Applied Evidence

Do you routinely check with new first-time mothers at a postpartum visit about changes in anal continence? They are at particular risk for obstetric anal sphincter injury and could be too embarrassed to raise the issue.

Sphincter injury following labor is the most common cause of anal incontinence (including flatus) in women, which can severely diminish quality of life and lead to considerable personal and financial costs.¹ Endoanal ultrasound can detect these injuries, even in the absence of clinically obvious damage to the anal sphincter (occult obstetric anal sphincter injury).²

In this article, we review measures to reduce the occurrence of obstetric anal sphincter injury, proper primary repair when it does occur, and appropriate long-term follow-up. Women with known obstetric anal sphincter injury must also be counseled about the risk of further damage during a future vaginal delivery.

Injury more common than symptoms would suggest

The conventional definitions of the 4 grades of perineal laceration in the US have been supplemented by more recent modifications included in a recent British Royal College of Obstetricians and Gynaecologists (RCOG) guideline (TABLE 1).³ The definition of third-degree laceration now reflects the various degrees of anal sphincter injury that may occur: partial (3a), full-thickness (3b), external anal sphincter injury, with or without injury to the internal anal sphincter (3c).

The incidence of clinical third- and fourth-degree lacerations varies widely; it is reported at between 0.5% and 3.0% in Europe and between 5.85% and 8.9% in the US.^2,4-6 A landmark British paper from 1993 revealed that though only 3% had a clinical third- or fourth-degree perineal laceration, 35% of primiparous women (none of whom had any defect before delivery) had ultrasound evidence of varying degrees of anal sphincter defect at 6 weeks postpartum that persisted at 6 months.² However, only about a third of these women had symptoms of bowel disturbance during the time of study.

These findings are supported by a meta-analysis in which 70% of women with a documented obstetric anal sphincter injury were asymptomatic.⁷ This meta-analysis concluded that clinical or occult obstetric anal sphincter injury occurs in 27% of primigravid women, and in 8.5% of multiparous women.

The long-term significance of occult obstetric anal sphincter injury and any relationship with geriatric fecal incontinence is unknown, although 71% of a sample of women with late-onset fecal incontinence were found to have ultrasound evidence of an anal sphincter defect thought to have occurred at a previous vaginal delivery.⁸ A recent English study⁹ reveals that when women were carefully re-examined after delivery by a skilled obstetrician looking specifically at the anal sphincter, the prevalence of clinically diagnosed third-degree lacerations rose sharply from the 11% initially diagnosed by the delivering physician or midwife to 24.5%. A subsequent endoanal ultrasound detected only an additional 1.2% (3 injuries, 2 of which were in the internal anal sphincter and therefore clinically undetectable). This strongly suggests that the vast majority of obstetric anal sphincter injuries can be detected clinically by a careful exam and that, when this is done, true occult injuries will be a rare finding.

TABLE 1
Classification of perineal injury⁹

Mechanisms of injury

Maintenance of fecal continence involves the coordinated action of several anatomical and physiological elements (FIGURE 1).¹⁰ An intact, innervated anal sphincter complex (both external and internal) is necessary. The sphincter complex can be damaged during childbirth in 3 ways.

Direct mechanical injury. Direct external or internal anal sphincter muscle disruption can occur, as with a clinically obvious third- or fourth-degree perineal laceration or an occult injury subsequently noted on ultrasound.

Neurologic injury. Neuropathy of the pudendal nerve may result from forceps delivery or persistent nerve compression from the fetal head.¹⁴ Traction neuropathy may also occur with fetal macrosomia and with prolonged pushing during Stage 2 in successive pregnancies, or with prolonged stretching of the nerve due to persistent poor postpartum pelvic floor tone. Injured nerves often undergo demyelination but usually recover with time.

Combined mechanical and neurologic trauma. Isolated neurologic injury, as described above, is believed to be rare. Neuropathy more commonly accompanies mechanical damage.¹⁵

Who is at risk?

Several risk factors are unavoidable. One of these is primiparity, a consistently reported independent variable also associated with other risk factors for obstetric anal sphincter injury, such as instrument delivery (TABLE 2).

TABLE 2
Major risk factors for obstetric anal sphincter injury

Preventing obstetric anal sphincter injury

Sphincter injury can occur even when obstetrical management is optimal. Although evidence from RCT data is often lacking, sufficient observational and retrospective data support the following recommendations to reduce the likelihood of injury.

Choose vacuum delivery before forceps

Any form of instrument delivery increases the risk of obstetric anal sphincter injury and altered fecal continence by between 2- and 7-fold.^2,16,24 An RCT found clinical third-degree tears in 16% of women with forceps-assisted deliveries, compared with 7% of vacuum-assisted deliveries; the authors concluded that, when circumstances allow, vacuum delivery should be attempted first (acknowledging however that 23% of vacuum deliveries failed and proceeded to a forceps extraction, a sequence associated with increased injury).¹⁷ A meta-analysis confirmed that vacuum extraction is preferred when instrumental delivery is necessary (SOR: A).²⁵

When midline episiotomy was performed during instrument delivery, the risk of obstetric anal sphincter injury approximately doubled again, such that, in one study, forceps delivery with episiotomy caused a 25-fold increase in obstetric anal sphincter injury.²⁴

Any steps that may safely reduce the need for instrument delivery should be supported. Toward this end, the Canadian Clinical Practices Obstetrics committee has recommended evidence-based labor interventions such as one-to-one support in labor, the increased use of a partogram in labor and appropriate oxytocin use, all in an effort to reduce needs for operative interventions.²⁶

If episiotomy necessary, mediolateral less risky than midline

Episiotomy was long promoted as a means of preserving the integrity of the perineal musculature and of avoiding damage to the anal sphincter, and it has been practiced routinely by some.²⁷ Strong evidence now indicates that routine episiotomy (midline or mediolateral) is unhelpful and should be abandoned.^25,27-29

Observational evidence overwhelmingly shows that midline episiotomy is strongly associated with obstetric anal sphincter injury.^{19,22,23,30,31} One of the few RCTs comparing midline with mediolateral episiotomy, although flawed in its design, noted that a clinical third-degree laceration occurred as an extension of episiotomy in 11.6% of midline incisions compared with just 2% of mediolateral cuts.³²

Another RCT, designed to examine routine versus restrictive episiotomy, noted that all but 1 (98%) of the 47 third- or fourth-degree lacerations in a group of 700 women followed midline episiotomy.²⁹ A retrospective database analysis noted a 6-fold higher risk of third-degree perineal lacerations for women undergoing midline episiotomy compared with mediolateral incision.²³ Elsewhere, midline episiotomy was associated with a 5-fold increase in symptoms of fecal incontinence at 3 months postpartum when compared with women with an intact perineum.²⁴

Even when midline episiotomies do not extend into clinical third-degree lacerations, the incidence of resultant postpartum fecal incontinence triples when compared with spontaneous second-degree perineal lacerations.³⁰ The authors postulate that a perineum cut by midline episiotomy allows for more direct contact to occur between the fetal hard parts and the anal sphincter complex during delivery, thereby increasing occult obstetric anal sphincter injury.

Observational data conflict as to whether mediolateral episiotomy contributes to, or protects against, obstetric anal sphincter injury—although the burden of evidence favors it as a risk factor that should be avoided when possible.^16,23,33 An angle of mediolateral incision cut closer to 45 degrees from the midline has been associated with less obstetric anal sphincter injury than incisions cut at closer angles to the midline.³⁴

Repairing sphincter injury

Detecting injury in labor

With any severe perineal laceration, closely inspect the external and, if exposed, internal anal sphincter and perform a rectal exam, particularly for women with numerous risk factors (although no good evidence supports the role of the rectal exam in diagnosing obstetric anal sphincter injury). Colorectal surgeons have advocated the use of a muscle stimulator to assist in identifying the ends of the external sphincter, but this has not become common practice.³⁵

Immediate vs delayed repair

It is standard practice to repair a damaged anal sphincter immediately or soon after delivery. However, given that a repair should be well done, and since a short delay does not appear to adversely affect healing, be prepared to wait for assistance for up to 24 hours rather than risk a suboptimal repair.³⁶

Better training is needed

Even among trained obstetricians and obgyn residents, 64% have reported no training or unsatisfactory training in management of obstetric anal sphincter injury; 94% of physicians felt inadequately prepared at the time of their first independent repair of the anal sphincter.^37,38 To improve your repair skills in a workshop setting, consult the following sources—www.aafp.org/also.xml in the US, or www.perineum.net in UK.

Analgesia and setting

Adequate analgesia is an essential element in a good repair. Complete relaxation of the anesthetized anal sphincter complex facilitates bringing torn ends of the sphincter together without tension.³⁹ Though theoretically this can be attained with local anesthetic infiltration, RCOG recommends that regional or general anesthesia be considered to provide complete analgesia.³⁷ It is further recommended that repair of the anal sphincter occur in an operating room, given the degree of contamination present in the labor room after delivery and the devastating effects of an infected repair (SOR: C).⁴⁰

Repair technique

There are 2 commonly used methods of external anal sphincter repair: one, the traditionally taught end-to-end approximation of the cut ends, and the other, overlapping the cut ends of the external sphincter and suturing through the overlapped portions (FIGURE 2).³⁶ Though an RCT from 2000 noted no significant difference in outcomes between these methods,⁴¹ other authors have suggested that an overlapping technique is preferred, and it remains the method most often used by colorectal surgeons in elective, secondary anal sphincter repairs.^36,39,42

A Cochrane review of which technique is better has been registered in the Clinical Trials Database. General agreement is that closure using interrupted sutures of a monofilament material, such as 2-0 polydioxanone sulfate (PDS), is the preferred closure method for the external sphincter (SOR: C).^36,40 It is recommended that a damaged internal sphincter be repaired with a running continuous suture of a material such as 2-0 polyglactin 910 (Vicryl) (SOR: C).³⁶

FIGURE 2
2 methods of anal sphincter repair

Immediate post-repair management

Use a stool softener

It had long been thought that constipation following obstetric anal sphincter injury allowed the sphincter to heal more effectively. However, new evidence from RCTs shows that using a laxative instead of a constipating regimen is more helpful in the immediate postpartum phase.⁴³ Toward this end, use a stool softener, such as lactulose, for 3 to 10 days postpartum for women with obstetric anal sphincter injury.⁴⁰

Should you prescribe an antibiotic?

Given the devastating effects of post-repair infection, most authorities consider it prudent to prescribe a course of broad-spectrum antibiotics, possibly including metronidazole (SOR: C)^37,40 A Cochrane review is registered to further examine this issue. A separate Cochrane review of the use of antibiotics for instrument vaginal delivery concluded that quality data were insufficient to make any recommendations.⁴⁴

Refer for physical therapy

Some authorities consider an early referral to physical therapy for pelvic floor exercises helpful in the immediate post-partum for all patients with obstetric anal sphincter injury (SOR: C).⁴⁵

Long-term management

Ask patients about incontinence

Given that some women are too embarrassed to seek assistance, ask those with obstetric anal sphincter injury specific questions about any symptoms of anal incontinence at a follow-up visit, such as the 6-week postpartum visit (SOR: C).^37,40 In some practices, all women who have sustained a third- or fourth-degree laceration are routinely scheduled for a 3-month follow-up visit to a dedicated clinic, irrespective of symptoms. Given the prevalence of occult obstetric anal sphincter injury for primigravid women, you may find it best to survey all women postnatally concerning any changes in anal continence. TABLE 3 demonstrates a validated, modified patient survey of anal incontinence.^37,40 A score of 6 is often used as a cutoff.

TABLE 3
Fecal Continence Scoring scale

When additional evaluation is needed

Patients who have symptoms of altered continence at 3 months (or who score above 6 on the Wexner scale) should be seen at a dedicated gynecologic or colorectal surgery clinic,⁴⁶ where they can receive a more detailed clinical evaluation and undergo anal manometry (during resting and forced squeezing) or endoanal ultrasonography. Some patients respond well to physical therapy, though a few patients ultimately require reconstructive colorectal surgery and temporary colostomy.

Management in a subsequent pregnancy

Women who have had an obstetric anal sphincter injury are at increased risk for repeat injury in a future pregnancy.⁴⁸ At some units, all such women are routinely offered a prenatal visit at the end of the second trimester to review their symptoms and to evaluate the anal sphincter with manometry or ultrasound. A large prospective study, however, found that recurrence of obstetric anal sphincter injury could not be predicted and that 95% of women with prior injury did not sustain further overt sphincter damage during a subsequent vaginal delivery.⁴⁹

However, for some women, a repeat anal sphincter laceration could prove devastating. For these women—eg, those with previous severe symptoms that required secondary surgical repair—initiate an in-depth discussion concerning the risks and benefits of elective cesarean delivery versus vaginal delivery.^37,40

CORRESPONDENCE
David Power, MD, MPH, Department of Family Medicine and Community Health, University of Minnesota, Mayo Mail Code 381, 516 Delaware St SE, Minneapolis, MN 55455. E-mail: [email protected]

Do you routinely check with new first-time mothers at a postpartum visit about changes in anal continence? They are at particular risk for obstetric anal sphincter injury and could be too embarrassed to raise the issue.

Sphincter injury following labor is the most common cause of anal incontinence (including flatus) in women, which can severely diminish quality of life and lead to considerable personal and financial costs.¹ Endoanal ultrasound can detect these injuries, even in the absence of clinically obvious damage to the anal sphincter (occult obstetric anal sphincter injury).²

In this article, we review measures to reduce the occurrence of obstetric anal sphincter injury, proper primary repair when it does occur, and appropriate long-term follow-up. Women with known obstetric anal sphincter injury must also be counseled about the risk of further damage during a future vaginal delivery.

Injury more common than symptoms would suggest

The conventional definitions of the 4 grades of perineal laceration in the US have been supplemented by more recent modifications included in a recent British Royal College of Obstetricians and Gynaecologists (RCOG) guideline (TABLE 1).³ The definition of third-degree laceration now reflects the various degrees of anal sphincter injury that may occur: partial (3a), full-thickness (3b), external anal sphincter injury, with or without injury to the internal anal sphincter (3c).

The incidence of clinical third- and fourth-degree lacerations varies widely; it is reported at between 0.5% and 3.0% in Europe and between 5.85% and 8.9% in the US.^2,4-6 A landmark British paper from 1993 revealed that though only 3% had a clinical third- or fourth-degree perineal laceration, 35% of primiparous women (none of whom had any defect before delivery) had ultrasound evidence of varying degrees of anal sphincter defect at 6 weeks postpartum that persisted at 6 months.² However, only about a third of these women had symptoms of bowel disturbance during the time of study.

These findings are supported by a meta-analysis in which 70% of women with a documented obstetric anal sphincter injury were asymptomatic.⁷ This meta-analysis concluded that clinical or occult obstetric anal sphincter injury occurs in 27% of primigravid women, and in 8.5% of multiparous women.

The long-term significance of occult obstetric anal sphincter injury and any relationship with geriatric fecal incontinence is unknown, although 71% of a sample of women with late-onset fecal incontinence were found to have ultrasound evidence of an anal sphincter defect thought to have occurred at a previous vaginal delivery.⁸ A recent English study⁹ reveals that when women were carefully re-examined after delivery by a skilled obstetrician looking specifically at the anal sphincter, the prevalence of clinically diagnosed third-degree lacerations rose sharply from the 11% initially diagnosed by the delivering physician or midwife to 24.5%. A subsequent endoanal ultrasound detected only an additional 1.2% (3 injuries, 2 of which were in the internal anal sphincter and therefore clinically undetectable). This strongly suggests that the vast majority of obstetric anal sphincter injuries can be detected clinically by a careful exam and that, when this is done, true occult injuries will be a rare finding.

TABLE 1
Classification of perineal injury⁹

Mechanisms of injury

Maintenance of fecal continence involves the coordinated action of several anatomical and physiological elements (FIGURE 1).¹⁰ An intact, innervated anal sphincter complex (both external and internal) is necessary. The sphincter complex can be damaged during childbirth in 3 ways.

Direct mechanical injury. Direct external or internal anal sphincter muscle disruption can occur, as with a clinically obvious third- or fourth-degree perineal laceration or an occult injury subsequently noted on ultrasound.

Neurologic injury. Neuropathy of the pudendal nerve may result from forceps delivery or persistent nerve compression from the fetal head.¹⁴ Traction neuropathy may also occur with fetal macrosomia and with prolonged pushing during Stage 2 in successive pregnancies, or with prolonged stretching of the nerve due to persistent poor postpartum pelvic floor tone. Injured nerves often undergo demyelination but usually recover with time.

Combined mechanical and neurologic trauma. Isolated neurologic injury, as described above, is believed to be rare. Neuropathy more commonly accompanies mechanical damage.¹⁵

Who is at risk?

Several risk factors are unavoidable. One of these is primiparity, a consistently reported independent variable also associated with other risk factors for obstetric anal sphincter injury, such as instrument delivery (TABLE 2).

TABLE 2
Major risk factors for obstetric anal sphincter injury

Preventing obstetric anal sphincter injury

Sphincter injury can occur even when obstetrical management is optimal. Although evidence from RCT data is often lacking, sufficient observational and retrospective data support the following recommendations to reduce the likelihood of injury.

Choose vacuum delivery before forceps

Any form of instrument delivery increases the risk of obstetric anal sphincter injury and altered fecal continence by between 2- and 7-fold.^2,16,24 An RCT found clinical third-degree tears in 16% of women with forceps-assisted deliveries, compared with 7% of vacuum-assisted deliveries; the authors concluded that, when circumstances allow, vacuum delivery should be attempted first (acknowledging however that 23% of vacuum deliveries failed and proceeded to a forceps extraction, a sequence associated with increased injury).¹⁷ A meta-analysis confirmed that vacuum extraction is preferred when instrumental delivery is necessary (SOR: A).²⁵

When midline episiotomy was performed during instrument delivery, the risk of obstetric anal sphincter injury approximately doubled again, such that, in one study, forceps delivery with episiotomy caused a 25-fold increase in obstetric anal sphincter injury.²⁴

Any steps that may safely reduce the need for instrument delivery should be supported. Toward this end, the Canadian Clinical Practices Obstetrics committee has recommended evidence-based labor interventions such as one-to-one support in labor, the increased use of a partogram in labor and appropriate oxytocin use, all in an effort to reduce needs for operative interventions.²⁶

If episiotomy necessary, mediolateral less risky than midline

Episiotomy was long promoted as a means of preserving the integrity of the perineal musculature and of avoiding damage to the anal sphincter, and it has been practiced routinely by some.²⁷ Strong evidence now indicates that routine episiotomy (midline or mediolateral) is unhelpful and should be abandoned.^25,27-29

Observational evidence overwhelmingly shows that midline episiotomy is strongly associated with obstetric anal sphincter injury.^{19,22,23,30,31} One of the few RCTs comparing midline with mediolateral episiotomy, although flawed in its design, noted that a clinical third-degree laceration occurred as an extension of episiotomy in 11.6% of midline incisions compared with just 2% of mediolateral cuts.³²

Another RCT, designed to examine routine versus restrictive episiotomy, noted that all but 1 (98%) of the 47 third- or fourth-degree lacerations in a group of 700 women followed midline episiotomy.²⁹ A retrospective database analysis noted a 6-fold higher risk of third-degree perineal lacerations for women undergoing midline episiotomy compared with mediolateral incision.²³ Elsewhere, midline episiotomy was associated with a 5-fold increase in symptoms of fecal incontinence at 3 months postpartum when compared with women with an intact perineum.²⁴

Even when midline episiotomies do not extend into clinical third-degree lacerations, the incidence of resultant postpartum fecal incontinence triples when compared with spontaneous second-degree perineal lacerations.³⁰ The authors postulate that a perineum cut by midline episiotomy allows for more direct contact to occur between the fetal hard parts and the anal sphincter complex during delivery, thereby increasing occult obstetric anal sphincter injury.

