Lauren DeAlleaume, MD

Evidence summary

A Cochrane review found DM to be modestly effective in 2 of 3 studies.¹ In the first study—a meta-analysis of 6 industry-sponsored RCTs of 710 adults—a single 30-mg dose of DM decreased coughing bouts by 12% (P=.004) and increased the time between bouts by 17 % (P=.002) in the 3 hours after treatment.

A second study of 3 successive industry-sponsored, blinded RCTs enrolling a total of 451 adults found that 30 mg of DM decreased cough counts between 19% and 36% (P<.05) over a 3-hour follow-up period. Neither of the 2 studies specified whether the outcome assessors were blinded to treatment groups.

A third double-blinded RCT evaluating a single 30-mg dose of DM in 43 adults during a 3-hour follow-up period showed no statistically significant improvement in cough outcomes compared with placebo.

A split decision on guaifenesin

The same Cochrane review also evaluated other medications for cough related to upper respiratory infection in adults.¹ The results of 2 guaifenesin trials were split. In a double-blinded RCT of 239 adults, more patients taking guaifenesin reported decreased cough frequency and intensity (75% vs 31%; P<.01). However, another double-blinded RCT of 65 patients found guaifenesin to be no more effective than placebo in reducing subjective cough frequency.

Antihistamines don’t help, adding a decongestant isn’t much better

Three trials of antihistamines in a total of 1900 adults found that the drugs didn’t relieve cough symptoms more effectively than placebo. Antihistamine-decongestant combination trials produced split results. In 1 double-blinded RCT of 283 adults, loratadine-pseudoephedrine (5 and 120 mg, respectively) twice daily for 4 days didn’t decrease subjective cough scores more than placebo and was associated with more dry mouth, dizziness, headache, and insomnia (30% vs 21%; P value not reported).

Another partially double-blinded RCT of 73 adults reported that dexbrompheniramine-pseudoephedrine (6 and 120 mg, respectively) twice daily for 1 week decreased subjective cough severity (1.4 vs 2.0; P<.05) on a scale of 0 to 4 during days 3 to 5 of treatment. The combination was associated with increased dizziness and dry mouth, however (exact data not reported; P≤.01).

Codeine works no better than placebo

Two partially double-blinded RCTs of 163 adults found codeine (sold OTC in Canada) to be no more effective than placebo in relieving cough caused by the common cold.¹

Zinc lozenges show mixed results

Zinc lozenges containing 13.3 mg of zinc acetate taken every 2 to 3 hours decreased the duration of cough from 5.35 to 2.14 days (P<.001) in a double-blinded placebo-controlled RCT of 50 adults.² The most recent systematic review showed mixed results: Half the studies found no benefit for zinc in treating upper respiratory infection.³

In children, forget DM, antihistamines, decongestants

The previously mentioned Cochrane review¹ also summarized studies in children. DM was no more effective than placebo for decreasing cough in 2 RCTs enrolling a total of 107 children. Another single-blinded RCT of 100 children showed that neither DM nor diphenhydramine relieved cough better than placebo.

Two RCTs involving a total of 237 children compared antihistamines with placebo. One double-blinded trial reported that clemastine and chlorpheniramine were no more effective than placebo. The other partially double-blinded trial found that diphenhydramine didn’t decrease cough frequency more than placebo.

Two double-blinded RCTs (total of 155 children) showed that antihistamine-decongestant combinations (brompheniramine-phenylpropanolamine and brompheniramine-phenylephrine-propanolamine) didn’t reduce cough more than placebo. No studies have evaluated guai-fenesin in children.¹

Honey appears to help

In a partially double-blinded RCT (the “no treatment” group was not blinded) of 105 children, a single dose of buckwheat honey decreased cough frequency, as assessed by parents, by 1.89 points on the 7-point Likert scale compared with DM (1.39) and no treatment (0.92; P<.001). Overall improvement in symptom score averaged 10.71 out of a total 30 points for honey compared with 6.41 for no treatment (P=.04). Cough frequency and overall symptom scores for DM didn’t differ significantly from no treatment. Hyperactivity, nervousness, and insomnia were reported more often with honey (5 patients) than DM (2 patients) or no treatment (0 patients); (P=.04).⁴

Placebos work, but why?

In a review of 8 RCTs, the average reduction in cough in the placebo group (both capsules and syrups) was approximately 85% of that seen in the active medication group (range 56%-105%).⁵ Several factors may account for the efficacy of placebo, including lubrication of the pharynx by increased salivation caused by sweet or bitter vehicles. Sweet vehicles also may stimulate endogenous opioids that may suppress cough. In an unblinded RCT of 54 patients, a capsule placebo significantly decreased the number of coughs during a 15-minute follow-up compared with no treatment (18 vs 3; P=.0003).⁶

Recommendations

The American College of Chest Physicians recommends a first-generation antihistamine-decongestant combination or naproxen for acute cough in the common cold (SOR: A). Newer-generation, nonsedating antihistamines are not recommended (SOR: D).⁷

The US Food and Drug Administration (FDA) advises against using OTC cough and cold medicines in children younger than 2 years because of the risk of “serious and potentially life-threatening side effects.” The FDA also recommends taking significant precautions if these products are used in children older than 2 years, pending a complete FDA review of the medications for children 2 to 11 years. Manufacturers of children’s cough and cold remedies have changed the labels on the medications voluntarily to recommend that they not be given to children younger than 4 years.⁸

The American Academy of Pediatrics recommends that physicians clearly educate parents about the potential risks and lack of benefits of DM- and codeine-containing cough remedies.⁹

Evidence summary

A Cochrane review found DM to be modestly effective in 2 of 3 studies.¹ In the first study—a meta-analysis of 6 industry-sponsored RCTs of 710 adults—a single 30-mg dose of DM decreased coughing bouts by 12% (P=.004) and increased the time between bouts by 17 % (P=.002) in the 3 hours after treatment.

