debris and internal echoes.
Diagnosis: Tubo-ovarian abscess.
Treatment: Admission to the hospital for intravenous antibiotic therapy.

It is important to note that TOAs can be seen in patients who have not been sexually active, as well as in cases related not to an ascending infection but rather to a history of pelvic surgery or complex structural anomaly.² The majority of the time a TOA is a result of pelvic inflammatory disease (PID). Often, patients with uncomplicated PID can be treated on an outpatient basis if they meet strict criteria, but patients with a TOA need to be treated as an inpatient due to the severity of this infection.

Clinical pearl. If a patient has an IUD in place, close clinical follow up is critical to determine response to therapy. The Centers for Disease Control and Prevention’s sexually transmitted infection treatment guidelines state that removal of the IUD is not mandatory, but if the patient is not responding to treatment removal ultimately may be necessary. The IUD should be removed if there is no improvement in the patient’s symptoms with antibiotic therapy, there is no decrease in size of the TOA with antibiotic therapy, or if there is no positive test of cure after treatment for the TOA is completed.

If a patient has progressive abdominal pain and other findings consistent with infectious etiology, consider that a ruptured appendix could have a very similar appearance to a TOA. Computed tomography can be a useful tool to aid in firm diagnosis in cases in which gastroenterologic entities must be ruled out, but ultimately the gold standard of diagnosis for both of these procedures is diagnostic laparoscopy. Diagnostic surgery can be performed if the patient does not respond to medical therapy. In an effort to avoid surgical intervention, interventional radiology may be an option to drain the TOA. If this is performed, it is useful to repeat the ultrasound to confirm resolution prior to removal of the drain.

Case 2: Acute-onset severe pelvic pain in a young adolescent
A 12-year-old girl presents to the emergency department with acute-onset right lower quadrant pain. She states that about 2 hours ago she was playing in the yard and suddenly doubled over with pain. She also has had nausea and vomiting since that time.

She is in obvious distress and is resting in the fetal position. Examination reveals normal vital signs and tenderness to palpation over the right lower pelvic quadrant. There is no palpable abdominal mass. Genital examination reveals Tanner stage 4 external genitalia with normal introitus and patent, intact, annular hymen.

An abdominal ultrasound reveals a normal appendix, and pelvic ultrasonography reveals that the right ovary is enlarged (volume = 25 mL). The left ovary shows no obvious mass and a volume of 8 mL.

She is taken to the operating room, where you perform diagnostic laparoscopy.
Diagnosis: Adnexal torsion (FIGURE 1).
Treatment: Surgical detorsion with or without cystectomy.

Ultrasonography certainly can be useful in determining the size of an adnexal mass. An adnexal volume of less than 20 mL is strong evidence against adnexal torsion in an adolescent. This information, in addition to the remainder of the clinical picture, can be used to determine if surgery is immediately necessary or can be delayed.³

Several studies have attempted to draw a link between size of an ovarian mass and risk of malignancy. Unfortunately, such attempts have been unsuccessful, especially for large and small masses.⁴ In addition, many studies have explored the use of Doppler technology to confirm a diagnosis of torsion found on sonography. Studies have shown, however, that diminished or absent Doppler flow is not a reliable finding and that ovarian blood flow can be preserved in cases of surgically confirmed adnexal torsion.⁵

Torsion ultimately is a clinical diagnosis, and medical history and physical examination are critical in the decision-making process. The decision to go to the operating room for further evaluation never should be made based on ultrasound findings alone, as not all ovarian torsions result in a mass greater than 20 mL.

Clinical pearl. In the setting of a known adnexal cyst, it is important to impress upon patients and their parents the warning signs of torsion and the need to proceed directly to the emergency center if acute pelvic pain occurs.

Historically, adnexal torsion is correlated with oophorectomy, but recent studies indicate that ovarian function can be preserved in the majority of cases with detorsion and cystectomy alone.^6,7 In cases in which no cyst is present, detorsion is therapeutic.

In addition, studies have shown that the appearance of the ovary is not indicative of damage to the ovary. Regardless of “necrotic” appearance, the adnexa should be preserved.^8,9 Ovarian function after detorsion also has been assessed in a case series that showed normal follicular development on ultrasonography in more than 90% of patients after detorsion. In this group, 6 of 102 patients with torsion eventually underwent in vitro fertilization and, in all 6 patients, oocytes retrieved from the ischemic ovary were fertilized.¹⁰

Case 3: 15-year-old girl with nontender, palpable abdominal mass
A 15-year-old adolescent comes to your office for a well woman exam and to establish gynecologic care. Abdominal examination reveals a mass below the umbilicus that is nontender to palpation. The remainder of the examination is normal, and the patient is sent for ultrasonography. Results reveal a complex-appearing mass in the left ovary that measures 8 cm x 7 cm x 8 cm. The report states that the mass shows areas of fat and calcifications. There are no other abnormalities noted in the abdomen or pelvis.
Diagnosis: Dermoid cyst.
Treatment: Surgical intervention versus expectant management.

Germ cell ovarian tumors are a diverse category of tumors that include both benign and malignant disease. The ovarian teratoma (“dermoid”) is the most common and perhaps best-known example of a benign ovarian germ cell tumor (FIGURE 2). While the incidence in the general population is unknown, dermoids account for approximately 65% of adnexal masses in pediatric patients presenting for treatment.¹¹ Most of the time, patients with ovarian dermoids will be asymptomatic; however, pain is the most common presenting symptom.

The spectrum of sonographic features includes:

The notation of internal vascularity is concerning for malignancy.¹²

Fortunately, malignant ovarian germ cell tumors are much less common than benign lesions. Of these, the most common pediatric ovarian germ cell tumor is dysgerminoma (FIGURE 3).¹³

Clinical pearl. A unique consideration in patients with dysgerminoma or choriocarcinoma is the possible diagnosis of XY gonadal dysgenesis, or Swyer syndrome.^14,15 This may be seen in young girls with female external genitalia and primary amenorrhea. Depending on the level of concern, obtaining a karyotype also may be beneficial.¹³

Sex cord−stromal tumors also are relatively common in the pediatric population.¹⁶ Of these, granulosa cell tumors are the most common and account for 2% to 5% of ovarian malignancies regardless of age at diagnosis. Juvenile-type granulosa ovarian cancers occur mainly in premenarchal girls and comprise roughly 5% of all granulosa cell tumors.¹⁷ The presenting problem usually is precocious puberty. Therefore, in any situation in which a prepubertal girl is developing too early and peripheral precocious puberty is suspected, sonography should be obtained to rule out a hormone-producing ovarian mass. Tumor markers most helpful in this situation include estradiol, testosterone, and inhibin B.¹⁷

When an adolescent is diagnosed with ovarian cancer, it is ideal to perform a fertility-sparing procedure whenever it is reasonable.¹⁸ While dermoid cysts can look concerning on sonography because of their heterogeneous appearance, the vast majority can be safely and effectively resected without oophorectomy in order to preserve fertility, as in most cases they are benign. Nonetheless, cystectomy does have a small, theoretic risk of cyst rupture, with the potential for pelvic peritonitis from dermoid content spillage.¹⁹ In the vast majority of cases in which a benign adnexal mass is identified, ovarian cystectomy is appropriate and oophorectomy is not indicated.²⁰

Another very rare presentation of mature cystic teratoma can include acute neurologic decline in cases of paraneoplastic anti-N-methyl-D-aspartate receptor encephalitis. Frequently, these teratomas will be small in size and discovered only incidentally during the work-up of a patient with altered mental status. Resection is indicated as soon as possible to stop the neurologic decline.²¹

Case 4: Cyclic pelvic pain and a pelvic mass in a 16-year-old
A 16-year-old adolescent presents to your office for evaluation of cyclic pelvic pain. She states that menarche occurred at age 12 years and menses have been irregular ever since, occurring every few months and associated with significant pain with the onset of bleeding.

Physical examination reveals Tanner stage 4 breasts and Tanner stage 4 external genitalia. The introitus is normal with a visible intact, annular hymen. A mildly tender palpable mass at the level of the umbilicus is noted. The patient consents to having a Q-tip placed in the vagina, which reveals a bulge in the left vaginal wall that is nontender and fluctuant. Pelvic ultrasonography reveals uterine didelphys and an obstructed left hemivagina. A renal ultrasound reveals an absent left kidney.
Diagnosis: OHVIRA (obstructed hemivagina and ipsilateral renal anomaly) syndrome.
Treatment: Surgical resection of the vaginal septum.

Masses that appear complex on imaging can be deceiving, as they also can be related to obstructive reproductive tract anomalies. “Pelvic mass” is often the working diagnosis in cases of imperforate hymen, vaginal atresia, cervical agenesis, and uterine didelphys with obstructed hemivagina. This underscores the importance of taking an accurate menstrual history as well as performing a thorough physical examination. Usually this does not require an internal vaginal examination if the patient is unable to tolerate one, but a rectal examination can provide similar information in a patient presenting with a “pelvic mass” who will consent to this portion of the exam.

Clinical pearl. If a patient is not comfortable consenting to a rectal exam, a lubricated Q-tip can be used to palpate the vagina to minimize patient discomfort.

Before performing surgery…
Vaginal surgery can be corrective in the majority of these cases; however, magnetic resonance imaging is the gold standard for diagnosis and should be performed prior to surgical planning to further characterize the anomaly.²² Because Müllerian anomalies are associated with renal malformation such as absent kidney, pelvic kidney, collecting system duplication, or ectopic ureteral insertion around 40% of the time, imaging studies to assess for these structures is important prior to surgical intervention.²³ If the patient is symptomatic and surgery cannot be performed immediately in a safe manner, she may require admission for pain control and placement of a Foley catheter (if the mass is obstructing urinary flow) until surgery can be performed safely.

A comprehensive review of Müllerian anomalies is beyond the scope of this article; it is important to note that these clinical scenarios are always unique and treatment should be individualized.

Conclusion
There are many sources of pelvic masses in children, adolescents, and young women; not all sources will be gynecologic. To avoid unnecessary surgical intervention, it is important to obtain as much information as possible from the patient’s history, physical examination, and laboratory and imaging studies.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Pelvic masses frequently are the reason for the medical evaluation of young women and girls. Regardless of what prompted the work-up that led to the mass’ discovery, the patient inevitably will be sent to a gynecologist for further evaluation, and such a practitioner should be involved whenever there is suspicion for a mass involving the reproductive tract.

While it does not happen often, it is possible that a mass diagnosed as ovarian, based on imaging, is then determined to be of another organ system at the time of surgery. Most frequently, this occurs with ruptured appendicitis, as the presence of an appendiceal abscess can mimic a complex ovarian mass or tubo-ovarian abscess (TOA).¹

The full differential diagnosis of non-gynecologic pelvic masses is extensive and includes mesenteric duplication cysts, presacral masses, pelvic kidney, peritoneal inclusion cysts, and urachal cysts (TABLE). It can be difficult to distinguish pathology as gynecologic or nongynecologic even if a thorough work-up is performed.

In this review, we offer several cases involving varying presentations of pelvic masses related to the reproductive tract.

Case 1: Severe pelvic pain in an 18-year-old
An 18-year-old adolescent presents to your office reporting worsening pelvic pain over the past 3 days. The pain is severe in the left lower quadrant. She reports a foul discharge and thinks she has a fever but hasn’t checked her temperature. She says she has been sexually active in the past few months with 2 different male partners. She has not been using condoms consistently and hasn’t been tested for sexually transmitted infections. Physical examination reveals a mucopurulent discharge at the cervix and copious white blood cells noted on wet mount. Bimanual examination reveals cervical motion tenderness and tenderness over the left adnexa. Ultrasound reveals a mass in the left adnexa with debris and internal echoes.
Diagnosis: Tubo-ovarian abscess.
Treatment: Admission to the hospital for intravenous antibiotic therapy.

It is important to note that TOAs can be seen in patients who have not been sexually active, as well as in cases related not to an ascending infection but rather to a history of pelvic surgery or complex structural anomaly.² The majority of the time a TOA is a result of pelvic inflammatory disease (PID). Often, patients with uncomplicated PID can be treated on an outpatient basis if they meet strict criteria, but patients with a TOA need to be treated as an inpatient due to the severity of this infection.

Clinical pearl. If a patient has an IUD in place, close clinical follow up is critical to determine response to therapy. The Centers for Disease Control and Prevention’s sexually transmitted infection treatment guidelines state that removal of the IUD is not mandatory, but if the patient is not responding to treatment removal ultimately may be necessary. The IUD should be removed if there is no improvement in the patient’s symptoms with antibiotic therapy, there is no decrease in size of the TOA with antibiotic therapy, or if there is no positive test of cure after treatment for the TOA is completed.

If a patient has progressive abdominal pain and other findings consistent with infectious etiology, consider that a ruptured appendix could have a very similar appearance to a TOA. Computed tomography can be a useful tool to aid in firm diagnosis in cases in which gastroenterologic entities must be ruled out, but ultimately the gold standard of diagnosis for both of these procedures is diagnostic laparoscopy. Diagnostic surgery can be performed if the patient does not respond to medical therapy. In an effort to avoid surgical intervention, interventional radiology may be an option to drain the TOA. If this is performed, it is useful to repeat the ultrasound to confirm resolution prior to removal of the drain.

Case 2: Acute-onset severe pelvic pain in a young adolescent
A 12-year-old girl presents to the emergency department with acute-onset right lower quadrant pain. She states that about 2 hours ago she was playing in the yard and suddenly doubled over with pain. She also has had nausea and vomiting since that time.

She is in obvious distress and is resting in the fetal position. Examination reveals normal vital signs and tenderness to palpation over the right lower pelvic quadrant. There is no palpable abdominal mass. Genital examination reveals Tanner stage 4 external genitalia with normal introitus and patent, intact, annular hymen.

An abdominal ultrasound reveals a normal appendix, and pelvic ultrasonography reveals that the right ovary is enlarged (volume = 25 mL). The left ovary shows no obvious mass and a volume of 8 mL.

She is taken to the operating room, where you perform diagnostic laparoscopy.
Diagnosis: Adnexal torsion (FIGURE 1).
Treatment: Surgical detorsion with or without cystectomy.

Ultrasonography certainly can be useful in determining the size of an adnexal mass. An adnexal volume of less than 20 mL is strong evidence against adnexal torsion in an adolescent. This information, in addition to the remainder of the clinical picture, can be used to determine if surgery is immediately necessary or can be delayed.³

Several studies have attempted to draw a link between size of an ovarian mass and risk of malignancy. Unfortunately, such attempts have been unsuccessful, especially for large and small masses.⁴ In addition, many studies have explored the use of Doppler technology to confirm a diagnosis of torsion found on sonography. Studies have shown, however, that diminished or absent Doppler flow is not a reliable finding and that ovarian blood flow can be preserved in cases of surgically confirmed adnexal torsion.⁵

Torsion ultimately is a clinical diagnosis, and medical history and physical examination are critical in the decision-making process. The decision to go to the operating room for further evaluation never should be made based on ultrasound findings alone, as not all ovarian torsions result in a mass greater than 20 mL.

Clinical pearl. In the setting of a known adnexal cyst, it is important to impress upon patients and their parents the warning signs of torsion and the need to proceed directly to the emergency center if acute pelvic pain occurs.

Historically, adnexal torsion is correlated with oophorectomy, but recent studies indicate that ovarian function can be preserved in the majority of cases with detorsion and cystectomy alone.^6,7 In cases in which no cyst is present, detorsion is therapeutic.

In addition, studies have shown that the appearance of the ovary is not indicative of damage to the ovary. Regardless of “necrotic” appearance, the adnexa should be preserved.^8,9 Ovarian function after detorsion also has been assessed in a case series that showed normal follicular development on ultrasonography in more than 90% of patients after detorsion. In this group, 6 of 102 patients with torsion eventually underwent in vitro fertilization and, in all 6 patients, oocytes retrieved from the ischemic ovary were fertilized.¹⁰

Case 3: 15-year-old girl with nontender, palpable abdominal mass
A 15-year-old adolescent comes to your office for a well woman exam and to establish gynecologic care. Abdominal examination reveals a mass below the umbilicus that is nontender to palpation. The remainder of the examination is normal, and the patient is sent for ultrasonography. Results reveal a complex-appearing mass in the left ovary that measures 8 cm x 7 cm x 8 cm. The report states that the mass shows areas of fat and calcifications. There are no other abnormalities noted in the abdomen or pelvis.
Diagnosis: Dermoid cyst.
Treatment: Surgical intervention versus expectant management.