Observational data conflict as to whether mediolateral episiotomy contributes to, or protects against, obstetric anal sphincter injury—although the burden of evidence favors it as a risk factor that should be avoided when possible.^16,23,33 An angle of mediolateral incision cut closer to 45 degrees from the midline has been associated with less obstetric anal sphincter injury than incisions cut at closer angles to the midline.³⁴

Repairing sphincter injury

Detecting injury in labor

With any severe perineal laceration, closely inspect the external and, if exposed, internal anal sphincter and perform a rectal exam, particularly for women with numerous risk factors (although no good evidence supports the role of the rectal exam in diagnosing obstetric anal sphincter injury). Colorectal surgeons have advocated the use of a muscle stimulator to assist in identifying the ends of the external sphincter, but this has not become common practice.³⁵

Immediate vs delayed repair

It is standard practice to repair a damaged anal sphincter immediately or soon after delivery. However, given that a repair should be well done, and since a short delay does not appear to adversely affect healing, be prepared to wait for assistance for up to 24 hours rather than risk a suboptimal repair.³⁶

Better training is needed

Even among trained obstetricians and obgyn residents, 64% have reported no training or unsatisfactory training in management of obstetric anal sphincter injury; 94% of physicians felt inadequately prepared at the time of their first independent repair of the anal sphincter.^37,38 To improve your repair skills in a workshop setting, consult the following sources—www.aafp.org/also.xml in the US, or www.perineum.net in UK.

Analgesia and setting

Adequate analgesia is an essential element in a good repair. Complete relaxation of the anesthetized anal sphincter complex facilitates bringing torn ends of the sphincter together without tension.³⁹ Though theoretically this can be attained with local anesthetic infiltration, RCOG recommends that regional or general anesthesia be considered to provide complete analgesia.³⁷ It is further recommended that repair of the anal sphincter occur in an operating room, given the degree of contamination present in the labor room after delivery and the devastating effects of an infected repair (SOR: C).⁴⁰

Repair technique

There are 2 commonly used methods of external anal sphincter repair: one, the traditionally taught end-to-end approximation of the cut ends, and the other, overlapping the cut ends of the external sphincter and suturing through the overlapped portions (FIGURE 2).³⁶ Though an RCT from 2000 noted no significant difference in outcomes between these methods,⁴¹ other authors have suggested that an overlapping technique is preferred, and it remains the method most often used by colorectal surgeons in elective, secondary anal sphincter repairs.^36,39,42

A Cochrane review of which technique is better has been registered in the Clinical Trials Database. General agreement is that closure using interrupted sutures of a monofilament material, such as 2-0 polydioxanone sulfate (PDS), is the preferred closure method for the external sphincter (SOR: C).^36,40 It is recommended that a damaged internal sphincter be repaired with a running continuous suture of a material such as 2-0 polyglactin 910 (Vicryl) (SOR: C).³⁶

FIGURE 2
2 methods of anal sphincter repair

Immediate post-repair management

Use a stool softener

It had long been thought that constipation following obstetric anal sphincter injury allowed the sphincter to heal more effectively. However, new evidence from RCTs shows that using a laxative instead of a constipating regimen is more helpful in the immediate postpartum phase.⁴³ Toward this end, use a stool softener, such as lactulose, for 3 to 10 days postpartum for women with obstetric anal sphincter injury.⁴⁰

Should you prescribe an antibiotic?

Given the devastating effects of post-repair infection, most authorities consider it prudent to prescribe a course of broad-spectrum antibiotics, possibly including metronidazole (SOR: C)^37,40 A Cochrane review is registered to further examine this issue. A separate Cochrane review of the use of antibiotics for instrument vaginal delivery concluded that quality data were insufficient to make any recommendations.⁴⁴

Refer for physical therapy

Some authorities consider an early referral to physical therapy for pelvic floor exercises helpful in the immediate post-partum for all patients with obstetric anal sphincter injury (SOR: C).⁴⁵

Long-term management

Ask patients about incontinence

Given that some women are too embarrassed to seek assistance, ask those with obstetric anal sphincter injury specific questions about any symptoms of anal incontinence at a follow-up visit, such as the 6-week postpartum visit (SOR: C).^37,40 In some practices, all women who have sustained a third- or fourth-degree laceration are routinely scheduled for a 3-month follow-up visit to a dedicated clinic, irrespective of symptoms. Given the prevalence of occult obstetric anal sphincter injury for primigravid women, you may find it best to survey all women postnatally concerning any changes in anal continence. TABLE 3 demonstrates a validated, modified patient survey of anal incontinence.^37,40 A score of 6 is often used as a cutoff.

TABLE 3
Fecal Continence Scoring scale

When additional evaluation is needed

Patients who have symptoms of altered continence at 3 months (or who score above 6 on the Wexner scale) should be seen at a dedicated gynecologic or colorectal surgery clinic,⁴⁶ where they can receive a more detailed clinical evaluation and undergo anal manometry (during resting and forced squeezing) or endoanal ultrasonography. Some patients respond well to physical therapy, though a few patients ultimately require reconstructive colorectal surgery and temporary colostomy.

Management in a subsequent pregnancy

Women who have had an obstetric anal sphincter injury are at increased risk for repeat injury in a future pregnancy.⁴⁸ At some units, all such women are routinely offered a prenatal visit at the end of the second trimester to review their symptoms and to evaluate the anal sphincter with manometry or ultrasound. A large prospective study, however, found that recurrence of obstetric anal sphincter injury could not be predicted and that 95% of women with prior injury did not sustain further overt sphincter damage during a subsequent vaginal delivery.⁴⁹

However, for some women, a repeat anal sphincter laceration could prove devastating. For these women—eg, those with previous severe symptoms that required secondary surgical repair—initiate an in-depth discussion concerning the risks and benefits of elective cesarean delivery versus vaginal delivery.^37,40

CORRESPONDENCE
David Power, MD, MPH, Department of Family Medicine and Community Health, University of Minnesota, Mayo Mail Code 381, 516 Delaware St SE, Minneapolis, MN 55455. E-mail: [email protected]

Do you routinely check with new first-time mothers at a postpartum visit about changes in anal continence? They are at particular risk for obstetric anal sphincter injury and could be too embarrassed to raise the issue.

Sphincter injury following labor is the most common cause of anal incontinence (including flatus) in women, which can severely diminish quality of life and lead to considerable personal and financial costs.¹ Endoanal ultrasound can detect these injuries, even in the absence of clinically obvious damage to the anal sphincter (occult obstetric anal sphincter injury).²

In this article, we review measures to reduce the occurrence of obstetric anal sphincter injury, proper primary repair when it does occur, and appropriate long-term follow-up. Women with known obstetric anal sphincter injury must also be counseled about the risk of further damage during a future vaginal delivery.

Injury more common than symptoms would suggest

The conventional definitions of the 4 grades of perineal laceration in the US have been supplemented by more recent modifications included in a recent British Royal College of Obstetricians and Gynaecologists (RCOG) guideline (TABLE 1).³ The definition of third-degree laceration now reflects the various degrees of anal sphincter injury that may occur: partial (3a), full-thickness (3b), external anal sphincter injury, with or without injury to the internal anal sphincter (3c).

The incidence of clinical third- and fourth-degree lacerations varies widely; it is reported at between 0.5% and 3.0% in Europe and between 5.85% and 8.9% in the US.^2,4-6 A landmark British paper from 1993 revealed that though only 3% had a clinical third- or fourth-degree perineal laceration, 35% of primiparous women (none of whom had any defect before delivery) had ultrasound evidence of varying degrees of anal sphincter defect at 6 weeks postpartum that persisted at 6 months.² However, only about a third of these women had symptoms of bowel disturbance during the time of study.

These findings are supported by a meta-analysis in which 70% of women with a documented obstetric anal sphincter injury were asymptomatic.⁷ This meta-analysis concluded that clinical or occult obstetric anal sphincter injury occurs in 27% of primigravid women, and in 8.5% of multiparous women.

The long-term significance of occult obstetric anal sphincter injury and any relationship with geriatric fecal incontinence is unknown, although 71% of a sample of women with late-onset fecal incontinence were found to have ultrasound evidence of an anal sphincter defect thought to have occurred at a previous vaginal delivery.⁸ A recent English study⁹ reveals that when women were carefully re-examined after delivery by a skilled obstetrician looking specifically at the anal sphincter, the prevalence of clinically diagnosed third-degree lacerations rose sharply from the 11% initially diagnosed by the delivering physician or midwife to 24.5%. A subsequent endoanal ultrasound detected only an additional 1.2% (3 injuries, 2 of which were in the internal anal sphincter and therefore clinically undetectable). This strongly suggests that the vast majority of obstetric anal sphincter injuries can be detected clinically by a careful exam and that, when this is done, true occult injuries will be a rare finding.

TABLE 1
Classification of perineal injury⁹

Mechanisms of injury

Maintenance of fecal continence involves the coordinated action of several anatomical and physiological elements (FIGURE 1).¹⁰ An intact, innervated anal sphincter complex (both external and internal) is necessary. The sphincter complex can be damaged during childbirth in 3 ways.

Direct mechanical injury. Direct external or internal anal sphincter muscle disruption can occur, as with a clinically obvious third- or fourth-degree perineal laceration or an occult injury subsequently noted on ultrasound.

Neurologic injury. Neuropathy of the pudendal nerve may result from forceps delivery or persistent nerve compression from the fetal head.¹⁴ Traction neuropathy may also occur with fetal macrosomia and with prolonged pushing during Stage 2 in successive pregnancies, or with prolonged stretching of the nerve due to persistent poor postpartum pelvic floor tone. Injured nerves often undergo demyelination but usually recover with time.

Combined mechanical and neurologic trauma. Isolated neurologic injury, as described above, is believed to be rare. Neuropathy more commonly accompanies mechanical damage.¹⁵

Who is at risk?

Several risk factors are unavoidable. One of these is primiparity, a consistently reported independent variable also associated with other risk factors for obstetric anal sphincter injury, such as instrument delivery (TABLE 2).

TABLE 2
Major risk factors for obstetric anal sphincter injury

Preventing obstetric anal sphincter injury

Sphincter injury can occur even when obstetrical management is optimal. Although evidence from RCT data is often lacking, sufficient observational and retrospective data support the following recommendations to reduce the likelihood of injury.

Choose vacuum delivery before forceps

Any form of instrument delivery increases the risk of obstetric anal sphincter injury and altered fecal continence by between 2- and 7-fold.^2,16,24 An RCT found clinical third-degree tears in 16% of women with forceps-assisted deliveries, compared with 7% of vacuum-assisted deliveries; the authors concluded that, when circumstances allow, vacuum delivery should be attempted first (acknowledging however that 23% of vacuum deliveries failed and proceeded to a forceps extraction, a sequence associated with increased injury).¹⁷ A meta-analysis confirmed that vacuum extraction is preferred when instrumental delivery is necessary (SOR: A).²⁵

When midline episiotomy was performed during instrument delivery, the risk of obstetric anal sphincter injury approximately doubled again, such that, in one study, forceps delivery with episiotomy caused a 25-fold increase in obstetric anal sphincter injury.²⁴

Any steps that may safely reduce the need for instrument delivery should be supported. Toward this end, the Canadian Clinical Practices Obstetrics committee has recommended evidence-based labor interventions such as one-to-one support in labor, the increased use of a partogram in labor and appropriate oxytocin use, all in an effort to reduce needs for operative interventions.²⁶

If episiotomy necessary, mediolateral less risky than midline

Episiotomy was long promoted as a means of preserving the integrity of the perineal musculature and of avoiding damage to the anal sphincter, and it has been practiced routinely by some.²⁷ Strong evidence now indicates that routine episiotomy (midline or mediolateral) is unhelpful and should be abandoned.^25,27-29

Observational evidence overwhelmingly shows that midline episiotomy is strongly associated with obstetric anal sphincter injury.^{19,22,23,30,31} One of the few RCTs comparing midline with mediolateral episiotomy, although flawed in its design, noted that a clinical third-degree laceration occurred as an extension of episiotomy in 11.6% of midline incisions compared with just 2% of mediolateral cuts.³²

Another RCT, designed to examine routine versus restrictive episiotomy, noted that all but 1 (98%) of the 47 third- or fourth-degree lacerations in a group of 700 women followed midline episiotomy.²⁹ A retrospective database analysis noted a 6-fold higher risk of third-degree perineal lacerations for women undergoing midline episiotomy compared with mediolateral incision.²³ Elsewhere, midline episiotomy was associated with a 5-fold increase in symptoms of fecal incontinence at 3 months postpartum when compared with women with an intact perineum.²⁴

Even when midline episiotomies do not extend into clinical third-degree lacerations, the incidence of resultant postpartum fecal incontinence triples when compared with spontaneous second-degree perineal lacerations.³⁰ The authors postulate that a perineum cut by midline episiotomy allows for more direct contact to occur between the fetal hard parts and the anal sphincter complex during delivery, thereby increasing occult obstetric anal sphincter injury.

Observational data conflict as to whether mediolateral episiotomy contributes to, or protects against, obstetric anal sphincter injury—although the burden of evidence favors it as a risk factor that should be avoided when possible.^16,23,33 An angle of mediolateral incision cut closer to 45 degrees from the midline has been associated with less obstetric anal sphincter injury than incisions cut at closer angles to the midline.³⁴

Repairing sphincter injury

Detecting injury in labor

With any severe perineal laceration, closely inspect the external and, if exposed, internal anal sphincter and perform a rectal exam, particularly for women with numerous risk factors (although no good evidence supports the role of the rectal exam in diagnosing obstetric anal sphincter injury). Colorectal surgeons have advocated the use of a muscle stimulator to assist in identifying the ends of the external sphincter, but this has not become common practice.³⁵

Immediate vs delayed repair

It is standard practice to repair a damaged anal sphincter immediately or soon after delivery. However, given that a repair should be well done, and since a short delay does not appear to adversely affect healing, be prepared to wait for assistance for up to 24 hours rather than risk a suboptimal repair.³⁶

Better training is needed

Even among trained obstetricians and obgyn residents, 64% have reported no training or unsatisfactory training in management of obstetric anal sphincter injury; 94% of physicians felt inadequately prepared at the time of their first independent repair of the anal sphincter.^37,38 To improve your repair skills in a workshop setting, consult the following sources—www.aafp.org/also.xml in the US, or www.perineum.net in UK.

Analgesia and setting

Adequate analgesia is an essential element in a good repair. Complete relaxation of the anesthetized anal sphincter complex facilitates bringing torn ends of the sphincter together without tension.³⁹ Though theoretically this can be attained with local anesthetic infiltration, RCOG recommends that regional or general anesthesia be considered to provide complete analgesia.³⁷ It is further recommended that repair of the anal sphincter occur in an operating room, given the degree of contamination present in the labor room after delivery and the devastating effects of an infected repair (SOR: C).⁴⁰

Repair technique

There are 2 commonly used methods of external anal sphincter repair: one, the traditionally taught end-to-end approximation of the cut ends, and the other, overlapping the cut ends of the external sphincter and suturing through the overlapped portions (FIGURE 2).³⁶ Though an RCT from 2000 noted no significant difference in outcomes between these methods,⁴¹ other authors have suggested that an overlapping technique is preferred, and it remains the method most often used by colorectal surgeons in elective, secondary anal sphincter repairs.^36,39,42

A Cochrane review of which technique is better has been registered in the Clinical Trials Database. General agreement is that closure using interrupted sutures of a monofilament material, such as 2-0 polydioxanone sulfate (PDS), is the preferred closure method for the external sphincter (SOR: C).^36,40 It is recommended that a damaged internal sphincter be repaired with a running continuous suture of a material such as 2-0 polyglactin 910 (Vicryl) (SOR: C).³⁶

FIGURE 2
2 methods of anal sphincter repair

Immediate post-repair management

Use a stool softener

It had long been thought that constipation following obstetric anal sphincter injury allowed the sphincter to heal more effectively. However, new evidence from RCTs shows that using a laxative instead of a constipating regimen is more helpful in the immediate postpartum phase.⁴³ Toward this end, use a stool softener, such as lactulose, for 3 to 10 days postpartum for women with obstetric anal sphincter injury.⁴⁰

Should you prescribe an antibiotic?

Given the devastating effects of post-repair infection, most authorities consider it prudent to prescribe a course of broad-spectrum antibiotics, possibly including metronidazole (SOR: C)^37,40 A Cochrane review is registered to further examine this issue. A separate Cochrane review of the use of antibiotics for instrument vaginal delivery concluded that quality data were insufficient to make any recommendations.⁴⁴

Refer for physical therapy

Some authorities consider an early referral to physical therapy for pelvic floor exercises helpful in the immediate post-partum for all patients with obstetric anal sphincter injury (SOR: C).⁴⁵

Long-term management

Ask patients about incontinence

Given that some women are too embarrassed to seek assistance, ask those with obstetric anal sphincter injury specific questions about any symptoms of anal incontinence at a follow-up visit, such as the 6-week postpartum visit (SOR: C).^37,40 In some practices, all women who have sustained a third- or fourth-degree laceration are routinely scheduled for a 3-month follow-up visit to a dedicated clinic, irrespective of symptoms. Given the prevalence of occult obstetric anal sphincter injury for primigravid women, you may find it best to survey all women postnatally concerning any changes in anal continence. TABLE 3 demonstrates a validated, modified patient survey of anal incontinence.^37,40 A score of 6 is often used as a cutoff.

TABLE 3
Fecal Continence Scoring scale

When additional evaluation is needed

Patients who have symptoms of altered continence at 3 months (or who score above 6 on the Wexner scale) should be seen at a dedicated gynecologic or colorectal surgery clinic,⁴⁶ where they can receive a more detailed clinical evaluation and undergo anal manometry (during resting and forced squeezing) or endoanal ultrasonography. Some patients respond well to physical therapy, though a few patients ultimately require reconstructive colorectal surgery and temporary colostomy.

Management in a subsequent pregnancy

Women who have had an obstetric anal sphincter injury are at increased risk for repeat injury in a future pregnancy.⁴⁸ At some units, all such women are routinely offered a prenatal visit at the end of the second trimester to review their symptoms and to evaluate the anal sphincter with manometry or ultrasound. A large prospective study, however, found that recurrence of obstetric anal sphincter injury could not be predicted and that 95% of women with prior injury did not sustain further overt sphincter damage during a subsequent vaginal delivery.⁴⁹

However, for some women, a repeat anal sphincter laceration could prove devastating. For these women—eg, those with previous severe symptoms that required secondary surgical repair—initiate an in-depth discussion concerning the risks and benefits of elective cesarean delivery versus vaginal delivery.^37,40

CORRESPONDENCE
David Power, MD, MPH, Department of Family Medicine and Community Health, University of Minnesota, Mayo Mail Code 381, 516 Delaware St SE, Minneapolis, MN 55455. E-mail: [email protected]

Orthotopic liver transplantation (OLT) is the replacement of a whole diseased liver with a healthy donor liver. The number of persons receiving OLT is increasing. Though it is unlikely you will be involved in the care of a patient immediately after OLT, you ’ ll need to know about the complications that occur in this period as they may impact the long-term care of the patient.