A second study of 3 successive industry-sponsored, blinded RCTs enrolling a total of 451 adults found that 30 mg of DM decreased cough counts between 19% and 36% (P<.05) over a 3-hour follow-up period. Neither of the 2 studies specified whether the outcome assessors were blinded to treatment groups.

A third double-blinded RCT evaluating a single 30-mg dose of DM in 43 adults during a 3-hour follow-up period showed no statistically significant improvement in cough outcomes compared with placebo.

A split decision on guaifenesin

The same Cochrane review also evaluated other medications for cough related to upper respiratory infection in adults.¹ The results of 2 guaifenesin trials were split. In a double-blinded RCT of 239 adults, more patients taking guaifenesin reported decreased cough frequency and intensity (75% vs 31%; P<.01). However, another double-blinded RCT of 65 patients found guaifenesin to be no more effective than placebo in reducing subjective cough frequency.

Antihistamines don’t help, adding a decongestant isn’t much better

Three trials of antihistamines in a total of 1900 adults found that the drugs didn’t relieve cough symptoms more effectively than placebo. Antihistamine-decongestant combination trials produced split results. In 1 double-blinded RCT of 283 adults, loratadine-pseudoephedrine (5 and 120 mg, respectively) twice daily for 4 days didn’t decrease subjective cough scores more than placebo and was associated with more dry mouth, dizziness, headache, and insomnia (30% vs 21%; P value not reported).

Another partially double-blinded RCT of 73 adults reported that dexbrompheniramine-pseudoephedrine (6 and 120 mg, respectively) twice daily for 1 week decreased subjective cough severity (1.4 vs 2.0; P<.05) on a scale of 0 to 4 during days 3 to 5 of treatment. The combination was associated with increased dizziness and dry mouth, however (exact data not reported; P≤.01).

Codeine works no better than placebo

Two partially double-blinded RCTs of 163 adults found codeine (sold OTC in Canada) to be no more effective than placebo in relieving cough caused by the common cold.¹

Zinc lozenges show mixed results

Zinc lozenges containing 13.3 mg of zinc acetate taken every 2 to 3 hours decreased the duration of cough from 5.35 to 2.14 days (P<.001) in a double-blinded placebo-controlled RCT of 50 adults.² The most recent systematic review showed mixed results: Half the studies found no benefit for zinc in treating upper respiratory infection.³

In children, forget DM, antihistamines, decongestants

The previously mentioned Cochrane review¹ also summarized studies in children. DM was no more effective than placebo for decreasing cough in 2 RCTs enrolling a total of 107 children. Another single-blinded RCT of 100 children showed that neither DM nor diphenhydramine relieved cough better than placebo.

Two RCTs involving a total of 237 children compared antihistamines with placebo. One double-blinded trial reported that clemastine and chlorpheniramine were no more effective than placebo. The other partially double-blinded trial found that diphenhydramine didn’t decrease cough frequency more than placebo.

Two double-blinded RCTs (total of 155 children) showed that antihistamine-decongestant combinations (brompheniramine-phenylpropanolamine and brompheniramine-phenylephrine-propanolamine) didn’t reduce cough more than placebo. No studies have evaluated guai-fenesin in children.¹

Honey appears to help

In a partially double-blinded RCT (the “no treatment” group was not blinded) of 105 children, a single dose of buckwheat honey decreased cough frequency, as assessed by parents, by 1.89 points on the 7-point Likert scale compared with DM (1.39) and no treatment (0.92; P<.001). Overall improvement in symptom score averaged 10.71 out of a total 30 points for honey compared with 6.41 for no treatment (P=.04). Cough frequency and overall symptom scores for DM didn’t differ significantly from no treatment. Hyperactivity, nervousness, and insomnia were reported more often with honey (5 patients) than DM (2 patients) or no treatment (0 patients); (P=.04).⁴

Placebos work, but why?

In a review of 8 RCTs, the average reduction in cough in the placebo group (both capsules and syrups) was approximately 85% of that seen in the active medication group (range 56%-105%).⁵ Several factors may account for the efficacy of placebo, including lubrication of the pharynx by increased salivation caused by sweet or bitter vehicles. Sweet vehicles also may stimulate endogenous opioids that may suppress cough. In an unblinded RCT of 54 patients, a capsule placebo significantly decreased the number of coughs during a 15-minute follow-up compared with no treatment (18 vs 3; P=.0003).⁶

Recommendations

The American College of Chest Physicians recommends a first-generation antihistamine-decongestant combination or naproxen for acute cough in the common cold (SOR: A). Newer-generation, nonsedating antihistamines are not recommended (SOR: D).⁷

The US Food and Drug Administration (FDA) advises against using OTC cough and cold medicines in children younger than 2 years because of the risk of “serious and potentially life-threatening side effects.” The FDA also recommends taking significant precautions if these products are used in children older than 2 years, pending a complete FDA review of the medications for children 2 to 11 years. Manufacturers of children’s cough and cold remedies have changed the labels on the medications voluntarily to recommend that they not be given to children younger than 4 years.⁸

The American Academy of Pediatrics recommends that physicians clearly educate parents about the potential risks and lack of benefits of DM- and codeine-containing cough remedies.⁹

Evidence summary

A Cochrane review found DM to be modestly effective in 2 of 3 studies.¹ In the first study—a meta-analysis of 6 industry-sponsored RCTs of 710 adults—a single 30-mg dose of DM decreased coughing bouts by 12% (P=.004) and increased the time between bouts by 17 % (P=.002) in the 3 hours after treatment.

