Clinical Inquiries

Evidence summary

Transient ischemic attack (Figure) is a temporary, focal brain or retinal deficit caused by vascular disease that clears completely in less than 24 hours.¹ A large prospective cohort study recently estimated the risk of stroke after a TIA or minor stroke to be 8% to 12% at 7 days and 11% to 15% at 1 month.²

In a large retrospective cohort study, 5% of TIA patients returned to the emergency department with a stroke within the first 2 days after TIA.³ Another 6% returned with a stroke within 90 days. Five independent risk factors were identified: age >60 years, diabetes mellitus, duration of TIA longer than 10 minutes, signs or symptoms of weakness, and speech impairment. Thirty-four percent of patients with all 5 risk factors, and none of the patients without any risk factors, had a stroke within 90 days. Of note, 13% of the TIA patients had an arrhythmia, congestive heart failure, unstable angina, myocardial infarction, stroke, or recurrent TIA within 4 days of initial presenting with a TIA. Twenty-five percent of the patients experienced 1 of these cardiovascular events during the 3 months of follow-up.

In a retrospective case review of TIA and stroke patients, the hospital admissions of 4 of 21 TIA patients were retrospectively categorized as medically justified.⁴ Admission was categorized as medically justified if the patient had 1 or more of the following criteria: another diagnosis that warranted admission, inadequate home situation, altered mental status, an adverse event during hospitalization including worsening of the deficit, and if the patient underwent some hospital-based treatment that could not be provided on an out-patient basis. Ease and rapidity of evaluation was not considered medically justifiable and outcome improvement (stroke prevention) was not studied.

Two retrospective chart reviews of TIA found considerable practice variability in the evaluation of TIA patient. In 1 study of TIA patients presenting to an emergency department, 81% had a computed tomography scan, 75% had electrocardiogram, and 74% had a complete blood count.⁵ Carotid Doppler imaging was performed in the emergency department in 16%, and 26% were referred for outpatient Doppler studies. One percent had an ECG in the emergency department, and 16% were given ECGs as outpatients. Seventy-five percent of patients were discharged home. Those hospitalized had a median length of stay of 1 day. In the second study, 31% of the TIA patients had no diagnostic studies performed during the first month after presenting to their primary care physician.⁶

FIGURE
Expeditious evaluation of TIA is imperative

Recommendations from others

The American Heart Association (AHA) recommends that physicians use a stepwise approach to TIA evaluation as outlined in the Table. The AHA also recommends that the diagnostic evaluation of patients seen within 7 days of a TIA should be completed within 1 week or less. The AHA leaves the decision whether to hospitalize a patient up to the physician based on a patient’s circumstances. The goals of diagnostic testing are to identify or exclude causes of TIA requiring specific therapy, to assess modifiable risk factors, and to determine prognosis.⁷

The National Stroke Association recommends that patients with known high-grade stenosis in a vascular territory appropriate to the symptoms, and patients with recurrent symptoms, undergo urgent evaluation. Evaluation includes imaging and ruling out other causes of TIA. Patients should be admitted to the hospital if imaging is not immediately available. If indicated, carotid endarterectomy should be performed without delay.⁸

TABLE
Stepwise diagnostic evaluation for patients with transient ischemic attack

Evidence summary

Transient ischemic attack (Figure) is a temporary, focal brain or retinal deficit caused by vascular disease that clears completely in less than 24 hours.¹ A large prospective cohort study recently estimated the risk of stroke after a TIA or minor stroke to be 8% to 12% at 7 days and 11% to 15% at 1 month.²

In a large retrospective cohort study, 5% of TIA patients returned to the emergency department with a stroke within the first 2 days after TIA.³ Another 6% returned with a stroke within 90 days. Five independent risk factors were identified: age >60 years, diabetes mellitus, duration of TIA longer than 10 minutes, signs or symptoms of weakness, and speech impairment. Thirty-four percent of patients with all 5 risk factors, and none of the patients without any risk factors, had a stroke within 90 days. Of note, 13% of the TIA patients had an arrhythmia, congestive heart failure, unstable angina, myocardial infarction, stroke, or recurrent TIA within 4 days of initial presenting with a TIA. Twenty-five percent of the patients experienced 1 of these cardiovascular events during the 3 months of follow-up.

In a retrospective case review of TIA and stroke patients, the hospital admissions of 4 of 21 TIA patients were retrospectively categorized as medically justified.⁴ Admission was categorized as medically justified if the patient had 1 or more of the following criteria: another diagnosis that warranted admission, inadequate home situation, altered mental status, an adverse event during hospitalization including worsening of the deficit, and if the patient underwent some hospital-based treatment that could not be provided on an out-patient basis. Ease and rapidity of evaluation was not considered medically justifiable and outcome improvement (stroke prevention) was not studied.

Two retrospective chart reviews of TIA found considerable practice variability in the evaluation of TIA patient. In 1 study of TIA patients presenting to an emergency department, 81% had a computed tomography scan, 75% had electrocardiogram, and 74% had a complete blood count.⁵ Carotid Doppler imaging was performed in the emergency department in 16%, and 26% were referred for outpatient Doppler studies. One percent had an ECG in the emergency department, and 16% were given ECGs as outpatients. Seventy-five percent of patients were discharged home. Those hospitalized had a median length of stay of 1 day. In the second study, 31% of the TIA patients had no diagnostic studies performed during the first month after presenting to their primary care physician.⁶

FIGURE
Expeditious evaluation of TIA is imperative

Recommendations from others

The American Heart Association (AHA) recommends that physicians use a stepwise approach to TIA evaluation as outlined in the Table. The AHA also recommends that the diagnostic evaluation of patients seen within 7 days of a TIA should be completed within 1 week or less. The AHA leaves the decision whether to hospitalize a patient up to the physician based on a patient’s circumstances. The goals of diagnostic testing are to identify or exclude causes of TIA requiring specific therapy, to assess modifiable risk factors, and to determine prognosis.⁷

The National Stroke Association recommends that patients with known high-grade stenosis in a vascular territory appropriate to the symptoms, and patients with recurrent symptoms, undergo urgent evaluation. Evaluation includes imaging and ruling out other causes of TIA. Patients should be admitted to the hospital if imaging is not immediately available. If indicated, carotid endarterectomy should be performed without delay.⁸

TABLE
Stepwise diagnostic evaluation for patients with transient ischemic attack

Evidence summary

Transient ischemic attack (Figure) is a temporary, focal brain or retinal deficit caused by vascular disease that clears completely in less than 24 hours.¹ A large prospective cohort study recently estimated the risk of stroke after a TIA or minor stroke to be 8% to 12% at 7 days and 11% to 15% at 1 month.²

In a large retrospective cohort study, 5% of TIA patients returned to the emergency department with a stroke within the first 2 days after TIA.³ Another 6% returned with a stroke within 90 days. Five independent risk factors were identified: age >60 years, diabetes mellitus, duration of TIA longer than 10 minutes, signs or symptoms of weakness, and speech impairment. Thirty-four percent of patients with all 5 risk factors, and none of the patients without any risk factors, had a stroke within 90 days. Of note, 13% of the TIA patients had an arrhythmia, congestive heart failure, unstable angina, myocardial infarction, stroke, or recurrent TIA within 4 days of initial presenting with a TIA. Twenty-five percent of the patients experienced 1 of these cardiovascular events during the 3 months of follow-up.

In a retrospective case review of TIA and stroke patients, the hospital admissions of 4 of 21 TIA patients were retrospectively categorized as medically justified.⁴ Admission was categorized as medically justified if the patient had 1 or more of the following criteria: another diagnosis that warranted admission, inadequate home situation, altered mental status, an adverse event during hospitalization including worsening of the deficit, and if the patient underwent some hospital-based treatment that could not be provided on an out-patient basis. Ease and rapidity of evaluation was not considered medically justifiable and outcome improvement (stroke prevention) was not studied.

Two retrospective chart reviews of TIA found considerable practice variability in the evaluation of TIA patient. In 1 study of TIA patients presenting to an emergency department, 81% had a computed tomography scan, 75% had electrocardiogram, and 74% had a complete blood count.⁵ Carotid Doppler imaging was performed in the emergency department in 16%, and 26% were referred for outpatient Doppler studies. One percent had an ECG in the emergency department, and 16% were given ECGs as outpatients. Seventy-five percent of patients were discharged home. Those hospitalized had a median length of stay of 1 day. In the second study, 31% of the TIA patients had no diagnostic studies performed during the first month after presenting to their primary care physician.⁶

FIGURE
Expeditious evaluation of TIA is imperative

Recommendations from others

The American Heart Association (AHA) recommends that physicians use a stepwise approach to TIA evaluation as outlined in the Table. The AHA also recommends that the diagnostic evaluation of patients seen within 7 days of a TIA should be completed within 1 week or less. The AHA leaves the decision whether to hospitalize a patient up to the physician based on a patient’s circumstances. The goals of diagnostic testing are to identify or exclude causes of TIA requiring specific therapy, to assess modifiable risk factors, and to determine prognosis.⁷

The National Stroke Association recommends that patients with known high-grade stenosis in a vascular territory appropriate to the symptoms, and patients with recurrent symptoms, undergo urgent evaluation. Evaluation includes imaging and ruling out other causes of TIA. Patients should be admitted to the hospital if imaging is not immediately available. If indicated, carotid endarterectomy should be performed without delay.⁸

TABLE
Stepwise diagnostic evaluation for patients with transient ischemic attack

Evidence summary

Clinical guidelines for osteoarthritis recommend acetaminophen as first-line therapy followed by an NSAID or cyclooxygenase-2 (COX-2) inhibitor, and many patients are treated with combination therapy.