Long-term issues — such as cardiovascular disease, bone disease, malignancy, anemia, psychiatric disorders, and financial stressors — put these patients at higher risk for problems more than the average patient. Perhaps the most important task is for you to keep in contact with the transplant center when questions or concerns arise. Over time, you will once again become the primary physician and advocate for these patients.

Complications after transplant (less than 1 year)

Within 1 month post-OLT, the most frequent complications are acute graft rejection, vascular thrombosis, biliary leak or stricture, and infection. Between 1 and 3 months, acute and chronic graft rejection can occur, but medication toxicity and opportunistic infections become more common (TABLE 1). ¹

A broad range of infections may develop, including cytomegalovirus, Epstein-Barr virus, herpes simplex virus, varicella zoster virus, adenovirus, tuberculosis, Pneumocystis , toxoplasmosis, Listeria spp, Candida spp, Aspergillus spp, and Cryptococcus spp. During this time, doses of immunosuppressive agents are lowered and corticosteroids are discontinued in many patients.

Once patients are considered stable after OLT, they will likely come under your supervision again. While opportunistic infections, surgical issues, and acute rejection become less common between 3 and 12 months, other complications related to OLT may occur.

Graft reinfection with hepatitis C virus (HCV) is universal, and 50% to 80% patients will develop biopsy-proven hepatitis. ² Many will require treatment for recurrent HCV to avoid progression to cirrhosis.

Recurrent hepatitis B infection is much less common due to prophylactic therapy with hepatitis B immunoglobulin and antiviral medications, although 10% of transplant recipients will develop hepatitis despite prophylaxis.

Other causes of recurrent liver disease post-OLT include liver injury due to recurrent drug or alcohol abuse, non-alcoholic steatohepatitis, cholestatic and autoimmune liver disease, and liver cancer.

Toxicity due to immunosuppressive medications is also common in this time frame (TABLE 2). ³ Be alert to the potential for hepatotoxicity and drug interactions with any new pharmacologic agent. Other drugs (eg, lipid-lowering agents, antibiotics, antifungals) may cause liver injury on their own and need to be closely monitored.

Lastly, even though patients are at increased risk for such common infections as influenza, pneumonia, and urinary tract infections, opportunistic infections are uncommon in this period. Keep in mind that patients usually develop infections that are community-acquired and not opportunistic, particularly as time goes on.

TABLE 1
Common complications immediately after liver transplantation

TABLE 2
Immunosuppressive medications and interactions after liver transplantation

Long-term complications

Cardiovascular disease

Up to 20% of late deaths after OLT are caused by cardiovascular disease. ⁴ Uncontrollable factors, such as preexisting cardiac disease, male sex, family history of cardiac disease, and advanced age contribute to the incidence of cardiovascular disease. However, a number of potentially controllable factors, such as hypertension, hyperlipidemia, obesity, and diabetes are common after OLT and should be addressed.

Hypertension. Hypertension occurs in 40% to 75% of OLT patients. ⁵ Causes include calcineurin-inhibitor (cyclosporine, tacrolimus) therapy, high-dose corticosteroids, and renal insufficiency. Calcineurin inhibitors cause renal vasoconstriction, leading to sodium retention and hypertension. Reducing the doses of these medications by the transplant center typically improves blood pressure control.

Treatment of choice for hypertension depends on how recently the transplant was performed. In the first 6 months following the procedure, dihydropyridine calcium-channel blockers (eg, amlodipine) and alpha-blockers are the mainstay of therapy, although peripheral edema and orthostatic hypotension may affect their tolerability. Diuretics can also be used in volume-overloaded patients.

After 6 months, other pharmacologic agents, such as angiotensin-converting enzyme (ACE) inhibitors and beta-blockers, can be administered to patients with stable renal function and without other contraindications (strength of recommendation [SOR]: C ). Long-term management of hypertension does not differ significantly from that in non-transplant patients.

Hyperlipidemia/obesity. Obesity and hyperlipidemia may affect up to half of OLT patients. Factors that contribute to both disorders include immunosuppressive drugs, increased appetite, diabetes, pretransplant hyperlipidemia, and history of cholestatic liver disease.

For hyperlipidemia, lifestyle modifications, such as diet and exercise, are recommended. If these measures are ineffective, statins are first-line agents. Avoid bile acid binding resins, which may interfere with the absorption of all medications. For refractory cases, switching from cyclosporine to tacrolimus under the direction of the transplant center might be indicated.

Treatment of obesity following OLT should also focus on lifestyle changes, as the safety of pharmacotherapy and surgery for obesity is uncertain in these patients.

Glucose intolerance and diabetes. Many patients will have glucose intolerance that resolves after steroid withdrawal. Main risk factors are pre-OLT diabetes, episodes of steroid-resistant rejection, and obesity. Post-OLT onset of diabetes will persist for only a small percentage of patients. ⁶

Treatment of post-transplant diabetes is similar to that for any patient. Insulin is often required initially, but with reduction in immunosuppression and corticosteroids, patients can usually be switched to oral agents. Though there is no absolute contraindication to using any antidiabetes medications, most physicians try to avoid those with potential hepatotoxicity, such as the thiazolidinediones (SOR: C).

Weight loss is critical and often improves glucose tolerance. Transplant centers may switch patients from tacrolimus to cyclosporine to control hyperglycemia. Long-term screening for end organ complications (retinopathy, nephropathy, neuropathy) is as important for this population as it is for in non-transplant diabetics.

FIGURE
Indications for liver transplantation in the US

Renal disease

Up to 20% of OLT recipients develop end-stage renal disease, requiring hemodialysis or renal transplantation within 10 years after transplant. ⁵ If patients have renal dysfunction before OLT, lower-dose calcineurin inhibitors and using alternative immunosuppression post-OLT may improve renal function in the long term. A rise in creatinine in the first year after OLT is a strong risk factor for long-term development of renal insufficiency, while stable creatinine levels at 1 year usually indicate long-term maintenance of renal function.^9,10

Closely monitor patients with early renal dysfunction, avoid nephrotoxic agents, and reduce or withdraw calcineurin inhibitors (as directed by the transplant center). Occasionally renal transplantation will be indicated.

Be aware that all OLT recipients need adequate hydration during acute illnesses (influenza, common colds, gastroenteritis), especially if they have renal dysfunction. Potential nephrotoxic agents, such as nonsteroidal anti-inflammatory drugs (NSAIDs), aminoglycoside antibiotics, and intravenous contrast, should be avoided if possible.

Bone disease

Osteoporosis should be screened for and identified before OLT. Contributing factors for bone disease after transplant include preexisting osteoporosis, immobility, vitamin D deficiency, corticosteroid use, and hypogonadism. In the first 6 months after transplant, bone mineral density (BMD) significantly declines, often accelerated by immunosuppressive medications, corticosteroids, and immobility. ^{11 - 14} After 6 months, BMD increases rapidly and, by 12 months, approaches pre-OLT values. All patients should have bone densitometry performed before OLT or before hospital discharge and receive calcium (1500 mg/d) and vitamin D (800 IU/d) supplementation (SOR: C).

Unless significant risk factors for osteoporosis are present (eg, continued use of corticosteroids, history of bone loss, fracture, or cholestatic liver disease), it is unclear whether low-risk patients should have serial bone densitometry tests performed in the years following OLT. Patients with T-scores ≥ 2 standard deviations below mean should be considered for antiresorptive therapy. Given the recent concerns regarding estrogen use and cardiovascular disease, bisphosphonates and calcitonin are preferred. For patients who develop fractures or avascular necrosis from corticosteroids, joint replacement surgery appears to be safe and effective post-OLT (SOR: B).^15,16

Malignancy

Of all of the complications following OLT, malignancy causes the highest morbidity and mortality. The overall incidence of malignancy is between 2.3% and 12.9% and may be up to 5 times higher than in the general population.^17,18 The most common malignancies are post-transplant lymphoproliferative disorder (1% – 4.4%) and nonmelanoma skin cancer (0.5% – 4.3%); less common are gastrointestinal (0.4% – 1.0%), genitourinary (0.2% – 2.2%), lung (0.2% – 0.8%), and oropharyngeal (0.4% – 0.8%) malignancies.¹⁷ Many patients with small liver cancers (1 lesion < 5 cm or up to 3 lesions each < 3 cm) are receiving transplants and, despite the risk of recurrence post-OLT, have a similar survival rate as patients receiving OLT for other indications.

Though it is clear that OLT recipients are at higher risk than the general population for malignancy, there are no specific guidelines for screening. However, based on the risk of early malignancy, screening should resume within the first 2 to 3 years after OLT (SOR: B).^{19- 22} Most transplant centers will recommend either performing a more intense screening protocol than for non-transplant patients or individualizing screening protocols for each patient depending on risk factors.

Advise patients who spend time in the sun to wear sun block with a protective factor > 40 and to have routine skin examinations. It is unclear whether colorectal cancer screening in OLT recipients should occur more frequently than the general population. Colorectal adenomas may be more common among OLT recipients than among healthy controls, ²³ but until more data are available, screening should mirror that of the general population (SOR: B). Since hepatocellular carcinoma may recur after OLT, transplant centers typically request imaging (computed tomography, ultrasound, magnetic resonance imaging) at regular intervals after OLT.

Use your discretion when screening for other common malignancies, such as breast, cervical, and prostate cancer. It is unclear whether screening specific groups of patients (such as tobacco smokers) for oropharyngeal, lung, and genitourinary cancer will be cost-effective or impact survival.

Anemia

The prevalence of anemia after OLT reportedly is between 4.3% and 28.2%, depending on the population studied and time after transplantation. ^{24 , 25} Blood loss, sepsis, medications, renal dysfunction, or hypersplenism can contribute to immediate postoperative anemia. Beyond the immediate postoperative period, anemia may be related to different causes (TABLE 3). ²⁶ Medication-induced anemia is usually related to bone marrow suppression, although calcineurin inhibitors may cause microangiopathic hemolysis, hemolyticuremic syndrome, or pure redcell aplasia.

Viral infections often cause anemia in the first 12 weeks after transplantation. Aplastic anemia may be related to parvovirus B19 infection, although it is more commonly seen in patients who undergo liver transplantation for acute liver failure.^24,27 Posttransplant lymphoproliferative disorder ranges from polyclonal B-cell hyperplasia (related to Epstein-Barr virus) that responds to reduction in immunosuppression to aggressive lymphoma treated with high dose chemotherapy.

Graft-versus-host disease is a rare but important cause of pancytopenia after OLT and is diagnosed by establishing chimerism, donor and recipient lymphocytes, in the blood and bone marrow; mortality is high.²⁸

Lastly, renal failure and iron deficiency are other common causes of anemia after OLT that warrant investigation. Despite complete evaluation, half of adult patients do not have an identifiable cause of anemia and may respond to a therapeutic trial of erythropoietin (SOR: C).²⁶

TABLE 3
Evaluation of anemia after liver transplantation

Psychosocial and socioeconomic concerns

Liver transplantation is a tremendously stressful and life-altering procedure affecting patients and their families. In the initial postoperative period, the stress of the operation and other factors (immunosuppression, infection, prolonged hospital stay) can lead to a variety of psychiatric disorders, such as delirium, anxiety, depression, mania, and psychosis. A multidisciplinary approach, including psychiatry, social work, and nursing care, is required to help the patients and families through this period, as expectations for full recovery may be delayed by psychiatric conditions.

Psychiatric problems

Many transplant recipients have long-term psychiatric problems. Depression and anxiety diminish quality of life, particularly for patients whose transplant was for hepatitis C and those with post-transplant viral recurrence.^29,30 Most patients will respond to antidepressants and ongoing psychiatric care. The side-effect profile should be individualized for each patient, keeping in mind the potential interactions with the current medications.

Mania and hypomania, while less common than depression, are often related to higher doses of immunosuppression (eg, corticosteroids). Cyclosporine may increase lithium levels, leading to toxicity.³¹ Treatment with anticonvulsant medications, such as carbamazepine, may decrease calcineurin-inhibitor levels and should be monitored in coordination with the transplant team. Finally, some patients with encephalopathy prior to OLT have persistent cognitive deficits long after OLT.³²

Drug and alcohol recidivism are common post-OLT and typically occurs in about 20% of patients. It is important that active steps are taken to avoid recidivism immediately after OLT. Long-term psychiatric care and continued attendance at support groups help maintain sobriety. The important contributions you can make are maintaining a heightened awareness for recidivism, communicating with patients regularly about drug and alcohol abuse, and providing support and referral services.

Socioeconomic problems

While most transplant recipients maintain a good quality of life, some have long-term socioeconomic problems. One study showed that only one third of OLT recipients returned successfully to work, just slightly higher than the percentage working before OLT.³³ The economic situation improved in 11.9% of the recipients, worsened in 33.9%, and stayed the same in 54.2%. Concurrent illness, prolonged inactivity, psychiatric disorders, and the level of physical requirements at work are the main contributing factors to unemployment.

Another major stressor is medical cost. The average cost of immunosuppressive medications alone is $10,000 to $20,000 per year.³ Most of the charges are reimbursable, although this depends on the payer and time from transplantation. Medicare pays for immunosuppressive medications for only 36 months. Beyond that point, patients will require secondary insurance or other assistance. This expenditure is exacerbated by the cost of other medications, clinic visits with the transplant center, family physicians, and specialists, and time away from work.

Although patients are usually well informed of these concerns before OLT, they often do not appreciate the financial magnitude until after OLT. Encourage patients to return to work, stay active physically and mentally, and prepare for these financial considerations.

Sexual issues

Some patients have persistent sexual dysfunction that may have an organic basis (cardiovascular, renal, liver, endocrine) requiring investigation. The safety and efficacy of sildenafil (Viagra) in OLT recipients has not been investigated to date.

However, other patients regain their libido and gonadal function immediately after OLT; pregnancy may occur in this period. Advise patients to wait at least 2 years post-OLT before considering pregnancy (SOR: C). ³⁴ Contraception, preferably barrier-type, should be used during sexual intercourse. Hormonal contraceptives are not contraindicated but should probably not be administered until the patient ’ s transplant status is stable.

If pregnancy does occur, apprise the patient of potential complications and adverse outcomes. Hypertension and preeclampsia are more common in pregnant OLT recipients; life-threatening infections and acute rejection are rare. Fortunately, most patients deliver healthy babies; miscarriages, stillbirths, and malformations are uncommon. An obstetrician specializing in high-risk pregnancy should follow all pregnant OLT recipients.

Vaccination

Vaccination after OLT is controversial. Live vaccines are generally contraindicated post-OLT and their safety in patients with stable graft function and on low levels of immunosuppression is unclear (SOR: C). Patients should receive pneumococcal vaccination, hepatitis A and B vaccination if not already immune, and yearly influenza vaccination (SOR: C). For travel outside of the US or in uncertain situations or exposures, the best reference is the Centers for Disease Control web site: www.cdc.org.

Communication with the transplant center

Direct communication with the patient ’ s transplant center is extremely important. You and the transplant center should determine the most effective way (phone, fax or e-mail) to communicate.

When should you contact the transplant center? First, obtain the center ’ s approval for any new medications that may be used long-term or have the potential for nephrotoxicity, hepatotoxicity, or immunosuppression. Second, notify the transplant center in the event of new signs or symptoms, such as fever, weight loss, abdominal pain, or jaundice. Being cautious by communicating early is often the most prudent course. Third, alert the transplant center of any hospitalizations. Transfer to the transplant center for any transplant-related problem or prolonged hospitalization usually provides the best outcome for the patient.

On the flip side, you are the primary caretaker, and the transplant center should regularly communicate with you regarding general medical concerns and any new diagnoses, interventions, or treatments. The transplant center should also regularly communicate with you regarding general medical concerns and any new diagnoses, interventions, or treatments. A strong, mutual relationship between you and the transplant center will have great impact on the recipient ’ s long-term care.

CORRESPONDENCE
Josh Levitsky, MD, Northwestern Memorial Hospital, 675 North St. Clair St, Suite 15-250, Chicago, IL 60611. E-mail: [email protected]

Orthotopic liver transplantation (OLT) is the replacement of a whole diseased liver with a healthy donor liver. The number of persons receiving OLT is increasing. Though it is unlikely you will be involved in the care of a patient immediately after OLT, you ’ ll need to know about the complications that occur in this period as they may impact the long-term care of the patient.

Long-term issues — such as cardiovascular disease, bone disease, malignancy, anemia, psychiatric disorders, and financial stressors — put these patients at higher risk for problems more than the average patient. Perhaps the most important task is for you to keep in contact with the transplant center when questions or concerns arise. Over time, you will once again become the primary physician and advocate for these patients.

Complications after transplant (less than 1 year)

Within 1 month post-OLT, the most frequent complications are acute graft rejection, vascular thrombosis, biliary leak or stricture, and infection. Between 1 and 3 months, acute and chronic graft rejection can occur, but medication toxicity and opportunistic infections become more common (TABLE 1). ¹

A broad range of infections may develop, including cytomegalovirus, Epstein-Barr virus, herpes simplex virus, varicella zoster virus, adenovirus, tuberculosis, Pneumocystis , toxoplasmosis, Listeria spp, Candida spp, Aspergillus spp, and Cryptococcus spp. During this time, doses of immunosuppressive agents are lowered and corticosteroids are discontinued in many patients.

Once patients are considered stable after OLT, they will likely come under your supervision again. While opportunistic infections, surgical issues, and acute rejection become less common between 3 and 12 months, other complications related to OLT may occur.

Graft reinfection with hepatitis C virus (HCV) is universal, and 50% to 80% patients will develop biopsy-proven hepatitis. ² Many will require treatment for recurrent HCV to avoid progression to cirrhosis.

Recurrent hepatitis B infection is much less common due to prophylactic therapy with hepatitis B immunoglobulin and antiviral medications, although 10% of transplant recipients will develop hepatitis despite prophylaxis.

Other causes of recurrent liver disease post-OLT include liver injury due to recurrent drug or alcohol abuse, non-alcoholic steatohepatitis, cholestatic and autoimmune liver disease, and liver cancer.

Toxicity due to immunosuppressive medications is also common in this time frame (TABLE 2). ³ Be alert to the potential for hepatotoxicity and drug interactions with any new pharmacologic agent. Other drugs (eg, lipid-lowering agents, antibiotics, antifungals) may cause liver injury on their own and need to be closely monitored.

Lastly, even though patients are at increased risk for such common infections as influenza, pneumonia, and urinary tract infections, opportunistic infections are uncommon in this period. Keep in mind that patients usually develop infections that are community-acquired and not opportunistic, particularly as time goes on.

TABLE 1
Common complications immediately after liver transplantation

TABLE 2
Immunosuppressive medications and interactions after liver transplantation

Long-term complications

Cardiovascular disease

Up to 20% of late deaths after OLT are caused by cardiovascular disease. ⁴ Uncontrollable factors, such as preexisting cardiac disease, male sex, family history of cardiac disease, and advanced age contribute to the incidence of cardiovascular disease. However, a number of potentially controllable factors, such as hypertension, hyperlipidemia, obesity, and diabetes are common after OLT and should be addressed.

Hypertension. Hypertension occurs in 40% to 75% of OLT patients. ⁵ Causes include calcineurin-inhibitor (cyclosporine, tacrolimus) therapy, high-dose corticosteroids, and renal insufficiency. Calcineurin inhibitors cause renal vasoconstriction, leading to sodium retention and hypertension. Reducing the doses of these medications by the transplant center typically improves blood pressure control.

Treatment of choice for hypertension depends on how recently the transplant was performed. In the first 6 months following the procedure, dihydropyridine calcium-channel blockers (eg, amlodipine) and alpha-blockers are the mainstay of therapy, although peripheral edema and orthostatic hypotension may affect their tolerability. Diuretics can also be used in volume-overloaded patients.