A second study of 3 successive industry-sponsored, blinded RCTs enrolling a total of 451 adults found that 30 mg of DM decreased cough counts between 19% and 36% (P<.05) over a 3-hour follow-up period. Neither of the 2 studies specified whether the outcome assessors were blinded to treatment groups.

A third double-blinded RCT evaluating a single 30-mg dose of DM in 43 adults during a 3-hour follow-up period showed no statistically significant improvement in cough outcomes compared with placebo.

A split decision on guaifenesin

The same Cochrane review also evaluated other medications for cough related to upper respiratory infection in adults.¹ The results of 2 guaifenesin trials were split. In a double-blinded RCT of 239 adults, more patients taking guaifenesin reported decreased cough frequency and intensity (75% vs 31%; P<.01). However, another double-blinded RCT of 65 patients found guaifenesin to be no more effective than placebo in reducing subjective cough frequency.

Antihistamines don’t help, adding a decongestant isn’t much better

Three trials of antihistamines in a total of 1900 adults found that the drugs didn’t relieve cough symptoms more effectively than placebo. Antihistamine-decongestant combination trials produced split results. In 1 double-blinded RCT of 283 adults, loratadine-pseudoephedrine (5 and 120 mg, respectively) twice daily for 4 days didn’t decrease subjective cough scores more than placebo and was associated with more dry mouth, dizziness, headache, and insomnia (30% vs 21%; P value not reported).

Another partially double-blinded RCT of 73 adults reported that dexbrompheniramine-pseudoephedrine (6 and 120 mg, respectively) twice daily for 1 week decreased subjective cough severity (1.4 vs 2.0; P<.05) on a scale of 0 to 4 during days 3 to 5 of treatment. The combination was associated with increased dizziness and dry mouth, however (exact data not reported; P≤.01).

Codeine works no better than placebo

Two partially double-blinded RCTs of 163 adults found codeine (sold OTC in Canada) to be no more effective than placebo in relieving cough caused by the common cold.¹

Zinc lozenges show mixed results

Zinc lozenges containing 13.3 mg of zinc acetate taken every 2 to 3 hours decreased the duration of cough from 5.35 to 2.14 days (P<.001) in a double-blinded placebo-controlled RCT of 50 adults.² The most recent systematic review showed mixed results: Half the studies found no benefit for zinc in treating upper respiratory infection.³

In children, forget DM, antihistamines, decongestants

The previously mentioned Cochrane review¹ also summarized studies in children. DM was no more effective than placebo for decreasing cough in 2 RCTs enrolling a total of 107 children. Another single-blinded RCT of 100 children showed that neither DM nor diphenhydramine relieved cough better than placebo.

Two RCTs involving a total of 237 children compared antihistamines with placebo. One double-blinded trial reported that clemastine and chlorpheniramine were no more effective than placebo. The other partially double-blinded trial found that diphenhydramine didn’t decrease cough frequency more than placebo.

Two double-blinded RCTs (total of 155 children) showed that antihistamine-decongestant combinations (brompheniramine-phenylpropanolamine and brompheniramine-phenylephrine-propanolamine) didn’t reduce cough more than placebo. No studies have evaluated guai-fenesin in children.¹

Honey appears to help

In a partially double-blinded RCT (the “no treatment” group was not blinded) of 105 children, a single dose of buckwheat honey decreased cough frequency, as assessed by parents, by 1.89 points on the 7-point Likert scale compared with DM (1.39) and no treatment (0.92; P<.001). Overall improvement in symptom score averaged 10.71 out of a total 30 points for honey compared with 6.41 for no treatment (P=.04). Cough frequency and overall symptom scores for DM didn’t differ significantly from no treatment. Hyperactivity, nervousness, and insomnia were reported more often with honey (5 patients) than DM (2 patients) or no treatment (0 patients); (P=.04).⁴

Placebos work, but why?

In a review of 8 RCTs, the average reduction in cough in the placebo group (both capsules and syrups) was approximately 85% of that seen in the active medication group (range 56%-105%).⁵ Several factors may account for the efficacy of placebo, including lubrication of the pharynx by increased salivation caused by sweet or bitter vehicles. Sweet vehicles also may stimulate endogenous opioids that may suppress cough. In an unblinded RCT of 54 patients, a capsule placebo significantly decreased the number of coughs during a 15-minute follow-up compared with no treatment (18 vs 3; P=.0003).⁶

Recommendations

The American College of Chest Physicians recommends a first-generation antihistamine-decongestant combination or naproxen for acute cough in the common cold (SOR: A). Newer-generation, nonsedating antihistamines are not recommended (SOR: D).⁷

The US Food and Drug Administration (FDA) advises against using OTC cough and cold medicines in children younger than 2 years because of the risk of “serious and potentially life-threatening side effects.” The FDA also recommends taking significant precautions if these products are used in children older than 2 years, pending a complete FDA review of the medications for children 2 to 11 years. Manufacturers of children’s cough and cold remedies have changed the labels on the medications voluntarily to recommend that they not be given to children younger than 4 years.⁸

The American Academy of Pediatrics recommends that physicians clearly educate parents about the potential risks and lack of benefits of DM- and codeine-containing cough remedies.⁹

TABLE 1
Contraindications to empiric antibiotics for urinary tract infection (telephone treatment)