Several small randomized controlled trials have compared the individual efficacy of NSAIDs and acetaminophen in osteoarthritis and have found that both provide more pain relief than placebo.^1-3There is a trend toward improved pain relief with NSAIDs compared with acetaminophen in the initial treatment period; however, few long-term studies of efficacy have been reported. One randomized controlled trial comparing 750 mg/d naproxen (Aleve, Naprosyn) with 2600 mg/d acetaminophen for 2 years found similar pain relief for both medications and a dropout rate of 65% in both groups.² Similar numbers of persons taking acetaminophen or naproxen dropped out because of adverse effects (20%) or lack of efficacy (19%), and no difference was seen in functional improvement between the 2 groups.

A 6-week randomized double-blind crossover trial of 227 patients comparing 75 mg diclofenac and 200 mg misoprostol (Arthrotec) with acetaminophen 4 g/d found the diclofenac-misoprostol combination provided more pain control than acetaminophen alone. Adverse events were slightly more common in the diclofenac group (54% vs 46%; P=.046).⁴

The COX-2 inhibitors rofecoxib (Vioxx) and celecoxib (Celebrex) have been shown to provide equal pain relief compared with naproxen for patients with osteoarthritis.⁵ One industry-sponsored randomized trial found rofecoxib superior to celecoxib, and both superior to acetaminophen in treatment of osteoarthritis pain.⁶ There was no difference in the incidence of side effects among the 3 medications. Thirty percent of patients taking 4 g/d acetaminophen discontinued the study because of lack of efficacy, compared with 20% of those taking either celecoxib or rofecoxib.⁶

Few studies have evaluated the safety or efficacy of the combination of NSAIDs and acetaminophen in osteoarthritis. One double-blind, double-dummy crossover trial of 18 patients with osteoarthritis of the hip compared naproxen at doses of 500 mg and 1000 mg, with and without 4 g/d of acetaminophen, and 1500 mg/d of naproxen alone over 5 one-week trial periods.⁷Adding acetaminophen improved patient-reported pain scores compared with naproxen alone. Higher doses of naproxen alone provided less pain relief than a lower dose of naproxen combined with acetaminophen. GI side effects increased with the increase in naproxen dose, but were unaffected by the addition of acetaminophen. Functional ability was not affected during this short study. A similar study by the same researchers of patients with rheumatoid arthritis found similar results.⁷

One randomized, double-blind, crossover trial compared single doses of tolmetin (Tolectin, 100, 150, 200 mg) and acetaminophen (400 mg) alone and in combination with placebo in the control of experimentally induced pain (thermal and electrical stimulation). Acetaminophen alone did not differ from placebo in pain control; however, the combinations of acetaminophen with tolmetin provided similar pain relief to higher doses of tolmetin alone.⁸ No studies have evaluated the efficacy or safety of acetaminophen combined with rofecoxib or celecoxib.

Regarding the risks of combining acetaminophen with NSAIDs, 1 nested case-control study based on the entire enrollment panel of the British National Health Service characterized the risk of upper GI side effects among persons taking NSAIDs or acetaminophen alone or in combination. The study evaluated medications in use at the time of an upper GI bleed, controlling for age, sex, and concomitant medications (corticosteroids, H₂ receptor antagonists, omeprazole, anticoagulants, and others) and excluding patients with varices, alcohol-related disorders, liver disease, and cancer; no attempt was made to control other comorbidities. The relative risk of upper GI perforation or bleeding for patients taking 2g/d acetaminophen or high-dose NSAIDs was 2.4 (95% confidence interval [CI], 1.7–3.5) and 3.6 (95% CI, 2.9–4.3), respectively. Concomitant use of an NSAID with 2 g/d of acetaminophen showed a relative risk of upper GI perforation or bleed of 16.6 (95% CI, 11.0–24.9). Acetaminophen doses <2 g/d conferred no additional risk for serious upper GI side effects.⁹

A systematic review of selective COX-2 inhibitors vs naproxen found fewer endoscopically detected ulcers in patients taking celecoxib but no difference in serious gastrointestinal bleeds.⁵ A meta-analysis of randomized controlled trials found a higher incidence of serious thrombotic cardiovascular events among patients taking COX-2 inhibitors compared with naprosyn.¹⁰ The safety profile of rofecoxib and celecoxib in the long-term treatment of pain is not fully understood at this time.

Recommendations from others

The American College of Rheumatology (ACR) recommends acetaminophen up to 4 g/d as a first-line pharmacologic treatment for osteoarthritis of the hip and knee, and advises NSAIDs be used at the lowest effective dose if they are necessary for pain control.¹¹ The ACR does not specifically comment on combining NSAID and acetaminophen use. The American Academy of Orthopaedic Surgeons recommends initial use of an NSAID or acetaminophen, but does not comment on the combination of NSAIDs and acetaminophen.¹²

Evidence summary

Clinical guidelines for osteoarthritis recommend acetaminophen as first-line therapy followed by an NSAID or cyclooxygenase-2 (COX-2) inhibitor, and many patients are treated with combination therapy.

Several small randomized controlled trials have compared the individual efficacy of NSAIDs and acetaminophen in osteoarthritis and have found that both provide more pain relief than placebo.^1-3There is a trend toward improved pain relief with NSAIDs compared with acetaminophen in the initial treatment period; however, few long-term studies of efficacy have been reported. One randomized controlled trial comparing 750 mg/d naproxen (Aleve, Naprosyn) with 2600 mg/d acetaminophen for 2 years found similar pain relief for both medications and a dropout rate of 65% in both groups.² Similar numbers of persons taking acetaminophen or naproxen dropped out because of adverse effects (20%) or lack of efficacy (19%), and no difference was seen in functional improvement between the 2 groups.

A 6-week randomized double-blind crossover trial of 227 patients comparing 75 mg diclofenac and 200 mg misoprostol (Arthrotec) with acetaminophen 4 g/d found the diclofenac-misoprostol combination provided more pain control than acetaminophen alone. Adverse events were slightly more common in the diclofenac group (54% vs 46%; P=.046).⁴

The COX-2 inhibitors rofecoxib (Vioxx) and celecoxib (Celebrex) have been shown to provide equal pain relief compared with naproxen for patients with osteoarthritis.⁵ One industry-sponsored randomized trial found rofecoxib superior to celecoxib, and both superior to acetaminophen in treatment of osteoarthritis pain.⁶ There was no difference in the incidence of side effects among the 3 medications. Thirty percent of patients taking 4 g/d acetaminophen discontinued the study because of lack of efficacy, compared with 20% of those taking either celecoxib or rofecoxib.⁶

Few studies have evaluated the safety or efficacy of the combination of NSAIDs and acetaminophen in osteoarthritis. One double-blind, double-dummy crossover trial of 18 patients with osteoarthritis of the hip compared naproxen at doses of 500 mg and 1000 mg, with and without 4 g/d of acetaminophen, and 1500 mg/d of naproxen alone over 5 one-week trial periods.⁷Adding acetaminophen improved patient-reported pain scores compared with naproxen alone. Higher doses of naproxen alone provided less pain relief than a lower dose of naproxen combined with acetaminophen. GI side effects increased with the increase in naproxen dose, but were unaffected by the addition of acetaminophen. Functional ability was not affected during this short study. A similar study by the same researchers of patients with rheumatoid arthritis found similar results.⁷

One randomized, double-blind, crossover trial compared single doses of tolmetin (Tolectin, 100, 150, 200 mg) and acetaminophen (400 mg) alone and in combination with placebo in the control of experimentally induced pain (thermal and electrical stimulation). Acetaminophen alone did not differ from placebo in pain control; however, the combinations of acetaminophen with tolmetin provided similar pain relief to higher doses of tolmetin alone.⁸ No studies have evaluated the efficacy or safety of acetaminophen combined with rofecoxib or celecoxib.

Regarding the risks of combining acetaminophen with NSAIDs, 1 nested case-control study based on the entire enrollment panel of the British National Health Service characterized the risk of upper GI side effects among persons taking NSAIDs or acetaminophen alone or in combination. The study evaluated medications in use at the time of an upper GI bleed, controlling for age, sex, and concomitant medications (corticosteroids, H₂ receptor antagonists, omeprazole, anticoagulants, and others) and excluding patients with varices, alcohol-related disorders, liver disease, and cancer; no attempt was made to control other comorbidities. The relative risk of upper GI perforation or bleeding for patients taking 2g/d acetaminophen or high-dose NSAIDs was 2.4 (95% confidence interval [CI], 1.7–3.5) and 3.6 (95% CI, 2.9–4.3), respectively. Concomitant use of an NSAID with 2 g/d of acetaminophen showed a relative risk of upper GI perforation or bleed of 16.6 (95% CI, 11.0–24.9). Acetaminophen doses <2 g/d conferred no additional risk for serious upper GI side effects.⁹

A systematic review of selective COX-2 inhibitors vs naproxen found fewer endoscopically detected ulcers in patients taking celecoxib but no difference in serious gastrointestinal bleeds.⁵ A meta-analysis of randomized controlled trials found a higher incidence of serious thrombotic cardiovascular events among patients taking COX-2 inhibitors compared with naprosyn.¹⁰ The safety profile of rofecoxib and celecoxib in the long-term treatment of pain is not fully understood at this time.