After 6 months, other pharmacologic agents, such as angiotensin-converting enzyme (ACE) inhibitors and beta-blockers, can be administered to patients with stable renal function and without other contraindications (strength of recommendation [SOR]: C ). Long-term management of hypertension does not differ significantly from that in non-transplant patients.

Hyperlipidemia/obesity. Obesity and hyperlipidemia may affect up to half of OLT patients. Factors that contribute to both disorders include immunosuppressive drugs, increased appetite, diabetes, pretransplant hyperlipidemia, and history of cholestatic liver disease.

For hyperlipidemia, lifestyle modifications, such as diet and exercise, are recommended. If these measures are ineffective, statins are first-line agents. Avoid bile acid binding resins, which may interfere with the absorption of all medications. For refractory cases, switching from cyclosporine to tacrolimus under the direction of the transplant center might be indicated.

Treatment of obesity following OLT should also focus on lifestyle changes, as the safety of pharmacotherapy and surgery for obesity is uncertain in these patients.

Glucose intolerance and diabetes. Many patients will have glucose intolerance that resolves after steroid withdrawal. Main risk factors are pre-OLT diabetes, episodes of steroid-resistant rejection, and obesity. Post-OLT onset of diabetes will persist for only a small percentage of patients. ⁶

Treatment of post-transplant diabetes is similar to that for any patient. Insulin is often required initially, but with reduction in immunosuppression and corticosteroids, patients can usually be switched to oral agents. Though there is no absolute contraindication to using any antidiabetes medications, most physicians try to avoid those with potential hepatotoxicity, such as the thiazolidinediones (SOR: C).

Weight loss is critical and often improves glucose tolerance. Transplant centers may switch patients from tacrolimus to cyclosporine to control hyperglycemia. Long-term screening for end organ complications (retinopathy, nephropathy, neuropathy) is as important for this population as it is for in non-transplant diabetics.

FIGURE
Indications for liver transplantation in the US

Renal disease

Up to 20% of OLT recipients develop end-stage renal disease, requiring hemodialysis or renal transplantation within 10 years after transplant. ⁵ If patients have renal dysfunction before OLT, lower-dose calcineurin inhibitors and using alternative immunosuppression post-OLT may improve renal function in the long term. A rise in creatinine in the first year after OLT is a strong risk factor for long-term development of renal insufficiency, while stable creatinine levels at 1 year usually indicate long-term maintenance of renal function.^9,10

Closely monitor patients with early renal dysfunction, avoid nephrotoxic agents, and reduce or withdraw calcineurin inhibitors (as directed by the transplant center). Occasionally renal transplantation will be indicated.

Be aware that all OLT recipients need adequate hydration during acute illnesses (influenza, common colds, gastroenteritis), especially if they have renal dysfunction. Potential nephrotoxic agents, such as nonsteroidal anti-inflammatory drugs (NSAIDs), aminoglycoside antibiotics, and intravenous contrast, should be avoided if possible.

Bone disease

Osteoporosis should be screened for and identified before OLT. Contributing factors for bone disease after transplant include preexisting osteoporosis, immobility, vitamin D deficiency, corticosteroid use, and hypogonadism. In the first 6 months after transplant, bone mineral density (BMD) significantly declines, often accelerated by immunosuppressive medications, corticosteroids, and immobility. ^{11 - 14} After 6 months, BMD increases rapidly and, by 12 months, approaches pre-OLT values. All patients should have bone densitometry performed before OLT or before hospital discharge and receive calcium (1500 mg/d) and vitamin D (800 IU/d) supplementation (SOR: C).

Unless significant risk factors for osteoporosis are present (eg, continued use of corticosteroids, history of bone loss, fracture, or cholestatic liver disease), it is unclear whether low-risk patients should have serial bone densitometry tests performed in the years following OLT. Patients with T-scores ≥ 2 standard deviations below mean should be considered for antiresorptive therapy. Given the recent concerns regarding estrogen use and cardiovascular disease, bisphosphonates and calcitonin are preferred. For patients who develop fractures or avascular necrosis from corticosteroids, joint replacement surgery appears to be safe and effective post-OLT (SOR: B).^15,16

Malignancy

Of all of the complications following OLT, malignancy causes the highest morbidity and mortality. The overall incidence of malignancy is between 2.3% and 12.9% and may be up to 5 times higher than in the general population.^17,18 The most common malignancies are post-transplant lymphoproliferative disorder (1% – 4.4%) and nonmelanoma skin cancer (0.5% – 4.3%); less common are gastrointestinal (0.4% – 1.0%), genitourinary (0.2% – 2.2%), lung (0.2% – 0.8%), and oropharyngeal (0.4% – 0.8%) malignancies.¹⁷ Many patients with small liver cancers (1 lesion < 5 cm or up to 3 lesions each < 3 cm) are receiving transplants and, despite the risk of recurrence post-OLT, have a similar survival rate as patients receiving OLT for other indications.

Though it is clear that OLT recipients are at higher risk than the general population for malignancy, there are no specific guidelines for screening. However, based on the risk of early malignancy, screening should resume within the first 2 to 3 years after OLT (SOR: B).^{19- 22} Most transplant centers will recommend either performing a more intense screening protocol than for non-transplant patients or individualizing screening protocols for each patient depending on risk factors.

Advise patients who spend time in the sun to wear sun block with a protective factor > 40 and to have routine skin examinations. It is unclear whether colorectal cancer screening in OLT recipients should occur more frequently than the general population. Colorectal adenomas may be more common among OLT recipients than among healthy controls, ²³ but until more data are available, screening should mirror that of the general population (SOR: B). Since hepatocellular carcinoma may recur after OLT, transplant centers typically request imaging (computed tomography, ultrasound, magnetic resonance imaging) at regular intervals after OLT.

Use your discretion when screening for other common malignancies, such as breast, cervical, and prostate cancer. It is unclear whether screening specific groups of patients (such as tobacco smokers) for oropharyngeal, lung, and genitourinary cancer will be cost-effective or impact survival.

Anemia

The prevalence of anemia after OLT reportedly is between 4.3% and 28.2%, depending on the population studied and time after transplantation. ^{24 , 25} Blood loss, sepsis, medications, renal dysfunction, or hypersplenism can contribute to immediate postoperative anemia. Beyond the immediate postoperative period, anemia may be related to different causes (TABLE 3). ²⁶ Medication-induced anemia is usually related to bone marrow suppression, although calcineurin inhibitors may cause microangiopathic hemolysis, hemolyticuremic syndrome, or pure redcell aplasia.

Viral infections often cause anemia in the first 12 weeks after transplantation. Aplastic anemia may be related to parvovirus B19 infection, although it is more commonly seen in patients who undergo liver transplantation for acute liver failure.^24,27 Posttransplant lymphoproliferative disorder ranges from polyclonal B-cell hyperplasia (related to Epstein-Barr virus) that responds to reduction in immunosuppression to aggressive lymphoma treated with high dose chemotherapy.

Graft-versus-host disease is a rare but important cause of pancytopenia after OLT and is diagnosed by establishing chimerism, donor and recipient lymphocytes, in the blood and bone marrow; mortality is high.²⁸

Lastly, renal failure and iron deficiency are other common causes of anemia after OLT that warrant investigation. Despite complete evaluation, half of adult patients do not have an identifiable cause of anemia and may respond to a therapeutic trial of erythropoietin (SOR: C).²⁶

TABLE 3
Evaluation of anemia after liver transplantation

Psychosocial and socioeconomic concerns

Liver transplantation is a tremendously stressful and life-altering procedure affecting patients and their families. In the initial postoperative period, the stress of the operation and other factors (immunosuppression, infection, prolonged hospital stay) can lead to a variety of psychiatric disorders, such as delirium, anxiety, depression, mania, and psychosis. A multidisciplinary approach, including psychiatry, social work, and nursing care, is required to help the patients and families through this period, as expectations for full recovery may be delayed by psychiatric conditions.

Psychiatric problems

Many transplant recipients have long-term psychiatric problems. Depression and anxiety diminish quality of life, particularly for patients whose transplant was for hepatitis C and those with post-transplant viral recurrence.^29,30 Most patients will respond to antidepressants and ongoing psychiatric care. The side-effect profile should be individualized for each patient, keeping in mind the potential interactions with the current medications.

Mania and hypomania, while less common than depression, are often related to higher doses of immunosuppression (eg, corticosteroids). Cyclosporine may increase lithium levels, leading to toxicity.³¹ Treatment with anticonvulsant medications, such as carbamazepine, may decrease calcineurin-inhibitor levels and should be monitored in coordination with the transplant team. Finally, some patients with encephalopathy prior to OLT have persistent cognitive deficits long after OLT.³²

Drug and alcohol recidivism are common post-OLT and typically occurs in about 20% of patients. It is important that active steps are taken to avoid recidivism immediately after OLT. Long-term psychiatric care and continued attendance at support groups help maintain sobriety. The important contributions you can make are maintaining a heightened awareness for recidivism, communicating with patients regularly about drug and alcohol abuse, and providing support and referral services.

Socioeconomic problems

While most transplant recipients maintain a good quality of life, some have long-term socioeconomic problems. One study showed that only one third of OLT recipients returned successfully to work, just slightly higher than the percentage working before OLT.³³ The economic situation improved in 11.9% of the recipients, worsened in 33.9%, and stayed the same in 54.2%. Concurrent illness, prolonged inactivity, psychiatric disorders, and the level of physical requirements at work are the main contributing factors to unemployment.

Another major stressor is medical cost. The average cost of immunosuppressive medications alone is $10,000 to $20,000 per year.³ Most of the charges are reimbursable, although this depends on the payer and time from transplantation. Medicare pays for immunosuppressive medications for only 36 months. Beyond that point, patients will require secondary insurance or other assistance. This expenditure is exacerbated by the cost of other medications, clinic visits with the transplant center, family physicians, and specialists, and time away from work.

Although patients are usually well informed of these concerns before OLT, they often do not appreciate the financial magnitude until after OLT. Encourage patients to return to work, stay active physically and mentally, and prepare for these financial considerations.

Sexual issues

Some patients have persistent sexual dysfunction that may have an organic basis (cardiovascular, renal, liver, endocrine) requiring investigation. The safety and efficacy of sildenafil (Viagra) in OLT recipients has not been investigated to date.

However, other patients regain their libido and gonadal function immediately after OLT; pregnancy may occur in this period. Advise patients to wait at least 2 years post-OLT before considering pregnancy (SOR: C). ³⁴ Contraception, preferably barrier-type, should be used during sexual intercourse. Hormonal contraceptives are not contraindicated but should probably not be administered until the patient ’ s transplant status is stable.

If pregnancy does occur, apprise the patient of potential complications and adverse outcomes. Hypertension and preeclampsia are more common in pregnant OLT recipients; life-threatening infections and acute rejection are rare. Fortunately, most patients deliver healthy babies; miscarriages, stillbirths, and malformations are uncommon. An obstetrician specializing in high-risk pregnancy should follow all pregnant OLT recipients.

Vaccination

Vaccination after OLT is controversial. Live vaccines are generally contraindicated post-OLT and their safety in patients with stable graft function and on low levels of immunosuppression is unclear (SOR: C). Patients should receive pneumococcal vaccination, hepatitis A and B vaccination if not already immune, and yearly influenza vaccination (SOR: C). For travel outside of the US or in uncertain situations or exposures, the best reference is the Centers for Disease Control web site: www.cdc.org.

Communication with the transplant center

Direct communication with the patient ’ s transplant center is extremely important. You and the transplant center should determine the most effective way (phone, fax or e-mail) to communicate.

When should you contact the transplant center? First, obtain the center ’ s approval for any new medications that may be used long-term or have the potential for nephrotoxicity, hepatotoxicity, or immunosuppression. Second, notify the transplant center in the event of new signs or symptoms, such as fever, weight loss, abdominal pain, or jaundice. Being cautious by communicating early is often the most prudent course. Third, alert the transplant center of any hospitalizations. Transfer to the transplant center for any transplant-related problem or prolonged hospitalization usually provides the best outcome for the patient.

On the flip side, you are the primary caretaker, and the transplant center should regularly communicate with you regarding general medical concerns and any new diagnoses, interventions, or treatments. The transplant center should also regularly communicate with you regarding general medical concerns and any new diagnoses, interventions, or treatments. A strong, mutual relationship between you and the transplant center will have great impact on the recipient ’ s long-term care.

CORRESPONDENCE
Josh Levitsky, MD, Northwestern Memorial Hospital, 675 North St. Clair St, Suite 15-250, Chicago, IL 60611. E-mail: [email protected]

Orthotopic liver transplantation (OLT) is the replacement of a whole diseased liver with a healthy donor liver. The number of persons receiving OLT is increasing. Though it is unlikely you will be involved in the care of a patient immediately after OLT, you ’ ll need to know about the complications that occur in this period as they may impact the long-term care of the patient.

Long-term issues — such as cardiovascular disease, bone disease, malignancy, anemia, psychiatric disorders, and financial stressors — put these patients at higher risk for problems more than the average patient. Perhaps the most important task is for you to keep in contact with the transplant center when questions or concerns arise. Over time, you will once again become the primary physician and advocate for these patients.

Complications after transplant (less than 1 year)

Within 1 month post-OLT, the most frequent complications are acute graft rejection, vascular thrombosis, biliary leak or stricture, and infection. Between 1 and 3 months, acute and chronic graft rejection can occur, but medication toxicity and opportunistic infections become more common (TABLE 1). ¹

A broad range of infections may develop, including cytomegalovirus, Epstein-Barr virus, herpes simplex virus, varicella zoster virus, adenovirus, tuberculosis, Pneumocystis , toxoplasmosis, Listeria spp, Candida spp, Aspergillus spp, and Cryptococcus spp. During this time, doses of immunosuppressive agents are lowered and corticosteroids are discontinued in many patients.

Once patients are considered stable after OLT, they will likely come under your supervision again. While opportunistic infections, surgical issues, and acute rejection become less common between 3 and 12 months, other complications related to OLT may occur.

Graft reinfection with hepatitis C virus (HCV) is universal, and 50% to 80% patients will develop biopsy-proven hepatitis. ² Many will require treatment for recurrent HCV to avoid progression to cirrhosis.

Recurrent hepatitis B infection is much less common due to prophylactic therapy with hepatitis B immunoglobulin and antiviral medications, although 10% of transplant recipients will develop hepatitis despite prophylaxis.

Other causes of recurrent liver disease post-OLT include liver injury due to recurrent drug or alcohol abuse, non-alcoholic steatohepatitis, cholestatic and autoimmune liver disease, and liver cancer.

Toxicity due to immunosuppressive medications is also common in this time frame (TABLE 2). ³ Be alert to the potential for hepatotoxicity and drug interactions with any new pharmacologic agent. Other drugs (eg, lipid-lowering agents, antibiotics, antifungals) may cause liver injury on their own and need to be closely monitored.

Lastly, even though patients are at increased risk for such common infections as influenza, pneumonia, and urinary tract infections, opportunistic infections are uncommon in this period. Keep in mind that patients usually develop infections that are community-acquired and not opportunistic, particularly as time goes on.

TABLE 1
Common complications immediately after liver transplantation

TABLE 2
Immunosuppressive medications and interactions after liver transplantation

Long-term complications

Cardiovascular disease

Up to 20% of late deaths after OLT are caused by cardiovascular disease. ⁴ Uncontrollable factors, such as preexisting cardiac disease, male sex, family history of cardiac disease, and advanced age contribute to the incidence of cardiovascular disease. However, a number of potentially controllable factors, such as hypertension, hyperlipidemia, obesity, and diabetes are common after OLT and should be addressed.

Hypertension. Hypertension occurs in 40% to 75% of OLT patients. ⁵ Causes include calcineurin-inhibitor (cyclosporine, tacrolimus) therapy, high-dose corticosteroids, and renal insufficiency. Calcineurin inhibitors cause renal vasoconstriction, leading to sodium retention and hypertension. Reducing the doses of these medications by the transplant center typically improves blood pressure control.

Treatment of choice for hypertension depends on how recently the transplant was performed. In the first 6 months following the procedure, dihydropyridine calcium-channel blockers (eg, amlodipine) and alpha-blockers are the mainstay of therapy, although peripheral edema and orthostatic hypotension may affect their tolerability. Diuretics can also be used in volume-overloaded patients.

After 6 months, other pharmacologic agents, such as angiotensin-converting enzyme (ACE) inhibitors and beta-blockers, can be administered to patients with stable renal function and without other contraindications (strength of recommendation [SOR]: C ). Long-term management of hypertension does not differ significantly from that in non-transplant patients.

Hyperlipidemia/obesity. Obesity and hyperlipidemia may affect up to half of OLT patients. Factors that contribute to both disorders include immunosuppressive drugs, increased appetite, diabetes, pretransplant hyperlipidemia, and history of cholestatic liver disease.

For hyperlipidemia, lifestyle modifications, such as diet and exercise, are recommended. If these measures are ineffective, statins are first-line agents. Avoid bile acid binding resins, which may interfere with the absorption of all medications. For refractory cases, switching from cyclosporine to tacrolimus under the direction of the transplant center might be indicated.

Treatment of obesity following OLT should also focus on lifestyle changes, as the safety of pharmacotherapy and surgery for obesity is uncertain in these patients.

Glucose intolerance and diabetes. Many patients will have glucose intolerance that resolves after steroid withdrawal. Main risk factors are pre-OLT diabetes, episodes of steroid-resistant rejection, and obesity. Post-OLT onset of diabetes will persist for only a small percentage of patients. ⁶

Treatment of post-transplant diabetes is similar to that for any patient. Insulin is often required initially, but with reduction in immunosuppression and corticosteroids, patients can usually be switched to oral agents. Though there is no absolute contraindication to using any antidiabetes medications, most physicians try to avoid those with potential hepatotoxicity, such as the thiazolidinediones (SOR: C).

Weight loss is critical and often improves glucose tolerance. Transplant centers may switch patients from tacrolimus to cyclosporine to control hyperglycemia. Long-term screening for end organ complications (retinopathy, nephropathy, neuropathy) is as important for this population as it is for in non-transplant diabetics.

FIGURE
Indications for liver transplantation in the US

Renal disease

Up to 20% of OLT recipients develop end-stage renal disease, requiring hemodialysis or renal transplantation within 10 years after transplant. ⁵ If patients have renal dysfunction before OLT, lower-dose calcineurin inhibitors and using alternative immunosuppression post-OLT may improve renal function in the long term. A rise in creatinine in the first year after OLT is a strong risk factor for long-term development of renal insufficiency, while stable creatinine levels at 1 year usually indicate long-term maintenance of renal function.^9,10

Closely monitor patients with early renal dysfunction, avoid nephrotoxic agents, and reduce or withdraw calcineurin inhibitors (as directed by the transplant center). Occasionally renal transplantation will be indicated.

Be aware that all OLT recipients need adequate hydration during acute illnesses (influenza, common colds, gastroenteritis), especially if they have renal dysfunction. Potential nephrotoxic agents, such as nonsteroidal anti-inflammatory drugs (NSAIDs), aminoglycoside antibiotics, and intravenous contrast, should be avoided if possible.

Bone disease

Osteoporosis should be screened for and identified before OLT. Contributing factors for bone disease after transplant include preexisting osteoporosis, immobility, vitamin D deficiency, corticosteroid use, and hypogonadism. In the first 6 months after transplant, bone mineral density (BMD) significantly declines, often accelerated by immunosuppressive medications, corticosteroids, and immobility. ^{11 - 14} After 6 months, BMD increases rapidly and, by 12 months, approaches pre-OLT values. All patients should have bone densitometry performed before OLT or before hospital discharge and receive calcium (1500 mg/d) and vitamin D (800 IU/d) supplementation (SOR: C).