Evidence summary

An evidence-based review¹ found 5 high-quality studies on the diagnosis of acute uncomplicated UTI among women. (“Uncomplicated” was defined as normal urinary tract and no contributing medical problems, such as diabetes, neurogenic bladder, renal stones.) UTIs were defined as the presence of significant bacteriuria (≥10⁴ to 10⁵ colony-forming units) on culture. A patient presenting to a clinician with 1 or more UTI symptoms had approximately a 50% chance of having significant bacteriuria on culture.¹ The authors estimated the pretest probability of UTI as 5% from the incidence of asymptomatic bacteriuria among healthy women.^1,2 This produced a LR+ of 19 simply for presenting to a clinician with 1 or more UTI symptoms.¹ The summary LRs for clinical signs and symptoms in the prediction of UTI after presentation to the office are found in TABLE 2. A history of a vaginal discharge or irritation has a LR– of 0.3, decreasing the probability of UTI for a patient presenting to the office from approximately 50% to 20%, so further testing would be indicated.¹

No single sign or symptom accurately predicted UTI. However, the triad of dysuria with frequency but without vaginal symptoms increased the probability of significant bacteriuria on culture from 50% to 96% (LR+=24.6).¹ In contrast, a 1999 review of 51 studies calculated that if both the nitrites and leukocyte esterase are positive on urine dipstick testing, the LR+ is 4.2; if both are negative the LR– is 0.3.^1,3 Since the probability of UTI for patients with the symptom triad is so high, dipstick urinalysis is unlikely to alter management regardless of whether nitrites and leukocyte esterase were both positive or negative (posttest probability=98%–99% and 80%, respectively). If urine dipstick or other office-based tests are not needed to make the diagnosis of uncomplicated UTI for a patient with the classic triad of symptoms, then telephone treatment based on symptoms may be reasonable. Women who have recurrent UTIs (2 or more culture positive UTIs over the previous 12 months) can accurately self-diagnose subsequent UTIs based on symptoms (LR+=4.0).^4,5

A recent retrospective case series⁶ evaluated a telephone guideline for the empiric treatment of UTI for 4177 women in a California HMO. UTI criteria were ≤10 days of dysuria; frequency, urgency, pressure, or increased nocturia; or gross hematuria. Women were excluded if they had any one of a variety of contraindications (TABLE 1). Upper tract infection occurred in 21 patients (1.1%) within 60 days of telephone treatment, two thirds of which likely represented treatment failures. This is similar to rates in control groups of other studies. Fourteen women (1.5%) received care for sexually transmitted diseases or other gynecologic conditions, primarily bacterial vaginitis, within 60 days of telephone treatment. Of note, 6% of the cohort were elderly, diabetic, taking glucocorticoids or early in pregnancy and are typically excluded from other studies. This higher-risk group did not have an increased incidence of either sepsis or pyelonephritis.⁶ No increase in adverse outcomes was seen in another study of a telephone treatment protocol.⁷

Several studies^6-8 have noted that telephone treatment protocols increase the use of protocol-recommended antibiotics (eg, generally less expensive agents such as trimethoprim-sulfamethoxazole), which may help limit resistance to fluoroquinolones. However, specific data are not available.

McIssac et al⁹ reviewed a cohort of 231 women presenting to family physicians’ offices with uncomplicated cystitis symptoms. Empiric therapy resulted in approximately 40% of women unnecessarily receiving antibiotics. Treating only women with classic cystitis symptoms and pyuria would have decreased the unnecessary use of antibiotics to 26.2%, but fewer women with confirmed cystitis would have received immediate antibiotics (66.4% vs 91.8%). They derived a clinical decision rule designed to balance false positives and false negatives. It recommends immediate antibiotic treatment if women have ≥2 of 4 signs or symptoms: dysuria, leukocyte esterase (greater than trace), positive nitrites, or blood (greater than trace) on dipstick (LR+=2.29). Otherwise the rule recommends a culture to guide antibiotic therapy. This decision rule would have reduced unnecessary antibiotic use by 27.5% while ensuring that more women with confirmed UTIs received immediate antibiotics (81.3%).

In 1999, Saint et al⁸ estimated savings of $367,000 for 147,000 women enrolled over 1 year after widespread guideline implementation. Two cost-effectiveness studies^10,11 of office treatment concluded that empiric treatment without additional testing is the least costly option in this setting. However, a recent, comprehensive cost-effectiveness study¹¹ concluded that if a patient presents to an office, the marginal cost of performing a pelvic examination and urine culture for women with a negative dipstick was relatively low ($4 to $32 per symptom day avoided).

TABLE 2
Diagnosis of urinary tract infection

Recommendations from others

A 2002 Institute for Clinical Systems Improvement guideline¹² advised offering telephone treatment of uncomplicated UTI for low-risk patients if preferred by both provider and patient.

TABLE 1
Contraindications to empiric antibiotics for urinary tract infection (telephone treatment)

Evidence summary

An evidence-based review¹ found 5 high-quality studies on the diagnosis of acute uncomplicated UTI among women. (“Uncomplicated” was defined as normal urinary tract and no contributing medical problems, such as diabetes, neurogenic bladder, renal stones.) UTIs were defined as the presence of significant bacteriuria (≥10⁴ to 10⁵ colony-forming units) on culture. A patient presenting to a clinician with 1 or more UTI symptoms had approximately a 50% chance of having significant bacteriuria on culture.¹ The authors estimated the pretest probability of UTI as 5% from the incidence of asymptomatic bacteriuria among healthy women.^1,2 This produced a LR+ of 19 simply for presenting to a clinician with 1 or more UTI symptoms.¹ The summary LRs for clinical signs and symptoms in the prediction of UTI after presentation to the office are found in TABLE 2. A history of a vaginal discharge or irritation has a LR– of 0.3, decreasing the probability of UTI for a patient presenting to the office from approximately 50% to 20%, so further testing would be indicated.¹