Recommendations from others

The American College of Rheumatology (ACR) recommends acetaminophen up to 4 g/d as a first-line pharmacologic treatment for osteoarthritis of the hip and knee, and advises NSAIDs be used at the lowest effective dose if they are necessary for pain control.¹¹ The ACR does not specifically comment on combining NSAID and acetaminophen use. The American Academy of Orthopaedic Surgeons recommends initial use of an NSAID or acetaminophen, but does not comment on the combination of NSAIDs and acetaminophen.¹²

Evidence summary

Clinical guidelines for osteoarthritis recommend acetaminophen as first-line therapy followed by an NSAID or cyclooxygenase-2 (COX-2) inhibitor, and many patients are treated with combination therapy.

Several small randomized controlled trials have compared the individual efficacy of NSAIDs and acetaminophen in osteoarthritis and have found that both provide more pain relief than placebo.^1-3There is a trend toward improved pain relief with NSAIDs compared with acetaminophen in the initial treatment period; however, few long-term studies of efficacy have been reported. One randomized controlled trial comparing 750 mg/d naproxen (Aleve, Naprosyn) with 2600 mg/d acetaminophen for 2 years found similar pain relief for both medications and a dropout rate of 65% in both groups.² Similar numbers of persons taking acetaminophen or naproxen dropped out because of adverse effects (20%) or lack of efficacy (19%), and no difference was seen in functional improvement between the 2 groups.

A 6-week randomized double-blind crossover trial of 227 patients comparing 75 mg diclofenac and 200 mg misoprostol (Arthrotec) with acetaminophen 4 g/d found the diclofenac-misoprostol combination provided more pain control than acetaminophen alone. Adverse events were slightly more common in the diclofenac group (54% vs 46%; P=.046).⁴

The COX-2 inhibitors rofecoxib (Vioxx) and celecoxib (Celebrex) have been shown to provide equal pain relief compared with naproxen for patients with osteoarthritis.⁵ One industry-sponsored randomized trial found rofecoxib superior to celecoxib, and both superior to acetaminophen in treatment of osteoarthritis pain.⁶ There was no difference in the incidence of side effects among the 3 medications. Thirty percent of patients taking 4 g/d acetaminophen discontinued the study because of lack of efficacy, compared with 20% of those taking either celecoxib or rofecoxib.⁶

Few studies have evaluated the safety or efficacy of the combination of NSAIDs and acetaminophen in osteoarthritis. One double-blind, double-dummy crossover trial of 18 patients with osteoarthritis of the hip compared naproxen at doses of 500 mg and 1000 mg, with and without 4 g/d of acetaminophen, and 1500 mg/d of naproxen alone over 5 one-week trial periods.⁷Adding acetaminophen improved patient-reported pain scores compared with naproxen alone. Higher doses of naproxen alone provided less pain relief than a lower dose of naproxen combined with acetaminophen. GI side effects increased with the increase in naproxen dose, but were unaffected by the addition of acetaminophen. Functional ability was not affected during this short study. A similar study by the same researchers of patients with rheumatoid arthritis found similar results.⁷

One randomized, double-blind, crossover trial compared single doses of tolmetin (Tolectin, 100, 150, 200 mg) and acetaminophen (400 mg) alone and in combination with placebo in the control of experimentally induced pain (thermal and electrical stimulation). Acetaminophen alone did not differ from placebo in pain control; however, the combinations of acetaminophen with tolmetin provided similar pain relief to higher doses of tolmetin alone.⁸ No studies have evaluated the efficacy or safety of acetaminophen combined with rofecoxib or celecoxib.

Regarding the risks of combining acetaminophen with NSAIDs, 1 nested case-control study based on the entire enrollment panel of the British National Health Service characterized the risk of upper GI side effects among persons taking NSAIDs or acetaminophen alone or in combination. The study evaluated medications in use at the time of an upper GI bleed, controlling for age, sex, and concomitant medications (corticosteroids, H₂ receptor antagonists, omeprazole, anticoagulants, and others) and excluding patients with varices, alcohol-related disorders, liver disease, and cancer; no attempt was made to control other comorbidities. The relative risk of upper GI perforation or bleeding for patients taking 2g/d acetaminophen or high-dose NSAIDs was 2.4 (95% confidence interval [CI], 1.7–3.5) and 3.6 (95% CI, 2.9–4.3), respectively. Concomitant use of an NSAID with 2 g/d of acetaminophen showed a relative risk of upper GI perforation or bleed of 16.6 (95% CI, 11.0–24.9). Acetaminophen doses <2 g/d conferred no additional risk for serious upper GI side effects.⁹

A systematic review of selective COX-2 inhibitors vs naproxen found fewer endoscopically detected ulcers in patients taking celecoxib but no difference in serious gastrointestinal bleeds.⁵ A meta-analysis of randomized controlled trials found a higher incidence of serious thrombotic cardiovascular events among patients taking COX-2 inhibitors compared with naprosyn.¹⁰ The safety profile of rofecoxib and celecoxib in the long-term treatment of pain is not fully understood at this time.

Recommendations from others

The American College of Rheumatology (ACR) recommends acetaminophen up to 4 g/d as a first-line pharmacologic treatment for osteoarthritis of the hip and knee, and advises NSAIDs be used at the lowest effective dose if they are necessary for pain control.¹¹ The ACR does not specifically comment on combining NSAID and acetaminophen use. The American Academy of Orthopaedic Surgeons recommends initial use of an NSAID or acetaminophen, but does not comment on the combination of NSAIDs and acetaminophen.¹²

Evidence summary

Recommendations about antibiotic prophylaxis are based on several premises. Reflux predisposes children to acute pyelonephritis; reflux plus infection leads to reflux nephropathy and ultimately to renal scarring. In theory, if antibiotics could be initiated at the appropriate time and be maintained until reflux resolves, we could successfully prevent infection and scarring.⁴

A recent systematic review evaluated the use of antibiotics to prevent UTI in children.⁵ This review of 5 randomized controlled trials included a total of 463 children between the ages of 2 months to 16 years. Three out of 5 trials evaluated the effectiveness of antibiotic treatment for 2 to 6 months to prevent subsequent off-treatment recurrence. The 2 smaller trials (n=71) evaluated the use of low-dose long-term antibiotics to prevent UTI.

There was a clinically, but not statistically, significant trend towards reduced risk of UTI during long-term antibiotic treatment (risk reduction [RR]=0.31; 95% confidence interval [CI]=0.10–1.00); however, no sustained benefit was seen once antibiotics were stopped (RR=0.79; 95% CI, 0.61–1.02). There were many problems with the methodological quality of these trials, including significant heterogeneity. The researchers concluded that well-designed randomized controlled trails are still needed to evaluate this commonly used intervention in the pediatric population.⁴ Benefits for long-term prophylaxis are even less clear in children with low-grade reflux (I–II).⁵ Furthermore, no randomized controlled trials assess whether prophylaxis prevents the development of new renal scars in children.⁶

In addition, a recent systematic review of studies done in children with normal urinary tracts, as well in children with neurogenic bladders, found that the available evidence is of low quality. Only 6 out of 31 potential studies fulfilled the inclusion criteria. These were small (mean sample size was 28), and the quality scores of all 6 trials were low, indicating that the evidence may be unreliable.⁷

Two of 3 studies done in children with normal urinary tracts demonstrated statistically significant higher rates of UTI recurrence in control groups compared with treatment groups receiving 6 to 10 months of either nitrofurantoin or cotrimoxazole (RR=24–31). The third study showed no difference between groups.

One of 2 trials in children with neurogenic bladder demonstrated higher recurrence rates of 2.9 per 10 patient years for patients receiving antibiotics compared with 1.5 in the untreated group. The other study showed lower recurrence rates of 17.1 for patients receiving antibiotics, compared with 33 in the untreated group.⁷Neither of these findings were statistically significant.

A different meta-analysis of 15 controlled clinical trials in children with neurogenic bladder due to spinal cord dysfunction. This analysis showed that antibiotic prophylaxis was associated with a reduction in asymptomatic bacteruria among children with acute spinal cord injury (P<.05), but there was no significant reduction in symptomatic infections. Prophylaxis resulted in an approximately twofold increase in antimicrobial-resistant bacteria. The researchers concluded that although a clinically important effect has not been excluded, the regular use of antimicrobial prophylaxis for most patients who have neurogenic bladder caused by spinal cord dysfunction is not supported at this time.⁸

Poor compliance may be an issue with long-term prophylaxis and may represent patient or parent practice.⁹One study found that in children taking low-dose trimethoprim, 97% of the parents reported giving antibiotics on daily basis, but in 31% of subjects, trimethoprim was not detectable in the urine.⁶Risk of prophylaxis includes nausea, vomiting, and rash in 8% to 10% of patients; development of resistant organisms; and change in indigenous microflora.⁶ One study of resistance found that children who received antibiotics for more than 4 weeks in the previous 6 months were more likely to have resistant Escherichia coli isolates than children who had not received prolonged antibiotic treatment (odds ratio [OR]=13.9; 95% CI, 8.2–23.5). Children with abnormalities of the genitourinary tract were approximately 4 times more likely to have resistant isolates of E coli than children without abnormalities of the genitourinary tract (OR=3.9; 95% CI, 2.7–5.7).¹¹

Recommendations from others

The American Academy of Pediatrics, American Urological Association, and the Swedish Medical Research Council guidelines recommend prophylaxis for children with reflux ( Table ), but they all acknowledge that the recommendations are not supported by well-designed randomized controlled trials.^1-3 No guidelines are available for children with neurogenic bladder and recurrent urinary tract infections.⁷

TABLE
Oral antibiotics for prophylaxis of urinary tract infections in children

Evidence summary

Recommendations about antibiotic prophylaxis are based on several premises. Reflux predisposes children to acute pyelonephritis; reflux plus infection leads to reflux nephropathy and ultimately to renal scarring. In theory, if antibiotics could be initiated at the appropriate time and be maintained until reflux resolves, we could successfully prevent infection and scarring.⁴

A recent systematic review evaluated the use of antibiotics to prevent UTI in children.⁵ This review of 5 randomized controlled trials included a total of 463 children between the ages of 2 months to 16 years. Three out of 5 trials evaluated the effectiveness of antibiotic treatment for 2 to 6 months to prevent subsequent off-treatment recurrence. The 2 smaller trials (n=71) evaluated the use of low-dose long-term antibiotics to prevent UTI.