Unless significant risk factors for osteoporosis are present (eg, continued use of corticosteroids, history of bone loss, fracture, or cholestatic liver disease), it is unclear whether low-risk patients should have serial bone densitometry tests performed in the years following OLT. Patients with T-scores ≥ 2 standard deviations below mean should be considered for antiresorptive therapy. Given the recent concerns regarding estrogen use and cardiovascular disease, bisphosphonates and calcitonin are preferred. For patients who develop fractures or avascular necrosis from corticosteroids, joint replacement surgery appears to be safe and effective post-OLT (SOR: B).^15,16

Malignancy

Of all of the complications following OLT, malignancy causes the highest morbidity and mortality. The overall incidence of malignancy is between 2.3% and 12.9% and may be up to 5 times higher than in the general population.^17,18 The most common malignancies are post-transplant lymphoproliferative disorder (1% – 4.4%) and nonmelanoma skin cancer (0.5% – 4.3%); less common are gastrointestinal (0.4% – 1.0%), genitourinary (0.2% – 2.2%), lung (0.2% – 0.8%), and oropharyngeal (0.4% – 0.8%) malignancies.¹⁷ Many patients with small liver cancers (1 lesion < 5 cm or up to 3 lesions each < 3 cm) are receiving transplants and, despite the risk of recurrence post-OLT, have a similar survival rate as patients receiving OLT for other indications.

Though it is clear that OLT recipients are at higher risk than the general population for malignancy, there are no specific guidelines for screening. However, based on the risk of early malignancy, screening should resume within the first 2 to 3 years after OLT (SOR: B).^{19- 22} Most transplant centers will recommend either performing a more intense screening protocol than for non-transplant patients or individualizing screening protocols for each patient depending on risk factors.

Advise patients who spend time in the sun to wear sun block with a protective factor > 40 and to have routine skin examinations. It is unclear whether colorectal cancer screening in OLT recipients should occur more frequently than the general population. Colorectal adenomas may be more common among OLT recipients than among healthy controls, ²³ but until more data are available, screening should mirror that of the general population (SOR: B). Since hepatocellular carcinoma may recur after OLT, transplant centers typically request imaging (computed tomography, ultrasound, magnetic resonance imaging) at regular intervals after OLT.

Use your discretion when screening for other common malignancies, such as breast, cervical, and prostate cancer. It is unclear whether screening specific groups of patients (such as tobacco smokers) for oropharyngeal, lung, and genitourinary cancer will be cost-effective or impact survival.

Anemia

The prevalence of anemia after OLT reportedly is between 4.3% and 28.2%, depending on the population studied and time after transplantation. ^{24 , 25} Blood loss, sepsis, medications, renal dysfunction, or hypersplenism can contribute to immediate postoperative anemia. Beyond the immediate postoperative period, anemia may be related to different causes (TABLE 3). ²⁶ Medication-induced anemia is usually related to bone marrow suppression, although calcineurin inhibitors may cause microangiopathic hemolysis, hemolyticuremic syndrome, or pure redcell aplasia.

Viral infections often cause anemia in the first 12 weeks after transplantation. Aplastic anemia may be related to parvovirus B19 infection, although it is more commonly seen in patients who undergo liver transplantation for acute liver failure.^24,27 Posttransplant lymphoproliferative disorder ranges from polyclonal B-cell hyperplasia (related to Epstein-Barr virus) that responds to reduction in immunosuppression to aggressive lymphoma treated with high dose chemotherapy.

Graft-versus-host disease is a rare but important cause of pancytopenia after OLT and is diagnosed by establishing chimerism, donor and recipient lymphocytes, in the blood and bone marrow; mortality is high.²⁸

Lastly, renal failure and iron deficiency are other common causes of anemia after OLT that warrant investigation. Despite complete evaluation, half of adult patients do not have an identifiable cause of anemia and may respond to a therapeutic trial of erythropoietin (SOR: C).²⁶

TABLE 3
Evaluation of anemia after liver transplantation

Psychosocial and socioeconomic concerns

Liver transplantation is a tremendously stressful and life-altering procedure affecting patients and their families. In the initial postoperative period, the stress of the operation and other factors (immunosuppression, infection, prolonged hospital stay) can lead to a variety of psychiatric disorders, such as delirium, anxiety, depression, mania, and psychosis. A multidisciplinary approach, including psychiatry, social work, and nursing care, is required to help the patients and families through this period, as expectations for full recovery may be delayed by psychiatric conditions.

Psychiatric problems

Many transplant recipients have long-term psychiatric problems. Depression and anxiety diminish quality of life, particularly for patients whose transplant was for hepatitis C and those with post-transplant viral recurrence.^29,30 Most patients will respond to antidepressants and ongoing psychiatric care. The side-effect profile should be individualized for each patient, keeping in mind the potential interactions with the current medications.

Mania and hypomania, while less common than depression, are often related to higher doses of immunosuppression (eg, corticosteroids). Cyclosporine may increase lithium levels, leading to toxicity.³¹ Treatment with anticonvulsant medications, such as carbamazepine, may decrease calcineurin-inhibitor levels and should be monitored in coordination with the transplant team. Finally, some patients with encephalopathy prior to OLT have persistent cognitive deficits long after OLT.³²

Drug and alcohol recidivism are common post-OLT and typically occurs in about 20% of patients. It is important that active steps are taken to avoid recidivism immediately after OLT. Long-term psychiatric care and continued attendance at support groups help maintain sobriety. The important contributions you can make are maintaining a heightened awareness for recidivism, communicating with patients regularly about drug and alcohol abuse, and providing support and referral services.

Socioeconomic problems

While most transplant recipients maintain a good quality of life, some have long-term socioeconomic problems. One study showed that only one third of OLT recipients returned successfully to work, just slightly higher than the percentage working before OLT.³³ The economic situation improved in 11.9% of the recipients, worsened in 33.9%, and stayed the same in 54.2%. Concurrent illness, prolonged inactivity, psychiatric disorders, and the level of physical requirements at work are the main contributing factors to unemployment.

Another major stressor is medical cost. The average cost of immunosuppressive medications alone is $10,000 to $20,000 per year.³ Most of the charges are reimbursable, although this depends on the payer and time from transplantation. Medicare pays for immunosuppressive medications for only 36 months. Beyond that point, patients will require secondary insurance or other assistance. This expenditure is exacerbated by the cost of other medications, clinic visits with the transplant center, family physicians, and specialists, and time away from work.

Although patients are usually well informed of these concerns before OLT, they often do not appreciate the financial magnitude until after OLT. Encourage patients to return to work, stay active physically and mentally, and prepare for these financial considerations.

Sexual issues

Some patients have persistent sexual dysfunction that may have an organic basis (cardiovascular, renal, liver, endocrine) requiring investigation. The safety and efficacy of sildenafil (Viagra) in OLT recipients has not been investigated to date.

However, other patients regain their libido and gonadal function immediately after OLT; pregnancy may occur in this period. Advise patients to wait at least 2 years post-OLT before considering pregnancy (SOR: C). ³⁴ Contraception, preferably barrier-type, should be used during sexual intercourse. Hormonal contraceptives are not contraindicated but should probably not be administered until the patient ’ s transplant status is stable.

If pregnancy does occur, apprise the patient of potential complications and adverse outcomes. Hypertension and preeclampsia are more common in pregnant OLT recipients; life-threatening infections and acute rejection are rare. Fortunately, most patients deliver healthy babies; miscarriages, stillbirths, and malformations are uncommon. An obstetrician specializing in high-risk pregnancy should follow all pregnant OLT recipients.

Vaccination

Vaccination after OLT is controversial. Live vaccines are generally contraindicated post-OLT and their safety in patients with stable graft function and on low levels of immunosuppression is unclear (SOR: C). Patients should receive pneumococcal vaccination, hepatitis A and B vaccination if not already immune, and yearly influenza vaccination (SOR: C). For travel outside of the US or in uncertain situations or exposures, the best reference is the Centers for Disease Control web site: www.cdc.org.

Communication with the transplant center

Direct communication with the patient ’ s transplant center is extremely important. You and the transplant center should determine the most effective way (phone, fax or e-mail) to communicate.

When should you contact the transplant center? First, obtain the center ’ s approval for any new medications that may be used long-term or have the potential for nephrotoxicity, hepatotoxicity, or immunosuppression. Second, notify the transplant center in the event of new signs or symptoms, such as fever, weight loss, abdominal pain, or jaundice. Being cautious by communicating early is often the most prudent course. Third, alert the transplant center of any hospitalizations. Transfer to the transplant center for any transplant-related problem or prolonged hospitalization usually provides the best outcome for the patient.

On the flip side, you are the primary caretaker, and the transplant center should regularly communicate with you regarding general medical concerns and any new diagnoses, interventions, or treatments. The transplant center should also regularly communicate with you regarding general medical concerns and any new diagnoses, interventions, or treatments. A strong, mutual relationship between you and the transplant center will have great impact on the recipient ’ s long-term care.

CORRESPONDENCE
Josh Levitsky, MD, Northwestern Memorial Hospital, 675 North St. Clair St, Suite 15-250, Chicago, IL 60611. E-mail: [email protected]

The objective of this review was to determine how well pulse palpation detects the classical sign of atrial fibrillation—an irregularly irregular pulse—and to estimate the test’s false-positive and false-negative rates. You may want to consider using pulse palpation opportunistically with certain high-risk patients identified in this article.

The scope of the problem. Atrial fibrillation in the elderly is common and potentially life-threatening. Its prevalence increases from 2% in the 65- to 69-year-old age group to 8% among those older than 85 years.¹

Atrial fibrillation is an independent risk factor for stroke, increasing a person’s risk 3- to 5-fold. While other major risk factors for stroke, such as hypertension, become less important with age, the risk of stroke from atrial fibrillation increases with age.

Stroke attributable to atrial fibrillation increases from 1.5% among patients 50 to 59 years to 23.5% for those aged 80 to 89.² Furthermore, stroke in the presence of atrial fibrillation is almost twice as likely to be fatal, results in more functional impairment among survivors, and recurs frequently.³

Stroke due to atrial fibrillation results in considerable mortality and morbidity, but treatment of atrial fibrillation reduces the risk. Warfarin reduces the risk of stroke in non-rheumatic atrial fibrillation by about 70%, an annual absolute risk reduction of 3% (number needed to treat: 30), though the benefits are somewhat offset by the risk of bleeding (annual risk of 0.6%).⁴

Methods

Search strategy and study selection

We searched the Medline and EMBASE electronic databases from 1966 to June 2005 for all studies comparing pulse palpation with ECG diagnosis. There was no restriction on the language of publication.

The search strategy (see Search strategy and inclusion criteria) included terms for pulse, atrial fibrillation, and a search filter for studies of diagnostic accuracy developed by Van der Weijden et al⁵ that is highly sensitive and precise.⁶ We also searched the reference lists of all possibly relevant studies, including review articles. We included all studies that tested patients with both pulse palpation and ECG, and that provided data suitable for calculating sensitivity or specificity (or could be provided by the study authors).

Two reviewers (GC and JD) independently screened the electronic lists of citations. We obtained full-text documents of relevant papers and the 2 reviewers independently reviewed the full-text articles for inclusion according to predetermined criteria (see Search strategy and inclusion criteria). The reviewers agreed on all inclusions.

Data extraction and analysis

We assessed study quality with the QUADAS tool.⁷ Two reviewers assessed study quality and extracted data independently. We calculated the pooled sensitivity and specificity using Metatest (New England Medical Center, Boston, Mass) and the pooled positive and negative likelihood ratios using the Der-Simonian and Laird random effects model of Revman 4.29 (Cochrane Collaboration), which also tests for heterogeneity.

Results

Search results and selection

The search strategy identified 110 citations. Six studies were selected for further assessment for inclusion; 3 met the inclusion criteria and covered 2385 patients (FIGURE). All 3 studies included patients older than 65 years enrolled from general practice clinics, and in each study the pulse was taken by a nurse in the clinic. TABLE 1 summarizes the methodological quality, patient characteristics, and results of the 3 studies.

FIGURE
Studies included in this review

TABLE 1
Detecting atrial fibrillation by pulse palpation: Pertinent studies

Methodological quality of included studies

The study by Morgan and Mant⁸ fulfilled all the major quality criteria. The studies by Sudlow et al⁹ and Somerville et al¹⁰ were of moderate quality. Sudlow et al did not report an overall sensitivity or specificity, nor the data for calculating these test attributes. The authors kindly provided us the data from their study to calculate the 2×2 table of results.

Prevalence of atrial fibrillation

We calculated the pooled prevalence of atrial fibrillation, excluding data from Somerville et al, as it was a case-control study. The pooled prevalence of atrial fibrillation in the 2 remaining studies was 5.3%, consistent with previous estimates of atrial fibrillation in general practice settings and data from the Framingham cohort.^1,11

Accuracy of pulse palpation for detecting atrial fibrillation

The sensitivities in the 3 studies ranged from 91% to 100% and the specificities ranged from 70% to 77%. The pooled sensitivity was 94% (95% CI, 84%–97%) and the pooled specificity was 72% (95% CI, 69%–75%). Low specificity reflects the pervasiveness of other pulse abnormalities, resulting in false-positive test results for atrial fibrillation (TABLE 2).

The pooled positive likelihood ratio was 3.4 (95% CI, 3.2–3.7), showing that pulse palpation was only moderately helpful for ruling in atrial fibrillation. The pooled negative likelihood ratio was 0.11 (95% CI, 0.06–0.20), showing that when no pulse irregularity is detected, the diagnosis of atrial fibrillation can be excluded with reasonable confidence. The chisquare test for heterogeneity for the positive likelihood ratio was 3.87 (df=2; P=.14) and for the negative likelihood ratio was 1.49 (df=2; P=.49), indicating consistency of results between the studies.

TABLE 2
Causes of an irregular pulse

Discussion

Though an irregularly irregular pulse is a classical clinical sign, we were able to find only 3 published studies evaluating its accuracy in the detection of atrial fibrillation. Study results showed a relatively high sensitivity but poor specificity.

Each study compared the diagnostic accuracy of pulse palpation by nurses in general practice against the reference test, ECG diagnosis by a cardiologist. No information was available on interobserver reliability of pulse palpation.

Working toward reliable screening. Atrial fibrillation is common, serious, and treatable. The initial clinical manifestation of undetected atrial fibrillation might be a dire consequence of the arrhythmia, such as stroke. If a suitably accurate diagnostic test or series of tests were available, it might be effective and cost-effective to screen for the disorder.¹² A randomized controlled trial to evaluate such screening (including a comparison of diagnostic strategies) is currently underway.¹³

Whom to screen in the meantime. It may be worthwhile considering opportunistic testing for atrial fibrillation particularly in patients with ischemic heart disease or heart failure, who are at particularly high risk. Even among these patients, cases of paroxysmal atrial fibrillation may be missed, whether using pulse palpation or ECG. You may therefore want to repeat the test periodically.

CORRESPONDENCE
Jenny Doust, Level 2 Edith Cavell Building, Royal Brisbane Hospital Complex, Herston, Queensland 4029, Australia. E-mail: [email protected]. Previous presentation of work: GP and PHC RED Research Conference 2-4 June 2004, Brisbane, Australia

The objective of this review was to determine how well pulse palpation detects the classical sign of atrial fibrillation—an irregularly irregular pulse—and to estimate the test’s false-positive and false-negative rates. You may want to consider using pulse palpation opportunistically with certain high-risk patients identified in this article.

The scope of the problem. Atrial fibrillation in the elderly is common and potentially life-threatening. Its prevalence increases from 2% in the 65- to 69-year-old age group to 8% among those older than 85 years.¹

Atrial fibrillation is an independent risk factor for stroke, increasing a person’s risk 3- to 5-fold. While other major risk factors for stroke, such as hypertension, become less important with age, the risk of stroke from atrial fibrillation increases with age.

Stroke attributable to atrial fibrillation increases from 1.5% among patients 50 to 59 years to 23.5% for those aged 80 to 89.² Furthermore, stroke in the presence of atrial fibrillation is almost twice as likely to be fatal, results in more functional impairment among survivors, and recurs frequently.³

Stroke due to atrial fibrillation results in considerable mortality and morbidity, but treatment of atrial fibrillation reduces the risk. Warfarin reduces the risk of stroke in non-rheumatic atrial fibrillation by about 70%, an annual absolute risk reduction of 3% (number needed to treat: 30), though the benefits are somewhat offset by the risk of bleeding (annual risk of 0.6%).⁴

Methods

Search strategy and study selection

We searched the Medline and EMBASE electronic databases from 1966 to June 2005 for all studies comparing pulse palpation with ECG diagnosis. There was no restriction on the language of publication.

The search strategy (see Search strategy and inclusion criteria) included terms for pulse, atrial fibrillation, and a search filter for studies of diagnostic accuracy developed by Van der Weijden et al⁵ that is highly sensitive and precise.⁶ We also searched the reference lists of all possibly relevant studies, including review articles. We included all studies that tested patients with both pulse palpation and ECG, and that provided data suitable for calculating sensitivity or specificity (or could be provided by the study authors).

Two reviewers (GC and JD) independently screened the electronic lists of citations. We obtained full-text documents of relevant papers and the 2 reviewers independently reviewed the full-text articles for inclusion according to predetermined criteria (see Search strategy and inclusion criteria). The reviewers agreed on all inclusions.

Data extraction and analysis

We assessed study quality with the QUADAS tool.⁷ Two reviewers assessed study quality and extracted data independently. We calculated the pooled sensitivity and specificity using Metatest (New England Medical Center, Boston, Mass) and the pooled positive and negative likelihood ratios using the Der-Simonian and Laird random effects model of Revman 4.29 (Cochrane Collaboration), which also tests for heterogeneity.

Results

Search results and selection

The search strategy identified 110 citations. Six studies were selected for further assessment for inclusion; 3 met the inclusion criteria and covered 2385 patients (FIGURE). All 3 studies included patients older than 65 years enrolled from general practice clinics, and in each study the pulse was taken by a nurse in the clinic. TABLE 1 summarizes the methodological quality, patient characteristics, and results of the 3 studies.

FIGURE
Studies included in this review

TABLE 1
Detecting atrial fibrillation by pulse palpation: Pertinent studies

Methodological quality of included studies

The study by Morgan and Mant⁸ fulfilled all the major quality criteria. The studies by Sudlow et al⁹ and Somerville et al¹⁰ were of moderate quality. Sudlow et al did not report an overall sensitivity or specificity, nor the data for calculating these test attributes. The authors kindly provided us the data from their study to calculate the 2×2 table of results.

Prevalence of atrial fibrillation

We calculated the pooled prevalence of atrial fibrillation, excluding data from Somerville et al, as it was a case-control study. The pooled prevalence of atrial fibrillation in the 2 remaining studies was 5.3%, consistent with previous estimates of atrial fibrillation in general practice settings and data from the Framingham cohort.^1,11

Accuracy of pulse palpation for detecting atrial fibrillation

The sensitivities in the 3 studies ranged from 91% to 100% and the specificities ranged from 70% to 77%. The pooled sensitivity was 94% (95% CI, 84%–97%) and the pooled specificity was 72% (95% CI, 69%–75%). Low specificity reflects the pervasiveness of other pulse abnormalities, resulting in false-positive test results for atrial fibrillation (TABLE 2).