Germ cell ovarian tumors are a diverse category of tumors that include both benign and malignant disease. The ovarian teratoma (“dermoid”) is the most common and perhaps best-known example of a benign ovarian germ cell tumor (FIGURE 2). While the incidence in the general population is unknown, dermoids account for approximately 65% of adnexal masses in pediatric patients presenting for treatment.¹¹ Most of the time, patients with ovarian dermoids will be asymptomatic; however, pain is the most common presenting symptom.

The spectrum of sonographic features includes:

The notation of internal vascularity is concerning for malignancy.¹²

Fortunately, malignant ovarian germ cell tumors are much less common than benign lesions. Of these, the most common pediatric ovarian germ cell tumor is dysgerminoma (FIGURE 3).¹³

Clinical pearl. A unique consideration in patients with dysgerminoma or choriocarcinoma is the possible diagnosis of XY gonadal dysgenesis, or Swyer syndrome.^14,15 This may be seen in young girls with female external genitalia and primary amenorrhea. Depending on the level of concern, obtaining a karyotype also may be beneficial.¹³

Sex cord−stromal tumors also are relatively common in the pediatric population.¹⁶ Of these, granulosa cell tumors are the most common and account for 2% to 5% of ovarian malignancies regardless of age at diagnosis. Juvenile-type granulosa ovarian cancers occur mainly in premenarchal girls and comprise roughly 5% of all granulosa cell tumors.¹⁷ The presenting problem usually is precocious puberty. Therefore, in any situation in which a prepubertal girl is developing too early and peripheral precocious puberty is suspected, sonography should be obtained to rule out a hormone-producing ovarian mass. Tumor markers most helpful in this situation include estradiol, testosterone, and inhibin B.¹⁷

When an adolescent is diagnosed with ovarian cancer, it is ideal to perform a fertility-sparing procedure whenever it is reasonable.¹⁸ While dermoid cysts can look concerning on sonography because of their heterogeneous appearance, the vast majority can be safely and effectively resected without oophorectomy in order to preserve fertility, as in most cases they are benign. Nonetheless, cystectomy does have a small, theoretic risk of cyst rupture, with the potential for pelvic peritonitis from dermoid content spillage.¹⁹ In the vast majority of cases in which a benign adnexal mass is identified, ovarian cystectomy is appropriate and oophorectomy is not indicated.²⁰

Another very rare presentation of mature cystic teratoma can include acute neurologic decline in cases of paraneoplastic anti-N-methyl-D-aspartate receptor encephalitis. Frequently, these teratomas will be small in size and discovered only incidentally during the work-up of a patient with altered mental status. Resection is indicated as soon as possible to stop the neurologic decline.²¹

Case 4: Cyclic pelvic pain and a pelvic mass in a 16-year-old
A 16-year-old adolescent presents to your office for evaluation of cyclic pelvic pain. She states that menarche occurred at age 12 years and menses have been irregular ever since, occurring every few months and associated with significant pain with the onset of bleeding.

Physical examination reveals Tanner stage 4 breasts and Tanner stage 4 external genitalia. The introitus is normal with a visible intact, annular hymen. A mildly tender palpable mass at the level of the umbilicus is noted. The patient consents to having a Q-tip placed in the vagina, which reveals a bulge in the left vaginal wall that is nontender and fluctuant. Pelvic ultrasonography reveals uterine didelphys and an obstructed left hemivagina. A renal ultrasound reveals an absent left kidney.
Diagnosis: OHVIRA (obstructed hemivagina and ipsilateral renal anomaly) syndrome.
Treatment: Surgical resection of the vaginal septum.

Masses that appear complex on imaging can be deceiving, as they also can be related to obstructive reproductive tract anomalies. “Pelvic mass” is often the working diagnosis in cases of imperforate hymen, vaginal atresia, cervical agenesis, and uterine didelphys with obstructed hemivagina. This underscores the importance of taking an accurate menstrual history as well as performing a thorough physical examination. Usually this does not require an internal vaginal examination if the patient is unable to tolerate one, but a rectal examination can provide similar information in a patient presenting with a “pelvic mass” who will consent to this portion of the exam.

Clinical pearl. If a patient is not comfortable consenting to a rectal exam, a lubricated Q-tip can be used to palpate the vagina to minimize patient discomfort.

Before performing surgery…
Vaginal surgery can be corrective in the majority of these cases; however, magnetic resonance imaging is the gold standard for diagnosis and should be performed prior to surgical planning to further characterize the anomaly.²² Because Müllerian anomalies are associated with renal malformation such as absent kidney, pelvic kidney, collecting system duplication, or ectopic ureteral insertion around 40% of the time, imaging studies to assess for these structures is important prior to surgical intervention.²³ If the patient is symptomatic and surgery cannot be performed immediately in a safe manner, she may require admission for pain control and placement of a Foley catheter (if the mass is obstructing urinary flow) until surgery can be performed safely.

A comprehensive review of Müllerian anomalies is beyond the scope of this article; it is important to note that these clinical scenarios are always unique and treatment should be individualized.

Conclusion
There are many sources of pelvic masses in children, adolescents, and young women; not all sources will be gynecologic. To avoid unnecessary surgical intervention, it is important to obtain as much information as possible from the patient’s history, physical examination, and laboratory and imaging studies.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Pregnant women may be more severelyaffected by certain microorganisms than nonpregnant individuals.¹ In a recent review, Kourtis and colleagues cited evidence for increased mortality risk for pregnant patients related to 5 specific infections.¹ What are those infections and why does pregnancy put a woman at greater risk for adverse outcomes? I review these topics in this article and, based on this evidence, I suggest 5 specific ways to avoid infection during pregnancy.

Five infections that can lead to detrimental outcomes during pregnancy
Influenza
During the pandemic of 1918, maternal mortality was 27%. During the 1957 pandemic, 50% of influenza-related deaths among women of reproductive age occurred among pregnant women. In the 2009 H1N1 influenza A pandemic, pregnant women were clearly at increased risk for severe disease, reflected by an increased frequency of hospitalization and increased likelihood of admission to an intensive care unit.

Hepatitis E
Compared with nonpregnant women and men, pregnant women also are at markedly increased risk for mortality due to hepa-titis E infection, especially in Southeast Asia, the Middle East, and Africa. The pathophysiologic basis for this increase is not well understood. Interestingly, in a report from India, 33% to 43% of pregnant women infected with hepatitis E had such severe disease that they developed hepatic failure.

Herpes simplex virus
Pregnant women with primary herpes simplex virus (HSV) infection are at increased risk for hepatitis and for disseminated infection, compared with other nonpregnant adults. Only patients with obvious immunodeficiency disorders are at greater risk for disseminated HSV infection.

Malaria
Pregnant women are at significantly increased risk for acquiring Plasmodium falciparum malaria and developing severe, life-threatening disease. In multiple studies from the Asia-Pacific region, pregnant women have been threefold more likely to acquire malaria compared with nonpregnant individuals. In India, during the period 2004 to 2006, malaria was the most common cause of maternal death.

The most likely explanation for the deleterious effect of this particular form of malaria is the fact that the P falciparum parasites accumulate selectively in the placenta because they bind avidly to syncytiotrophoblastic chondroitin sulfate A. Intense inflammation in the placenta, in turn, can lead to early pregnancy loss, preterm delivery, and fetal infection.

Listeria
Another important infection to which pregnant women are particularly susceptible is listeriosis. Listeria monocytogenes may contaminate several types of food such as uncooked meats and vegetables, unpasteurized milk, and soft cheeses. The organism has a predilection to attack the placenta and fetus and can cause spontaneous abortion, stillbirth, preterm delivery, and neonatal infection. Hispanic women may be at unusually high risk for listeria.

Immune system changes during pregnancy
Certain subtle changes occur in the immune system during pregnancy, which may help explain the increased risk of acquired infection and subsequent adverse effects. These changes include¹:

5 ways to reduce infection risk during pregnancy
Pregnant patients clearly are not as immunosuppressed as patients receiving chemotherapy or high doses of systemic glucocorticoids. Nevertheless, the subtle alterations in their immune system just described make pregnant women increasingly susceptible to certain infections. Therefore, I suggest these take-home messages for reducing infection risk in your pregnant patients.

1. Vaccinate against the flu
All pregnant women should be vaccinated each year for influenza. If your patient becomes infected despite vaccination, treat her promptly with an antiviral medication such as oseltamivir and observe her carefully for evidence of superimposed bacterial pneumonia. If the latter complication develops, hospitalize the patient immediately and treat her with appropriate broad spectrum antibiotics. (See, “Don’t wait for rapid flu test results. Treat your pregnant patient with antiviral therapy!”)

2. Avoid hepatitis E−endemic areas 
Ideally, our patients should avoid travel to areas of the world where hepatitis E is endemic. If travel cannot be avoided, the patient should receive the new hepatitis E vaccine. This vaccine is administered in a 3-dose series; in clinical trials, it has had an efficacy of 85% to 90%.²

If a patient acquires hepatitis E infection, she should receive aggressive supportive care, with hospitalization strongly considered because of the increased risk for hepatic failure. 

The clinical manifestations of hepatitis Eare very similar to those of hepatitis A: fever, malaise, anorexia, nausea, pain and tenderness in the right upper quadrant, jaundice, darkened urine, and clay-colored stools. Laboratory abnormalities in affected patients include elevated transaminase enzymes, elevated bilirubin, positive immunoglobulin M antibody against hepatitis E virus, a fourfold increase in a prior immunoglobulin G antibody titer against hepatitis E virus, and a positive test for hepatitis E RNA.²

3. Treat patients with HSV infection to avoid an outbreak during delivery
Pregnant women who develop primary or recurrent HSV infection should be treated promptly with therapeutic doses of acyclovir or valacyclovir. Patients with frequent recurrences should receive daily anti-HSV prophylaxis throughout pregnancy. Other patients should be treated prophylactically from week 36 until delivery.

4. Recommend malaria prophylaxis when appropriate
If your pregnant patient is traveling to an area of the world where malaria is endemic, she should receive appropriate prophylaxis, especially against P falciparum.

5. Vaccinate during pregnancy and promptly treat developed infections
All infections in pregnant women should be treated in a timely manner with appropriate antibiotics. Moreover, we should make a firm effort to provide all pregnant women with the following vaccinations: influenza, Tdap, and hepatitis B (if susceptible). Select patients also should receive pneumococcal vaccine (those who are immunosuppressed; have chronic medical illnessess, particulary cardiopulmonary disease; or have had a splenectomy).

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Pregnant women may be more severelyaffected by certain microorganisms than nonpregnant individuals.¹ In a recent review, Kourtis and colleagues cited evidence for increased mortality risk for pregnant patients related to 5 specific infections.¹ What are those infections and why does pregnancy put a woman at greater risk for adverse outcomes? I review these topics in this article and, based on this evidence, I suggest 5 specific ways to avoid infection during pregnancy.

Five infections that can lead to detrimental outcomes during pregnancy
Influenza
During the pandemic of 1918, maternal mortality was 27%. During the 1957 pandemic, 50% of influenza-related deaths among women of reproductive age occurred among pregnant women. In the 2009 H1N1 influenza A pandemic, pregnant women were clearly at increased risk for severe disease, reflected by an increased frequency of hospitalization and increased likelihood of admission to an intensive care unit.

Hepatitis E
Compared with nonpregnant women and men, pregnant women also are at markedly increased risk for mortality due to hepa-titis E infection, especially in Southeast Asia, the Middle East, and Africa. The pathophysiologic basis for this increase is not well understood. Interestingly, in a report from India, 33% to 43% of pregnant women infected with hepatitis E had such severe disease that they developed hepatic failure.

Herpes simplex virus
Pregnant women with primary herpes simplex virus (HSV) infection are at increased risk for hepatitis and for disseminated infection, compared with other nonpregnant adults. Only patients with obvious immunodeficiency disorders are at greater risk for disseminated HSV infection.

Malaria
Pregnant women are at significantly increased risk for acquiring Plasmodium falciparum malaria and developing severe, life-threatening disease. In multiple studies from the Asia-Pacific region, pregnant women have been threefold more likely to acquire malaria compared with nonpregnant individuals. In India, during the period 2004 to 2006, malaria was the most common cause of maternal death.

The most likely explanation for the deleterious effect of this particular form of malaria is the fact that the P falciparum parasites accumulate selectively in the placenta because they bind avidly to syncytiotrophoblastic chondroitin sulfate A. Intense inflammation in the placenta, in turn, can lead to early pregnancy loss, preterm delivery, and fetal infection.

Listeria
Another important infection to which pregnant women are particularly susceptible is listeriosis. Listeria monocytogenes may contaminate several types of food such as uncooked meats and vegetables, unpasteurized milk, and soft cheeses. The organism has a predilection to attack the placenta and fetus and can cause spontaneous abortion, stillbirth, preterm delivery, and neonatal infection. Hispanic women may be at unusually high risk for listeria.

Immune system changes during pregnancy
Certain subtle changes occur in the immune system during pregnancy, which may help explain the increased risk of acquired infection and subsequent adverse effects. These changes include¹:

5 ways to reduce infection risk during pregnancy
Pregnant patients clearly are not as immunosuppressed as patients receiving chemotherapy or high doses of systemic glucocorticoids. Nevertheless, the subtle alterations in their immune system just described make pregnant women increasingly susceptible to certain infections. Therefore, I suggest these take-home messages for reducing infection risk in your pregnant patients.

1. Vaccinate against the flu
All pregnant women should be vaccinated each year for influenza. If your patient becomes infected despite vaccination, treat her promptly with an antiviral medication such as oseltamivir and observe her carefully for evidence of superimposed bacterial pneumonia. If the latter complication develops, hospitalize the patient immediately and treat her with appropriate broad spectrum antibiotics. (See, “Don’t wait for rapid flu test results. Treat your pregnant patient with antiviral therapy!”)

2. Avoid hepatitis E−endemic areas 
Ideally, our patients should avoid travel to areas of the world where hepatitis E is endemic. If travel cannot be avoided, the patient should receive the new hepatitis E vaccine. This vaccine is administered in a 3-dose series; in clinical trials, it has had an efficacy of 85% to 90%.²

If a patient acquires hepatitis E infection, she should receive aggressive supportive care, with hospitalization strongly considered because of the increased risk for hepatic failure. 

The clinical manifestations of hepatitis Eare very similar to those of hepatitis A: fever, malaise, anorexia, nausea, pain and tenderness in the right upper quadrant, jaundice, darkened urine, and clay-colored stools. Laboratory abnormalities in affected patients include elevated transaminase enzymes, elevated bilirubin, positive immunoglobulin M antibody against hepatitis E virus, a fourfold increase in a prior immunoglobulin G antibody titer against hepatitis E virus, and a positive test for hepatitis E RNA.²

3. Treat patients with HSV infection to avoid an outbreak during delivery
Pregnant women who develop primary or recurrent HSV infection should be treated promptly with therapeutic doses of acyclovir or valacyclovir. Patients with frequent recurrences should receive daily anti-HSV prophylaxis throughout pregnancy. Other patients should be treated prophylactically from week 36 until delivery.

4. Recommend malaria prophylaxis when appropriate
If your pregnant patient is traveling to an area of the world where malaria is endemic, she should receive appropriate prophylaxis, especially against P falciparum.

5. Vaccinate during pregnancy and promptly treat developed infections
All infections in pregnant women should be treated in a timely manner with appropriate antibiotics. Moreover, we should make a firm effort to provide all pregnant women with the following vaccinations: influenza, Tdap, and hepatitis B (if susceptible). Select patients also should receive pneumococcal vaccine (those who are immunosuppressed; have chronic medical illnessess, particulary cardiopulmonary disease; or have had a splenectomy).

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Pregnant women may be more severelyaffected by certain microorganisms than nonpregnant individuals.¹ In a recent review, Kourtis and colleagues cited evidence for increased mortality risk for pregnant patients related to 5 specific infections.¹ What are those infections and why does pregnancy put a woman at greater risk for adverse outcomes? I review these topics in this article and, based on this evidence, I suggest 5 specific ways to avoid infection during pregnancy.