There was a clinically, but not statistically, significant trend towards reduced risk of UTI during long-term antibiotic treatment (risk reduction [RR]=0.31; 95% confidence interval [CI]=0.10–1.00); however, no sustained benefit was seen once antibiotics were stopped (RR=0.79; 95% CI, 0.61–1.02). There were many problems with the methodological quality of these trials, including significant heterogeneity. The researchers concluded that well-designed randomized controlled trails are still needed to evaluate this commonly used intervention in the pediatric population.⁴ Benefits for long-term prophylaxis are even less clear in children with low-grade reflux (I–II).⁵ Furthermore, no randomized controlled trials assess whether prophylaxis prevents the development of new renal scars in children.⁶

In addition, a recent systematic review of studies done in children with normal urinary tracts, as well in children with neurogenic bladders, found that the available evidence is of low quality. Only 6 out of 31 potential studies fulfilled the inclusion criteria. These were small (mean sample size was 28), and the quality scores of all 6 trials were low, indicating that the evidence may be unreliable.⁷

Two of 3 studies done in children with normal urinary tracts demonstrated statistically significant higher rates of UTI recurrence in control groups compared with treatment groups receiving 6 to 10 months of either nitrofurantoin or cotrimoxazole (RR=24–31). The third study showed no difference between groups.

One of 2 trials in children with neurogenic bladder demonstrated higher recurrence rates of 2.9 per 10 patient years for patients receiving antibiotics compared with 1.5 in the untreated group. The other study showed lower recurrence rates of 17.1 for patients receiving antibiotics, compared with 33 in the untreated group.⁷Neither of these findings were statistically significant.

A different meta-analysis of 15 controlled clinical trials in children with neurogenic bladder due to spinal cord dysfunction. This analysis showed that antibiotic prophylaxis was associated with a reduction in asymptomatic bacteruria among children with acute spinal cord injury (P<.05), but there was no significant reduction in symptomatic infections. Prophylaxis resulted in an approximately twofold increase in antimicrobial-resistant bacteria. The researchers concluded that although a clinically important effect has not been excluded, the regular use of antimicrobial prophylaxis for most patients who have neurogenic bladder caused by spinal cord dysfunction is not supported at this time.⁸

Poor compliance may be an issue with long-term prophylaxis and may represent patient or parent practice.⁹One study found that in children taking low-dose trimethoprim, 97% of the parents reported giving antibiotics on daily basis, but in 31% of subjects, trimethoprim was not detectable in the urine.⁶Risk of prophylaxis includes nausea, vomiting, and rash in 8% to 10% of patients; development of resistant organisms; and change in indigenous microflora.⁶ One study of resistance found that children who received antibiotics for more than 4 weeks in the previous 6 months were more likely to have resistant Escherichia coli isolates than children who had not received prolonged antibiotic treatment (odds ratio [OR]=13.9; 95% CI, 8.2–23.5). Children with abnormalities of the genitourinary tract were approximately 4 times more likely to have resistant isolates of E coli than children without abnormalities of the genitourinary tract (OR=3.9; 95% CI, 2.7–5.7).¹¹

Recommendations from others

The American Academy of Pediatrics, American Urological Association, and the Swedish Medical Research Council guidelines recommend prophylaxis for children with reflux ( Table ), but they all acknowledge that the recommendations are not supported by well-designed randomized controlled trials.^1-3 No guidelines are available for children with neurogenic bladder and recurrent urinary tract infections.⁷

TABLE
Oral antibiotics for prophylaxis of urinary tract infections in children

Evidence summary

Recommendations about antibiotic prophylaxis are based on several premises. Reflux predisposes children to acute pyelonephritis; reflux plus infection leads to reflux nephropathy and ultimately to renal scarring. In theory, if antibiotics could be initiated at the appropriate time and be maintained until reflux resolves, we could successfully prevent infection and scarring.⁴

A recent systematic review evaluated the use of antibiotics to prevent UTI in children.⁵ This review of 5 randomized controlled trials included a total of 463 children between the ages of 2 months to 16 years. Three out of 5 trials evaluated the effectiveness of antibiotic treatment for 2 to 6 months to prevent subsequent off-treatment recurrence. The 2 smaller trials (n=71) evaluated the use of low-dose long-term antibiotics to prevent UTI.

There was a clinically, but not statistically, significant trend towards reduced risk of UTI during long-term antibiotic treatment (risk reduction [RR]=0.31; 95% confidence interval [CI]=0.10–1.00); however, no sustained benefit was seen once antibiotics were stopped (RR=0.79; 95% CI, 0.61–1.02). There were many problems with the methodological quality of these trials, including significant heterogeneity. The researchers concluded that well-designed randomized controlled trails are still needed to evaluate this commonly used intervention in the pediatric population.⁴ Benefits for long-term prophylaxis are even less clear in children with low-grade reflux (I–II).⁵ Furthermore, no randomized controlled trials assess whether prophylaxis prevents the development of new renal scars in children.⁶

In addition, a recent systematic review of studies done in children with normal urinary tracts, as well in children with neurogenic bladders, found that the available evidence is of low quality. Only 6 out of 31 potential studies fulfilled the inclusion criteria. These were small (mean sample size was 28), and the quality scores of all 6 trials were low, indicating that the evidence may be unreliable.⁷

Two of 3 studies done in children with normal urinary tracts demonstrated statistically significant higher rates of UTI recurrence in control groups compared with treatment groups receiving 6 to 10 months of either nitrofurantoin or cotrimoxazole (RR=24–31). The third study showed no difference between groups.

One of 2 trials in children with neurogenic bladder demonstrated higher recurrence rates of 2.9 per 10 patient years for patients receiving antibiotics compared with 1.5 in the untreated group. The other study showed lower recurrence rates of 17.1 for patients receiving antibiotics, compared with 33 in the untreated group.⁷Neither of these findings were statistically significant.

A different meta-analysis of 15 controlled clinical trials in children with neurogenic bladder due to spinal cord dysfunction. This analysis showed that antibiotic prophylaxis was associated with a reduction in asymptomatic bacteruria among children with acute spinal cord injury (P<.05), but there was no significant reduction in symptomatic infections. Prophylaxis resulted in an approximately twofold increase in antimicrobial-resistant bacteria. The researchers concluded that although a clinically important effect has not been excluded, the regular use of antimicrobial prophylaxis for most patients who have neurogenic bladder caused by spinal cord dysfunction is not supported at this time.⁸

Poor compliance may be an issue with long-term prophylaxis and may represent patient or parent practice.⁹One study found that in children taking low-dose trimethoprim, 97% of the parents reported giving antibiotics on daily basis, but in 31% of subjects, trimethoprim was not detectable in the urine.⁶Risk of prophylaxis includes nausea, vomiting, and rash in 8% to 10% of patients; development of resistant organisms; and change in indigenous microflora.⁶ One study of resistance found that children who received antibiotics for more than 4 weeks in the previous 6 months were more likely to have resistant Escherichia coli isolates than children who had not received prolonged antibiotic treatment (odds ratio [OR]=13.9; 95% CI, 8.2–23.5). Children with abnormalities of the genitourinary tract were approximately 4 times more likely to have resistant isolates of E coli than children without abnormalities of the genitourinary tract (OR=3.9; 95% CI, 2.7–5.7).¹¹

Recommendations from others

The American Academy of Pediatrics, American Urological Association, and the Swedish Medical Research Council guidelines recommend prophylaxis for children with reflux ( Table ), but they all acknowledge that the recommendations are not supported by well-designed randomized controlled trials.^1-3 No guidelines are available for children with neurogenic bladder and recurrent urinary tract infections.⁷

TABLE
Oral antibiotics for prophylaxis of urinary tract infections in children

EVIDENCE-BASED ANSWER

Healthy, unimmunized children who have not had varicella infection should be vaccinated (strength of recommendation: A, based on randomized controlled trials). Use of the vaccine in immunocompromised children is still being studied and has not been approved by the Food and Drug Administration (FDA).

Evidence summary

Before the introduction of the varicella vaccine, almost 4 million cases of chickenpox occurred each year in the United States, resulting in 11,000 hospitalizations and 100 deaths.¹ Varicella is the leading cause of vaccine-preventable death in children.²

In a search of the literature from 1966 to 2000, a systematic review identified 24 randomized controlled trials and 18 cohort studies of varicella vaccination.³ In children aged 10 months to 14 years, 1 randomized controlled trial found protective efficacy of 100% over 9 months and 98% over 7 years.⁴ A second trial showed efficacy of 72% over 29 months.⁵Cohort studies of children report that the vaccine is 84% to 86% effective in preventing varicella and 100% effective in preventing moderate to severe infections.³

Cumulative results of all studies show the number needed to vaccinate to prevent 1 case of varicella ranges from 5.5 to 11.8, and the number needed to prevent 1 complicated case ranges from 550 to 1180.