No single sign or symptom accurately predicted UTI. However, the triad of dysuria with frequency but without vaginal symptoms increased the probability of significant bacteriuria on culture from 50% to 96% (LR+=24.6).¹ In contrast, a 1999 review of 51 studies calculated that if both the nitrites and leukocyte esterase are positive on urine dipstick testing, the LR+ is 4.2; if both are negative the LR– is 0.3.^1,3 Since the probability of UTI for patients with the symptom triad is so high, dipstick urinalysis is unlikely to alter management regardless of whether nitrites and leukocyte esterase were both positive or negative (posttest probability=98%–99% and 80%, respectively). If urine dipstick or other office-based tests are not needed to make the diagnosis of uncomplicated UTI for a patient with the classic triad of symptoms, then telephone treatment based on symptoms may be reasonable. Women who have recurrent UTIs (2 or more culture positive UTIs over the previous 12 months) can accurately self-diagnose subsequent UTIs based on symptoms (LR+=4.0).^4,5

A recent retrospective case series⁶ evaluated a telephone guideline for the empiric treatment of UTI for 4177 women in a California HMO. UTI criteria were ≤10 days of dysuria; frequency, urgency, pressure, or increased nocturia; or gross hematuria. Women were excluded if they had any one of a variety of contraindications (TABLE 1). Upper tract infection occurred in 21 patients (1.1%) within 60 days of telephone treatment, two thirds of which likely represented treatment failures. This is similar to rates in control groups of other studies. Fourteen women (1.5%) received care for sexually transmitted diseases or other gynecologic conditions, primarily bacterial vaginitis, within 60 days of telephone treatment. Of note, 6% of the cohort were elderly, diabetic, taking glucocorticoids or early in pregnancy and are typically excluded from other studies. This higher-risk group did not have an increased incidence of either sepsis or pyelonephritis.⁶ No increase in adverse outcomes was seen in another study of a telephone treatment protocol.⁷

Several studies^6-8 have noted that telephone treatment protocols increase the use of protocol-recommended antibiotics (eg, generally less expensive agents such as trimethoprim-sulfamethoxazole), which may help limit resistance to fluoroquinolones. However, specific data are not available.

McIssac et al⁹ reviewed a cohort of 231 women presenting to family physicians’ offices with uncomplicated cystitis symptoms. Empiric therapy resulted in approximately 40% of women unnecessarily receiving antibiotics. Treating only women with classic cystitis symptoms and pyuria would have decreased the unnecessary use of antibiotics to 26.2%, but fewer women with confirmed cystitis would have received immediate antibiotics (66.4% vs 91.8%). They derived a clinical decision rule designed to balance false positives and false negatives. It recommends immediate antibiotic treatment if women have ≥2 of 4 signs or symptoms: dysuria, leukocyte esterase (greater than trace), positive nitrites, or blood (greater than trace) on dipstick (LR+=2.29). Otherwise the rule recommends a culture to guide antibiotic therapy. This decision rule would have reduced unnecessary antibiotic use by 27.5% while ensuring that more women with confirmed UTIs received immediate antibiotics (81.3%).

In 1999, Saint et al⁸ estimated savings of $367,000 for 147,000 women enrolled over 1 year after widespread guideline implementation. Two cost-effectiveness studies^10,11 of office treatment concluded that empiric treatment without additional testing is the least costly option in this setting. However, a recent, comprehensive cost-effectiveness study¹¹ concluded that if a patient presents to an office, the marginal cost of performing a pelvic examination and urine culture for women with a negative dipstick was relatively low ($4 to $32 per symptom day avoided).

TABLE 2
Diagnosis of urinary tract infection

Recommendations from others

A 2002 Institute for Clinical Systems Improvement guideline¹² advised offering telephone treatment of uncomplicated UTI for low-risk patients if preferred by both provider and patient.

TABLE 1
Contraindications to empiric antibiotics for urinary tract infection (telephone treatment)

Evidence summary

An evidence-based review¹ found 5 high-quality studies on the diagnosis of acute uncomplicated UTI among women. (“Uncomplicated” was defined as normal urinary tract and no contributing medical problems, such as diabetes, neurogenic bladder, renal stones.) UTIs were defined as the presence of significant bacteriuria (≥10⁴ to 10⁵ colony-forming units) on culture. A patient presenting to a clinician with 1 or more UTI symptoms had approximately a 50% chance of having significant bacteriuria on culture.¹ The authors estimated the pretest probability of UTI as 5% from the incidence of asymptomatic bacteriuria among healthy women.^1,2 This produced a LR+ of 19 simply for presenting to a clinician with 1 or more UTI symptoms.¹ The summary LRs for clinical signs and symptoms in the prediction of UTI after presentation to the office are found in TABLE 2. A history of a vaginal discharge or irritation has a LR– of 0.3, decreasing the probability of UTI for a patient presenting to the office from approximately 50% to 20%, so further testing would be indicated.¹

No single sign or symptom accurately predicted UTI. However, the triad of dysuria with frequency but without vaginal symptoms increased the probability of significant bacteriuria on culture from 50% to 96% (LR+=24.6).¹ In contrast, a 1999 review of 51 studies calculated that if both the nitrites and leukocyte esterase are positive on urine dipstick testing, the LR+ is 4.2; if both are negative the LR– is 0.3.^1,3 Since the probability of UTI for patients with the symptom triad is so high, dipstick urinalysis is unlikely to alter management regardless of whether nitrites and leukocyte esterase were both positive or negative (posttest probability=98%–99% and 80%, respectively). If urine dipstick or other office-based tests are not needed to make the diagnosis of uncomplicated UTI for a patient with the classic triad of symptoms, then telephone treatment based on symptoms may be reasonable. Women who have recurrent UTIs (2 or more culture positive UTIs over the previous 12 months) can accurately self-diagnose subsequent UTIs based on symptoms (LR+=4.0).^4,5