Five infections that can lead to detrimental outcomes during pregnancy
Influenza
During the pandemic of 1918, maternal mortality was 27%. During the 1957 pandemic, 50% of influenza-related deaths among women of reproductive age occurred among pregnant women. In the 2009 H1N1 influenza A pandemic, pregnant women were clearly at increased risk for severe disease, reflected by an increased frequency of hospitalization and increased likelihood of admission to an intensive care unit.

Hepatitis E
Compared with nonpregnant women and men, pregnant women also are at markedly increased risk for mortality due to hepa-titis E infection, especially in Southeast Asia, the Middle East, and Africa. The pathophysiologic basis for this increase is not well understood. Interestingly, in a report from India, 33% to 43% of pregnant women infected with hepatitis E had such severe disease that they developed hepatic failure.

Herpes simplex virus
Pregnant women with primary herpes simplex virus (HSV) infection are at increased risk for hepatitis and for disseminated infection, compared with other nonpregnant adults. Only patients with obvious immunodeficiency disorders are at greater risk for disseminated HSV infection.

Malaria
Pregnant women are at significantly increased risk for acquiring Plasmodium falciparum malaria and developing severe, life-threatening disease. In multiple studies from the Asia-Pacific region, pregnant women have been threefold more likely to acquire malaria compared with nonpregnant individuals. In India, during the period 2004 to 2006, malaria was the most common cause of maternal death.

The most likely explanation for the deleterious effect of this particular form of malaria is the fact that the P falciparum parasites accumulate selectively in the placenta because they bind avidly to syncytiotrophoblastic chondroitin sulfate A. Intense inflammation in the placenta, in turn, can lead to early pregnancy loss, preterm delivery, and fetal infection.

Listeria
Another important infection to which pregnant women are particularly susceptible is listeriosis. Listeria monocytogenes may contaminate several types of food such as uncooked meats and vegetables, unpasteurized milk, and soft cheeses. The organism has a predilection to attack the placenta and fetus and can cause spontaneous abortion, stillbirth, preterm delivery, and neonatal infection. Hispanic women may be at unusually high risk for listeria.

Immune system changes during pregnancy
Certain subtle changes occur in the immune system during pregnancy, which may help explain the increased risk of acquired infection and subsequent adverse effects. These changes include¹:

5 ways to reduce infection risk during pregnancy
Pregnant patients clearly are not as immunosuppressed as patients receiving chemotherapy or high doses of systemic glucocorticoids. Nevertheless, the subtle alterations in their immune system just described make pregnant women increasingly susceptible to certain infections. Therefore, I suggest these take-home messages for reducing infection risk in your pregnant patients.

1. Vaccinate against the flu
All pregnant women should be vaccinated each year for influenza. If your patient becomes infected despite vaccination, treat her promptly with an antiviral medication such as oseltamivir and observe her carefully for evidence of superimposed bacterial pneumonia. If the latter complication develops, hospitalize the patient immediately and treat her with appropriate broad spectrum antibiotics. (See, “Don’t wait for rapid flu test results. Treat your pregnant patient with antiviral therapy!”)

2. Avoid hepatitis E−endemic areas 
Ideally, our patients should avoid travel to areas of the world where hepatitis E is endemic. If travel cannot be avoided, the patient should receive the new hepatitis E vaccine. This vaccine is administered in a 3-dose series; in clinical trials, it has had an efficacy of 85% to 90%.²

If a patient acquires hepatitis E infection, she should receive aggressive supportive care, with hospitalization strongly considered because of the increased risk for hepatic failure. 

The clinical manifestations of hepatitis Eare very similar to those of hepatitis A: fever, malaise, anorexia, nausea, pain and tenderness in the right upper quadrant, jaundice, darkened urine, and clay-colored stools. Laboratory abnormalities in affected patients include elevated transaminase enzymes, elevated bilirubin, positive immunoglobulin M antibody against hepatitis E virus, a fourfold increase in a prior immunoglobulin G antibody titer against hepatitis E virus, and a positive test for hepatitis E RNA.²

3. Treat patients with HSV infection to avoid an outbreak during delivery
Pregnant women who develop primary or recurrent HSV infection should be treated promptly with therapeutic doses of acyclovir or valacyclovir. Patients with frequent recurrences should receive daily anti-HSV prophylaxis throughout pregnancy. Other patients should be treated prophylactically from week 36 until delivery.

4. Recommend malaria prophylaxis when appropriate
If your pregnant patient is traveling to an area of the world where malaria is endemic, she should receive appropriate prophylaxis, especially against P falciparum.

5. Vaccinate during pregnancy and promptly treat developed infections
All infections in pregnant women should be treated in a timely manner with appropriate antibiotics. Moreover, we should make a firm effort to provide all pregnant women with the following vaccinations: influenza, Tdap, and hepatitis B (if susceptible). Select patients also should receive pneumococcal vaccine (those who are immunosuppressed; have chronic medical illnessess, particulary cardiopulmonary disease; or have had a splenectomy).

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Of the approximately 1 million benign breast biopsies obtained annually from US women, some 10% yield a diagnosis of atypical hyperplasia, microscopically classified as ductal or lobular. Atypical hyperplasia represents a “proliferation of dysplastic, mono­tonous epithelial-cell populations that include clonal subpopulations. In models of breast carcinogenesis, atypical hyperplasia occupies a transitional zone between benign and malignant disease,” write Hartmann and colleagues, the authors of a recent special report in the New England Journal of Medicine.¹

Long-term follow-up studies have found atypical hyperplasia to confer a relative risk for breast cancer of 4.0. Although these findings are well established, the cumulative absolute risk for breast cancer conferred by a diagnosis of atypical hyperplasia only recently has been described. Hartmann and colleagues note that it approaches 30% over 25 years.¹

Recommendations for clinical practice
The authors of this special report do a service to women and their clinicians by pointing out the high long-term risk of malignancy faced by women with atypical hyperplasia of the breast. They also make a number of important recommendations for practice:

Most women will not develop breast malignancy
As Hartmann and colleagues point out, all is not dire once a woman is diagnosed with atypical hyperplasia of the breast. In most of these women, breast cancer will not develop—and if it does develop, it may occur at an age when mortality from other causes is more likely than from breast cancer. In this respect, women with atypical hyperplasia of the breast are different from carriers of BRCA mutations. Although women with atypical hyperplasia as well as mutation carriers are both at high lifetime risk for breast cancer, breast malignancies occur at an earlier age in mutation carriers. Accordingly, as the authors of this special report advise, in general, a diagnosis of atypical hyperplasia should not be considered an indication for risk-reducing bilateral mastectomy.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Of the approximately 1 million benign breast biopsies obtained annually from US women, some 10% yield a diagnosis of atypical hyperplasia, microscopically classified as ductal or lobular. Atypical hyperplasia represents a “proliferation of dysplastic, mono­tonous epithelial-cell populations that include clonal subpopulations. In models of breast carcinogenesis, atypical hyperplasia occupies a transitional zone between benign and malignant disease,” write Hartmann and colleagues, the authors of a recent special report in the New England Journal of Medicine.¹

Long-term follow-up studies have found atypical hyperplasia to confer a relative risk for breast cancer of 4.0. Although these findings are well established, the cumulative absolute risk for breast cancer conferred by a diagnosis of atypical hyperplasia only recently has been described. Hartmann and colleagues note that it approaches 30% over 25 years.¹

Recommendations for clinical practice
The authors of this special report do a service to women and their clinicians by pointing out the high long-term risk of malignancy faced by women with atypical hyperplasia of the breast. They also make a number of important recommendations for practice:

Most women will not develop breast malignancy
As Hartmann and colleagues point out, all is not dire once a woman is diagnosed with atypical hyperplasia of the breast. In most of these women, breast cancer will not develop—and if it does develop, it may occur at an age when mortality from other causes is more likely than from breast cancer. In this respect, women with atypical hyperplasia of the breast are different from carriers of BRCA mutations. Although women with atypical hyperplasia as well as mutation carriers are both at high lifetime risk for breast cancer, breast malignancies occur at an earlier age in mutation carriers. Accordingly, as the authors of this special report advise, in general, a diagnosis of atypical hyperplasia should not be considered an indication for risk-reducing bilateral mastectomy.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Of the approximately 1 million benign breast biopsies obtained annually from US women, some 10% yield a diagnosis of atypical hyperplasia, microscopically classified as ductal or lobular. Atypical hyperplasia represents a “proliferation of dysplastic, mono­tonous epithelial-cell populations that include clonal subpopulations. In models of breast carcinogenesis, atypical hyperplasia occupies a transitional zone between benign and malignant disease,” write Hartmann and colleagues, the authors of a recent special report in the New England Journal of Medicine.¹

Long-term follow-up studies have found atypical hyperplasia to confer a relative risk for breast cancer of 4.0. Although these findings are well established, the cumulative absolute risk for breast cancer conferred by a diagnosis of atypical hyperplasia only recently has been described. Hartmann and colleagues note that it approaches 30% over 25 years.¹

Recommendations for clinical practice
The authors of this special report do a service to women and their clinicians by pointing out the high long-term risk of malignancy faced by women with atypical hyperplasia of the breast. They also make a number of important recommendations for practice:

Most women will not develop breast malignancy
As Hartmann and colleagues point out, all is not dire once a woman is diagnosed with atypical hyperplasia of the breast. In most of these women, breast cancer will not develop—and if it does develop, it may occur at an age when mortality from other causes is more likely than from breast cancer. In this respect, women with atypical hyperplasia of the breast are different from carriers of BRCA mutations. Although women with atypical hyperplasia as well as mutation carriers are both at high lifetime risk for breast cancer, breast malignancies occur at an earlier age in mutation carriers. Accordingly, as the authors of this special report advise, in general, a diagnosis of atypical hyperplasia should not be considered an indication for risk-reducing bilateral mastectomy.

Share your thoughts on this article! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Proximal humerus fractures account for 4% to 5% of all fractures.¹ These fractures occur most frequently in the elderly—patients older than 60 years sustain 71% of these injuries²—and in females.^1,3 Given an aging population, this incidence is predicted to increase 3-fold over the next 30 years.⁴ There is much debate regarding management of acute, displaced proximal humerus fractures. A recent Cochrane Review of published outcomes of operative and nonoperative treatment of displaced proximal humerus fractures found insufficient evidence supporting either modality, though surgery was associated with additional procedures.⁵ A review of 1000 proximal humerus fractures found that 49% had less than 1 cm of displacement of the major fragments or angulation of less than 45°.³ Other authors have reported similar findings.^6,7 Although the incidence of proximal humerus fractures has remained stable over the past decade, from 1999 to 2005 there was a 25% relative increase in surgical management, including a relative increase of 29% in open reduction and internal fixation (ORIF) versus a 20% increase in arthroplasty.¹

Locking plates have consistently demonstrated biomechanical superiority over other forms of fixation in osteoporotic bone.^8-11 Egol and colleagues⁸ found that osteoporotic bone limited the torque of fixation to values less than what is required for adequate frictional force between the plate and the bone. This problem can be overcome with fixed-angle devices, such as locked plates.⁹ Compared with locked nail constructs, proximal humerus locking plates have demonstrated superiority in torsion, loading, and varus bending.^10,11 Compared with blade plates, proximal humerus locking plates exhibited increased stiffness and torsional fatigue resistance.¹² In a randomized clinical trial, Olerud and colleagues¹³ reported superior functional results with locking plate fixation compared with nonoperative treatment of displaced 3-part fractures in elderly patients with 2-year follow-up, though these clinical results were not supported by others.¹⁴ Two recent case–control studies comparing functional outcomes for 3- and 4-part fractures with follow-up of more than 2 years revealed higher Constant scores after locked plating compared with hemiarthroplasty, though complications were higher with locked plates.^15,16 Adoption of locked proximal humerus plating has been correlated with good clinical outcomes and union rates, though this has been accompanied by a higher rate of reoperation.⁷ Reoperation rates from 1999 to 2005 increased both in the immediate postoperative period (odds ratio, 3.36) and at 1 year (odds ratio, 3.90).¹

Complications of Locked Plating

Regardless of fixation type, reduced humeral head bone mass and quality may lead to implant loosening, fracture redisplacement, and, ultimately, poor outcomes. Baseline osteoporosis may predict likelihood of fixation failure.¹⁷ Multiple studies have reported on the implant-related complications associated with locking plate fixation—most commonly, intra-articular screw penetration, postoperative fracture displacement, and avascular necrosis (AVN)^18-24 (Figure 1). A meta-analysis of 12 studies with a total of 514 proximal humerus fractures treated with locking plate fixation showed an overall complication rate of 49% and a 13.8% reoperation rate.²⁵ The most common indication for reoperation involved intra-articular screw perforation. The most common complications were varus malunion (16%), osteonecrosis (10%), intra-articular screw penetration (8%), subacromial impingement (6%), and infection (4%).

Suboptimal intraoperative fracture reduction, specifically with residual varus, has been correlated with loss of fracture fixation. In a series of 153 fractures, loss of fixation occurred in 13.7% of cases, with the leading risk factor being varus malreduction.¹⁹ Failure rates were 30.4% and 11% when the head shaft angle was less than 120° and when it was 120° or more, respectively. Solberg and colleagues¹⁶ found that initial postoperative varus angulation of more than 20° resulted in universal loss of fixation. Conversion of these cases to hemiarthroplasty resulted in poor outcomes. Preoperative fracture alignment may also predict fixation failure.²² In one series, initial varus angulation healed with a mean 16° varus and a Constant score of 63, whereas initial valgus alignment healed with 6° varus and a Constant score of 71.²² Complications occurred in fractures that were initially in varus 79% of the time and initially in valgus 19% of the time. Screw perforation has been associated with loss of reduction 44% of the time.²⁰

In an analysis of locking plate constructs revised after early (<4 weeks) failure in 8 patients with osteoporosis, Micic and colleagues²¹ found implant pullout leading to varus malalignment. All cases lacked medial support and subchondral screw purchase; 3 were initially malreduced. Owsley and Gorczyca²³ retrospectively reviewed 53 cases of displaced proximal humerus fractures treated with locked plating. Despite the high rate of radiographic union, 36% developed complications, including screw cutout (23%), varus displacement of more than 10° (25%), and AVN (4%); 13% required revision. These complications disproportionately affected patients older than 60 years (57%) and negatively affected functional outcomes.

Augmentation Techniques

Despite its reported complications, proximal humerus locked plating remains the most widely used type of fixation.¹ Advancements in locking plate design, improved understanding of fixation principles, and adoption of techniques augmenting proximal humerus locking plate fixation, particularly in osteoporotic bone, have reduced postoperative complications (Table 1).

Rotator Cuff Sutures

A widely adopted technique for neutralizing rotator cuff–deforming forces, which theoretically can cause fracture displacement, is incorporation of heavy nonabsorbable sutures. These sutures are placed through the rotator cuff–tuberosity junction and tied down after being passed through the plate. Obtaining and maintaining tuberosity reduction are essential in achieving good functional outcomes after fixation. In addition, tension band sutures may be particularly useful in the setting of initial varus deformity.²⁶

Although clinical use of these sutures is common, biomechanical studies of their adjunctive contribution to fracture stability are lacking.²⁷ The rotator cuff musculature has a maximal contractile force of 3.5 kg/cm².²⁸ Ricchetti and colleagues²⁹ described a technique that involves using a locked plate and tagging the rotator cuff with heavy nonabsorbable sutures. Selective traction on the sutures can help obtain and maintain fracture reduction. Multiple studies have reported on suture use with locked plating for proximal humerus fractures.^29-34 Badman and colleagues³⁰ retrospectively reviewed 81 cases of metaphyseal defects or medial comminution treated with locked plating, rotator cuff sutures, and structural allograft. All cases healed within 6 months after surgery. The incidence of screw cutout was 3.7%, the incidence of AVN was 6.2%, and the incidence of varus collapse was 6%. A cadaveric study that used specimens (mean age, 77 years) with a simulated 3-part proximal humerus fracture treated with a locked plate both with and without cerclage sutures found no difference in interfragmentary motion between the groups.²⁷ The authors concluded that additive sutures are not required for anatomically reduced fractures. Multiple sutures may counteract the deforming forces that act on bony segments that cannot be adequately maintained with screws, such as an osteoporotic greater tuberosity.