The pooled positive likelihood ratio was 3.4 (95% CI, 3.2–3.7), showing that pulse palpation was only moderately helpful for ruling in atrial fibrillation. The pooled negative likelihood ratio was 0.11 (95% CI, 0.06–0.20), showing that when no pulse irregularity is detected, the diagnosis of atrial fibrillation can be excluded with reasonable confidence. The chisquare test for heterogeneity for the positive likelihood ratio was 3.87 (df=2; P=.14) and for the negative likelihood ratio was 1.49 (df=2; P=.49), indicating consistency of results between the studies.

TABLE 2
Causes of an irregular pulse

Discussion

Though an irregularly irregular pulse is a classical clinical sign, we were able to find only 3 published studies evaluating its accuracy in the detection of atrial fibrillation. Study results showed a relatively high sensitivity but poor specificity.

Each study compared the diagnostic accuracy of pulse palpation by nurses in general practice against the reference test, ECG diagnosis by a cardiologist. No information was available on interobserver reliability of pulse palpation.

Working toward reliable screening. Atrial fibrillation is common, serious, and treatable. The initial clinical manifestation of undetected atrial fibrillation might be a dire consequence of the arrhythmia, such as stroke. If a suitably accurate diagnostic test or series of tests were available, it might be effective and cost-effective to screen for the disorder.¹² A randomized controlled trial to evaluate such screening (including a comparison of diagnostic strategies) is currently underway.¹³

Whom to screen in the meantime. It may be worthwhile considering opportunistic testing for atrial fibrillation particularly in patients with ischemic heart disease or heart failure, who are at particularly high risk. Even among these patients, cases of paroxysmal atrial fibrillation may be missed, whether using pulse palpation or ECG. You may therefore want to repeat the test periodically.

CORRESPONDENCE
Jenny Doust, Level 2 Edith Cavell Building, Royal Brisbane Hospital Complex, Herston, Queensland 4029, Australia. E-mail: [email protected]. Previous presentation of work: GP and PHC RED Research Conference 2-4 June 2004, Brisbane, Australia

The objective of this review was to determine how well pulse palpation detects the classical sign of atrial fibrillation—an irregularly irregular pulse—and to estimate the test’s false-positive and false-negative rates. You may want to consider using pulse palpation opportunistically with certain high-risk patients identified in this article.

The scope of the problem. Atrial fibrillation in the elderly is common and potentially life-threatening. Its prevalence increases from 2% in the 65- to 69-year-old age group to 8% among those older than 85 years.¹

Atrial fibrillation is an independent risk factor for stroke, increasing a person’s risk 3- to 5-fold. While other major risk factors for stroke, such as hypertension, become less important with age, the risk of stroke from atrial fibrillation increases with age.

Stroke attributable to atrial fibrillation increases from 1.5% among patients 50 to 59 years to 23.5% for those aged 80 to 89.² Furthermore, stroke in the presence of atrial fibrillation is almost twice as likely to be fatal, results in more functional impairment among survivors, and recurs frequently.³

Stroke due to atrial fibrillation results in considerable mortality and morbidity, but treatment of atrial fibrillation reduces the risk. Warfarin reduces the risk of stroke in non-rheumatic atrial fibrillation by about 70%, an annual absolute risk reduction of 3% (number needed to treat: 30), though the benefits are somewhat offset by the risk of bleeding (annual risk of 0.6%).⁴

Methods

Search strategy and study selection

We searched the Medline and EMBASE electronic databases from 1966 to June 2005 for all studies comparing pulse palpation with ECG diagnosis. There was no restriction on the language of publication.

The search strategy (see Search strategy and inclusion criteria) included terms for pulse, atrial fibrillation, and a search filter for studies of diagnostic accuracy developed by Van der Weijden et al⁵ that is highly sensitive and precise.⁶ We also searched the reference lists of all possibly relevant studies, including review articles. We included all studies that tested patients with both pulse palpation and ECG, and that provided data suitable for calculating sensitivity or specificity (or could be provided by the study authors).

Two reviewers (GC and JD) independently screened the electronic lists of citations. We obtained full-text documents of relevant papers and the 2 reviewers independently reviewed the full-text articles for inclusion according to predetermined criteria (see Search strategy and inclusion criteria). The reviewers agreed on all inclusions.

Data extraction and analysis

We assessed study quality with the QUADAS tool.⁷ Two reviewers assessed study quality and extracted data independently. We calculated the pooled sensitivity and specificity using Metatest (New England Medical Center, Boston, Mass) and the pooled positive and negative likelihood ratios using the Der-Simonian and Laird random effects model of Revman 4.29 (Cochrane Collaboration), which also tests for heterogeneity.

Results

Search results and selection

The search strategy identified 110 citations. Six studies were selected for further assessment for inclusion; 3 met the inclusion criteria and covered 2385 patients (FIGURE). All 3 studies included patients older than 65 years enrolled from general practice clinics, and in each study the pulse was taken by a nurse in the clinic. TABLE 1 summarizes the methodological quality, patient characteristics, and results of the 3 studies.

FIGURE
Studies included in this review

TABLE 1
Detecting atrial fibrillation by pulse palpation: Pertinent studies

Methodological quality of included studies

The study by Morgan and Mant⁸ fulfilled all the major quality criteria. The studies by Sudlow et al⁹ and Somerville et al¹⁰ were of moderate quality. Sudlow et al did not report an overall sensitivity or specificity, nor the data for calculating these test attributes. The authors kindly provided us the data from their study to calculate the 2×2 table of results.

Prevalence of atrial fibrillation

We calculated the pooled prevalence of atrial fibrillation, excluding data from Somerville et al, as it was a case-control study. The pooled prevalence of atrial fibrillation in the 2 remaining studies was 5.3%, consistent with previous estimates of atrial fibrillation in general practice settings and data from the Framingham cohort.^1,11

Accuracy of pulse palpation for detecting atrial fibrillation

The sensitivities in the 3 studies ranged from 91% to 100% and the specificities ranged from 70% to 77%. The pooled sensitivity was 94% (95% CI, 84%–97%) and the pooled specificity was 72% (95% CI, 69%–75%). Low specificity reflects the pervasiveness of other pulse abnormalities, resulting in false-positive test results for atrial fibrillation (TABLE 2).

The pooled positive likelihood ratio was 3.4 (95% CI, 3.2–3.7), showing that pulse palpation was only moderately helpful for ruling in atrial fibrillation. The pooled negative likelihood ratio was 0.11 (95% CI, 0.06–0.20), showing that when no pulse irregularity is detected, the diagnosis of atrial fibrillation can be excluded with reasonable confidence. The chisquare test for heterogeneity for the positive likelihood ratio was 3.87 (df=2; P=.14) and for the negative likelihood ratio was 1.49 (df=2; P=.49), indicating consistency of results between the studies.

TABLE 2
Causes of an irregular pulse

Discussion

Though an irregularly irregular pulse is a classical clinical sign, we were able to find only 3 published studies evaluating its accuracy in the detection of atrial fibrillation. Study results showed a relatively high sensitivity but poor specificity.

Each study compared the diagnostic accuracy of pulse palpation by nurses in general practice against the reference test, ECG diagnosis by a cardiologist. No information was available on interobserver reliability of pulse palpation.

Working toward reliable screening. Atrial fibrillation is common, serious, and treatable. The initial clinical manifestation of undetected atrial fibrillation might be a dire consequence of the arrhythmia, such as stroke. If a suitably accurate diagnostic test or series of tests were available, it might be effective and cost-effective to screen for the disorder.¹² A randomized controlled trial to evaluate such screening (including a comparison of diagnostic strategies) is currently underway.¹³

Whom to screen in the meantime. It may be worthwhile considering opportunistic testing for atrial fibrillation particularly in patients with ischemic heart disease or heart failure, who are at particularly high risk. Even among these patients, cases of paroxysmal atrial fibrillation may be missed, whether using pulse palpation or ECG. You may therefore want to repeat the test periodically.

CORRESPONDENCE
Jenny Doust, Level 2 Edith Cavell Building, Royal Brisbane Hospital Complex, Herston, Queensland 4029, Australia. E-mail: [email protected]. Previous presentation of work: GP and PHC RED Research Conference 2-4 June 2004, Brisbane, Australia

Undoubtedly you have patients who say they are allergic to penicillin but have difficulty recalling details of the reactions they experienced. To be safe, we often label these patients as penicillin-allergic without further questioning and withhold not only penicillins but cephalosporins due to concerns about potential cross-reactivity and resultant IgE-mediated, type I reactions. But even for patients truly allergic to penicillin, is the concern over cephalosporins justified? It depends on the specific agent. What is certain is that a blanket dismissal of all cephalosporins is unfounded.

The truth about the myth

Despite myriad studies spanning decades and involving varied patient populations, results have not conclusively established that penicillin allergy increases the risk of an allergic reaction to cephalosporins, compared with the incidence of a primary (and unrelated) cephalosporin allergy. Most people produce IgG and IgM antibodies in response to exposure to penicillin¹ that may cross-react with cephalosporin antigens.² The presence of these antibodies does not predict allergic, IgE cross-sensitivity to a cephalosporin. Even penicillin skin testing is generally not predictive of cephalosporin allergy.³

Reliably predicting cross-reactivity

A comprehensive review of the evidence shows that the attributable risk of a cross-reactive allergic reaction varies and is strongest when the chemical side chain of the specific cephalosporin is similar to that of penicillin or amoxicillin.

Administration of cephalothin, cephalexin, cefadroxil, and cefazolin in penicillin-allergic patients is associated with a significant increase in the rate of allergic reactions; whereas administration of cefprozil, cefuroxime, cefpodoxime, ceftazidime, and ceftriaxone is not.

Penicillin skin testing can accurately predict a penicillin-allergic reaction, but is not predictive for cephalosporin allergy unless the side chain of the penicillin or ampicillin testing reagent is similar to the cephalosporin side chain being evaluated. Patients who have a reaction to a penicillin or a cephalosporin that is not IgE mediated and not serious may receive repeated courses of that antibiotic and related antibiotics.

This article provides a comprehensive review of the frequency of allergic cross-reactivity between penicillin/amoxicillin and cephalosporin antibiotics, supporting the recent American Academy of Family Physicians evidence-based clinical practice guideline on treatment of acute otitis media recommending the use of cefuroxime, cefpodoxime, cefdinir, and ceftriaxone cephalosporins for patients allergic to penicillin.

Methods

We searched Medline and EMBASE databases for English-language articles using the keywords cephalosporin, penicillin, allergy, and cross-sensitivity for the years 1960 to 2005. Among 219 articles identified, 101 were included as source material for this review. Articles we excluded were reviews, republication of results, or ones irrelevant to our purpose.

Five articles described the rate of rashes following use of penicillin and cephalosporins,^4-8 and 4 articles described rates of anaphylaxis.^5,9-11 We included 26 articles for the evidence base evaluating penicillin/amoxicillin cross-allergy.^3,12-36 Eleven articles relied on patient history of penicillin/amoxicillin allergy to categorize results and establish reaction rates and relative risks for the penicillin/amoxicillin allergic vs nonallergic when receiving cephalosporins.^{12-15,17-20,27,28,31} Fourteen articles relied on patient history of penicillin/amoxicillin allergy plus skin testing results to penicillin/amoxicillin to categorize patients.^{16,21-25,29,30,32-37} One article³ provided data on a subset where penicillin/amoxicillin allergy was established based on history, and a separate subset where penicillin/amoxicillin allergy was established by skin testing. Other articles related to antibiotic chemical structures, animal studies, monoclonal antibody studies, cross-reactive antibody studies, and antibiotic skin testing were also reviewed.

Results

True incidence of reactions to cephalosporins

The most frequent reactions to cephalosporins are non-pruritic, non-urticarial rashes, which occur in 1.0% to 2.8% of patients;^4-8 for most, the mechanism is idiopathic and not a contraindication for future use.³⁸ Retrospective studies suggest a 1% to 3% incidence of immune or allergic reactions to cephalosporins independent of any history of penicillin/amoxicillin allergy.³¹ Anaphylactic reactions from cephalosporins are extremely rare, with the risk estimated at 0.0001% to 0.1%.^31,38 A seminal study suggested approximately 0.004% to 0.015% of treatment courses with penicillin results in anaphylaxis.^5,9-11 Several studies suggest that cephalosporin-induced anaphylaxis occurs no more frequently among patients with known penicillin allergy than among those without such allergy.^23,27,38-41

Determining cross-reactivity

Penicillins and cephalosporins both possess a beta-lactam ring for antimicrobial activity. They differ in that the 5-membered thiazolidine ring of penicillin is replaced in the cephalosporins with a 6-membered dihydrothiazine ring. After degradation, penicillin forms a stable ring, whereas cephalosporins undergo rapid fragmentation of their rings.⁴² Immunologic cross-reactivity between the penicillin and cephalosporin beta-lactam rings is, therefore, very unlikely—an observation confirmed by monoclonal antibody analysis.⁴³

How the “10% cross-reactivity” myth took hold. When the first-generation cephalosporins cephaloridine and cephalothin were introduced in the 1960s, allergic and anaphylactic reactions were reported in patients with previous allergic reactions to penicillins. Subsequent reports, which attributed up to 10% cross-reactivity between the 2 drug classes, involved these same first-generation cephalosporins plus cephalexin and cefadroxil and a second-generation drug, cefamandole. However, these studies were flawed because the penicillin test compounds had been contaminated with cephalosporins. Until 1982, penicillin was produced commercially using the cephalosporium mold.³⁸

Many recent studies have established that the rate of cross-reactivity between penicillin and cephalosporins has been grossly overestimated. In fact, the rate of cross-reactivity between penicillin/amoxicillin and second- or third-generation cephalosporins is very low and may actually be lower than that between penicillins and other classes of antibiotics.⁴⁴

The evidence for limited cross-reactivity. A summary of publications evaluating 38,846 children and adults with and without a history of penicillin allergy is presented in TABLE 1. The database included 2435 patients with a history of penicillin allergy and 961 patients with a history of penicillin allergy and positive skin-test results for penicillin or amoxicillin (total penicillin-allergic patients=3396). The allergic reaction rate is compared with 34,047 patients without a history of penicillin allergy and 1403 patients without a history of penicillin allergy and negative skin-test results for penicillin or amoxicillin (total penicillin-nonallergic patients=35,450).

When patients with a positive history of penicillin-allergy received first-generation cephalosporins, which share a chemical side chain similar to penicillin or amoxicillin (cephalothin, cephaloridine, cephalexin, cefadroxil, and cefazolin, plus the early second-generation cephalosporin, cefamandole), they exhibited a significant increased risk of an allergic reaction to the cephalosporin.

Second- and third-generation cephalosporins modified in size and the complexity of their side chains (eg, cefprozil, cefuroxime, ceftazidime, cefpodoxime, and ceftriaxone) were different enough from penicillin and ampicillin that they did not increase risk of allergic cross-reactivity (TABLE 1).

Many other studies have suggested that cross-reactive immune responses to cephalosporins depend on side chain structure;^{22,23,27,32,37,38,44-49} that is, cephalosporins with a 7-position side chain similar to benzylpenicillin are more likely to cross-react with penicillin (TABLE 2). Cephalosporins that share a similar 7-position or 3-position side chain are more likely to cross-react with each other.

Cephalosporin/penicillin cross-reactivity

Few studies have evaluated whether patients with primary hypersensitivity to cephalosporins will experience cross-reactivity with penicillin. Romano et al⁴⁹ conducted skin tests and RASTs in patients with immediate allergic reactions to cephalosporins to examine responses to other cephalosporins and to classic penicillin determinants. About 1 in 5 patients allergic to a cephalosporin reacted to penicillin determinants, while most had positive results to other cephalosporins with the same or similar side-chains.

Limitations of skin testing

Penicillin skin testing in patients with a history of penicillin allergy does not reliably predict allergy to a cephalosporin unless the side chain of the penicillin or ampicillin reagent is similar to the cephalosporin side chain being tested.³ The positive and negative predictive values of skin testing results for cephalosporins are not well established; if the haptens that cause cephalosporin allergy were known, cross-reactivity with penicillins could be assessed directly. Cephalosporin skin testing works only for the specific drug and drugs with the same side chains, and can be done only if the drug is available in an IV or IM formulation.

Even a positive result does not guarantee a clinical reaction. When penicillin and cephalosporin skin tests or radioallergosorbent tests (RASTs) are positive, a clinical reaction is observed in only 10% to 60% of patients, depending on the reagent and study.⁵⁰ For example, among 19 well-characterized patients allergic to penicillin who were studied for their sensitivity to the cephalosporins, cephaloridine and cefamandole (which have identical or very similar side chains to penicillin and were therefore potentially cross-reactive) only 2 (10.5%) reacted to cefamandole, while the other 17 patients tolerated both agents.²⁶ In another study of clinical cross-reactivity between amoxicillin and cefadroxil in patients allergic to amoxicillin with good tolerance of penicillin, only 12% had an immediate allergic reaction to cefadroxil, despite the 2 drugs sharing an identical side chain.³³ In a third study, allergenic cross-reactivity with cefadroxil and cefamandole was studied among 21 patients selectively allergic to amoxicillin; 8 (38%) had a positive response to cefadroxil (same side chain) and none to cefamandole (different side chain).³²

Discussion

Sensible approach to penicillin-allergic patients

Question patients who report penicillin allergy. In many cases, penicillin may not actually have been taken, or patients may have had non-immunologic adverse events such as vomiting, diarrhea, or nonspecific rash; toxic effects; or contemporaneous side effects inappropriately attributed to the drug. These patients can receive penicillin, amoxicillin, or the cephalosporins.

Without the ability to detect patients with IgE antibody to penicillin prospectively or to distinguish true IgE immunologic reactions from idiopathic reactions in patients receiving cephalosporins, it is impossible to definitely claim that increased immune or IgE-mediated reactions to cephalosporins occur in true penicillinallergic (IgE) patients.

When a cephalosporin is/is not safe for a penicillin-allergic patient. Only IgE-mediated reactions—such as anaphylaxis or hypotension, laryngeal edema, wheezing, angioedema, or urticaria—are likely to become more severe with time. Therefore, with a patient who has had a true IgE-mediated reaction to a penicillin, avoid using cephalosporins with a similar side chain. You may, however, give cephalosporins that have different side chains. Cephalosporins may also be used for patients who have had non-IgE-mediated adverse reactions (“non-type I allergy”)²¹ to a penicillin, such as a non-pruritic, non-urticarial morbilliform or maculopapular rash.

How prevalent is primary cephalosporin allergy? Even if the patient is not allergic to penicillin, cephalosporins can cause allergic or immune-mediated reactions in approximately 1% to 3% of patients. A patient who had an allergic reaction to a specific cephalosporin probably should not receive that cephalosporin again. The risk of a reaction with a different cephalosporin is very low to nonexistent if the side chains of the 2 drugs are dissimilar.

Bottom line. Penicillin-allergic patients have indeed shown an increased incidence of allergic reactions to cephalothin, cephaloridine, cephalexin, cefadroxil, cefazolin, and cefamandole. However, the risk has been overestimated because most studies reporting this cross-reactivity were flawed (because penicillins were contaminated with cephalosporins) and then failed to account for the fact that penicillin-allergic patients have a 3-fold increased risk of allergic reactions even to nonrelated drugs.⁵¹

For patients truly allergic to penicillin, the risk of a reaction from a cephalosporin with side chains that differ from penicillin/amoxicillin (cefuroxime, cefpodoxime, cefdinir, and ceftriaxone, as endorsed by the AAFP) is so low that use is justified and medico-legally defensible by the currently available evidence.