A recent retrospective case series⁶ evaluated a telephone guideline for the empiric treatment of UTI for 4177 women in a California HMO. UTI criteria were ≤10 days of dysuria; frequency, urgency, pressure, or increased nocturia; or gross hematuria. Women were excluded if they had any one of a variety of contraindications (TABLE 1). Upper tract infection occurred in 21 patients (1.1%) within 60 days of telephone treatment, two thirds of which likely represented treatment failures. This is similar to rates in control groups of other studies. Fourteen women (1.5%) received care for sexually transmitted diseases or other gynecologic conditions, primarily bacterial vaginitis, within 60 days of telephone treatment. Of note, 6% of the cohort were elderly, diabetic, taking glucocorticoids or early in pregnancy and are typically excluded from other studies. This higher-risk group did not have an increased incidence of either sepsis or pyelonephritis.⁶ No increase in adverse outcomes was seen in another study of a telephone treatment protocol.⁷

Several studies^6-8 have noted that telephone treatment protocols increase the use of protocol-recommended antibiotics (eg, generally less expensive agents such as trimethoprim-sulfamethoxazole), which may help limit resistance to fluoroquinolones. However, specific data are not available.

McIssac et al⁹ reviewed a cohort of 231 women presenting to family physicians’ offices with uncomplicated cystitis symptoms. Empiric therapy resulted in approximately 40% of women unnecessarily receiving antibiotics. Treating only women with classic cystitis symptoms and pyuria would have decreased the unnecessary use of antibiotics to 26.2%, but fewer women with confirmed cystitis would have received immediate antibiotics (66.4% vs 91.8%). They derived a clinical decision rule designed to balance false positives and false negatives. It recommends immediate antibiotic treatment if women have ≥2 of 4 signs or symptoms: dysuria, leukocyte esterase (greater than trace), positive nitrites, or blood (greater than trace) on dipstick (LR+=2.29). Otherwise the rule recommends a culture to guide antibiotic therapy. This decision rule would have reduced unnecessary antibiotic use by 27.5% while ensuring that more women with confirmed UTIs received immediate antibiotics (81.3%).

In 1999, Saint et al⁸ estimated savings of $367,000 for 147,000 women enrolled over 1 year after widespread guideline implementation. Two cost-effectiveness studies^10,11 of office treatment concluded that empiric treatment without additional testing is the least costly option in this setting. However, a recent, comprehensive cost-effectiveness study¹¹ concluded that if a patient presents to an office, the marginal cost of performing a pelvic examination and urine culture for women with a negative dipstick was relatively low ($4 to $32 per symptom day avoided).

TABLE 2
Diagnosis of urinary tract infection

Recommendations from others

A 2002 Institute for Clinical Systems Improvement guideline¹² advised offering telephone treatment of uncomplicated UTI for low-risk patients if preferred by both provider and patient.

Evidence summary

In one series, 87% of patients with the common cold had an abnormal computed tomography (CT) scan of the sinuses 48 to 96 hours after onset. Abnormalities visible on the CT scan persisted in 20% of patients at 2 weeks, yet epidemiological studies have shown that acute bacterial rhinosinusitis develops in only 0.5% to 2% of upper respiratory infections in adults. In primary care, only half of patients with a clinical diagnosis of acute bacterial rhinosinusitis have it proven upon aspiration.⁵

Two studies compared clinical findings with sinus puncture, the reference standard for acute bacterial rhinosinusitis. Berg found 4 independent predictors of aspirate purulence in Swedish emergency room patients with “paranasal” symptoms lasting <3 months (Table).¹ Together, unilateral purulent nasal discharge and predominantly unilateral pain predicated purulence on aspiration (sensitivity 79%, specificity 83%, positive predictive value [PPV], 80%). Clinical exam by an otolaryngologist had a PPV of 72%.

While emergency and primary care patients may differ, this study’s rate of aspiration-proven sinusitis (43%) is closer to that seen in primary care (50%) than in referral practices (70%–80%). This study’s limitations included unclear referral criteria, overlapping clinical predictors, and lack of culture data.

In a study of general practice patients in the United Kingdom with clinically diagnosed acute maxillary sinusitis, no signs or symptoms were independently associated with their illness.⁶ The authors concluded that the clinical examination was more or less worthless. Only patients with positive findings on CT scan underwent aspiration in this study. Less differentiated, less severe symptoms and a less stringent definition of positive aspiration in this study may account for the different results. Additionally, one third of patients eligible for this study refused participation or withdrew prior to sinus puncture.⁶

Other primary care studies used less accurate reference tests such as CT² (sensitivity and specificity unknown),⁵ x-ray³ (sensitivity 41%–90%, specificity 61%–85%),⁵ and ultrasound⁴ (sensitivity 76%, specificity 76%).⁷

Williams found 5 independent predictors of x-ray findings consistent with sinusitis in 247 male veterans:

• maxillary toothache (LR+, 2.5)
• no improvement with decongestants (LR+, 2.1)
• purulent secretions on exam (LR+, 2.1)
• abnormal transillumination (LR+, 1.6)
• colored nasal discharge (LR+, 1.5).³

In at least 2 of these 4 studies, purulent secretions in the nasal cavity (LR+, 2.1–5.5),^2,3 maxillary tooth pain (LR+, 2.1–2.5)^3,4 and purulent rhinorrhea (LR+, 1.5–1.9)^2,3,4 increased the likelihood of acute bacterial rhinosinusitis.