Medial Column Restoration

The importance of reducing and maintaining the medial calcar to provide biomechanical support for a laterally placed plate has been recognized.^26,34-37 Gardner and colleagues²⁶ suggested that medial support was achieved if the medial cortex was anatomically reduced, if the proximal fragment was impacted laterally onto the shaft, or if 1 or more inferomedial screws were placed. Cases that did not achieve medial support developed significantly more humeral head subsidence (5.8 mm vs 1.2 mm) and screw penetration. Krappinger and colleagues³⁶ found that factors leading to fixation failure included age, local bone mineral density, anatomical reduction, and restoration of the medial cortical support. The authors concluded that anatomical reduction and restoration of the medial cortex were important in minimizing mechanical loads at the bone–implant interface. Biomechanically, Lescheid and colleagues³⁷ found that the most stable construct was anatomical reduction with medial cortical contact. In the setting of comminution, however, it may be preferable to intentionally perform varus malreduction to achieve medial contact than to achieve anatomical reduction with a fracture gap. Badman and colleagues³⁰ found that the incidence of screw penetration was 6% in patients with an intact medial calcar versus 29% in patients without medial support. In a retrospective analysis of patients treated with a locking plate and suture augmentation, Jung and colleagues³⁵ concluded that restoring medial support was the most reliable factor in the prevention of loss of reduction with or without screw perforation. Last, Solberg and colleagues¹⁶ reported better clinical outcomes when the length of the metaphyseal segment attached to the articular fragment was more than 2 mm. A length of less than 2 mm was predictive of developing AVN.

Use of Bone Void Fillers

Allograft. Allograft is cancellous or corticocancellous chips or tricortical graft used as osteoconductive filler for metaphyseal defects.³⁸ An increasingly popular technique involves using an endosteal fibular allograft strut to indirectly reduce the fracture and help support the medial calcar.^39-42 Hettrich and colleagues⁴⁰ reported on radiographic outcomes of displaced proximal humerus fractures with medial comminution treated with a locked plate and an endosteal fibular allograft or semitubular plate. The reduction was maintained in 96% of cases; there was 1 varus collapse. There were no cases of implant failure, screw perforation, or AVN. Other authors have also reported on successful use of fibular allograft in conjunction with a locked plate; the rate of reduction loss was low, and there were no cases of screw cutout or intra-articular screw penetration.^30,41,42 These clinical outcomes are supported by results of biomechanical studies of the added benefit of intramedullary fibular allograft.^43-46 Mathison and colleagues⁴³ reported that a construct with fibular allograft and a locking plate increased the failure load by 1.72 times and the stiffness by 3.84 times compared with a control group of locking plate only. Bae and colleagues⁴⁶ found significantly higher maximum failure load and construct stiffness with no varus collapse in specimens prepared with locked plate and fibular strut augmentation compared with a control group.

Others have successfully used cancellous allograft to fill humeral head bone defects.^29,32,47-49 Duralde and Leddy⁴⁷ reported 100% radiographic union and 81% good to excellent results in cases treated with a locking plate and morselized cancellous allograft to fill bone voids. Varus collapse and screw cutout did not occur, but there were 2 cases of AVN. Ricchetti and colleagues²⁹ reviewed 54 cases treated with a locking plate and rotator cuff suture construct. Allograft cancellous chips and demineralized bone matrix were used in 3- and 4-part fractures (70% of cases) along with shorter screws in the humeral head. Major complications included AVN (1), fixation failure (3), and varus malunion (5). Others investigators have had less favorable results with use of cancellous bone graft. Schliemann and colleagues¹⁹ reported on 27 patients who were older than 65 years when they underwent ORIF with rotator cuff sutures to stabilize the tuberosities and either cancellous graft or a synthetic bone substitute in patients with massive metaphyseal defects. Patient-reported outcomes were superior to Constant scores. Complications included screw penetration (22.2%), reduction loss (44.4%), implant failure (3.7%), and AVN (29.6%).

Autograft. Autograft has both osteoconductive and osteoinductive properties and has been successfully used for metaphyseal defects.^32,50 Kim and colleagues⁵⁰ reported on patients with 4-part proximal humerus fractures treated with a locking plate and autologous iliac graft. All cases achieved union and had good or excellent outcomes. There were no cases of AVN, varus collapse, or hardware-related complications.

Bone Cement. Calcium phosphate cement has osteoconductive properties and enhances screw purchase in cancellous bone (Figures 2A–2F). It can be injected or molded into bone voids to provide improved compressive strength. It is resorbed through cell-mediated processes resembling bone remodeling and does not disappear until new bone forms. (Calcium sulfate cement, on the other hand, resorbs through a chemical process independent of new bone formation.⁵¹) Egol and colleagues⁵² reviewed the cases of 92 patients (mean age, 61 years) with 2-, 3-, and 4-part proximal humerus fractures treated with locked plate fixation. Metaphyseal defects were treated with no augmentation, augmentation with cancellous chips, or augmentation with calcium phosphate cement. Adding calcium phosphate cement was associated with lower incidence of intra-articular screw penetration and humeral head settling. In a recent cadaveric biomechanical study using 2-part proximal humerus fractures with metaphyseal comminution, the group augmented with calcium phosphate cement had enhanced axial stiffness and load to failure with reduced screw penetration.⁵³ Other biomechanical studies have found increased screw pullout strength⁵⁴ and decreased interfragmentary motion when specimens were augmented with calcium phosphate cement.⁵⁵

Similar good clinical and radiographic outcomes have been observed with use of calcium sulfate cement.^56,57 Somasundaram and colleagues⁵⁶ reported good clinical outcomes in 82% of patients treated with locking plates and calcium sulfate cement used to fill metaphyseal voids. All fractures united without infection, fixation failure, subsequent malunion, tuberosity failure, or AVN. Lee and Shin⁵⁷ compared outcomes of 14 patients who received calcium sulfate augmentation with outcomes of patients who did not receive this augmentation. Overall, 89% of patients had good or excellent results. Calcium sulfate cement did not affect the reduction failure rate or clinical outcomes in cases in which medial cortical reduction was achieved. However, postoperative displacement caused by lack of medial support was associated with poor outcomes.

Screw Placement

Screws optimally should be placed in the posterior-medial-inferior aspect of the humeral head to provide medial support for the fracture and mechanical stability.⁵⁸ Cadaveric studies have shown the highest cancellous bone density in the proximal, posterior, and medial portions of the humeral head.^59-63 Similarly, in a cadaveric study, Liew and colleagues⁶¹ found greater screw purchase and higher pullout strength when the screw was placed in the center of the humeral head, within subchondral bone; fixation was poorest when the screw was placed in the anterosuperior region of the humeral head. Tingart and colleagues⁶² reported that humeral head trabecular density significantly affected pullout strength of cancellous screws. In addition, the most pullout strength was at the center of the head, and the least within the anterosuperior head. Trabecular density was higher in the inferior and posterior regions than in the superior and anterior regions.

Most locking plate designs allow screws to be placed at the level of the medial calcar—the goal being to provide medial column support (Table 2). Zhang and colleagues⁵⁸ treated 2-, 3-, and 4-part fractures with a locking plate and randomized them into receiving the plate with or without medial support screws. For 3- and 4-part fractures, the group with these screws had a significantly greater final neck-shaft angle and smaller angulation loss compared with the group without screws. No additional benefit was found for 2-part fractures. Erhardt and colleagues⁶³ simulated unstable proximal humerus fractures using cadavers and testing different fixation methods using a polyaxial locking plate. They found that 5 screws in the head fragment and an inferomedial support screw significantly reduced the risk of screw perforation. Other authors have concluded that placing 1 or more inferomedial screws is important in cases of medial comminution or medial column malreduction.²⁶ Interestingly, compared with use of a polyaxial implant, which allows for adjustment of screw direction, use of a monoaxial locking plate did not lead to a clinically different outcome or complication profile.⁶⁴

Techniques have been used to achieve subchondral purchase of locking screws while reducing iatrogenic articular perforation.⁶⁵ However, given the incidence of fracture settling and subsequent postoperative screw penetration, many authors currently recommend using shorter divergent screws combined with other augmentation techniques, described previously.^17,29,32

Physical Therapy

There is no standardized physiotherapy regimen for postoperative management of proximal humerus fractures treated with locking plates.²⁵ In older patients, immediate active range of motion (ROM) exercises should be delayed until early callus is noted, though there is a risk for stiffness. Lee and Shin⁵⁷ found that a delay in rehabilitation after ORIF was an independent risk for poor clinical outcome. Namdari and colleagues¹⁷ recommended sling use only for comfort and initiated non-load-bearing activities and pendulum exercises immediately after surgery. Patients with adequate reduction at 4 to 6 weeks were advanced to full weight-bearing. Badman and colleagues³⁰ initiated passive-assisted ROM exercises when the wound was healed at 2 weeks in 2-part fractures, whereas patients with 3- and 4-part fractures were immobilized until radiographic healing. Formal therapy was started after 6 weeks. Stiffness was reported in 5% of patients. For patients with stable fixation, Ricchetti and colleagues²⁹ recommended passive shoulder ROM exercises on postoperative day 1; at 4 to 6 weeks, patients should start active shoulder ROM exercises, and then resistance exercises at 10 to 12 weeks. Other authors are more conservative—only sling immobilization and pendulum exercises the first month.⁶⁶ Barlow and colleagues³² immobilized their patients (age, >75 years) for 6 weeks. No patient developed disabling stiffness. The authors suggested that patients older than 75 years may not be prone to stiffness.

Our Preferred Treatment Method

All proximal humerus fractures are approached anteriorly through the deltopectoral interval (Figure 3A). The long head biceps is identified and truncated for later tenodesis. Multiple No. 5 Ethibond sutures (Ethicon) are placed at the bone–tendon interface. The fracture is reduced with a Cobb elevator (Figure 3B), and provisional Kirschner wires are placed within the head (Figure 3C). The plate is affixed to the humeral head with its anterior border paralleling the posterior aspect of the bicipital groove. Multiple locking screws are placed within the superior and posterior humeral head. Nonlocking screws are then used to fix the plate to the shaft to reduce the specific deformity. Under fluoroscopy, any metaphyseal void is filled with calcium phosphate cement (Figure 3D). The remaining inferior screws are placed within the humeral head. Dr. Gruson uses screws 4 to 6 mm short of subchondral bone to reduce the risk for joint penetration. The rotator cuff sutures are tied down through the plate. Patients are started on progressive supine passive ROM exercises at 7 days, followed by supine active-assisted ROM exercises 6 weeks after fracture healing is confirmed radiographically.

Conclusion

Use of locked plating for proximal humerus fractures has increased, particularly in the elderly. Resulting complications include intra-articular screw penetration, postoperative fracture displacement, and AVN. Recognition of the importance of reducing and supporting the medial calcar, filling any metaphyseal defects, and selectively placing screws within the humeral head has lowered the incidence of these complications. Further comparative studies evaluating the efficacy of individual augmentation techniques are needed to determine their contribution to successful fracture healing and their cost-effectiveness. Results of such studies may help in the development of protocols for more standardized implementation of these techniques and in understanding which specific fracture patterns and patients would benefit from their use.

Proximal humerus fractures account for 4% to 5% of all fractures.¹ These fractures occur most frequently in the elderly—patients older than 60 years sustain 71% of these injuries²—and in females.^1,3 Given an aging population, this incidence is predicted to increase 3-fold over the next 30 years.⁴ There is much debate regarding management of acute, displaced proximal humerus fractures. A recent Cochrane Review of published outcomes of operative and nonoperative treatment of displaced proximal humerus fractures found insufficient evidence supporting either modality, though surgery was associated with additional procedures.⁵ A review of 1000 proximal humerus fractures found that 49% had less than 1 cm of displacement of the major fragments or angulation of less than 45°.³ Other authors have reported similar findings.^6,7 Although the incidence of proximal humerus fractures has remained stable over the past decade, from 1999 to 2005 there was a 25% relative increase in surgical management, including a relative increase of 29% in open reduction and internal fixation (ORIF) versus a 20% increase in arthroplasty.¹

Locking plates have consistently demonstrated biomechanical superiority over other forms of fixation in osteoporotic bone.^8-11 Egol and colleagues⁸ found that osteoporotic bone limited the torque of fixation to values less than what is required for adequate frictional force between the plate and the bone. This problem can be overcome with fixed-angle devices, such as locked plates.⁹ Compared with locked nail constructs, proximal humerus locking plates have demonstrated superiority in torsion, loading, and varus bending.^10,11 Compared with blade plates, proximal humerus locking plates exhibited increased stiffness and torsional fatigue resistance.¹² In a randomized clinical trial, Olerud and colleagues¹³ reported superior functional results with locking plate fixation compared with nonoperative treatment of displaced 3-part fractures in elderly patients with 2-year follow-up, though these clinical results were not supported by others.¹⁴ Two recent case–control studies comparing functional outcomes for 3- and 4-part fractures with follow-up of more than 2 years revealed higher Constant scores after locked plating compared with hemiarthroplasty, though complications were higher with locked plates.^15,16 Adoption of locked proximal humerus plating has been correlated with good clinical outcomes and union rates, though this has been accompanied by a higher rate of reoperation.⁷ Reoperation rates from 1999 to 2005 increased both in the immediate postoperative period (odds ratio, 3.36) and at 1 year (odds ratio, 3.90).¹

Complications of Locked Plating

Regardless of fixation type, reduced humeral head bone mass and quality may lead to implant loosening, fracture redisplacement, and, ultimately, poor outcomes. Baseline osteoporosis may predict likelihood of fixation failure.¹⁷ Multiple studies have reported on the implant-related complications associated with locking plate fixation—most commonly, intra-articular screw penetration, postoperative fracture displacement, and avascular necrosis (AVN)^18-24 (Figure 1). A meta-analysis of 12 studies with a total of 514 proximal humerus fractures treated with locking plate fixation showed an overall complication rate of 49% and a 13.8% reoperation rate.²⁵ The most common indication for reoperation involved intra-articular screw perforation. The most common complications were varus malunion (16%), osteonecrosis (10%), intra-articular screw penetration (8%), subacromial impingement (6%), and infection (4%).

Suboptimal intraoperative fracture reduction, specifically with residual varus, has been correlated with loss of fracture fixation. In a series of 153 fractures, loss of fixation occurred in 13.7% of cases, with the leading risk factor being varus malreduction.¹⁹ Failure rates were 30.4% and 11% when the head shaft angle was less than 120° and when it was 120° or more, respectively. Solberg and colleagues¹⁶ found that initial postoperative varus angulation of more than 20° resulted in universal loss of fixation. Conversion of these cases to hemiarthroplasty resulted in poor outcomes. Preoperative fracture alignment may also predict fixation failure.²² In one series, initial varus angulation healed with a mean 16° varus and a Constant score of 63, whereas initial valgus alignment healed with 6° varus and a Constant score of 71.²² Complications occurred in fractures that were initially in varus 79% of the time and initially in valgus 19% of the time. Screw perforation has been associated with loss of reduction 44% of the time.²⁰

In an analysis of locking plate constructs revised after early (<4 weeks) failure in 8 patients with osteoporosis, Micic and colleagues²¹ found implant pullout leading to varus malalignment. All cases lacked medial support and subchondral screw purchase; 3 were initially malreduced. Owsley and Gorczyca²³ retrospectively reviewed 53 cases of displaced proximal humerus fractures treated with locked plating. Despite the high rate of radiographic union, 36% developed complications, including screw cutout (23%), varus displacement of more than 10° (25%), and AVN (4%); 13% required revision. These complications disproportionately affected patients older than 60 years (57%) and negatively affected functional outcomes.

Augmentation Techniques

Despite its reported complications, proximal humerus locked plating remains the most widely used type of fixation.¹ Advancements in locking plate design, improved understanding of fixation principles, and adoption of techniques augmenting proximal humerus locking plate fixation, particularly in osteoporotic bone, have reduced postoperative complications (Table 1).