No direct evidence supports or refutes a reduction in varicella mortality or rates of hospitalization due to vaccination. Randomized controlled trials show no increase in rates of fever or rash among those receiving vaccine; however, cohort studies report fever (0%–36%), local injection site reactions (7%–30%), and rash (5%).³ No clinical trials have shown transmission of vaccine-related varicella zoster virus in immunocompetent patients, and only 3 proven cases of transmission of vaccine virus to susceptible contacts have been documented.⁶ Some evidence suggests the incidence of herpes zoster is reduced in immunocompromised vaccine recipients, but long-term observation is needed to assess the effect on healthy recipients.⁷

One concern about the vaccine is that waning immunity over time could result in increased incidence of varicella infection during adulthood. While existing studies document persistence of antibodies for up to 20 years following immunization,³ long-term effectiveness should continue to be monitored.

The FDA has not approved this live-virus vaccine for use in pregnant women and immunocom-promised persons, including transplant recipients and persons receiving corticosteroid therapy. However, the vaccine has been very well-studied in children with leukemia. A review of these studies found that optimal seroconversion requires 2 sequential vaccine doses (86% efficacy). A rash of varying severity was the predominant adverse event in 20% to 50% of vacinees.⁷ Study of vaccine use in other immunocompromised children has been limited. Early results from a trial in HIV-infected children who were not severely immunocompromised suggests similar tolerance and efficacy compared with children without HIV.⁸ A systemic review of cost-effectiveness of varicella vaccine is based predominantly on mathematical models.⁹These models show societal savings due to decrease in unproductive days for parents, but fail to demonstrate actual healthcare savings.

Recommendations from others

The American Academy of Pediatrics (AAP), Advisory Committee on Immunization Practices (ACIP), and American Academy of Family Medicine all recommend vaccinating unimmunized children aged 12 months and older who have not had varicella infection, and not vaccinating children with cellular immunodeficiencies.^2,10,11 The AAP suggests the vaccine could be considered for children with acute lymphocytic leukemia and for HIV-infected children with mild or no signs or symptoms. The ACIP guidelines are similar, with the addition that children with impaired humoral immunity may now be vaccinated.

EVIDENCE-BASED ANSWER

Healthy, unimmunized children who have not had varicella infection should be vaccinated (strength of recommendation: A, based on randomized controlled trials). Use of the vaccine in immunocompromised children is still being studied and has not been approved by the Food and Drug Administration (FDA).

Evidence summary

Before the introduction of the varicella vaccine, almost 4 million cases of chickenpox occurred each year in the United States, resulting in 11,000 hospitalizations and 100 deaths.¹ Varicella is the leading cause of vaccine-preventable death in children.²

In a search of the literature from 1966 to 2000, a systematic review identified 24 randomized controlled trials and 18 cohort studies of varicella vaccination.³ In children aged 10 months to 14 years, 1 randomized controlled trial found protective efficacy of 100% over 9 months and 98% over 7 years.⁴ A second trial showed efficacy of 72% over 29 months.⁵Cohort studies of children report that the vaccine is 84% to 86% effective in preventing varicella and 100% effective in preventing moderate to severe infections.³

Cumulative results of all studies show the number needed to vaccinate to prevent 1 case of varicella ranges from 5.5 to 11.8, and the number needed to prevent 1 complicated case ranges from 550 to 1180.

No direct evidence supports or refutes a reduction in varicella mortality or rates of hospitalization due to vaccination. Randomized controlled trials show no increase in rates of fever or rash among those receiving vaccine; however, cohort studies report fever (0%–36%), local injection site reactions (7%–30%), and rash (5%).³ No clinical trials have shown transmission of vaccine-related varicella zoster virus in immunocompetent patients, and only 3 proven cases of transmission of vaccine virus to susceptible contacts have been documented.⁶ Some evidence suggests the incidence of herpes zoster is reduced in immunocompromised vaccine recipients, but long-term observation is needed to assess the effect on healthy recipients.⁷

One concern about the vaccine is that waning immunity over time could result in increased incidence of varicella infection during adulthood. While existing studies document persistence of antibodies for up to 20 years following immunization,³ long-term effectiveness should continue to be monitored.

The FDA has not approved this live-virus vaccine for use in pregnant women and immunocom-promised persons, including transplant recipients and persons receiving corticosteroid therapy. However, the vaccine has been very well-studied in children with leukemia. A review of these studies found that optimal seroconversion requires 2 sequential vaccine doses (86% efficacy). A rash of varying severity was the predominant adverse event in 20% to 50% of vacinees.⁷ Study of vaccine use in other immunocompromised children has been limited. Early results from a trial in HIV-infected children who were not severely immunocompromised suggests similar tolerance and efficacy compared with children without HIV.⁸ A systemic review of cost-effectiveness of varicella vaccine is based predominantly on mathematical models.⁹These models show societal savings due to decrease in unproductive days for parents, but fail to demonstrate actual healthcare savings.

Recommendations from others

The American Academy of Pediatrics (AAP), Advisory Committee on Immunization Practices (ACIP), and American Academy of Family Medicine all recommend vaccinating unimmunized children aged 12 months and older who have not had varicella infection, and not vaccinating children with cellular immunodeficiencies.^2,10,11 The AAP suggests the vaccine could be considered for children with acute lymphocytic leukemia and for HIV-infected children with mild or no signs or symptoms. The ACIP guidelines are similar, with the addition that children with impaired humoral immunity may now be vaccinated.

EVIDENCE-BASED ANSWER

Healthy, unimmunized children who have not had varicella infection should be vaccinated (strength of recommendation: A, based on randomized controlled trials). Use of the vaccine in immunocompromised children is still being studied and has not been approved by the Food and Drug Administration (FDA).

Evidence summary

Before the introduction of the varicella vaccine, almost 4 million cases of chickenpox occurred each year in the United States, resulting in 11,000 hospitalizations and 100 deaths.¹ Varicella is the leading cause of vaccine-preventable death in children.²

In a search of the literature from 1966 to 2000, a systematic review identified 24 randomized controlled trials and 18 cohort studies of varicella vaccination.³ In children aged 10 months to 14 years, 1 randomized controlled trial found protective efficacy of 100% over 9 months and 98% over 7 years.⁴ A second trial showed efficacy of 72% over 29 months.⁵Cohort studies of children report that the vaccine is 84% to 86% effective in preventing varicella and 100% effective in preventing moderate to severe infections.³

Cumulative results of all studies show the number needed to vaccinate to prevent 1 case of varicella ranges from 5.5 to 11.8, and the number needed to prevent 1 complicated case ranges from 550 to 1180.

No direct evidence supports or refutes a reduction in varicella mortality or rates of hospitalization due to vaccination. Randomized controlled trials show no increase in rates of fever or rash among those receiving vaccine; however, cohort studies report fever (0%–36%), local injection site reactions (7%–30%), and rash (5%).³ No clinical trials have shown transmission of vaccine-related varicella zoster virus in immunocompetent patients, and only 3 proven cases of transmission of vaccine virus to susceptible contacts have been documented.⁶ Some evidence suggests the incidence of herpes zoster is reduced in immunocompromised vaccine recipients, but long-term observation is needed to assess the effect on healthy recipients.⁷

One concern about the vaccine is that waning immunity over time could result in increased incidence of varicella infection during adulthood. While existing studies document persistence of antibodies for up to 20 years following immunization,³ long-term effectiveness should continue to be monitored.

The FDA has not approved this live-virus vaccine for use in pregnant women and immunocom-promised persons, including transplant recipients and persons receiving corticosteroid therapy. However, the vaccine has been very well-studied in children with leukemia. A review of these studies found that optimal seroconversion requires 2 sequential vaccine doses (86% efficacy). A rash of varying severity was the predominant adverse event in 20% to 50% of vacinees.⁷ Study of vaccine use in other immunocompromised children has been limited. Early results from a trial in HIV-infected children who were not severely immunocompromised suggests similar tolerance and efficacy compared with children without HIV.⁸ A systemic review of cost-effectiveness of varicella vaccine is based predominantly on mathematical models.⁹These models show societal savings due to decrease in unproductive days for parents, but fail to demonstrate actual healthcare savings.

Recommendations from others

The American Academy of Pediatrics (AAP), Advisory Committee on Immunization Practices (ACIP), and American Academy of Family Medicine all recommend vaccinating unimmunized children aged 12 months and older who have not had varicella infection, and not vaccinating children with cellular immunodeficiencies.^2,10,11 The AAP suggests the vaccine could be considered for children with acute lymphocytic leukemia and for HIV-infected children with mild or no signs or symptoms. The ACIP guidelines are similar, with the addition that children with impaired humoral immunity may now be vaccinated.

Evidence summary

A review of 33 double-blind randomized placebo-controlled studies showed a significant effect of bronchodilator therapy on exercise capacity in COPD patients in about one half of studies. Anticholinergic agents had significant beneficial effects in the majority, and these effects tended to be somewhat dose-dependent. Short-acting β₂-agonists improved exercise capacity in more than two thirds of the studies, but long-acting agents led to mixed outcomes. The researchers identified no superior agent between the 2 classes, citing a lack of adequate studies making a direct comparison.¹

A recent Cochrane Review comparing the short-term effects of ipratropium to β₂-agonists in changes in FEV₁and arterial oxygen pressure (PaO₂) concluded there was no evidence that the degree of bronchodilation from ipratropium was greater than that from short-acting β₂-agonists.²Subjective endpoints such as dyspnea and quality of life were not assessed, and neither of the above reviews included studies focusing on long-term outcomes.