Finding purulent secretions in the nasal cavity is highly specific for acute bacterial rhinosinusitis (specificity 79%–100%)^1,2,3 but is uncommon and difficult to assess, requiring the use of a nasal speculum and possibly topical decongestants. The primary care physician’s overall clinical impression was accurate in Williams’ study but not in others.^2,4,6 Illness starting as the common cold and pain on bending forward were not predictors of acute bacterial rhinosinusitis.^2,3,4 Headache, bilateral maxillary pain, frontal sinus pain, fever, sinus tenderness on exam, and purulent pharyngeal discharge have not been shown to be useful in acute bacterial rhinosinusitis diagnosis.⁷

TABLE
Clinical prediction rule for acute bacterial rhinosinusitis

Recommendations from others

A recommendation from the Agency for Health Care Policy and Research suggests using symptomatic treatment initially when the prevalence of acute bacterial rhinosinusitis in patients with upper respiratory infection is <25%, and using clinical criteria (see Table) for acute bacterial rhinosinusitis diagnosis when prevalence is higher.⁵

The Centers for Disease Control and Prevention recommends reserving the diagnosis of acute bacterial rhinosinusitis for patients with symptoms lasting ≥7 days with maxillary pain or tenderness in the face or teeth (especially unilateral) and purulent nasal secretions.⁸

An otolaryngology guideline recommends considering acute bacterial rhinosinusitis when viral upper respiratory infection persists beyond 10 days or worsens after 5 to 7 days with similar symptoms.⁹ The 7-to-10-day specification is based on the natural history of rhinovirus infections (SOR: C).

A Canadian Medical Association evidence-based review recommended a score based on Williams’ 5 independent predictor symptoms:

• fewer than 2 symptoms rule out acute bacterial rhinosinusitis (PPV, <40%)
• 4 or more symptoms rule in acute bacterial rhinosinusitis (PPV, 81%) (level of evidence [LOE]: 4)
• 2 or 3 symptoms (PPV, 40%–63%) may benefit from radiography (SOR: C).¹⁰

Evidence summary

In one series, 87% of patients with the common cold had an abnormal computed tomography (CT) scan of the sinuses 48 to 96 hours after onset. Abnormalities visible on the CT scan persisted in 20% of patients at 2 weeks, yet epidemiological studies have shown that acute bacterial rhinosinusitis develops in only 0.5% to 2% of upper respiratory infections in adults. In primary care, only half of patients with a clinical diagnosis of acute bacterial rhinosinusitis have it proven upon aspiration.⁵

Two studies compared clinical findings with sinus puncture, the reference standard for acute bacterial rhinosinusitis. Berg found 4 independent predictors of aspirate purulence in Swedish emergency room patients with “paranasal” symptoms lasting <3 months (Table).¹ Together, unilateral purulent nasal discharge and predominantly unilateral pain predicated purulence on aspiration (sensitivity 79%, specificity 83%, positive predictive value [PPV], 80%). Clinical exam by an otolaryngologist had a PPV of 72%.

While emergency and primary care patients may differ, this study’s rate of aspiration-proven sinusitis (43%) is closer to that seen in primary care (50%) than in referral practices (70%–80%). This study’s limitations included unclear referral criteria, overlapping clinical predictors, and lack of culture data.

In a study of general practice patients in the United Kingdom with clinically diagnosed acute maxillary sinusitis, no signs or symptoms were independently associated with their illness.⁶ The authors concluded that the clinical examination was more or less worthless. Only patients with positive findings on CT scan underwent aspiration in this study. Less differentiated, less severe symptoms and a less stringent definition of positive aspiration in this study may account for the different results. Additionally, one third of patients eligible for this study refused participation or withdrew prior to sinus puncture.⁶

Other primary care studies used less accurate reference tests such as CT² (sensitivity and specificity unknown),⁵ x-ray³ (sensitivity 41%–90%, specificity 61%–85%),⁵ and ultrasound⁴ (sensitivity 76%, specificity 76%).⁷

Williams found 5 independent predictors of x-ray findings consistent with sinusitis in 247 male veterans:

• maxillary toothache (LR+, 2.5)
• no improvement with decongestants (LR+, 2.1)
• purulent secretions on exam (LR+, 2.1)
• abnormal transillumination (LR+, 1.6)
• colored nasal discharge (LR+, 1.5).³

In at least 2 of these 4 studies, purulent secretions in the nasal cavity (LR+, 2.1–5.5),^2,3 maxillary tooth pain (LR+, 2.1–2.5)^3,4 and purulent rhinorrhea (LR+, 1.5–1.9)^2,3,4 increased the likelihood of acute bacterial rhinosinusitis.

Finding purulent secretions in the nasal cavity is highly specific for acute bacterial rhinosinusitis (specificity 79%–100%)^1,2,3 but is uncommon and difficult to assess, requiring the use of a nasal speculum and possibly topical decongestants. The primary care physician’s overall clinical impression was accurate in Williams’ study but not in others.^2,4,6 Illness starting as the common cold and pain on bending forward were not predictors of acute bacterial rhinosinusitis.^2,3,4 Headache, bilateral maxillary pain, frontal sinus pain, fever, sinus tenderness on exam, and purulent pharyngeal discharge have not been shown to be useful in acute bacterial rhinosinusitis diagnosis.⁷

TABLE
Clinical prediction rule for acute bacterial rhinosinusitis

Recommendations from others

A recommendation from the Agency for Health Care Policy and Research suggests using symptomatic treatment initially when the prevalence of acute bacterial rhinosinusitis in patients with upper respiratory infection is <25%, and using clinical criteria (see Table) for acute bacterial rhinosinusitis diagnosis when prevalence is higher.⁵

The Centers for Disease Control and Prevention recommends reserving the diagnosis of acute bacterial rhinosinusitis for patients with symptoms lasting ≥7 days with maxillary pain or tenderness in the face or teeth (especially unilateral) and purulent nasal secretions.⁸

An otolaryngology guideline recommends considering acute bacterial rhinosinusitis when viral upper respiratory infection persists beyond 10 days or worsens after 5 to 7 days with similar symptoms.⁹ The 7-to-10-day specification is based on the natural history of rhinovirus infections (SOR: C).