Rotator Cuff Sutures

A widely adopted technique for neutralizing rotator cuff–deforming forces, which theoretically can cause fracture displacement, is incorporation of heavy nonabsorbable sutures. These sutures are placed through the rotator cuff–tuberosity junction and tied down after being passed through the plate. Obtaining and maintaining tuberosity reduction are essential in achieving good functional outcomes after fixation. In addition, tension band sutures may be particularly useful in the setting of initial varus deformity.²⁶

Although clinical use of these sutures is common, biomechanical studies of their adjunctive contribution to fracture stability are lacking.²⁷ The rotator cuff musculature has a maximal contractile force of 3.5 kg/cm².²⁸ Ricchetti and colleagues²⁹ described a technique that involves using a locked plate and tagging the rotator cuff with heavy nonabsorbable sutures. Selective traction on the sutures can help obtain and maintain fracture reduction. Multiple studies have reported on suture use with locked plating for proximal humerus fractures.^29-34 Badman and colleagues³⁰ retrospectively reviewed 81 cases of metaphyseal defects or medial comminution treated with locked plating, rotator cuff sutures, and structural allograft. All cases healed within 6 months after surgery. The incidence of screw cutout was 3.7%, the incidence of AVN was 6.2%, and the incidence of varus collapse was 6%. A cadaveric study that used specimens (mean age, 77 years) with a simulated 3-part proximal humerus fracture treated with a locked plate both with and without cerclage sutures found no difference in interfragmentary motion between the groups.²⁷ The authors concluded that additive sutures are not required for anatomically reduced fractures. Multiple sutures may counteract the deforming forces that act on bony segments that cannot be adequately maintained with screws, such as an osteoporotic greater tuberosity.

Medial Column Restoration

The importance of reducing and maintaining the medial calcar to provide biomechanical support for a laterally placed plate has been recognized.^26,34-37 Gardner and colleagues²⁶ suggested that medial support was achieved if the medial cortex was anatomically reduced, if the proximal fragment was impacted laterally onto the shaft, or if 1 or more inferomedial screws were placed. Cases that did not achieve medial support developed significantly more humeral head subsidence (5.8 mm vs 1.2 mm) and screw penetration. Krappinger and colleagues³⁶ found that factors leading to fixation failure included age, local bone mineral density, anatomical reduction, and restoration of the medial cortical support. The authors concluded that anatomical reduction and restoration of the medial cortex were important in minimizing mechanical loads at the bone–implant interface. Biomechanically, Lescheid and colleagues³⁷ found that the most stable construct was anatomical reduction with medial cortical contact. In the setting of comminution, however, it may be preferable to intentionally perform varus malreduction to achieve medial contact than to achieve anatomical reduction with a fracture gap. Badman and colleagues³⁰ found that the incidence of screw penetration was 6% in patients with an intact medial calcar versus 29% in patients without medial support. In a retrospective analysis of patients treated with a locking plate and suture augmentation, Jung and colleagues³⁵ concluded that restoring medial support was the most reliable factor in the prevention of loss of reduction with or without screw perforation. Last, Solberg and colleagues¹⁶ reported better clinical outcomes when the length of the metaphyseal segment attached to the articular fragment was more than 2 mm. A length of less than 2 mm was predictive of developing AVN.

Use of Bone Void Fillers

Allograft. Allograft is cancellous or corticocancellous chips or tricortical graft used as osteoconductive filler for metaphyseal defects.³⁸ An increasingly popular technique involves using an endosteal fibular allograft strut to indirectly reduce the fracture and help support the medial calcar.^39-42 Hettrich and colleagues⁴⁰ reported on radiographic outcomes of displaced proximal humerus fractures with medial comminution treated with a locked plate and an endosteal fibular allograft or semitubular plate. The reduction was maintained in 96% of cases; there was 1 varus collapse. There were no cases of implant failure, screw perforation, or AVN. Other authors have also reported on successful use of fibular allograft in conjunction with a locked plate; the rate of reduction loss was low, and there were no cases of screw cutout or intra-articular screw penetration.^30,41,42 These clinical outcomes are supported by results of biomechanical studies of the added benefit of intramedullary fibular allograft.^43-46 Mathison and colleagues⁴³ reported that a construct with fibular allograft and a locking plate increased the failure load by 1.72 times and the stiffness by 3.84 times compared with a control group of locking plate only. Bae and colleagues⁴⁶ found significantly higher maximum failure load and construct stiffness with no varus collapse in specimens prepared with locked plate and fibular strut augmentation compared with a control group.

Others have successfully used cancellous allograft to fill humeral head bone defects.^29,32,47-49 Duralde and Leddy⁴⁷ reported 100% radiographic union and 81% good to excellent results in cases treated with a locking plate and morselized cancellous allograft to fill bone voids. Varus collapse and screw cutout did not occur, but there were 2 cases of AVN. Ricchetti and colleagues²⁹ reviewed 54 cases treated with a locking plate and rotator cuff suture construct. Allograft cancellous chips and demineralized bone matrix were used in 3- and 4-part fractures (70% of cases) along with shorter screws in the humeral head. Major complications included AVN (1), fixation failure (3), and varus malunion (5). Others investigators have had less favorable results with use of cancellous bone graft. Schliemann and colleagues¹⁹ reported on 27 patients who were older than 65 years when they underwent ORIF with rotator cuff sutures to stabilize the tuberosities and either cancellous graft or a synthetic bone substitute in patients with massive metaphyseal defects. Patient-reported outcomes were superior to Constant scores. Complications included screw penetration (22.2%), reduction loss (44.4%), implant failure (3.7%), and AVN (29.6%).

Autograft. Autograft has both osteoconductive and osteoinductive properties and has been successfully used for metaphyseal defects.^32,50 Kim and colleagues⁵⁰ reported on patients with 4-part proximal humerus fractures treated with a locking plate and autologous iliac graft. All cases achieved union and had good or excellent outcomes. There were no cases of AVN, varus collapse, or hardware-related complications.

Bone Cement. Calcium phosphate cement has osteoconductive properties and enhances screw purchase in cancellous bone (Figures 2A–2F). It can be injected or molded into bone voids to provide improved compressive strength. It is resorbed through cell-mediated processes resembling bone remodeling and does not disappear until new bone forms. (Calcium sulfate cement, on the other hand, resorbs through a chemical process independent of new bone formation.⁵¹) Egol and colleagues⁵² reviewed the cases of 92 patients (mean age, 61 years) with 2-, 3-, and 4-part proximal humerus fractures treated with locked plate fixation. Metaphyseal defects were treated with no augmentation, augmentation with cancellous chips, or augmentation with calcium phosphate cement. Adding calcium phosphate cement was associated with lower incidence of intra-articular screw penetration and humeral head settling. In a recent cadaveric biomechanical study using 2-part proximal humerus fractures with metaphyseal comminution, the group augmented with calcium phosphate cement had enhanced axial stiffness and load to failure with reduced screw penetration.⁵³ Other biomechanical studies have found increased screw pullout strength⁵⁴ and decreased interfragmentary motion when specimens were augmented with calcium phosphate cement.⁵⁵

Similar good clinical and radiographic outcomes have been observed with use of calcium sulfate cement.^56,57 Somasundaram and colleagues⁵⁶ reported good clinical outcomes in 82% of patients treated with locking plates and calcium sulfate cement used to fill metaphyseal voids. All fractures united without infection, fixation failure, subsequent malunion, tuberosity failure, or AVN. Lee and Shin⁵⁷ compared outcomes of 14 patients who received calcium sulfate augmentation with outcomes of patients who did not receive this augmentation. Overall, 89% of patients had good or excellent results. Calcium sulfate cement did not affect the reduction failure rate or clinical outcomes in cases in which medial cortical reduction was achieved. However, postoperative displacement caused by lack of medial support was associated with poor outcomes.

Screw Placement

Screws optimally should be placed in the posterior-medial-inferior aspect of the humeral head to provide medial support for the fracture and mechanical stability.⁵⁸ Cadaveric studies have shown the highest cancellous bone density in the proximal, posterior, and medial portions of the humeral head.^59-63 Similarly, in a cadaveric study, Liew and colleagues⁶¹ found greater screw purchase and higher pullout strength when the screw was placed in the center of the humeral head, within subchondral bone; fixation was poorest when the screw was placed in the anterosuperior region of the humeral head. Tingart and colleagues⁶² reported that humeral head trabecular density significantly affected pullout strength of cancellous screws. In addition, the most pullout strength was at the center of the head, and the least within the anterosuperior head. Trabecular density was higher in the inferior and posterior regions than in the superior and anterior regions.

Most locking plate designs allow screws to be placed at the level of the medial calcar—the goal being to provide medial column support (Table 2). Zhang and colleagues⁵⁸ treated 2-, 3-, and 4-part fractures with a locking plate and randomized them into receiving the plate with or without medial support screws. For 3- and 4-part fractures, the group with these screws had a significantly greater final neck-shaft angle and smaller angulation loss compared with the group without screws. No additional benefit was found for 2-part fractures. Erhardt and colleagues⁶³ simulated unstable proximal humerus fractures using cadavers and testing different fixation methods using a polyaxial locking plate. They found that 5 screws in the head fragment and an inferomedial support screw significantly reduced the risk of screw perforation. Other authors have concluded that placing 1 or more inferomedial screws is important in cases of medial comminution or medial column malreduction.²⁶ Interestingly, compared with use of a polyaxial implant, which allows for adjustment of screw direction, use of a monoaxial locking plate did not lead to a clinically different outcome or complication profile.⁶⁴

Techniques have been used to achieve subchondral purchase of locking screws while reducing iatrogenic articular perforation.⁶⁵ However, given the incidence of fracture settling and subsequent postoperative screw penetration, many authors currently recommend using shorter divergent screws combined with other augmentation techniques, described previously.^17,29,32

Physical Therapy

There is no standardized physiotherapy regimen for postoperative management of proximal humerus fractures treated with locking plates.²⁵ In older patients, immediate active range of motion (ROM) exercises should be delayed until early callus is noted, though there is a risk for stiffness. Lee and Shin⁵⁷ found that a delay in rehabilitation after ORIF was an independent risk for poor clinical outcome. Namdari and colleagues¹⁷ recommended sling use only for comfort and initiated non-load-bearing activities and pendulum exercises immediately after surgery. Patients with adequate reduction at 4 to 6 weeks were advanced to full weight-bearing. Badman and colleagues³⁰ initiated passive-assisted ROM exercises when the wound was healed at 2 weeks in 2-part fractures, whereas patients with 3- and 4-part fractures were immobilized until radiographic healing. Formal therapy was started after 6 weeks. Stiffness was reported in 5% of patients. For patients with stable fixation, Ricchetti and colleagues²⁹ recommended passive shoulder ROM exercises on postoperative day 1; at 4 to 6 weeks, patients should start active shoulder ROM exercises, and then resistance exercises at 10 to 12 weeks. Other authors are more conservative—only sling immobilization and pendulum exercises the first month.⁶⁶ Barlow and colleagues³² immobilized their patients (age, >75 years) for 6 weeks. No patient developed disabling stiffness. The authors suggested that patients older than 75 years may not be prone to stiffness.

Our Preferred Treatment Method

All proximal humerus fractures are approached anteriorly through the deltopectoral interval (Figure 3A). The long head biceps is identified and truncated for later tenodesis. Multiple No. 5 Ethibond sutures (Ethicon) are placed at the bone–tendon interface. The fracture is reduced with a Cobb elevator (Figure 3B), and provisional Kirschner wires are placed within the head (Figure 3C). The plate is affixed to the humeral head with its anterior border paralleling the posterior aspect of the bicipital groove. Multiple locking screws are placed within the superior and posterior humeral head. Nonlocking screws are then used to fix the plate to the shaft to reduce the specific deformity. Under fluoroscopy, any metaphyseal void is filled with calcium phosphate cement (Figure 3D). The remaining inferior screws are placed within the humeral head. Dr. Gruson uses screws 4 to 6 mm short of subchondral bone to reduce the risk for joint penetration. The rotator cuff sutures are tied down through the plate. Patients are started on progressive supine passive ROM exercises at 7 days, followed by supine active-assisted ROM exercises 6 weeks after fracture healing is confirmed radiographically.

Conclusion

Use of locked plating for proximal humerus fractures has increased, particularly in the elderly. Resulting complications include intra-articular screw penetration, postoperative fracture displacement, and AVN. Recognition of the importance of reducing and supporting the medial calcar, filling any metaphyseal defects, and selectively placing screws within the humeral head has lowered the incidence of these complications. Further comparative studies evaluating the efficacy of individual augmentation techniques are needed to determine their contribution to successful fracture healing and their cost-effectiveness. Results of such studies may help in the development of protocols for more standardized implementation of these techniques and in understanding which specific fracture patterns and patients would benefit from their use.

Proximal humerus fractures account for 4% to 5% of all fractures.¹ These fractures occur most frequently in the elderly—patients older than 60 years sustain 71% of these injuries²—and in females.^1,3 Given an aging population, this incidence is predicted to increase 3-fold over the next 30 years.⁴ There is much debate regarding management of acute, displaced proximal humerus fractures. A recent Cochrane Review of published outcomes of operative and nonoperative treatment of displaced proximal humerus fractures found insufficient evidence supporting either modality, though surgery was associated with additional procedures.⁵ A review of 1000 proximal humerus fractures found that 49% had less than 1 cm of displacement of the major fragments or angulation of less than 45°.³ Other authors have reported similar findings.^6,7 Although the incidence of proximal humerus fractures has remained stable over the past decade, from 1999 to 2005 there was a 25% relative increase in surgical management, including a relative increase of 29% in open reduction and internal fixation (ORIF) versus a 20% increase in arthroplasty.¹

Locking plates have consistently demonstrated biomechanical superiority over other forms of fixation in osteoporotic bone.^8-11 Egol and colleagues⁸ found that osteoporotic bone limited the torque of fixation to values less than what is required for adequate frictional force between the plate and the bone. This problem can be overcome with fixed-angle devices, such as locked plates.⁹ Compared with locked nail constructs, proximal humerus locking plates have demonstrated superiority in torsion, loading, and varus bending.^10,11 Compared with blade plates, proximal humerus locking plates exhibited increased stiffness and torsional fatigue resistance.¹² In a randomized clinical trial, Olerud and colleagues¹³ reported superior functional results with locking plate fixation compared with nonoperative treatment of displaced 3-part fractures in elderly patients with 2-year follow-up, though these clinical results were not supported by others.¹⁴ Two recent case–control studies comparing functional outcomes for 3- and 4-part fractures with follow-up of more than 2 years revealed higher Constant scores after locked plating compared with hemiarthroplasty, though complications were higher with locked plates.^15,16 Adoption of locked proximal humerus plating has been correlated with good clinical outcomes and union rates, though this has been accompanied by a higher rate of reoperation.⁷ Reoperation rates from 1999 to 2005 increased both in the immediate postoperative period (odds ratio, 3.36) and at 1 year (odds ratio, 3.90).¹

Complications of Locked Plating

Regardless of fixation type, reduced humeral head bone mass and quality may lead to implant loosening, fracture redisplacement, and, ultimately, poor outcomes. Baseline osteoporosis may predict likelihood of fixation failure.¹⁷ Multiple studies have reported on the implant-related complications associated with locking plate fixation—most commonly, intra-articular screw penetration, postoperative fracture displacement, and avascular necrosis (AVN)^18-24 (Figure 1). A meta-analysis of 12 studies with a total of 514 proximal humerus fractures treated with locking plate fixation showed an overall complication rate of 49% and a 13.8% reoperation rate.²⁵ The most common indication for reoperation involved intra-articular screw perforation. The most common complications were varus malunion (16%), osteonecrosis (10%), intra-articular screw penetration (8%), subacromial impingement (6%), and infection (4%).

Suboptimal intraoperative fracture reduction, specifically with residual varus, has been correlated with loss of fracture fixation. In a series of 153 fractures, loss of fixation occurred in 13.7% of cases, with the leading risk factor being varus malreduction.¹⁹ Failure rates were 30.4% and 11% when the head shaft angle was less than 120° and when it was 120° or more, respectively. Solberg and colleagues¹⁶ found that initial postoperative varus angulation of more than 20° resulted in universal loss of fixation. Conversion of these cases to hemiarthroplasty resulted in poor outcomes. Preoperative fracture alignment may also predict fixation failure.²² In one series, initial varus angulation healed with a mean 16° varus and a Constant score of 63, whereas initial valgus alignment healed with 6° varus and a Constant score of 71.²² Complications occurred in fractures that were initially in varus 79% of the time and initially in valgus 19% of the time. Screw perforation has been associated with loss of reduction 44% of the time.²⁰

In an analysis of locking plate constructs revised after early (<4 weeks) failure in 8 patients with osteoporosis, Micic and colleagues²¹ found implant pullout leading to varus malalignment. All cases lacked medial support and subchondral screw purchase; 3 were initially malreduced. Owsley and Gorczyca²³ retrospectively reviewed 53 cases of displaced proximal humerus fractures treated with locked plating. Despite the high rate of radiographic union, 36% developed complications, including screw cutout (23%), varus displacement of more than 10° (25%), and AVN (4%); 13% required revision. These complications disproportionately affected patients older than 60 years (57%) and negatively affected functional outcomes.