A 12-week double-blind, double-placebo-controlled parallel group study published in 2000 followed 144 patients (age 64 ± 7 years with a FEV₁of 44 ±11% predicted) randomized to receive salmeterol 50 μg twice daily alone, salmeterol 50 μg twice daily plus ipratropium 40 μg 4 times daily, or placebo. Patients were assessed for changes in FEV₁, daytime symptom scores, specific airway conductance, and the need for rescue medication. The study demonstrated a significant benefit from the addition of ipratropium to salmeterol in terms of reduction of airway obstruction, but not in symptom control or need for rescue medication.³ However, no patients were randomized to receive ipratropium alone, so comparison of the relative contribution of the 2 classes is limited.

A 6-month, randomized double-blind placebo-controlled study evaluating the efficacy of salmeterol 50 μg twice daily vs tiotropium (a new long-acting inhaled anticholinergic) 18 μg once daily was published in 2002. Endpoints in 623 patients were assessed using 12-hour spirometric performance, transition dyspnea index (TDI), and the St. George Respiratory Questionnaire (SGRQ). (SGRQ is a validated disease-specific instrument designed to measure impact on overall health, daily life, and perceived well-being. It measures activity limitations, symptoms, and psychosocial impact.) Tiotropium showed superiority over salmeterol in all endpoints assessed (0.14 L increase in morning FEV₁vs 0.09 L, 1.02 U improvement in TDI score vs 0.24, and –5.14 U improvement of SGRQ total score from baseline vs –3.54). However, it should be noted that a difference of 1 on the TDI score was necessary to suggest a clinical benefit. While the overall difference in SGQR between tiotropium and salmeterol did not reach statistical significance, the proportion of patients in the tiotropium group that reached the clinically significant threshold of 4 units improvement in SGRQ score was significantly higher than in the salmeterol group (51% vs 40%; P<.05).⁴

In a similar study in 2003, 1207 patients were randomized to receive the above doses of salmeterol, tiotropium, or placebo. Over the course of 6 months, tiotropium was associated with a significant delay in onset of the first exacerbation compared with placebo, and overall it led to the fewest exacerbations per patient-year. Fewer hospital admissions were also demonstrated in the tiotropium group per patient-year, and the number of days that patients were unable to perform usual activities was lowest for the tiotropium group. Again, improvement in TDI and SGRQ scores was significantly greater with tiotropium than placebo. In almost all outcomes, the salmeterol results were intermediate between those of tiotropium and placebo, and were not statistically different from placebo.⁵

Recommendations from others

The GOLD (Global Strategy for the Diagnosis, Management, and Prevention of COPD) Report states that the choice between β₂-agonist, anti-cholinergics, or combination therapy depends on the availability and the response of a given patient in terms of symptom relief and side effects. The 2003 GOLD Workshop Report update further recommends the use of regular treatment with long-acting bronchodilators, including tiotropium, rather than short-acting bronchodilators for moderate-to-severe COPD.⁶

A separate report for the Joint Expert Panel on Chronic Obstructive Pulmonary Disease of the American College of Chest Physicians and the American College of Physicians—American Society of Internal Medicine states that both are beneficial for management of acute exacerbations, but that anticholinergics should be considered first because they are associated with fewer and more benign side effects.⁷

Evidence summary

A review of 33 double-blind randomized placebo-controlled studies showed a significant effect of bronchodilator therapy on exercise capacity in COPD patients in about one half of studies. Anticholinergic agents had significant beneficial effects in the majority, and these effects tended to be somewhat dose-dependent. Short-acting β₂-agonists improved exercise capacity in more than two thirds of the studies, but long-acting agents led to mixed outcomes. The researchers identified no superior agent between the 2 classes, citing a lack of adequate studies making a direct comparison.¹

A recent Cochrane Review comparing the short-term effects of ipratropium to β₂-agonists in changes in FEV₁and arterial oxygen pressure (PaO₂) concluded there was no evidence that the degree of bronchodilation from ipratropium was greater than that from short-acting β₂-agonists.²Subjective endpoints such as dyspnea and quality of life were not assessed, and neither of the above reviews included studies focusing on long-term outcomes.

A 12-week double-blind, double-placebo-controlled parallel group study published in 2000 followed 144 patients (age 64 ± 7 years with a FEV₁of 44 ±11% predicted) randomized to receive salmeterol 50 μg twice daily alone, salmeterol 50 μg twice daily plus ipratropium 40 μg 4 times daily, or placebo. Patients were assessed for changes in FEV₁, daytime symptom scores, specific airway conductance, and the need for rescue medication. The study demonstrated a significant benefit from the addition of ipratropium to salmeterol in terms of reduction of airway obstruction, but not in symptom control or need for rescue medication.³ However, no patients were randomized to receive ipratropium alone, so comparison of the relative contribution of the 2 classes is limited.

A 6-month, randomized double-blind placebo-controlled study evaluating the efficacy of salmeterol 50 μg twice daily vs tiotropium (a new long-acting inhaled anticholinergic) 18 μg once daily was published in 2002. Endpoints in 623 patients were assessed using 12-hour spirometric performance, transition dyspnea index (TDI), and the St. George Respiratory Questionnaire (SGRQ). (SGRQ is a validated disease-specific instrument designed to measure impact on overall health, daily life, and perceived well-being. It measures activity limitations, symptoms, and psychosocial impact.) Tiotropium showed superiority over salmeterol in all endpoints assessed (0.14 L increase in morning FEV₁vs 0.09 L, 1.02 U improvement in TDI score vs 0.24, and –5.14 U improvement of SGRQ total score from baseline vs –3.54). However, it should be noted that a difference of 1 on the TDI score was necessary to suggest a clinical benefit. While the overall difference in SGQR between tiotropium and salmeterol did not reach statistical significance, the proportion of patients in the tiotropium group that reached the clinically significant threshold of 4 units improvement in SGRQ score was significantly higher than in the salmeterol group (51% vs 40%; P<.05).⁴

In a similar study in 2003, 1207 patients were randomized to receive the above doses of salmeterol, tiotropium, or placebo. Over the course of 6 months, tiotropium was associated with a significant delay in onset of the first exacerbation compared with placebo, and overall it led to the fewest exacerbations per patient-year. Fewer hospital admissions were also demonstrated in the tiotropium group per patient-year, and the number of days that patients were unable to perform usual activities was lowest for the tiotropium group. Again, improvement in TDI and SGRQ scores was significantly greater with tiotropium than placebo. In almost all outcomes, the salmeterol results were intermediate between those of tiotropium and placebo, and were not statistically different from placebo.⁵

Recommendations from others

The GOLD (Global Strategy for the Diagnosis, Management, and Prevention of COPD) Report states that the choice between β₂-agonist, anti-cholinergics, or combination therapy depends on the availability and the response of a given patient in terms of symptom relief and side effects. The 2003 GOLD Workshop Report update further recommends the use of regular treatment with long-acting bronchodilators, including tiotropium, rather than short-acting bronchodilators for moderate-to-severe COPD.⁶

A separate report for the Joint Expert Panel on Chronic Obstructive Pulmonary Disease of the American College of Chest Physicians and the American College of Physicians—American Society of Internal Medicine states that both are beneficial for management of acute exacerbations, but that anticholinergics should be considered first because they are associated with fewer and more benign side effects.⁷

Evidence summary

A review of 33 double-blind randomized placebo-controlled studies showed a significant effect of bronchodilator therapy on exercise capacity in COPD patients in about one half of studies. Anticholinergic agents had significant beneficial effects in the majority, and these effects tended to be somewhat dose-dependent. Short-acting β₂-agonists improved exercise capacity in more than two thirds of the studies, but long-acting agents led to mixed outcomes. The researchers identified no superior agent between the 2 classes, citing a lack of adequate studies making a direct comparison.¹

A recent Cochrane Review comparing the short-term effects of ipratropium to β₂-agonists in changes in FEV₁and arterial oxygen pressure (PaO₂) concluded there was no evidence that the degree of bronchodilation from ipratropium was greater than that from short-acting β₂-agonists.²Subjective endpoints such as dyspnea and quality of life were not assessed, and neither of the above reviews included studies focusing on long-term outcomes.

A 12-week double-blind, double-placebo-controlled parallel group study published in 2000 followed 144 patients (age 64 ± 7 years with a FEV₁of 44 ±11% predicted) randomized to receive salmeterol 50 μg twice daily alone, salmeterol 50 μg twice daily plus ipratropium 40 μg 4 times daily, or placebo. Patients were assessed for changes in FEV₁, daytime symptom scores, specific airway conductance, and the need for rescue medication. The study demonstrated a significant benefit from the addition of ipratropium to salmeterol in terms of reduction of airway obstruction, but not in symptom control or need for rescue medication.³ However, no patients were randomized to receive ipratropium alone, so comparison of the relative contribution of the 2 classes is limited.