CORRESPONDENCE
Michael E. Pichichero, MD, University of Rochester Medical Center, 601 Elmwood Avenue, Box 672, Rochester, New York 14642. E-mail: [email protected]

Undoubtedly you have patients who say they are allergic to penicillin but have difficulty recalling details of the reactions they experienced. To be safe, we often label these patients as penicillin-allergic without further questioning and withhold not only penicillins but cephalosporins due to concerns about potential cross-reactivity and resultant IgE-mediated, type I reactions. But even for patients truly allergic to penicillin, is the concern over cephalosporins justified? It depends on the specific agent. What is certain is that a blanket dismissal of all cephalosporins is unfounded.

The truth about the myth

Despite myriad studies spanning decades and involving varied patient populations, results have not conclusively established that penicillin allergy increases the risk of an allergic reaction to cephalosporins, compared with the incidence of a primary (and unrelated) cephalosporin allergy. Most people produce IgG and IgM antibodies in response to exposure to penicillin¹ that may cross-react with cephalosporin antigens.² The presence of these antibodies does not predict allergic, IgE cross-sensitivity to a cephalosporin. Even penicillin skin testing is generally not predictive of cephalosporin allergy.³

Reliably predicting cross-reactivity

A comprehensive review of the evidence shows that the attributable risk of a cross-reactive allergic reaction varies and is strongest when the chemical side chain of the specific cephalosporin is similar to that of penicillin or amoxicillin.

Administration of cephalothin, cephalexin, cefadroxil, and cefazolin in penicillin-allergic patients is associated with a significant increase in the rate of allergic reactions; whereas administration of cefprozil, cefuroxime, cefpodoxime, ceftazidime, and ceftriaxone is not.

Penicillin skin testing can accurately predict a penicillin-allergic reaction, but is not predictive for cephalosporin allergy unless the side chain of the penicillin or ampicillin testing reagent is similar to the cephalosporin side chain being evaluated. Patients who have a reaction to a penicillin or a cephalosporin that is not IgE mediated and not serious may receive repeated courses of that antibiotic and related antibiotics.

This article provides a comprehensive review of the frequency of allergic cross-reactivity between penicillin/amoxicillin and cephalosporin antibiotics, supporting the recent American Academy of Family Physicians evidence-based clinical practice guideline on treatment of acute otitis media recommending the use of cefuroxime, cefpodoxime, cefdinir, and ceftriaxone cephalosporins for patients allergic to penicillin.

Methods

We searched Medline and EMBASE databases for English-language articles using the keywords cephalosporin, penicillin, allergy, and cross-sensitivity for the years 1960 to 2005. Among 219 articles identified, 101 were included as source material for this review. Articles we excluded were reviews, republication of results, or ones irrelevant to our purpose.

Five articles described the rate of rashes following use of penicillin and cephalosporins,^4-8 and 4 articles described rates of anaphylaxis.^5,9-11 We included 26 articles for the evidence base evaluating penicillin/amoxicillin cross-allergy.^3,12-36 Eleven articles relied on patient history of penicillin/amoxicillin allergy to categorize results and establish reaction rates and relative risks for the penicillin/amoxicillin allergic vs nonallergic when receiving cephalosporins.^{12-15,17-20,27,28,31} Fourteen articles relied on patient history of penicillin/amoxicillin allergy plus skin testing results to penicillin/amoxicillin to categorize patients.^{16,21-25,29,30,32-37} One article³ provided data on a subset where penicillin/amoxicillin allergy was established based on history, and a separate subset where penicillin/amoxicillin allergy was established by skin testing. Other articles related to antibiotic chemical structures, animal studies, monoclonal antibody studies, cross-reactive antibody studies, and antibiotic skin testing were also reviewed.

Results

True incidence of reactions to cephalosporins

The most frequent reactions to cephalosporins are non-pruritic, non-urticarial rashes, which occur in 1.0% to 2.8% of patients;^4-8 for most, the mechanism is idiopathic and not a contraindication for future use.³⁸ Retrospective studies suggest a 1% to 3% incidence of immune or allergic reactions to cephalosporins independent of any history of penicillin/amoxicillin allergy.³¹ Anaphylactic reactions from cephalosporins are extremely rare, with the risk estimated at 0.0001% to 0.1%.^31,38 A seminal study suggested approximately 0.004% to 0.015% of treatment courses with penicillin results in anaphylaxis.^5,9-11 Several studies suggest that cephalosporin-induced anaphylaxis occurs no more frequently among patients with known penicillin allergy than among those without such allergy.^23,27,38-41

Determining cross-reactivity

Penicillins and cephalosporins both possess a beta-lactam ring for antimicrobial activity. They differ in that the 5-membered thiazolidine ring of penicillin is replaced in the cephalosporins with a 6-membered dihydrothiazine ring. After degradation, penicillin forms a stable ring, whereas cephalosporins undergo rapid fragmentation of their rings.⁴² Immunologic cross-reactivity between the penicillin and cephalosporin beta-lactam rings is, therefore, very unlikely—an observation confirmed by monoclonal antibody analysis.⁴³

How the “10% cross-reactivity” myth took hold. When the first-generation cephalosporins cephaloridine and cephalothin were introduced in the 1960s, allergic and anaphylactic reactions were reported in patients with previous allergic reactions to penicillins. Subsequent reports, which attributed up to 10% cross-reactivity between the 2 drug classes, involved these same first-generation cephalosporins plus cephalexin and cefadroxil and a second-generation drug, cefamandole. However, these studies were flawed because the penicillin test compounds had been contaminated with cephalosporins. Until 1982, penicillin was produced commercially using the cephalosporium mold.³⁸

Many recent studies have established that the rate of cross-reactivity between penicillin and cephalosporins has been grossly overestimated. In fact, the rate of cross-reactivity between penicillin/amoxicillin and second- or third-generation cephalosporins is very low and may actually be lower than that between penicillins and other classes of antibiotics.⁴⁴

The evidence for limited cross-reactivity. A summary of publications evaluating 38,846 children and adults with and without a history of penicillin allergy is presented in TABLE 1. The database included 2435 patients with a history of penicillin allergy and 961 patients with a history of penicillin allergy and positive skin-test results for penicillin or amoxicillin (total penicillin-allergic patients=3396). The allergic reaction rate is compared with 34,047 patients without a history of penicillin allergy and 1403 patients without a history of penicillin allergy and negative skin-test results for penicillin or amoxicillin (total penicillin-nonallergic patients=35,450).

When patients with a positive history of penicillin-allergy received first-generation cephalosporins, which share a chemical side chain similar to penicillin or amoxicillin (cephalothin, cephaloridine, cephalexin, cefadroxil, and cefazolin, plus the early second-generation cephalosporin, cefamandole), they exhibited a significant increased risk of an allergic reaction to the cephalosporin.

Second- and third-generation cephalosporins modified in size and the complexity of their side chains (eg, cefprozil, cefuroxime, ceftazidime, cefpodoxime, and ceftriaxone) were different enough from penicillin and ampicillin that they did not increase risk of allergic cross-reactivity (TABLE 1).

Many other studies have suggested that cross-reactive immune responses to cephalosporins depend on side chain structure;^{22,23,27,32,37,38,44-49} that is, cephalosporins with a 7-position side chain similar to benzylpenicillin are more likely to cross-react with penicillin (TABLE 2). Cephalosporins that share a similar 7-position or 3-position side chain are more likely to cross-react with each other.

Cephalosporin/penicillin cross-reactivity

Few studies have evaluated whether patients with primary hypersensitivity to cephalosporins will experience cross-reactivity with penicillin. Romano et al⁴⁹ conducted skin tests and RASTs in patients with immediate allergic reactions to cephalosporins to examine responses to other cephalosporins and to classic penicillin determinants. About 1 in 5 patients allergic to a cephalosporin reacted to penicillin determinants, while most had positive results to other cephalosporins with the same or similar side-chains.

Limitations of skin testing

Penicillin skin testing in patients with a history of penicillin allergy does not reliably predict allergy to a cephalosporin unless the side chain of the penicillin or ampicillin reagent is similar to the cephalosporin side chain being tested.³ The positive and negative predictive values of skin testing results for cephalosporins are not well established; if the haptens that cause cephalosporin allergy were known, cross-reactivity with penicillins could be assessed directly. Cephalosporin skin testing works only for the specific drug and drugs with the same side chains, and can be done only if the drug is available in an IV or IM formulation.

Even a positive result does not guarantee a clinical reaction. When penicillin and cephalosporin skin tests or radioallergosorbent tests (RASTs) are positive, a clinical reaction is observed in only 10% to 60% of patients, depending on the reagent and study.⁵⁰ For example, among 19 well-characterized patients allergic to penicillin who were studied for their sensitivity to the cephalosporins, cephaloridine and cefamandole (which have identical or very similar side chains to penicillin and were therefore potentially cross-reactive) only 2 (10.5%) reacted to cefamandole, while the other 17 patients tolerated both agents.²⁶ In another study of clinical cross-reactivity between amoxicillin and cefadroxil in patients allergic to amoxicillin with good tolerance of penicillin, only 12% had an immediate allergic reaction to cefadroxil, despite the 2 drugs sharing an identical side chain.³³ In a third study, allergenic cross-reactivity with cefadroxil and cefamandole was studied among 21 patients selectively allergic to amoxicillin; 8 (38%) had a positive response to cefadroxil (same side chain) and none to cefamandole (different side chain).³²

Discussion

Sensible approach to penicillin-allergic patients

Question patients who report penicillin allergy. In many cases, penicillin may not actually have been taken, or patients may have had non-immunologic adverse events such as vomiting, diarrhea, or nonspecific rash; toxic effects; or contemporaneous side effects inappropriately attributed to the drug. These patients can receive penicillin, amoxicillin, or the cephalosporins.

Without the ability to detect patients with IgE antibody to penicillin prospectively or to distinguish true IgE immunologic reactions from idiopathic reactions in patients receiving cephalosporins, it is impossible to definitely claim that increased immune or IgE-mediated reactions to cephalosporins occur in true penicillinallergic (IgE) patients.

When a cephalosporin is/is not safe for a penicillin-allergic patient. Only IgE-mediated reactions—such as anaphylaxis or hypotension, laryngeal edema, wheezing, angioedema, or urticaria—are likely to become more severe with time. Therefore, with a patient who has had a true IgE-mediated reaction to a penicillin, avoid using cephalosporins with a similar side chain. You may, however, give cephalosporins that have different side chains. Cephalosporins may also be used for patients who have had non-IgE-mediated adverse reactions (“non-type I allergy”)²¹ to a penicillin, such as a non-pruritic, non-urticarial morbilliform or maculopapular rash.

How prevalent is primary cephalosporin allergy? Even if the patient is not allergic to penicillin, cephalosporins can cause allergic or immune-mediated reactions in approximately 1% to 3% of patients. A patient who had an allergic reaction to a specific cephalosporin probably should not receive that cephalosporin again. The risk of a reaction with a different cephalosporin is very low to nonexistent if the side chains of the 2 drugs are dissimilar.

Bottom line. Penicillin-allergic patients have indeed shown an increased incidence of allergic reactions to cephalothin, cephaloridine, cephalexin, cefadroxil, cefazolin, and cefamandole. However, the risk has been overestimated because most studies reporting this cross-reactivity were flawed (because penicillins were contaminated with cephalosporins) and then failed to account for the fact that penicillin-allergic patients have a 3-fold increased risk of allergic reactions even to nonrelated drugs.⁵¹

For patients truly allergic to penicillin, the risk of a reaction from a cephalosporin with side chains that differ from penicillin/amoxicillin (cefuroxime, cefpodoxime, cefdinir, and ceftriaxone, as endorsed by the AAFP) is so low that use is justified and medico-legally defensible by the currently available evidence.

CORRESPONDENCE
Michael E. Pichichero, MD, University of Rochester Medical Center, 601 Elmwood Avenue, Box 672, Rochester, New York 14642. E-mail: [email protected]

Undoubtedly you have patients who say they are allergic to penicillin but have difficulty recalling details of the reactions they experienced. To be safe, we often label these patients as penicillin-allergic without further questioning and withhold not only penicillins but cephalosporins due to concerns about potential cross-reactivity and resultant IgE-mediated, type I reactions. But even for patients truly allergic to penicillin, is the concern over cephalosporins justified? It depends on the specific agent. What is certain is that a blanket dismissal of all cephalosporins is unfounded.

The truth about the myth

Despite myriad studies spanning decades and involving varied patient populations, results have not conclusively established that penicillin allergy increases the risk of an allergic reaction to cephalosporins, compared with the incidence of a primary (and unrelated) cephalosporin allergy. Most people produce IgG and IgM antibodies in response to exposure to penicillin¹ that may cross-react with cephalosporin antigens.² The presence of these antibodies does not predict allergic, IgE cross-sensitivity to a cephalosporin. Even penicillin skin testing is generally not predictive of cephalosporin allergy.³

Reliably predicting cross-reactivity

A comprehensive review of the evidence shows that the attributable risk of a cross-reactive allergic reaction varies and is strongest when the chemical side chain of the specific cephalosporin is similar to that of penicillin or amoxicillin.

Administration of cephalothin, cephalexin, cefadroxil, and cefazolin in penicillin-allergic patients is associated with a significant increase in the rate of allergic reactions; whereas administration of cefprozil, cefuroxime, cefpodoxime, ceftazidime, and ceftriaxone is not.

Penicillin skin testing can accurately predict a penicillin-allergic reaction, but is not predictive for cephalosporin allergy unless the side chain of the penicillin or ampicillin testing reagent is similar to the cephalosporin side chain being evaluated. Patients who have a reaction to a penicillin or a cephalosporin that is not IgE mediated and not serious may receive repeated courses of that antibiotic and related antibiotics.

This article provides a comprehensive review of the frequency of allergic cross-reactivity between penicillin/amoxicillin and cephalosporin antibiotics, supporting the recent American Academy of Family Physicians evidence-based clinical practice guideline on treatment of acute otitis media recommending the use of cefuroxime, cefpodoxime, cefdinir, and ceftriaxone cephalosporins for patients allergic to penicillin.

Methods

We searched Medline and EMBASE databases for English-language articles using the keywords cephalosporin, penicillin, allergy, and cross-sensitivity for the years 1960 to 2005. Among 219 articles identified, 101 were included as source material for this review. Articles we excluded were reviews, republication of results, or ones irrelevant to our purpose.

Five articles described the rate of rashes following use of penicillin and cephalosporins,^4-8 and 4 articles described rates of anaphylaxis.^5,9-11 We included 26 articles for the evidence base evaluating penicillin/amoxicillin cross-allergy.^3,12-36 Eleven articles relied on patient history of penicillin/amoxicillin allergy to categorize results and establish reaction rates and relative risks for the penicillin/amoxicillin allergic vs nonallergic when receiving cephalosporins.^{12-15,17-20,27,28,31} Fourteen articles relied on patient history of penicillin/amoxicillin allergy plus skin testing results to penicillin/amoxicillin to categorize patients.^{16,21-25,29,30,32-37} One article³ provided data on a subset where penicillin/amoxicillin allergy was established based on history, and a separate subset where penicillin/amoxicillin allergy was established by skin testing. Other articles related to antibiotic chemical structures, animal studies, monoclonal antibody studies, cross-reactive antibody studies, and antibiotic skin testing were also reviewed.

Results

True incidence of reactions to cephalosporins

The most frequent reactions to cephalosporins are non-pruritic, non-urticarial rashes, which occur in 1.0% to 2.8% of patients;^4-8 for most, the mechanism is idiopathic and not a contraindication for future use.³⁸ Retrospective studies suggest a 1% to 3% incidence of immune or allergic reactions to cephalosporins independent of any history of penicillin/amoxicillin allergy.³¹ Anaphylactic reactions from cephalosporins are extremely rare, with the risk estimated at 0.0001% to 0.1%.^31,38 A seminal study suggested approximately 0.004% to 0.015% of treatment courses with penicillin results in anaphylaxis.^5,9-11 Several studies suggest that cephalosporin-induced anaphylaxis occurs no more frequently among patients with known penicillin allergy than among those without such allergy.^23,27,38-41

Determining cross-reactivity

Penicillins and cephalosporins both possess a beta-lactam ring for antimicrobial activity. They differ in that the 5-membered thiazolidine ring of penicillin is replaced in the cephalosporins with a 6-membered dihydrothiazine ring. After degradation, penicillin forms a stable ring, whereas cephalosporins undergo rapid fragmentation of their rings.⁴² Immunologic cross-reactivity between the penicillin and cephalosporin beta-lactam rings is, therefore, very unlikely—an observation confirmed by monoclonal antibody analysis.⁴³

How the “10% cross-reactivity” myth took hold. When the first-generation cephalosporins cephaloridine and cephalothin were introduced in the 1960s, allergic and anaphylactic reactions were reported in patients with previous allergic reactions to penicillins. Subsequent reports, which attributed up to 10% cross-reactivity between the 2 drug classes, involved these same first-generation cephalosporins plus cephalexin and cefadroxil and a second-generation drug, cefamandole. However, these studies were flawed because the penicillin test compounds had been contaminated with cephalosporins. Until 1982, penicillin was produced commercially using the cephalosporium mold.³⁸

Many recent studies have established that the rate of cross-reactivity between penicillin and cephalosporins has been grossly overestimated. In fact, the rate of cross-reactivity between penicillin/amoxicillin and second- or third-generation cephalosporins is very low and may actually be lower than that between penicillins and other classes of antibiotics.⁴⁴

The evidence for limited cross-reactivity. A summary of publications evaluating 38,846 children and adults with and without a history of penicillin allergy is presented in TABLE 1. The database included 2435 patients with a history of penicillin allergy and 961 patients with a history of penicillin allergy and positive skin-test results for penicillin or amoxicillin (total penicillin-allergic patients=3396). The allergic reaction rate is compared with 34,047 patients without a history of penicillin allergy and 1403 patients without a history of penicillin allergy and negative skin-test results for penicillin or amoxicillin (total penicillin-nonallergic patients=35,450).

When patients with a positive history of penicillin-allergy received first-generation cephalosporins, which share a chemical side chain similar to penicillin or amoxicillin (cephalothin, cephaloridine, cephalexin, cefadroxil, and cefazolin, plus the early second-generation cephalosporin, cefamandole), they exhibited a significant increased risk of an allergic reaction to the cephalosporin.

Second- and third-generation cephalosporins modified in size and the complexity of their side chains (eg, cefprozil, cefuroxime, ceftazidime, cefpodoxime, and ceftriaxone) were different enough from penicillin and ampicillin that they did not increase risk of allergic cross-reactivity (TABLE 1).

Many other studies have suggested that cross-reactive immune responses to cephalosporins depend on side chain structure;^{22,23,27,32,37,38,44-49} that is, cephalosporins with a 7-position side chain similar to benzylpenicillin are more likely to cross-react with penicillin (TABLE 2). Cephalosporins that share a similar 7-position or 3-position side chain are more likely to cross-react with each other.

Cephalosporin/penicillin cross-reactivity

Few studies have evaluated whether patients with primary hypersensitivity to cephalosporins will experience cross-reactivity with penicillin. Romano et al⁴⁹ conducted skin tests and RASTs in patients with immediate allergic reactions to cephalosporins to examine responses to other cephalosporins and to classic penicillin determinants. About 1 in 5 patients allergic to a cephalosporin reacted to penicillin determinants, while most had positive results to other cephalosporins with the same or similar side-chains.