Augmentation Techniques

Despite its reported complications, proximal humerus locked plating remains the most widely used type of fixation.¹ Advancements in locking plate design, improved understanding of fixation principles, and adoption of techniques augmenting proximal humerus locking plate fixation, particularly in osteoporotic bone, have reduced postoperative complications (Table 1).

Rotator Cuff Sutures

A widely adopted technique for neutralizing rotator cuff–deforming forces, which theoretically can cause fracture displacement, is incorporation of heavy nonabsorbable sutures. These sutures are placed through the rotator cuff–tuberosity junction and tied down after being passed through the plate. Obtaining and maintaining tuberosity reduction are essential in achieving good functional outcomes after fixation. In addition, tension band sutures may be particularly useful in the setting of initial varus deformity.²⁶

Although clinical use of these sutures is common, biomechanical studies of their adjunctive contribution to fracture stability are lacking.²⁷ The rotator cuff musculature has a maximal contractile force of 3.5 kg/cm².²⁸ Ricchetti and colleagues²⁹ described a technique that involves using a locked plate and tagging the rotator cuff with heavy nonabsorbable sutures. Selective traction on the sutures can help obtain and maintain fracture reduction. Multiple studies have reported on suture use with locked plating for proximal humerus fractures.^29-34 Badman and colleagues³⁰ retrospectively reviewed 81 cases of metaphyseal defects or medial comminution treated with locked plating, rotator cuff sutures, and structural allograft. All cases healed within 6 months after surgery. The incidence of screw cutout was 3.7%, the incidence of AVN was 6.2%, and the incidence of varus collapse was 6%. A cadaveric study that used specimens (mean age, 77 years) with a simulated 3-part proximal humerus fracture treated with a locked plate both with and without cerclage sutures found no difference in interfragmentary motion between the groups.²⁷ The authors concluded that additive sutures are not required for anatomically reduced fractures. Multiple sutures may counteract the deforming forces that act on bony segments that cannot be adequately maintained with screws, such as an osteoporotic greater tuberosity.

Medial Column Restoration

The importance of reducing and maintaining the medial calcar to provide biomechanical support for a laterally placed plate has been recognized.^26,34-37 Gardner and colleagues²⁶ suggested that medial support was achieved if the medial cortex was anatomically reduced, if the proximal fragment was impacted laterally onto the shaft, or if 1 or more inferomedial screws were placed. Cases that did not achieve medial support developed significantly more humeral head subsidence (5.8 mm vs 1.2 mm) and screw penetration. Krappinger and colleagues³⁶ found that factors leading to fixation failure included age, local bone mineral density, anatomical reduction, and restoration of the medial cortical support. The authors concluded that anatomical reduction and restoration of the medial cortex were important in minimizing mechanical loads at the bone–implant interface. Biomechanically, Lescheid and colleagues³⁷ found that the most stable construct was anatomical reduction with medial cortical contact. In the setting of comminution, however, it may be preferable to intentionally perform varus malreduction to achieve medial contact than to achieve anatomical reduction with a fracture gap. Badman and colleagues³⁰ found that the incidence of screw penetration was 6% in patients with an intact medial calcar versus 29% in patients without medial support. In a retrospective analysis of patients treated with a locking plate and suture augmentation, Jung and colleagues³⁵ concluded that restoring medial support was the most reliable factor in the prevention of loss of reduction with or without screw perforation. Last, Solberg and colleagues¹⁶ reported better clinical outcomes when the length of the metaphyseal segment attached to the articular fragment was more than 2 mm. A length of less than 2 mm was predictive of developing AVN.

Use of Bone Void Fillers

Allograft. Allograft is cancellous or corticocancellous chips or tricortical graft used as osteoconductive filler for metaphyseal defects.³⁸ An increasingly popular technique involves using an endosteal fibular allograft strut to indirectly reduce the fracture and help support the medial calcar.^39-42 Hettrich and colleagues⁴⁰ reported on radiographic outcomes of displaced proximal humerus fractures with medial comminution treated with a locked plate and an endosteal fibular allograft or semitubular plate. The reduction was maintained in 96% of cases; there was 1 varus collapse. There were no cases of implant failure, screw perforation, or AVN. Other authors have also reported on successful use of fibular allograft in conjunction with a locked plate; the rate of reduction loss was low, and there were no cases of screw cutout or intra-articular screw penetration.^30,41,42 These clinical outcomes are supported by results of biomechanical studies of the added benefit of intramedullary fibular allograft.^43-46 Mathison and colleagues⁴³ reported that a construct with fibular allograft and a locking plate increased the failure load by 1.72 times and the stiffness by 3.84 times compared with a control group of locking plate only. Bae and colleagues⁴⁶ found significantly higher maximum failure load and construct stiffness with no varus collapse in specimens prepared with locked plate and fibular strut augmentation compared with a control group.

Others have successfully used cancellous allograft to fill humeral head bone defects.^29,32,47-49 Duralde and Leddy⁴⁷ reported 100% radiographic union and 81% good to excellent results in cases treated with a locking plate and morselized cancellous allograft to fill bone voids. Varus collapse and screw cutout did not occur, but there were 2 cases of AVN. Ricchetti and colleagues²⁹ reviewed 54 cases treated with a locking plate and rotator cuff suture construct. Allograft cancellous chips and demineralized bone matrix were used in 3- and 4-part fractures (70% of cases) along with shorter screws in the humeral head. Major complications included AVN (1), fixation failure (3), and varus malunion (5). Others investigators have had less favorable results with use of cancellous bone graft. Schliemann and colleagues¹⁹ reported on 27 patients who were older than 65 years when they underwent ORIF with rotator cuff sutures to stabilize the tuberosities and either cancellous graft or a synthetic bone substitute in patients with massive metaphyseal defects. Patient-reported outcomes were superior to Constant scores. Complications included screw penetration (22.2%), reduction loss (44.4%), implant failure (3.7%), and AVN (29.6%).

Autograft. Autograft has both osteoconductive and osteoinductive properties and has been successfully used for metaphyseal defects.^32,50 Kim and colleagues⁵⁰ reported on patients with 4-part proximal humerus fractures treated with a locking plate and autologous iliac graft. All cases achieved union and had good or excellent outcomes. There were no cases of AVN, varus collapse, or hardware-related complications.

Bone Cement. Calcium phosphate cement has osteoconductive properties and enhances screw purchase in cancellous bone (Figures 2A–2F). It can be injected or molded into bone voids to provide improved compressive strength. It is resorbed through cell-mediated processes resembling bone remodeling and does not disappear until new bone forms. (Calcium sulfate cement, on the other hand, resorbs through a chemical process independent of new bone formation.⁵¹) Egol and colleagues⁵² reviewed the cases of 92 patients (mean age, 61 years) with 2-, 3-, and 4-part proximal humerus fractures treated with locked plate fixation. Metaphyseal defects were treated with no augmentation, augmentation with cancellous chips, or augmentation with calcium phosphate cement. Adding calcium phosphate cement was associated with lower incidence of intra-articular screw penetration and humeral head settling. In a recent cadaveric biomechanical study using 2-part proximal humerus fractures with metaphyseal comminution, the group augmented with calcium phosphate cement had enhanced axial stiffness and load to failure with reduced screw penetration.⁵³ Other biomechanical studies have found increased screw pullout strength⁵⁴ and decreased interfragmentary motion when specimens were augmented with calcium phosphate cement.⁵⁵

Similar good clinical and radiographic outcomes have been observed with use of calcium sulfate cement.^56,57 Somasundaram and colleagues⁵⁶ reported good clinical outcomes in 82% of patients treated with locking plates and calcium sulfate cement used to fill metaphyseal voids. All fractures united without infection, fixation failure, subsequent malunion, tuberosity failure, or AVN. Lee and Shin⁵⁷ compared outcomes of 14 patients who received calcium sulfate augmentation with outcomes of patients who did not receive this augmentation. Overall, 89% of patients had good or excellent results. Calcium sulfate cement did not affect the reduction failure rate or clinical outcomes in cases in which medial cortical reduction was achieved. However, postoperative displacement caused by lack of medial support was associated with poor outcomes.

Screw Placement

Screws optimally should be placed in the posterior-medial-inferior aspect of the humeral head to provide medial support for the fracture and mechanical stability.⁵⁸ Cadaveric studies have shown the highest cancellous bone density in the proximal, posterior, and medial portions of the humeral head.^59-63 Similarly, in a cadaveric study, Liew and colleagues⁶¹ found greater screw purchase and higher pullout strength when the screw was placed in the center of the humeral head, within subchondral bone; fixation was poorest when the screw was placed in the anterosuperior region of the humeral head. Tingart and colleagues⁶² reported that humeral head trabecular density significantly affected pullout strength of cancellous screws. In addition, the most pullout strength was at the center of the head, and the least within the anterosuperior head. Trabecular density was higher in the inferior and posterior regions than in the superior and anterior regions.

Most locking plate designs allow screws to be placed at the level of the medial calcar—the goal being to provide medial column support (Table 2). Zhang and colleagues⁵⁸ treated 2-, 3-, and 4-part fractures with a locking plate and randomized them into receiving the plate with or without medial support screws. For 3- and 4-part fractures, the group with these screws had a significantly greater final neck-shaft angle and smaller angulation loss compared with the group without screws. No additional benefit was found for 2-part fractures. Erhardt and colleagues⁶³ simulated unstable proximal humerus fractures using cadavers and testing different fixation methods using a polyaxial locking plate. They found that 5 screws in the head fragment and an inferomedial support screw significantly reduced the risk of screw perforation. Other authors have concluded that placing 1 or more inferomedial screws is important in cases of medial comminution or medial column malreduction.²⁶ Interestingly, compared with use of a polyaxial implant, which allows for adjustment of screw direction, use of a monoaxial locking plate did not lead to a clinically different outcome or complication profile.⁶⁴

Techniques have been used to achieve subchondral purchase of locking screws while reducing iatrogenic articular perforation.⁶⁵ However, given the incidence of fracture settling and subsequent postoperative screw penetration, many authors currently recommend using shorter divergent screws combined with other augmentation techniques, described previously.^17,29,32

Physical Therapy

There is no standardized physiotherapy regimen for postoperative management of proximal humerus fractures treated with locking plates.²⁵ In older patients, immediate active range of motion (ROM) exercises should be delayed until early callus is noted, though there is a risk for stiffness. Lee and Shin⁵⁷ found that a delay in rehabilitation after ORIF was an independent risk for poor clinical outcome. Namdari and colleagues¹⁷ recommended sling use only for comfort and initiated non-load-bearing activities and pendulum exercises immediately after surgery. Patients with adequate reduction at 4 to 6 weeks were advanced to full weight-bearing. Badman and colleagues³⁰ initiated passive-assisted ROM exercises when the wound was healed at 2 weeks in 2-part fractures, whereas patients with 3- and 4-part fractures were immobilized until radiographic healing. Formal therapy was started after 6 weeks. Stiffness was reported in 5% of patients. For patients with stable fixation, Ricchetti and colleagues²⁹ recommended passive shoulder ROM exercises on postoperative day 1; at 4 to 6 weeks, patients should start active shoulder ROM exercises, and then resistance exercises at 10 to 12 weeks. Other authors are more conservative—only sling immobilization and pendulum exercises the first month.⁶⁶ Barlow and colleagues³² immobilized their patients (age, >75 years) for 6 weeks. No patient developed disabling stiffness. The authors suggested that patients older than 75 years may not be prone to stiffness.

Our Preferred Treatment Method

All proximal humerus fractures are approached anteriorly through the deltopectoral interval (Figure 3A). The long head biceps is identified and truncated for later tenodesis. Multiple No. 5 Ethibond sutures (Ethicon) are placed at the bone–tendon interface. The fracture is reduced with a Cobb elevator (Figure 3B), and provisional Kirschner wires are placed within the head (Figure 3C). The plate is affixed to the humeral head with its anterior border paralleling the posterior aspect of the bicipital groove. Multiple locking screws are placed within the superior and posterior humeral head. Nonlocking screws are then used to fix the plate to the shaft to reduce the specific deformity. Under fluoroscopy, any metaphyseal void is filled with calcium phosphate cement (Figure 3D). The remaining inferior screws are placed within the humeral head. Dr. Gruson uses screws 4 to 6 mm short of subchondral bone to reduce the risk for joint penetration. The rotator cuff sutures are tied down through the plate. Patients are started on progressive supine passive ROM exercises at 7 days, followed by supine active-assisted ROM exercises 6 weeks after fracture healing is confirmed radiographically.

Conclusion

Use of locked plating for proximal humerus fractures has increased, particularly in the elderly. Resulting complications include intra-articular screw penetration, postoperative fracture displacement, and AVN. Recognition of the importance of reducing and supporting the medial calcar, filling any metaphyseal defects, and selectively placing screws within the humeral head has lowered the incidence of these complications. Further comparative studies evaluating the efficacy of individual augmentation techniques are needed to determine their contribution to successful fracture healing and their cost-effectiveness. Results of such studies may help in the development of protocols for more standardized implementation of these techniques and in understanding which specific fracture patterns and patients would benefit from their use.

First-line treatments for osteoarthritis (OA) are targeted at the inflammatory reaction that occurs after breakdown of articular cartilage through regular use of nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroid injections, or surgical intervention. Associated activity restrictions and chronic pain have spurred a search for alternative treatments, commonly daily supplements such as glucosamine, chondroitin, and fish oil, to name a select few of the innumerable products reported to benefit patients with OA.

Background

Fish oil is 1 of the 2 most popular supplements among patients with OA. However, its effectiveness and precise benefit are still debated,^1,2 and there is confusion about the definition of the product, the nature of investigations into its effectiveness, and the standardization of research unique to OA. Most fish oil research relates to patients with rheumatoid arthritis (RA). The anti-inflammatory benefits seen in patients with RA are generally applied to characterize fish oils as anti-inflammatory agents with a logical benefit in reducing OA symptoms. However, there is a dearth of independent and focused clinical results justifying that assumption. Further, lack of federal regulation of the supplement industry hinders conducting generalizable studies regarding medical benefit in a regulated and verified dose and form.³

The benefits of fish oil in RA treatment are well supported and accepted. In patients with RA, daily fish oil supplementation has been shown to reduce use of other medications and improve pain scores reported by both physicians and patients.^4-10 The clinical efficacy of fish oil use in RA has been determined to be “reasonably strong,” with multiple studies confirming suppression of inflammatory cytokines in vitro and in vivo.^11,12 The mechanism by which the inflammatory processes are augmented by fish oil supplementation suggests potential benefit to patients with OA, though review articles as recent as 2011 have concluded that research in that capacity is not sufficient to warrant recommendation.^13,14

Most studies of OA-specific use of fish oils have been conducted in in vitro models. Treatment of bovine chondrocytes with omega-3 fatty acids causes reductions in inflammatory markers induced by interleukin 1, one of several proinflammatory cytokines that induce inflammation in OA at the gene and plasma levels, and these reductions have been reproduced.^15-17 Although a preventive benefit was found in a study of pig medial collateral ligament fibroblasts, findings of later studies have been inconsistent.¹⁸ It also appears that fish oils may alter lipid composition in membranes, favoring incorporation of anti-inflammatory precursor n-3 fatty acids over proinflammatory precursor n-6 fatty acids in these model systems.^19,20

Animal in vivo models have also been used to describe the effects of fish oil supplementation on OA. Assessment of dogs with OA before and after supplementation with the omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) revealed improvement in clinical signs observed by owners, improvement in weight-bearing measured by veterinary clinicians, and decreased use of NSAIDs.^21-24

Fish oil studies using osteoarthritic cartilage samples harvested during surgical procedures have demonstrated results consistent with other model systems described thus far. They have demonstrated a dose-dependent decrease in induced inflammatory destruction of tissue associated with fish oil supplementation. In addition, finding a lack of cellular toxicity, they have validated the safety of supplements.^25,26 Proposed but unproven mechanisms for the anti-inflammatory actions of EPA and DHA include competition with n-6 fatty acids; presence of resolvins (anti-inflammatory molecules derived from EPA and DHA); presence of n-3 products that compete with proinflammatory molecules for receptors; reduction in gene expression of cytokines, cyclo-oxygenase 2, and degrading proteinases; interference in the signaling pathways of inflammation; and reduction in lymphocyte proliferation.^26,27

Reduction in the n-6/n-3 ratio has been correlated with reduced inflammatory conditions such as OA, stemming from the epidemiologic evidence that higher n-3 intake in Eastern diets and lower intake of n-6 result in a lower incidence of these diseases.^18,28,29 Studies have found sufficient evidence to suggest that this ratio has a role in OA, though not sufficient to recommend supplement use over diet modification.¹⁹ One study demonstrated an ability to favorably alter bone marrow lipid composition with n-3 fatty acid supplementation.¹⁰

The evidence leads to a conclusion of anti-inflammatory benefits from fish oils in these abstracted models. The multitude of basic science studies conducted on the anti-inflammatory properties of omega-3 fatty acids, only briefly reviewed here, supports the potential benefits colloquially ascribed to fish oil in the treatment of OA yet also implies the need for human clinical trials to address these properties clinically.