A 6-month, randomized double-blind placebo-controlled study evaluating the efficacy of salmeterol 50 μg twice daily vs tiotropium (a new long-acting inhaled anticholinergic) 18 μg once daily was published in 2002. Endpoints in 623 patients were assessed using 12-hour spirometric performance, transition dyspnea index (TDI), and the St. George Respiratory Questionnaire (SGRQ). (SGRQ is a validated disease-specific instrument designed to measure impact on overall health, daily life, and perceived well-being. It measures activity limitations, symptoms, and psychosocial impact.) Tiotropium showed superiority over salmeterol in all endpoints assessed (0.14 L increase in morning FEV₁vs 0.09 L, 1.02 U improvement in TDI score vs 0.24, and –5.14 U improvement of SGRQ total score from baseline vs –3.54). However, it should be noted that a difference of 1 on the TDI score was necessary to suggest a clinical benefit. While the overall difference in SGQR between tiotropium and salmeterol did not reach statistical significance, the proportion of patients in the tiotropium group that reached the clinically significant threshold of 4 units improvement in SGRQ score was significantly higher than in the salmeterol group (51% vs 40%; P<.05).⁴

In a similar study in 2003, 1207 patients were randomized to receive the above doses of salmeterol, tiotropium, or placebo. Over the course of 6 months, tiotropium was associated with a significant delay in onset of the first exacerbation compared with placebo, and overall it led to the fewest exacerbations per patient-year. Fewer hospital admissions were also demonstrated in the tiotropium group per patient-year, and the number of days that patients were unable to perform usual activities was lowest for the tiotropium group. Again, improvement in TDI and SGRQ scores was significantly greater with tiotropium than placebo. In almost all outcomes, the salmeterol results were intermediate between those of tiotropium and placebo, and were not statistically different from placebo.⁵

Recommendations from others

The GOLD (Global Strategy for the Diagnosis, Management, and Prevention of COPD) Report states that the choice between β₂-agonist, anti-cholinergics, or combination therapy depends on the availability and the response of a given patient in terms of symptom relief and side effects. The 2003 GOLD Workshop Report update further recommends the use of regular treatment with long-acting bronchodilators, including tiotropium, rather than short-acting bronchodilators for moderate-to-severe COPD.⁶

A separate report for the Joint Expert Panel on Chronic Obstructive Pulmonary Disease of the American College of Chest Physicians and the American College of Physicians—American Society of Internal Medicine states that both are beneficial for management of acute exacerbations, but that anticholinergics should be considered first because they are associated with fewer and more benign side effects.⁷

Evidence summary

Treatment of maternal hypertension during pregnancy is based on maternal and fetal outcomes. Multiple meta-analyses of randomized controlled trials show that the major maternal outcomes improved by treating mild to moderate hypertension are decreased progression to severe hypertension (number needed to treat [NNT]=12; 95% confidence interval [CI], 9–17) and decreased need for additional antihypertensive therapy.^1,2 The relative risk (RR) for preventing preeclampsia was 0.99 (95% CI, 0.84–1.18). The risk of preterm delivery was 1.00 (95% CI, 0.87–1.15).

The data for fetal outcomes are important, as the maternal benefits of treatment remain small.³ Much of the debate centers on decreasing uteroplacental perfusion, which may lead to decreased fetal growth. One meta-analysis reviewed 45 trials to evaluate the potential increase in SGA infants caused by any antihypertensive treatment, through quantifying the fall in mean arterial pressure. The analysis found an average decrease in birthweight of 145 g for a 10 mm Hg fall in mean arterial pressurewith no increased perinatal morbidity.⁴ The clinical significance of this is unclear.

In comparing one agent with another, methyldopa was the most commonly tested agent, with 14 randomized controlled trials of more than 1010 subjects demonstrating its efficacy at reducing blood pressure. Other antihypertensive agents appear better than methyldopa in terms of reducing the risk of infant mortality (RR=0.49; 95% CI, 0.24–0.99),¹ but the studies were small and used weak methods, and this finding may be due to bias.⁵ Meta-analyses of beta-blocker trials show a borderline increase in SGA infants, with no related increase in perinatal mortality, as well as a decrease in the incidence of respiratory distress syndrome.⁶

Diuretics are effective antihypertensives, especially when combined with other agents, but they are known to decrease the circulating plasma volume, potentially decreasing uteroplacental perfusion. They are generally viewed as safe, as long as the mother is not already at increased risk for perfusion abnormalities (eg, preeclamptic states).⁷ Calcium channel blockers, though generally regarded as safe and effective, have mostly been evaluated for use late in pregnancy, so their benefit-to-risk ratio remains uncertain.⁸ACE inhibitors and, by extension, ARBs, due to their similar mechanisms of action, are contraindicated in pregnancy, having been linked to miscarriage, fetal death, fetal renal failure, and malformation.^5,9-11

Recommendations from others

The American College of Obstetricians and Gynecologists (ACOG) Practice Bulletin states there is no evidence that antihypertensive treatment for mild to moderate hypertension improves maternal or fetal outcomes, even for women who are already receiving hypertension treatment at the time of pregnancy. ACOG suggests treatment may be stopped during pregnancy, or not initiated until blood pressures reach 150–160 mm Hg systolic or 100–110 mm Hg diastolic, unless the mother has underlying renal or cardiovascular disease.⁹

The National High Blood Pressure Education Program recommends the same guidelines as ACOG,¹⁰ whereas the Canadian Hypertension Society consensus panel has chosen 140/90 mm Hg as the level at which treatment should be initiated.¹¹

The British Medical Journal Clinical Evidence Guidelines reiterate that the evidence does not support the benefit of treating mild to moderate hypertension, except in reducing the progression to severe hypertension.⁵ Methyldopa is consistently the drug of choice in all those making a specific recommendation,^9-11 although it should be noted these recommendations were published before the 2003 Cochrane Review.¹

Evidence summary

Treatment of maternal hypertension during pregnancy is based on maternal and fetal outcomes. Multiple meta-analyses of randomized controlled trials show that the major maternal outcomes improved by treating mild to moderate hypertension are decreased progression to severe hypertension (number needed to treat [NNT]=12; 95% confidence interval [CI], 9–17) and decreased need for additional antihypertensive therapy.^1,2 The relative risk (RR) for preventing preeclampsia was 0.99 (95% CI, 0.84–1.18). The risk of preterm delivery was 1.00 (95% CI, 0.87–1.15).

The data for fetal outcomes are important, as the maternal benefits of treatment remain small.³ Much of the debate centers on decreasing uteroplacental perfusion, which may lead to decreased fetal growth. One meta-analysis reviewed 45 trials to evaluate the potential increase in SGA infants caused by any antihypertensive treatment, through quantifying the fall in mean arterial pressure. The analysis found an average decrease in birthweight of 145 g for a 10 mm Hg fall in mean arterial pressurewith no increased perinatal morbidity.⁴ The clinical significance of this is unclear.

In comparing one agent with another, methyldopa was the most commonly tested agent, with 14 randomized controlled trials of more than 1010 subjects demonstrating its efficacy at reducing blood pressure. Other antihypertensive agents appear better than methyldopa in terms of reducing the risk of infant mortality (RR=0.49; 95% CI, 0.24–0.99),¹ but the studies were small and used weak methods, and this finding may be due to bias.⁵ Meta-analyses of beta-blocker trials show a borderline increase in SGA infants, with no related increase in perinatal mortality, as well as a decrease in the incidence of respiratory distress syndrome.⁶

Diuretics are effective antihypertensives, especially when combined with other agents, but they are known to decrease the circulating plasma volume, potentially decreasing uteroplacental perfusion. They are generally viewed as safe, as long as the mother is not already at increased risk for perfusion abnormalities (eg, preeclamptic states).⁷ Calcium channel blockers, though generally regarded as safe and effective, have mostly been evaluated for use late in pregnancy, so their benefit-to-risk ratio remains uncertain.⁸ACE inhibitors and, by extension, ARBs, due to their similar mechanisms of action, are contraindicated in pregnancy, having been linked to miscarriage, fetal death, fetal renal failure, and malformation.^5,9-11

Recommendations from others

The American College of Obstetricians and Gynecologists (ACOG) Practice Bulletin states there is no evidence that antihypertensive treatment for mild to moderate hypertension improves maternal or fetal outcomes, even for women who are already receiving hypertension treatment at the time of pregnancy. ACOG suggests treatment may be stopped during pregnancy, or not initiated until blood pressures reach 150–160 mm Hg systolic or 100–110 mm Hg diastolic, unless the mother has underlying renal or cardiovascular disease.⁹

The National High Blood Pressure Education Program recommends the same guidelines as ACOG,¹⁰ whereas the Canadian Hypertension Society consensus panel has chosen 140/90 mm Hg as the level at which treatment should be initiated.¹¹

The British Medical Journal Clinical Evidence Guidelines reiterate that the evidence does not support the benefit of treating mild to moderate hypertension, except in reducing the progression to severe hypertension.⁵ Methyldopa is consistently the drug of choice in all those making a specific recommendation,^9-11 although it should be noted these recommendations were published before the 2003 Cochrane Review.¹

Evidence summary

Treatment of maternal hypertension during pregnancy is based on maternal and fetal outcomes. Multiple meta-analyses of randomized controlled trials show that the major maternal outcomes improved by treating mild to moderate hypertension are decreased progression to severe hypertension (number needed to treat [NNT]=12; 95% confidence interval [CI], 9–17) and decreased need for additional antihypertensive therapy.^1,2 The relative risk (RR) for preventing preeclampsia was 0.99 (95% CI, 0.84–1.18). The risk of preterm delivery was 1.00 (95% CI, 0.87–1.15).

The data for fetal outcomes are important, as the maternal benefits of treatment remain small.³ Much of the debate centers on decreasing uteroplacental perfusion, which may lead to decreased fetal growth. One meta-analysis reviewed 45 trials to evaluate the potential increase in SGA infants caused by any antihypertensive treatment, through quantifying the fall in mean arterial pressure. The analysis found an average decrease in birthweight of 145 g for a 10 mm Hg fall in mean arterial pressurewith no increased perinatal morbidity.⁴ The clinical significance of this is unclear.