Limitations of skin testing

Penicillin skin testing in patients with a history of penicillin allergy does not reliably predict allergy to a cephalosporin unless the side chain of the penicillin or ampicillin reagent is similar to the cephalosporin side chain being tested.³ The positive and negative predictive values of skin testing results for cephalosporins are not well established; if the haptens that cause cephalosporin allergy were known, cross-reactivity with penicillins could be assessed directly. Cephalosporin skin testing works only for the specific drug and drugs with the same side chains, and can be done only if the drug is available in an IV or IM formulation.

Even a positive result does not guarantee a clinical reaction. When penicillin and cephalosporin skin tests or radioallergosorbent tests (RASTs) are positive, a clinical reaction is observed in only 10% to 60% of patients, depending on the reagent and study.⁵⁰ For example, among 19 well-characterized patients allergic to penicillin who were studied for their sensitivity to the cephalosporins, cephaloridine and cefamandole (which have identical or very similar side chains to penicillin and were therefore potentially cross-reactive) only 2 (10.5%) reacted to cefamandole, while the other 17 patients tolerated both agents.²⁶ In another study of clinical cross-reactivity between amoxicillin and cefadroxil in patients allergic to amoxicillin with good tolerance of penicillin, only 12% had an immediate allergic reaction to cefadroxil, despite the 2 drugs sharing an identical side chain.³³ In a third study, allergenic cross-reactivity with cefadroxil and cefamandole was studied among 21 patients selectively allergic to amoxicillin; 8 (38%) had a positive response to cefadroxil (same side chain) and none to cefamandole (different side chain).³²

Discussion

Sensible approach to penicillin-allergic patients

Question patients who report penicillin allergy. In many cases, penicillin may not actually have been taken, or patients may have had non-immunologic adverse events such as vomiting, diarrhea, or nonspecific rash; toxic effects; or contemporaneous side effects inappropriately attributed to the drug. These patients can receive penicillin, amoxicillin, or the cephalosporins.

Without the ability to detect patients with IgE antibody to penicillin prospectively or to distinguish true IgE immunologic reactions from idiopathic reactions in patients receiving cephalosporins, it is impossible to definitely claim that increased immune or IgE-mediated reactions to cephalosporins occur in true penicillinallergic (IgE) patients.

When a cephalosporin is/is not safe for a penicillin-allergic patient. Only IgE-mediated reactions—such as anaphylaxis or hypotension, laryngeal edema, wheezing, angioedema, or urticaria—are likely to become more severe with time. Therefore, with a patient who has had a true IgE-mediated reaction to a penicillin, avoid using cephalosporins with a similar side chain. You may, however, give cephalosporins that have different side chains. Cephalosporins may also be used for patients who have had non-IgE-mediated adverse reactions (“non-type I allergy”)²¹ to a penicillin, such as a non-pruritic, non-urticarial morbilliform or maculopapular rash.

How prevalent is primary cephalosporin allergy? Even if the patient is not allergic to penicillin, cephalosporins can cause allergic or immune-mediated reactions in approximately 1% to 3% of patients. A patient who had an allergic reaction to a specific cephalosporin probably should not receive that cephalosporin again. The risk of a reaction with a different cephalosporin is very low to nonexistent if the side chains of the 2 drugs are dissimilar.

Bottom line. Penicillin-allergic patients have indeed shown an increased incidence of allergic reactions to cephalothin, cephaloridine, cephalexin, cefadroxil, cefazolin, and cefamandole. However, the risk has been overestimated because most studies reporting this cross-reactivity were flawed (because penicillins were contaminated with cephalosporins) and then failed to account for the fact that penicillin-allergic patients have a 3-fold increased risk of allergic reactions even to nonrelated drugs.⁵¹

For patients truly allergic to penicillin, the risk of a reaction from a cephalosporin with side chains that differ from penicillin/amoxicillin (cefuroxime, cefpodoxime, cefdinir, and ceftriaxone, as endorsed by the AAFP) is so low that use is justified and medico-legally defensible by the currently available evidence.

CORRESPONDENCE
Michael E. Pichichero, MD, University of Rochester Medical Center, 601 Elmwood Avenue, Box 672, Rochester, New York 14642. E-mail: [email protected]

A Guideline Update¹ published in September 2003 summarized the Global Strategy for the management of chronic obstructive pulmonary disease (COPD).² The evidence in that report recommended against spirometry to “diagnose or assess severity of COPD.” However, the Agency for Healthcare Research and Quality (AHRQ) recently published new evidence³ that supports limited use of spirometry for assessing the condition of COPD patients.

AHRQ’s Minnesota Evidence-Based Practice Center reviewed articles published from 1966–2005. Pertinent studies assessed outcomes for adults in primary care settings who were at risk for COPD according to race, age, gender, tobacco use, symptoms, and spirometric status. Excluded from the review were children, persons with asthma, and those with alpha-1 antitrypsin deficiency. The 169-page report had 82 references. The evidence was not explicitly graded, which made it difficult to interpret the significance of each recommendation.

The authors cautioned against widespread spirometric testing of COPD patients. They cited expense of spirometry, resulting treatment costs, resource utilization expense, and personnel time. There is risk of labeling a large number of individuals as diseased who would not benefit from treatment.

CORRESPONDENCEKeith B. Holten, MD, 825 Locust Street, Wilmington, OH 45177. E-mail: [email protected]

A Guideline Update¹ published in September 2003 summarized the Global Strategy for the management of chronic obstructive pulmonary disease (COPD).² The evidence in that report recommended against spirometry to “diagnose or assess severity of COPD.” However, the Agency for Healthcare Research and Quality (AHRQ) recently published new evidence³ that supports limited use of spirometry for assessing the condition of COPD patients.

AHRQ’s Minnesota Evidence-Based Practice Center reviewed articles published from 1966–2005. Pertinent studies assessed outcomes for adults in primary care settings who were at risk for COPD according to race, age, gender, tobacco use, symptoms, and spirometric status. Excluded from the review were children, persons with asthma, and those with alpha-1 antitrypsin deficiency. The 169-page report had 82 references. The evidence was not explicitly graded, which made it difficult to interpret the significance of each recommendation.

The authors cautioned against widespread spirometric testing of COPD patients. They cited expense of spirometry, resulting treatment costs, resource utilization expense, and personnel time. There is risk of labeling a large number of individuals as diseased who would not benefit from treatment.

CORRESPONDENCEKeith B. Holten, MD, 825 Locust Street, Wilmington, OH 45177. E-mail: [email protected]

A Guideline Update¹ published in September 2003 summarized the Global Strategy for the management of chronic obstructive pulmonary disease (COPD).² The evidence in that report recommended against spirometry to “diagnose or assess severity of COPD.” However, the Agency for Healthcare Research and Quality (AHRQ) recently published new evidence³ that supports limited use of spirometry for assessing the condition of COPD patients.

AHRQ’s Minnesota Evidence-Based Practice Center reviewed articles published from 1966–2005. Pertinent studies assessed outcomes for adults in primary care settings who were at risk for COPD according to race, age, gender, tobacco use, symptoms, and spirometric status. Excluded from the review were children, persons with asthma, and those with alpha-1 antitrypsin deficiency. The 169-page report had 82 references. The evidence was not explicitly graded, which made it difficult to interpret the significance of each recommendation.

The authors cautioned against widespread spirometric testing of COPD patients. They cited expense of spirometry, resulting treatment costs, resource utilization expense, and personnel time. There is risk of labeling a large number of individuals as diseased who would not benefit from treatment.

CORRESPONDENCEKeith B. Holten, MD, 825 Locust Street, Wilmington, OH 45177. E-mail: [email protected]

African Americans depend on quality mental health services in primary care more so than whites, since they are more likely to seek care in this setting. However, accumulating evidence shows that African Americans in primary care settings face disparity in recognition and treatment of depression (see Race and mental health treatment: the divide).^1-5 Until now, the specific factors leading to this disparity have not been well described.¹⁶

In this article—a systematic review of studies on the evaluation and treatment of depression in African American patients—we draw attention to factors regarding patients, physicians, and clinical settings that raise barriers to diagnosis and treatment. We also recommend strategies to break through these barriers.

Assumptions that mask disparities

Disparities may not be apparent to healthcare providers in part because community-based epidemiologic studies report that African Americans have a lower rate of major depression than whites.^5,17-19 In primary care settings specifically, the prevalence of major depression among African Americans appears to be lower than for white or Hispanic Americans, mirroring rates in national and community studies.²⁰ Explanations for this relatively low prevalence of depression have included attributes of the African American experience, such as strong religious beliefs and community ties.

Moreover, according to analyses of data from the National Comorbidity Study, which included an examination of attitudes toward mental health services, African Americans with major depression are more likely to report that they would “definitely go” for mental health services than would whites.²¹

High-risk populations overlooked. Community-based studies, however, do not fully capture the mental health needs of African Americans, a large number of whom are in high-risk populations excluded from community epidemiologic studies of mental disorders. Studies of these populations show high rates of major depression and other mental disorders, due to factors including alcohol and substance use disorders, poor physical health, and poverty and homelessness.

An unsettling irony. For emotional distress, African Americans are more likely to seek help from primary care clinicians than from specialty mental health providers.^22-24 However, in primary care, depression in African Americans is less likely to be detected than it is in whites.³

In a nationally representative sample, African Americans have been found less likely than whites to receive effective care for depression in primary care.^12,25,26 In a study of Medicaid recipients who had been diagnosed with major depression, African Americans less often received antidepressant medications than did whites.²⁷ Such disparities in pharmacologic treatments were also found in another study of African American patients in primary care, using data from the National Ambulatory Medical Care Surveys.²⁴

Search methods

Data sources

We systematically searched for and retrieved articles in Medline (January 1966 to December 2004) using the MeSH (medical subject heading) terms African Americans and depressive disorder along with the MesH terms adult or adolescent. We refined our search results using the MeSH terms diagnosis, comorbidity, physician-patient relations, patient satisfaction, and patient acceptance of health care to focus on patient and physician factors.

To search for factors related to practice setting, we refined our initial search strategy with the MeSH terms primary health care, family practice, and health services. We also hand-checked bibliographies to find potential articles.

Study selection and data extraction

Two investigators (AKD, MO) independently reviewed abstracts of the articles retrieved and excluded articles that did not fulfill the criteria to be relevant to the topic of depression in African American patients. Articles were excluded if they did not focus on US populations, African Americans, depressive disorders, and clinical settings, or if they did not report original data. Differences between the 2 reviewers were resolved through consensus after full-text review of the article.

Additional articles were defined by 2 other investigators (HLC, MMW) who provided peer review and identified potential articles in bibliographies. For each article, we classified the barriers described as related to diagnosis, management, or both and to patient, physician, and treatment-setting factors.

Results

The initial search using the MeSH terms African Americans and depressive disorder with adult or adolescent returned 156 articles. We found a total of 73 unique articles by refining our Medline search with additional MeSH terms. Of these publications, 12 fulfilled our criteria as relevant to the topic of diagnosing and managing depression in African American patients.

Our hand-checking of bibliographic references found an additional 12 articles that met our relevancy criteria. We examined each of the 24 articles (TABLE 1) to determine the types of barrier studies and whether the factor was related to the detection of depression, its management, or both.

TABLE 1
Selected studies on managing depression in African Americans

Discussion

Barriers in evaluation

Research on depression in African Americans can be analyzed in terms of attributes of the provider, patient, and practice setting.

Provider attributes. The race of the physician has been linked to the diagnosis of depression in African-American patients, but studies of this factor have yielded conflicting results. McKinlay and colleagues used videotaped cases of depressed patients presented to 128 physicians to examine whether diagnosis was affected by physician characteristics (age, gender race, and medical specialty), patient characteristics (age, gender, race, and socioeconomic status), or their combination.³⁷ They found that physician characteristics affected the diagnosis of depression, but that patient attributes did not. In particular, white primary care physicians were twice as likely as African American physicians to diagnose depression. Since approximately 1 in 5 African American patients seeks care from a physician of their own race,⁴⁷ such differences may influence the overall rate of recognition in primary care. However, data from the Medical Outcome Study, which included demographic characteristics on 349 primary care physicians, did not find that the race of the physician influenced the detection of depression.³

In addition to physician demographic factors, physician communication style may affect ability to diagnose depression. Appropriate diagnosis and treatment of depression depends to a great extent on verbal communications between patient and provider about the nature, extent, severity, and consequences of symptoms. Problems with communication may lead to misunderstandings, misdiagnosis, inappropriate treatments, and premature termination of treatment. An analysis of patient-physician encounters indicates that physicians may be more likely to minimize emotional symptoms of African American than of whites.⁴² Relative to whites, African Americans are more likely to rate their visits with white physicians as less participatory.

This difference is overcome when African Americans see physicians of their own race.²³ In a population-based study among largely low-income African American women, primary care physicians who were rated as showing more respect by the participants were more likely to inquire about depressive symptoms during a clinical visit.⁴⁰ Thus, despite the fact that African American patients rate their encounters more satisfying and are more likely to disclose their problems to African American physicians, white physicians appear to diagnose depression more commonly in their African American patients. Clearly, more work is needed to clarify this seeming contradiction.

Patient attributes. During the primary care encounter, African Americans are as likely as whites to discuss mental health problems.³⁴ However, African Americans may be more likely to exhibit somatic and neurovegetative symptoms of depression than mood or cognitive symptoms,^33,36 which may complicate detection and diagnosis.

A recent national survey suggests that depression in African Americans may be commonly masked by self medication and somatic symptoms and so may pass undetected in primary care.⁴⁸ In a randomized clinical trial of depression treatment in primary care, depressed African American subjects were more likely than white subjects to have symptoms of poor physical health and pain and to have somatization.²⁹ However, another study found few differences between whites and African Americans in symptom presentation of depression when comorbid disorders and sociodemographic factors were controlled.⁴⁶

Do primarily somatic presentations of depression reduce physicians’ ability to accurately diagnose major depression in racial and ethnic minorities? This question has not been studied, though coexisting medical problems in primary care populations have been found to impede the diagnosis of depression,⁴⁹ presumably by competing for the physician’s attention.^3,50 In a variety of medical contexts, known medical disorders are associated with under-treatment of unrelated disorders.⁵¹

Practice-setting attributes. No studies have examined whether specific practice setting factors, such as insurance coverage, are related to the low detection rate of major depression. Among low-income women of multiethnic backgrounds attending primary care clinics, Van Hook found that perceived separation of primary care services and specialty mental health care was a self-reported barrier to seeking help for depression.⁴⁴

The Surgeon General’s report documents the overall poor access to general medical services faced by African Americans. They are more likely to receive health care in outpatient hospital and emergency departments, and their mental health services are also characterized by high rates of emergency care. As a result, they are less likely to receive the continuity of treatment provided in primary care, which may allow better detection of depression.¹⁵

Barriers to effective management

Research on the potential causes of disparities in treatment for African American patients in primary care has examined a variety of provider, patient, and practice-setting factors.

Provider attributes. Physician communication style may not only influence diagnosis but also a decision to treat. In a study that analyzed audiotaped communication between patients and primary care providers during clinic visits, the tendency of physicians to minimize emotional expression by African Americans, relative to whites, led to lower prescribing of antidepressant medication among African Americans.⁴² Further research is needed to understand why such communication styles exist between primary care providers and African American patients.

Patient attributes. A number of patient attributes have been examined that may explain the lower rate of mental health services provided in primary care. Significant research has been undertaken on African American attitudes toward and beliefs about mental health treatments. Cooper-Patrick and colleagues conducted focus groups for African American and white patients as well as health care professionals. In questions related to depression and treatment preferences, African American patients expressed more concerns about stigma and spirituality than did white patients.³²

In related research, a survey of African American patients recruited from primary care offices indicated they were less likely to find antidepressant medication acceptable than white patients in primary care.³¹ Such attitudinal differences may explain why African Americans use antidepressants less commonly than whites, even when primary care physicians make similar recommendations for both groups.⁵² As yet unstudied is the extent to which sensational reports about antidepressant side effects may cause persons already skeptical about the care they receive to discontinue treatment.

Another reason African Americans may avoid or discontinue antidepressant treatment is that they tend to tolerate certain classes of psychotropic medications poorly. Strickland and colleagues found that African Americans are more likely than whites to be “poor metabolizers” of tricyclic antidepressants.⁵³ African Americans treated with tricyclic antidepressants will therefore experience higher plasma levels per dose than whites, and an earlier onset of action. African Americans are also more likely to experience side effects, which may lead to treatment nonadherence.

Few studies have examined the tolerability of newer class of antidepressant medications, such as selective serotonin reuptake inhibitors (SSRIs), for African Americans. More research is required to determine whether antidepressant side effects or other experiences with psychotropic medications are a reason for the lower rates of antidepressant use among African American patients in primary care. Educating patients about antidepressants, their onset of action, and side effects may reduce some of these barriers and prevent early discontinuation of antidepressant therapy.⁵⁴

Poverty and its associated psychosocial factors may also contribute to the lower quality of mental health care among African Americans. Miranda and colleagues found this to be so for African American women receiving primary care in obstetric-gynecologic clinics.⁴⁷ They have argued that clinical case management is needed as a component of effective mental health treatment for primary care patients who are poor and likely to face significant negative life events.

Practice-setting attributes. Few practice-setting factors have been examined in relationship to the disparities in depression treatment faced by African Americans. O’Malley and colleagues examined whether primary care physicians who were evaluated as having comprehensive medical services by low-income African American women were more likely to provide treatment for depression.⁴⁰ Comprehensiveness was determined by the ability to meet all health needs, thoroughness of physical exam, and provision of counseling and screening services. In a survey study, they found that comprehensiveness of medical services was correlated to being asked about and being treated for depression.

Interventions in primary care

Concerted efforts to improve quality of care can reduce the mental health burden of undetected depression for African Americans as well as other ethnic and racial groups. African American and white primary care patients both appear to respond equally to standardized psychotherapy and pharmacotherapy for major depression.²⁹

In a randomized control trial to improve the quality of depression management in primary care, no significant differences existed between white and African American participants in the process of depression care or clinical outcomes, although both groups had less than optimal recovery rates.⁴¹ One randomized clinical trial included modified interventions to target the mental health needs of low-income minority primary care patients. The interventions included educating clinicians about depression; teaching nurses to educate, assess, and follow-up depressed patients; and making cognitive behavioral therapy available. Patients and physicians selected the treatment. Modification for minority patients was modest and included translations and cultural training for clinicians.

Using these approaches, African American participants were more likely than whites to have better depression outcome at 6 and 12 months.³⁹ These studies indicate that both general interventions to depression care and small modifications for minority patients can lead to improved health outcomes among African American patients.

It may be unreasonable to believe that simply seeking to improve the primary care assessment and treatment of depression in African Americans will eliminate racial differences in the health outcomes of depressed adults. Reform is also clearly needed in health care financing and in broader social welfare policy as it affects the lives of depressed minority populations. However, primary care providers who are aware of the risk of racial disparities in the recognition and treatment of depression and work to provide treatments that are tailored to the individual’s needs can help to reduce the significant burden of depression (TABLE 2).

TABLE 2
Recommendations for addressing barriers to the detection and treatment of depression in the African American patient

Acknowledgments

This review was funded by the Columbia Center for the Health of Urban Minorities (NCMHHD MD000206-019006) (Drs. Olfson and Weissman), an unrestricted grant from Eli Lilly & Company (Dr. Weissman) and a NIMH National Service Research Award Institutional Research Training Grant 5T32MH015144 (Dr. Das).

CORRESPONDENCE
Myrna M. Weissman, Ph.D. Columbia University College of Physicians and Surgeons, 1051 Riverside Drive, New York, NY 10032. E-mail: [email protected]