We reviewed the literature to address claims that fish oil supplementation can prevent or decrease severity of OA. We hypothesized there would be insufficient clinical studies to justify recommending supplementation to patients. Of note, the degree of heterogeneity in the evidence precluded performing a meta-analysis with any statistical validity.

Literature Review

In the PubMed database, we targeted the subject of fish oils and OA by using search terms that included omega-3, DHA, EPA, and alpha-linolenic acid. The MedLine and Google Scholar databases were searched as well. Results were limited to those reported in English and involving human subjects and clinical trials; results were excluded if they primarily involved patients with RA. Studies cited or mentioned in articles found through the PubMed search were evaluated according to the criteria mentioned, such that all relevant articles available at time of search are thought to be included, and these articles represent a reasonable presentation of the available evidence.

Findings

Our search revealed 6 clinical trials in which omega-3–containing supplements were used in the treatment of human OA with differing endpoints. We reviewed these trials in detail. One study, which used alteration of bone marrow lipids as an endpoint, was included for completeness of the evaluation of the relevant evidence.²⁰ In addition, the study by Wang and colleagues,³⁰ who assessed patients without clinical evidence of OA for development of bone marrow lesions, was reviewed. This study was deemed relevant to examine the process by which n-3 fatty acids alter knee structure, as subsequent risk of OA has not been elucidated, and effects on bone marrow lesions may indeed have a direct impact on the OA process. Results of the trials that were identified were varied between no significant difference in OA symptoms between treatment and control groups, implied benefits, and substantial benefits.

The first clinical study of omega-3 supplementation in OA treatment was conducted in 1992.³¹ The study compared 10 g of cod liver oil (containing 786 mg of EPA) with 10 g of olive oil, both taken daily over 24 weeks by 86 patients with OA. Effects were assessed by NSAID use (recorded in patient diary) and pain score (evaluated by clinician) every 4 weeks. The trial found no significant difference in effects between the oils.

Wang and colleagues³⁰ used a food questionnaire to measure the n-3 intake of 293 healthy adults and quantified their bone marrow lesions after 10 years in an effort to describe how n-3 intake correlates with development of OA or pre-OA lesions. Higher intake of n-6 fatty acids was positively associated with presence of bone marrow lesions; n-3 intake had no association.

In a study of 84 patients who had joint replacement, Pritchett²⁰ evaluated lipid alterations resulting from a regimen of 3 g of fish oil containing 11% DHA daily for a 6-month trial period, measuring lipids before and after the trial period. Pritchett²⁰ found a 20% increase in long-chain fatty acids and a corresponding decrease in saturated fatty acids, as measured in bone marrow.

The supplement Phytalgic (Phythea Laboratories), which is advertised for OA, includes n-3 fatty acids, n-6 fatty acids, extract from Urtica dioica (the common nettle), zinc, and vitamin E. In a study by Jacquet and colleagues,³² this supplement was given 3 times daily over 3 separate 4-week periods to 81 patients with knee or hip OA. Measuring NSAID use with patient diaries and assessing pain with the WOMAC (Western Ontario and McMaster Universities) Osteoarthritis Index every 4 weeks for 12 weeks, the authors found a significant decrease in NSAID use and, according to WOMAC results, a more than 50% reduction in pain and stiffness, and improved function.

One study compared the effects of glucosamine with and without omega-3 fatty acids in 182 patients with knee or hip OA.³³ Each day, patients took 500 mg of glucosamine plus 3 capsules each containing either 444 mg of omega-3 fatty acids or 444 mg of an oil mixture. Pain was assessed with visual analog scale and the WOMAC scale 3 times over the 26-week study. More than 90% reductions in morning stiffness and pain were found for the combination of fish oil and glucosamine.

The Multicenter Osteoarthritis Study (MOST), published in February 2012, demonstrated that plasma levels of n-3 and n-6 polyunsaturated fatty acids (PUFAs) may be related to knee structural findings.³⁴ This study confirmed that dietary modification of n-3 and n-6 PUFAs altered plasma concentration predictably. Higher DHA intake was associated with less evidence of OA on patellofemoral cartilage, though no association was found on tibiofemoral cartilage.³⁴

Discussion

The lack of human clinical trials detailing the effects of fish oil supplementation in patients with OA is arguably the most significant hindrance to fish oil being routinely recommended. Since 1992, only 6 studies have addressed this topic, and their endpoints and results were inconsistent. These interventional trials had their limitations, including short duration, insufficient dosage, inappropriate n-3 choice, dietary interactions, genotype, and medication interactions.¹⁸ The present review is limited as well, by the quantity of evidence on the topic and by the focus (of the majority of the studies) on short-term alterations in pain and mobility instead of on disease-modifying potential. Short-term evaluation is unlikely to capture such an effect, which may require long-term supplementation to become evident.

The results of the study by Stammers and colleagues³¹ must be examined critically, as the likelihood of detection bias is high. Highly subjective assessments of effect, lack of standardized NSAID treatments, and limitations in patient numbers and disease severity raise concerns about validity. In addition, confounding variables (eg, medication interactions, alternative treatments, olive oil use) undermine the design. It is therefore difficult to interpret the results of this trial.

The study by Wang and colleagues³⁰ did not involve supplementation, and intake was assessed only with food frequency questionnaires. It is therefore difficult to apply their results or findings to this review. In addition, the authors did not obtain baseline magnetic resonance imaging for comparison with that obtained at study completion—that is, they did not address any subclinical disease before dietary recording.

Pritchett²⁰ acknowledged study limitations of small sample size and use of 1 subject as both patient and control. Although the study seemed to demonstrate that omega-3 supplementation augmented the lipid profile of joints, it did not directly demonstrate improvement in or prevention of OA. Identification of bone marrow lesions is not definitive proof of OA but an alteration that may correlate with development. The logical supposition is that altering the local environment may alter development of disease within that environment, though this is not proven.

An article reviewing the Phytalgic study highlighted the suspect nature of its results—claims that the supplement is 76% more effective than gold-standard corticosteroid injection.³⁵ Also highlighted were lack of confirmed mechanism, questionable control, detection bias caused by aftertaste, and the high attrition rate in the placebo group. It is difficult to apply these results to fish oil supplementation, as Phytalgic contains other potentially confounding substances.

Of note, the findings of MOST were observational; n-3 and n-6 levels were not altered or supplemented. Altered disease process was demonstrated in patellofemoral cartilage but not in tibiofemoral cartilage in the same patient. The inconsistencies may be explained by the observational nature of the study and the lack of supplementation that would have produced a more significant increase in n-3 PUFA levels and thus more uniform conclusions, if in fact n-3 PUFAs were the significant factor in the altered cartilage structure. Although supportive of a preventive or disease-altering benefit, the results do not speak to supplementation.

Perhaps the most convincing evidence supporting fish oil for OA comes from a 2009 study by Gruenwald and colleagues.³³ However, this 2-supplement study addressing synergy was financed by Seven Seas, a company with industry ties. The study was not placebo-controlled and was registered only after completion. The authors omitted baseline values, apparently did not correct for baseline in the statistical analysis, and did not report the distribution of results. The implication is that the results were overstated, or that, at minimum, the supporting data were not reported. Nevertheless, this study demonstrated benefits consistent with the animal and human laboratory studies. However, research is needed to repeat and validate these results, elucidate the mechanism of action, and quantify the benefit unique to fish oil.

Conclusion

Despite the overwhelming popularity of fish oil supplements and the assumption of benefit for patients with arthritis, there appears to be insufficient clinical evidence to justify use of fish oils in the treatment or prevention of OA. Possible efficacy in laboratory and animal studies has yet to be sufficiently observed and verified in clinical trials. Although it is impossible to refute the promise of these agents as beneficial adjuncts to anti-inflammatory regimens, there remains a need for significant, well-designed clinical trials to evaluate the efficacy, safety, and clinical parameters of omega-3 fatty acids in a standardized form before they can in good faith be recommended to patients with OA.

First-line treatments for osteoarthritis (OA) are targeted at the inflammatory reaction that occurs after breakdown of articular cartilage through regular use of nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroid injections, or surgical intervention. Associated activity restrictions and chronic pain have spurred a search for alternative treatments, commonly daily supplements such as glucosamine, chondroitin, and fish oil, to name a select few of the innumerable products reported to benefit patients with OA.

Background

Fish oil is 1 of the 2 most popular supplements among patients with OA. However, its effectiveness and precise benefit are still debated,^1,2 and there is confusion about the definition of the product, the nature of investigations into its effectiveness, and the standardization of research unique to OA. Most fish oil research relates to patients with rheumatoid arthritis (RA). The anti-inflammatory benefits seen in patients with RA are generally applied to characterize fish oils as anti-inflammatory agents with a logical benefit in reducing OA symptoms. However, there is a dearth of independent and focused clinical results justifying that assumption. Further, lack of federal regulation of the supplement industry hinders conducting generalizable studies regarding medical benefit in a regulated and verified dose and form.³

The benefits of fish oil in RA treatment are well supported and accepted. In patients with RA, daily fish oil supplementation has been shown to reduce use of other medications and improve pain scores reported by both physicians and patients.^4-10 The clinical efficacy of fish oil use in RA has been determined to be “reasonably strong,” with multiple studies confirming suppression of inflammatory cytokines in vitro and in vivo.^11,12 The mechanism by which the inflammatory processes are augmented by fish oil supplementation suggests potential benefit to patients with OA, though review articles as recent as 2011 have concluded that research in that capacity is not sufficient to warrant recommendation.^13,14

Most studies of OA-specific use of fish oils have been conducted in in vitro models. Treatment of bovine chondrocytes with omega-3 fatty acids causes reductions in inflammatory markers induced by interleukin 1, one of several proinflammatory cytokines that induce inflammation in OA at the gene and plasma levels, and these reductions have been reproduced.^15-17 Although a preventive benefit was found in a study of pig medial collateral ligament fibroblasts, findings of later studies have been inconsistent.¹⁸ It also appears that fish oils may alter lipid composition in membranes, favoring incorporation of anti-inflammatory precursor n-3 fatty acids over proinflammatory precursor n-6 fatty acids in these model systems.^19,20

Animal in vivo models have also been used to describe the effects of fish oil supplementation on OA. Assessment of dogs with OA before and after supplementation with the omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) revealed improvement in clinical signs observed by owners, improvement in weight-bearing measured by veterinary clinicians, and decreased use of NSAIDs.^21-24

Fish oil studies using osteoarthritic cartilage samples harvested during surgical procedures have demonstrated results consistent with other model systems described thus far. They have demonstrated a dose-dependent decrease in induced inflammatory destruction of tissue associated with fish oil supplementation. In addition, finding a lack of cellular toxicity, they have validated the safety of supplements.^25,26 Proposed but unproven mechanisms for the anti-inflammatory actions of EPA and DHA include competition with n-6 fatty acids; presence of resolvins (anti-inflammatory molecules derived from EPA and DHA); presence of n-3 products that compete with proinflammatory molecules for receptors; reduction in gene expression of cytokines, cyclo-oxygenase 2, and degrading proteinases; interference in the signaling pathways of inflammation; and reduction in lymphocyte proliferation.^26,27

Reduction in the n-6/n-3 ratio has been correlated with reduced inflammatory conditions such as OA, stemming from the epidemiologic evidence that higher n-3 intake in Eastern diets and lower intake of n-6 result in a lower incidence of these diseases.^18,28,29 Studies have found sufficient evidence to suggest that this ratio has a role in OA, though not sufficient to recommend supplement use over diet modification.¹⁹ One study demonstrated an ability to favorably alter bone marrow lipid composition with n-3 fatty acid supplementation.¹⁰

The evidence leads to a conclusion of anti-inflammatory benefits from fish oils in these abstracted models. The multitude of basic science studies conducted on the anti-inflammatory properties of omega-3 fatty acids, only briefly reviewed here, supports the potential benefits colloquially ascribed to fish oil in the treatment of OA yet also implies the need for human clinical trials to address these properties clinically.

We reviewed the literature to address claims that fish oil supplementation can prevent or decrease severity of OA. We hypothesized there would be insufficient clinical studies to justify recommending supplementation to patients. Of note, the degree of heterogeneity in the evidence precluded performing a meta-analysis with any statistical validity.

Literature Review

In the PubMed database, we targeted the subject of fish oils and OA by using search terms that included omega-3, DHA, EPA, and alpha-linolenic acid. The MedLine and Google Scholar databases were searched as well. Results were limited to those reported in English and involving human subjects and clinical trials; results were excluded if they primarily involved patients with RA. Studies cited or mentioned in articles found through the PubMed search were evaluated according to the criteria mentioned, such that all relevant articles available at time of search are thought to be included, and these articles represent a reasonable presentation of the available evidence.

Findings

Our search revealed 6 clinical trials in which omega-3–containing supplements were used in the treatment of human OA with differing endpoints. We reviewed these trials in detail. One study, which used alteration of bone marrow lipids as an endpoint, was included for completeness of the evaluation of the relevant evidence.²⁰ In addition, the study by Wang and colleagues,³⁰ who assessed patients without clinical evidence of OA for development of bone marrow lesions, was reviewed. This study was deemed relevant to examine the process by which n-3 fatty acids alter knee structure, as subsequent risk of OA has not been elucidated, and effects on bone marrow lesions may indeed have a direct impact on the OA process. Results of the trials that were identified were varied between no significant difference in OA symptoms between treatment and control groups, implied benefits, and substantial benefits.

The first clinical study of omega-3 supplementation in OA treatment was conducted in 1992.³¹ The study compared 10 g of cod liver oil (containing 786 mg of EPA) with 10 g of olive oil, both taken daily over 24 weeks by 86 patients with OA. Effects were assessed by NSAID use (recorded in patient diary) and pain score (evaluated by clinician) every 4 weeks. The trial found no significant difference in effects between the oils.

Wang and colleagues³⁰ used a food questionnaire to measure the n-3 intake of 293 healthy adults and quantified their bone marrow lesions after 10 years in an effort to describe how n-3 intake correlates with development of OA or pre-OA lesions. Higher intake of n-6 fatty acids was positively associated with presence of bone marrow lesions; n-3 intake had no association.

In a study of 84 patients who had joint replacement, Pritchett²⁰ evaluated lipid alterations resulting from a regimen of 3 g of fish oil containing 11% DHA daily for a 6-month trial period, measuring lipids before and after the trial period. Pritchett²⁰ found a 20% increase in long-chain fatty acids and a corresponding decrease in saturated fatty acids, as measured in bone marrow.

The supplement Phytalgic (Phythea Laboratories), which is advertised for OA, includes n-3 fatty acids, n-6 fatty acids, extract from Urtica dioica (the common nettle), zinc, and vitamin E. In a study by Jacquet and colleagues,³² this supplement was given 3 times daily over 3 separate 4-week periods to 81 patients with knee or hip OA. Measuring NSAID use with patient diaries and assessing pain with the WOMAC (Western Ontario and McMaster Universities) Osteoarthritis Index every 4 weeks for 12 weeks, the authors found a significant decrease in NSAID use and, according to WOMAC results, a more than 50% reduction in pain and stiffness, and improved function.