A Canadian Medical Association evidence-based review recommended a score based on Williams’ 5 independent predictor symptoms:

• fewer than 2 symptoms rule out acute bacterial rhinosinusitis (PPV, <40%)
• 4 or more symptoms rule in acute bacterial rhinosinusitis (PPV, 81%) (level of evidence [LOE]: 4)
• 2 or 3 symptoms (PPV, 40%–63%) may benefit from radiography (SOR: C).¹⁰

Evidence summary

In one series, 87% of patients with the common cold had an abnormal computed tomography (CT) scan of the sinuses 48 to 96 hours after onset. Abnormalities visible on the CT scan persisted in 20% of patients at 2 weeks, yet epidemiological studies have shown that acute bacterial rhinosinusitis develops in only 0.5% to 2% of upper respiratory infections in adults. In primary care, only half of patients with a clinical diagnosis of acute bacterial rhinosinusitis have it proven upon aspiration.⁵

Two studies compared clinical findings with sinus puncture, the reference standard for acute bacterial rhinosinusitis. Berg found 4 independent predictors of aspirate purulence in Swedish emergency room patients with “paranasal” symptoms lasting <3 months (Table).¹ Together, unilateral purulent nasal discharge and predominantly unilateral pain predicated purulence on aspiration (sensitivity 79%, specificity 83%, positive predictive value [PPV], 80%). Clinical exam by an otolaryngologist had a PPV of 72%.

While emergency and primary care patients may differ, this study’s rate of aspiration-proven sinusitis (43%) is closer to that seen in primary care (50%) than in referral practices (70%–80%). This study’s limitations included unclear referral criteria, overlapping clinical predictors, and lack of culture data.

In a study of general practice patients in the United Kingdom with clinically diagnosed acute maxillary sinusitis, no signs or symptoms were independently associated with their illness.⁶ The authors concluded that the clinical examination was more or less worthless. Only patients with positive findings on CT scan underwent aspiration in this study. Less differentiated, less severe symptoms and a less stringent definition of positive aspiration in this study may account for the different results. Additionally, one third of patients eligible for this study refused participation or withdrew prior to sinus puncture.⁶

Other primary care studies used less accurate reference tests such as CT² (sensitivity and specificity unknown),⁵ x-ray³ (sensitivity 41%–90%, specificity 61%–85%),⁵ and ultrasound⁴ (sensitivity 76%, specificity 76%).⁷

Williams found 5 independent predictors of x-ray findings consistent with sinusitis in 247 male veterans:

• maxillary toothache (LR+, 2.5)
• no improvement with decongestants (LR+, 2.1)
• purulent secretions on exam (LR+, 2.1)
• abnormal transillumination (LR+, 1.6)
• colored nasal discharge (LR+, 1.5).³

In at least 2 of these 4 studies, purulent secretions in the nasal cavity (LR+, 2.1–5.5),^2,3 maxillary tooth pain (LR+, 2.1–2.5)^3,4 and purulent rhinorrhea (LR+, 1.5–1.9)^2,3,4 increased the likelihood of acute bacterial rhinosinusitis.

Finding purulent secretions in the nasal cavity is highly specific for acute bacterial rhinosinusitis (specificity 79%–100%)^1,2,3 but is uncommon and difficult to assess, requiring the use of a nasal speculum and possibly topical decongestants. The primary care physician’s overall clinical impression was accurate in Williams’ study but not in others.^2,4,6 Illness starting as the common cold and pain on bending forward were not predictors of acute bacterial rhinosinusitis.^2,3,4 Headache, bilateral maxillary pain, frontal sinus pain, fever, sinus tenderness on exam, and purulent pharyngeal discharge have not been shown to be useful in acute bacterial rhinosinusitis diagnosis.⁷

TABLE
Clinical prediction rule for acute bacterial rhinosinusitis

Recommendations from others

A recommendation from the Agency for Health Care Policy and Research suggests using symptomatic treatment initially when the prevalence of acute bacterial rhinosinusitis in patients with upper respiratory infection is <25%, and using clinical criteria (see Table) for acute bacterial rhinosinusitis diagnosis when prevalence is higher.⁵

The Centers for Disease Control and Prevention recommends reserving the diagnosis of acute bacterial rhinosinusitis for patients with symptoms lasting ≥7 days with maxillary pain or tenderness in the face or teeth (especially unilateral) and purulent nasal secretions.⁸

An otolaryngology guideline recommends considering acute bacterial rhinosinusitis when viral upper respiratory infection persists beyond 10 days or worsens after 5 to 7 days with similar symptoms.⁹ The 7-to-10-day specification is based on the natural history of rhinovirus infections (SOR: C).

A Canadian Medical Association evidence-based review recommended a score based on Williams’ 5 independent predictor symptoms:

• fewer than 2 symptoms rule out acute bacterial rhinosinusitis (PPV, <40%)
• 4 or more symptoms rule in acute bacterial rhinosinusitis (PPV, 81%) (level of evidence [LOE]: 4)
• 2 or 3 symptoms (PPV, 40%–63%) may benefit from radiography (SOR: C).¹⁰