One study compared the effects of glucosamine with and without omega-3 fatty acids in 182 patients with knee or hip OA.³³ Each day, patients took 500 mg of glucosamine plus 3 capsules each containing either 444 mg of omega-3 fatty acids or 444 mg of an oil mixture. Pain was assessed with visual analog scale and the WOMAC scale 3 times over the 26-week study. More than 90% reductions in morning stiffness and pain were found for the combination of fish oil and glucosamine.

The Multicenter Osteoarthritis Study (MOST), published in February 2012, demonstrated that plasma levels of n-3 and n-6 polyunsaturated fatty acids (PUFAs) may be related to knee structural findings.³⁴ This study confirmed that dietary modification of n-3 and n-6 PUFAs altered plasma concentration predictably. Higher DHA intake was associated with less evidence of OA on patellofemoral cartilage, though no association was found on tibiofemoral cartilage.³⁴

Discussion

The lack of human clinical trials detailing the effects of fish oil supplementation in patients with OA is arguably the most significant hindrance to fish oil being routinely recommended. Since 1992, only 6 studies have addressed this topic, and their endpoints and results were inconsistent. These interventional trials had their limitations, including short duration, insufficient dosage, inappropriate n-3 choice, dietary interactions, genotype, and medication interactions.¹⁸ The present review is limited as well, by the quantity of evidence on the topic and by the focus (of the majority of the studies) on short-term alterations in pain and mobility instead of on disease-modifying potential. Short-term evaluation is unlikely to capture such an effect, which may require long-term supplementation to become evident.

The results of the study by Stammers and colleagues³¹ must be examined critically, as the likelihood of detection bias is high. Highly subjective assessments of effect, lack of standardized NSAID treatments, and limitations in patient numbers and disease severity raise concerns about validity. In addition, confounding variables (eg, medication interactions, alternative treatments, olive oil use) undermine the design. It is therefore difficult to interpret the results of this trial.

The study by Wang and colleagues³⁰ did not involve supplementation, and intake was assessed only with food frequency questionnaires. It is therefore difficult to apply their results or findings to this review. In addition, the authors did not obtain baseline magnetic resonance imaging for comparison with that obtained at study completion—that is, they did not address any subclinical disease before dietary recording.

Pritchett²⁰ acknowledged study limitations of small sample size and use of 1 subject as both patient and control. Although the study seemed to demonstrate that omega-3 supplementation augmented the lipid profile of joints, it did not directly demonstrate improvement in or prevention of OA. Identification of bone marrow lesions is not definitive proof of OA but an alteration that may correlate with development. The logical supposition is that altering the local environment may alter development of disease within that environment, though this is not proven.

An article reviewing the Phytalgic study highlighted the suspect nature of its results—claims that the supplement is 76% more effective than gold-standard corticosteroid injection.³⁵ Also highlighted were lack of confirmed mechanism, questionable control, detection bias caused by aftertaste, and the high attrition rate in the placebo group. It is difficult to apply these results to fish oil supplementation, as Phytalgic contains other potentially confounding substances.

Of note, the findings of MOST were observational; n-3 and n-6 levels were not altered or supplemented. Altered disease process was demonstrated in patellofemoral cartilage but not in tibiofemoral cartilage in the same patient. The inconsistencies may be explained by the observational nature of the study and the lack of supplementation that would have produced a more significant increase in n-3 PUFA levels and thus more uniform conclusions, if in fact n-3 PUFAs were the significant factor in the altered cartilage structure. Although supportive of a preventive or disease-altering benefit, the results do not speak to supplementation.

Perhaps the most convincing evidence supporting fish oil for OA comes from a 2009 study by Gruenwald and colleagues.³³ However, this 2-supplement study addressing synergy was financed by Seven Seas, a company with industry ties. The study was not placebo-controlled and was registered only after completion. The authors omitted baseline values, apparently did not correct for baseline in the statistical analysis, and did not report the distribution of results. The implication is that the results were overstated, or that, at minimum, the supporting data were not reported. Nevertheless, this study demonstrated benefits consistent with the animal and human laboratory studies. However, research is needed to repeat and validate these results, elucidate the mechanism of action, and quantify the benefit unique to fish oil.

Conclusion

Despite the overwhelming popularity of fish oil supplements and the assumption of benefit for patients with arthritis, there appears to be insufficient clinical evidence to justify use of fish oils in the treatment or prevention of OA. Possible efficacy in laboratory and animal studies has yet to be sufficiently observed and verified in clinical trials. Although it is impossible to refute the promise of these agents as beneficial adjuncts to anti-inflammatory regimens, there remains a need for significant, well-designed clinical trials to evaluate the efficacy, safety, and clinical parameters of omega-3 fatty acids in a standardized form before they can in good faith be recommended to patients with OA.

First-line treatments for osteoarthritis (OA) are targeted at the inflammatory reaction that occurs after breakdown of articular cartilage through regular use of nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroid injections, or surgical intervention. Associated activity restrictions and chronic pain have spurred a search for alternative treatments, commonly daily supplements such as glucosamine, chondroitin, and fish oil, to name a select few of the innumerable products reported to benefit patients with OA.

Background

Fish oil is 1 of the 2 most popular supplements among patients with OA. However, its effectiveness and precise benefit are still debated,^1,2 and there is confusion about the definition of the product, the nature of investigations into its effectiveness, and the standardization of research unique to OA. Most fish oil research relates to patients with rheumatoid arthritis (RA). The anti-inflammatory benefits seen in patients with RA are generally applied to characterize fish oils as anti-inflammatory agents with a logical benefit in reducing OA symptoms. However, there is a dearth of independent and focused clinical results justifying that assumption. Further, lack of federal regulation of the supplement industry hinders conducting generalizable studies regarding medical benefit in a regulated and verified dose and form.³

The benefits of fish oil in RA treatment are well supported and accepted. In patients with RA, daily fish oil supplementation has been shown to reduce use of other medications and improve pain scores reported by both physicians and patients.^4-10 The clinical efficacy of fish oil use in RA has been determined to be “reasonably strong,” with multiple studies confirming suppression of inflammatory cytokines in vitro and in vivo.^11,12 The mechanism by which the inflammatory processes are augmented by fish oil supplementation suggests potential benefit to patients with OA, though review articles as recent as 2011 have concluded that research in that capacity is not sufficient to warrant recommendation.^13,14

Most studies of OA-specific use of fish oils have been conducted in in vitro models. Treatment of bovine chondrocytes with omega-3 fatty acids causes reductions in inflammatory markers induced by interleukin 1, one of several proinflammatory cytokines that induce inflammation in OA at the gene and plasma levels, and these reductions have been reproduced.^15-17 Although a preventive benefit was found in a study of pig medial collateral ligament fibroblasts, findings of later studies have been inconsistent.¹⁸ It also appears that fish oils may alter lipid composition in membranes, favoring incorporation of anti-inflammatory precursor n-3 fatty acids over proinflammatory precursor n-6 fatty acids in these model systems.^19,20

Animal in vivo models have also been used to describe the effects of fish oil supplementation on OA. Assessment of dogs with OA before and after supplementation with the omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) revealed improvement in clinical signs observed by owners, improvement in weight-bearing measured by veterinary clinicians, and decreased use of NSAIDs.^21-24

Fish oil studies using osteoarthritic cartilage samples harvested during surgical procedures have demonstrated results consistent with other model systems described thus far. They have demonstrated a dose-dependent decrease in induced inflammatory destruction of tissue associated with fish oil supplementation. In addition, finding a lack of cellular toxicity, they have validated the safety of supplements.^25,26 Proposed but unproven mechanisms for the anti-inflammatory actions of EPA and DHA include competition with n-6 fatty acids; presence of resolvins (anti-inflammatory molecules derived from EPA and DHA); presence of n-3 products that compete with proinflammatory molecules for receptors; reduction in gene expression of cytokines, cyclo-oxygenase 2, and degrading proteinases; interference in the signaling pathways of inflammation; and reduction in lymphocyte proliferation.^26,27

Reduction in the n-6/n-3 ratio has been correlated with reduced inflammatory conditions such as OA, stemming from the epidemiologic evidence that higher n-3 intake in Eastern diets and lower intake of n-6 result in a lower incidence of these diseases.^18,28,29 Studies have found sufficient evidence to suggest that this ratio has a role in OA, though not sufficient to recommend supplement use over diet modification.¹⁹ One study demonstrated an ability to favorably alter bone marrow lipid composition with n-3 fatty acid supplementation.¹⁰

The evidence leads to a conclusion of anti-inflammatory benefits from fish oils in these abstracted models. The multitude of basic science studies conducted on the anti-inflammatory properties of omega-3 fatty acids, only briefly reviewed here, supports the potential benefits colloquially ascribed to fish oil in the treatment of OA yet also implies the need for human clinical trials to address these properties clinically.

We reviewed the literature to address claims that fish oil supplementation can prevent or decrease severity of OA. We hypothesized there would be insufficient clinical studies to justify recommending supplementation to patients. Of note, the degree of heterogeneity in the evidence precluded performing a meta-analysis with any statistical validity.

Literature Review

In the PubMed database, we targeted the subject of fish oils and OA by using search terms that included omega-3, DHA, EPA, and alpha-linolenic acid. The MedLine and Google Scholar databases were searched as well. Results were limited to those reported in English and involving human subjects and clinical trials; results were excluded if they primarily involved patients with RA. Studies cited or mentioned in articles found through the PubMed search were evaluated according to the criteria mentioned, such that all relevant articles available at time of search are thought to be included, and these articles represent a reasonable presentation of the available evidence.

Findings

Our search revealed 6 clinical trials in which omega-3–containing supplements were used in the treatment of human OA with differing endpoints. We reviewed these trials in detail. One study, which used alteration of bone marrow lipids as an endpoint, was included for completeness of the evaluation of the relevant evidence.²⁰ In addition, the study by Wang and colleagues,³⁰ who assessed patients without clinical evidence of OA for development of bone marrow lesions, was reviewed. This study was deemed relevant to examine the process by which n-3 fatty acids alter knee structure, as subsequent risk of OA has not been elucidated, and effects on bone marrow lesions may indeed have a direct impact on the OA process. Results of the trials that were identified were varied between no significant difference in OA symptoms between treatment and control groups, implied benefits, and substantial benefits.

The first clinical study of omega-3 supplementation in OA treatment was conducted in 1992.³¹ The study compared 10 g of cod liver oil (containing 786 mg of EPA) with 10 g of olive oil, both taken daily over 24 weeks by 86 patients with OA. Effects were assessed by NSAID use (recorded in patient diary) and pain score (evaluated by clinician) every 4 weeks. The trial found no significant difference in effects between the oils.

Wang and colleagues³⁰ used a food questionnaire to measure the n-3 intake of 293 healthy adults and quantified their bone marrow lesions after 10 years in an effort to describe how n-3 intake correlates with development of OA or pre-OA lesions. Higher intake of n-6 fatty acids was positively associated with presence of bone marrow lesions; n-3 intake had no association.

In a study of 84 patients who had joint replacement, Pritchett²⁰ evaluated lipid alterations resulting from a regimen of 3 g of fish oil containing 11% DHA daily for a 6-month trial period, measuring lipids before and after the trial period. Pritchett²⁰ found a 20% increase in long-chain fatty acids and a corresponding decrease in saturated fatty acids, as measured in bone marrow.

The supplement Phytalgic (Phythea Laboratories), which is advertised for OA, includes n-3 fatty acids, n-6 fatty acids, extract from Urtica dioica (the common nettle), zinc, and vitamin E. In a study by Jacquet and colleagues,³² this supplement was given 3 times daily over 3 separate 4-week periods to 81 patients with knee or hip OA. Measuring NSAID use with patient diaries and assessing pain with the WOMAC (Western Ontario and McMaster Universities) Osteoarthritis Index every 4 weeks for 12 weeks, the authors found a significant decrease in NSAID use and, according to WOMAC results, a more than 50% reduction in pain and stiffness, and improved function.

One study compared the effects of glucosamine with and without omega-3 fatty acids in 182 patients with knee or hip OA.³³ Each day, patients took 500 mg of glucosamine plus 3 capsules each containing either 444 mg of omega-3 fatty acids or 444 mg of an oil mixture. Pain was assessed with visual analog scale and the WOMAC scale 3 times over the 26-week study. More than 90% reductions in morning stiffness and pain were found for the combination of fish oil and glucosamine.

The Multicenter Osteoarthritis Study (MOST), published in February 2012, demonstrated that plasma levels of n-3 and n-6 polyunsaturated fatty acids (PUFAs) may be related to knee structural findings.³⁴ This study confirmed that dietary modification of n-3 and n-6 PUFAs altered plasma concentration predictably. Higher DHA intake was associated with less evidence of OA on patellofemoral cartilage, though no association was found on tibiofemoral cartilage.³⁴

Discussion

The lack of human clinical trials detailing the effects of fish oil supplementation in patients with OA is arguably the most significant hindrance to fish oil being routinely recommended. Since 1992, only 6 studies have addressed this topic, and their endpoints and results were inconsistent. These interventional trials had their limitations, including short duration, insufficient dosage, inappropriate n-3 choice, dietary interactions, genotype, and medication interactions.¹⁸ The present review is limited as well, by the quantity of evidence on the topic and by the focus (of the majority of the studies) on short-term alterations in pain and mobility instead of on disease-modifying potential. Short-term evaluation is unlikely to capture such an effect, which may require long-term supplementation to become evident.

The results of the study by Stammers and colleagues³¹ must be examined critically, as the likelihood of detection bias is high. Highly subjective assessments of effect, lack of standardized NSAID treatments, and limitations in patient numbers and disease severity raise concerns about validity. In addition, confounding variables (eg, medication interactions, alternative treatments, olive oil use) undermine the design. It is therefore difficult to interpret the results of this trial.

The study by Wang and colleagues³⁰ did not involve supplementation, and intake was assessed only with food frequency questionnaires. It is therefore difficult to apply their results or findings to this review. In addition, the authors did not obtain baseline magnetic resonance imaging for comparison with that obtained at study completion—that is, they did not address any subclinical disease before dietary recording.

Pritchett²⁰ acknowledged study limitations of small sample size and use of 1 subject as both patient and control. Although the study seemed to demonstrate that omega-3 supplementation augmented the lipid profile of joints, it did not directly demonstrate improvement in or prevention of OA. Identification of bone marrow lesions is not definitive proof of OA but an alteration that may correlate with development. The logical supposition is that altering the local environment may alter development of disease within that environment, though this is not proven.

An article reviewing the Phytalgic study highlighted the suspect nature of its results—claims that the supplement is 76% more effective than gold-standard corticosteroid injection.³⁵ Also highlighted were lack of confirmed mechanism, questionable control, detection bias caused by aftertaste, and the high attrition rate in the placebo group. It is difficult to apply these results to fish oil supplementation, as Phytalgic contains other potentially confounding substances.

Of note, the findings of MOST were observational; n-3 and n-6 levels were not altered or supplemented. Altered disease process was demonstrated in patellofemoral cartilage but not in tibiofemoral cartilage in the same patient. The inconsistencies may be explained by the observational nature of the study and the lack of supplementation that would have produced a more significant increase in n-3 PUFA levels and thus more uniform conclusions, if in fact n-3 PUFAs were the significant factor in the altered cartilage structure. Although supportive of a preventive or disease-altering benefit, the results do not speak to supplementation.

Perhaps the most convincing evidence supporting fish oil for OA comes from a 2009 study by Gruenwald and colleagues.³³ However, this 2-supplement study addressing synergy was financed by Seven Seas, a company with industry ties. The study was not placebo-controlled and was registered only after completion. The authors omitted baseline values, apparently did not correct for baseline in the statistical analysis, and did not report the distribution of results. The implication is that the results were overstated, or that, at minimum, the supporting data were not reported. Nevertheless, this study demonstrated benefits consistent with the animal and human laboratory studies. However, research is needed to repeat and validate these results, elucidate the mechanism of action, and quantify the benefit unique to fish oil.

Conclusion

Despite the overwhelming popularity of fish oil supplements and the assumption of benefit for patients with arthritis, there appears to be insufficient clinical evidence to justify use of fish oils in the treatment or prevention of OA. Possible efficacy in laboratory and animal studies has yet to be sufficiently observed and verified in clinical trials. Although it is impossible to refute the promise of these agents as beneficial adjuncts to anti-inflammatory regimens, there remains a need for significant, well-designed clinical trials to evaluate the efficacy, safety, and clinical parameters of omega-3 fatty acids in a standardized form before they can in good faith be recommended to patients with OA.