In comparing one agent with another, methyldopa was the most commonly tested agent, with 14 randomized controlled trials of more than 1010 subjects demonstrating its efficacy at reducing blood pressure. Other antihypertensive agents appear better than methyldopa in terms of reducing the risk of infant mortality (RR=0.49; 95% CI, 0.24–0.99),¹ but the studies were small and used weak methods, and this finding may be due to bias.⁵ Meta-analyses of beta-blocker trials show a borderline increase in SGA infants, with no related increase in perinatal mortality, as well as a decrease in the incidence of respiratory distress syndrome.⁶

Diuretics are effective antihypertensives, especially when combined with other agents, but they are known to decrease the circulating plasma volume, potentially decreasing uteroplacental perfusion. They are generally viewed as safe, as long as the mother is not already at increased risk for perfusion abnormalities (eg, preeclamptic states).⁷ Calcium channel blockers, though generally regarded as safe and effective, have mostly been evaluated for use late in pregnancy, so their benefit-to-risk ratio remains uncertain.⁸ACE inhibitors and, by extension, ARBs, due to their similar mechanisms of action, are contraindicated in pregnancy, having been linked to miscarriage, fetal death, fetal renal failure, and malformation.^5,9-11

Recommendations from others

The American College of Obstetricians and Gynecologists (ACOG) Practice Bulletin states there is no evidence that antihypertensive treatment for mild to moderate hypertension improves maternal or fetal outcomes, even for women who are already receiving hypertension treatment at the time of pregnancy. ACOG suggests treatment may be stopped during pregnancy, or not initiated until blood pressures reach 150–160 mm Hg systolic or 100–110 mm Hg diastolic, unless the mother has underlying renal or cardiovascular disease.⁹

The National High Blood Pressure Education Program recommends the same guidelines as ACOG,¹⁰ whereas the Canadian Hypertension Society consensus panel has chosen 140/90 mm Hg as the level at which treatment should be initiated.¹¹

The British Medical Journal Clinical Evidence Guidelines reiterate that the evidence does not support the benefit of treating mild to moderate hypertension, except in reducing the progression to severe hypertension.⁵ Methyldopa is consistently the drug of choice in all those making a specific recommendation,^9-11 although it should be noted these recommendations were published before the 2003 Cochrane Review.¹

Evidence summary

A Cochrane Review of corticosteroid injections for shoulder pain found 7 randomized controlled trials comparing subacromial steroid injections with placebo.¹ The placebos were either injectable anesthetics alone or injectable anesthetics combined with oral placebo tablets. Six of the 7 studies used the anterolateral approach to inject under the acromion.

All studies used a clinical exam for diagnosis that showed pain with range of motion (especially abduction) or pain that was consistent with impingement syndrome. Most of the follow-up times were short, typically 4 to 12 weeks, and the longest study was 33 weeks. Meta-analyses often report the effect size using standard mean difference (SMD). A rule of thumb for interpretation of SMD is a value of 0.2 indicates a small effect, a value of 0.5 indicates a medium effect, and a value of 0.8 or larger indicates a large effect. If the 95% confidence interval [CI] does not include zero, then the SMD is statistically significant at the 5% level (P<.05).²

Two of the studies comparing steroid injection with placebo were methodologically suitable for meta-analysis; these studies showed thatsteroids provided a mild, short-term (4-week)benefit with respect to pain (SMD=0.83; 95% CI,0.39–1.26), function (SMD=0.63; 95% CI,0.20–1.06), and abductive range of motion(SMD=0.82; 95% CI, 0.39–1.25).^3,4

Results of the remaining, less rigorous trialswere conflicting and inconclusive. The reviewersalso found 3 randomized controlled trials comparing subacromial steroid injection with oralNSAIDs. The pooled results of these trials,encompassing 120 patients, found no differences in these 3 outcomes at 4 or 6 weeks. The review of an additional trial of 50 patients comparing subacromial steroid injection plus simultaneous oral NSAIDs with oral NSAIDs alone found no differences at 4 weeks. All 11 studies had small sample sizes, and suffered from variable methodological quality and heterogeneous results.

The reviewers concluded that steroids are probably better than placebo but provide little or no benefit in addition to NSAIDs, and that evidence is insufficient to guide treatment. Likewise, a Cochrane Review of multiple interventions for shoulder pain also found “little evidence to support or refute the efficacy of common interventions” and highlighted the need for new, well-designed trials.⁵

Another Cochrane Review examined 4 randomized controlled trials comparing physiotherapy interventions for shoulder pain.⁶ They found that steroid injections may be superior to physiotherapy for rotator cuff disease, but the type of physiotherapy and injection sites were not consistent across the studies, making creation of summary estimates inappropriate. The individual studies showed significant short-term benefits (3–7 weeks) of steroid injection over physiotherapy; however, long-term (6–52 weeks) benefits ranged from some benefit to no difference. These studies were consistent regarding age (mean age=53–55 years, SD ± 13–14 years) and complications reported, with the only side effect being postinjection soreness.

Hay et al⁷ conducted a multicenter, primary care–based randomized controlled trial with more than 200 patients, which was published too recently for inclusion in the Cochrane Review. They found no statistical difference in improvement between steroid injection without physiotherapy and physiotherapy alone at 6 weeks.

In 1996, van der Heijden et al⁸ systematically reviewed randomized clinical trials of steroid injections for shoulder disorders, including rota-tor cuff disease, adhesive capsulitis, rheumatoid conditions, and periarthritis. They screened more than 200 articles from searches in Medline (1966–1995) and EMBASE (1984–1995) and found 16 articles that met qualifying conditions for further review. Of these, 3 were methodologically adequate for final review. None of these 3 studies provided evidence showing the efficacy of steroid injections. The results of the major trials reviewed can be found in the Table .

TABLE
Major placebo-controlled trials of injectable steroids for shoulder pain

Recommendations from others

The American Academy of Orthopaedic Surgeons’ clinical guideline for shoulder pain⁹ recommends the following for rotator cuff disease: avoidance of irritating activity; anti-inflammatory medications if tolerated; exercises to recover and maintain passive range of motion; exercises to strengthen the rotator cuff once acute symptoms abated. If these are unsuccessful over several weeks, they recommend considering subacromial injection of local anesthetic and a short-acting corticosteroid. They gave their recommendation a “B” rating (some evidence exists to suggest benefit).

Evidence summary

A Cochrane Review of corticosteroid injections for shoulder pain found 7 randomized controlled trials comparing subacromial steroid injections with placebo.¹ The placebos were either injectable anesthetics alone or injectable anesthetics combined with oral placebo tablets. Six of the 7 studies used the anterolateral approach to inject under the acromion.

All studies used a clinical exam for diagnosis that showed pain with range of motion (especially abduction) or pain that was consistent with impingement syndrome. Most of the follow-up times were short, typically 4 to 12 weeks, and the longest study was 33 weeks. Meta-analyses often report the effect size using standard mean difference (SMD). A rule of thumb for interpretation of SMD is a value of 0.2 indicates a small effect, a value of 0.5 indicates a medium effect, and a value of 0.8 or larger indicates a large effect. If the 95% confidence interval [CI] does not include zero, then the SMD is statistically significant at the 5% level (P<.05).²

Two of the studies comparing steroid injection with placebo were methodologically suitable for meta-analysis; these studies showed thatsteroids provided a mild, short-term (4-week)benefit with respect to pain (SMD=0.83; 95% CI,0.39–1.26), function (SMD=0.63; 95% CI,0.20–1.06), and abductive range of motion(SMD=0.82; 95% CI, 0.39–1.25).^3,4

Results of the remaining, less rigorous trialswere conflicting and inconclusive. The reviewersalso found 3 randomized controlled trials comparing subacromial steroid injection with oralNSAIDs. The pooled results of these trials,encompassing 120 patients, found no differences in these 3 outcomes at 4 or 6 weeks. The review of an additional trial of 50 patients comparing subacromial steroid injection plus simultaneous oral NSAIDs with oral NSAIDs alone found no differences at 4 weeks. All 11 studies had small sample sizes, and suffered from variable methodological quality and heterogeneous results.

The reviewers concluded that steroids are probably better than placebo but provide little or no benefit in addition to NSAIDs, and that evidence is insufficient to guide treatment. Likewise, a Cochrane Review of multiple interventions for shoulder pain also found “little evidence to support or refute the efficacy of common interventions” and highlighted the need for new, well-designed trials.⁵

Another Cochrane Review examined 4 randomized controlled trials comparing physiotherapy interventions for shoulder pain.⁶ They found that steroid injections may be superior to physiotherapy for rotator cuff disease, but the type of physiotherapy and injection sites were not consistent across the studies, making creation of summary estimates inappropriate. The individual studies showed significant short-term benefits (3–7 weeks) of steroid injection over physiotherapy; however, long-term (6–52 weeks) benefits ranged from some benefit to no difference. These studies were consistent regarding age (mean age=53–55 years, SD ± 13–14 years) and complications reported, with the only side effect being postinjection soreness.

Hay et al⁷ conducted a multicenter, primary care–based randomized controlled trial with more than 200 patients, which was published too recently for inclusion in the Cochrane Review. They found no statistical difference in improvement between steroid injection without physiotherapy and physiotherapy alone at 6 weeks.

In 1996, van der Heijden et al⁸ systematically reviewed randomized clinical trials of steroid injections for shoulder disorders, including rota-tor cuff disease, adhesive capsulitis, rheumatoid conditions, and periarthritis. They screened more than 200 articles from searches in Medline (1966–1995) and EMBASE (1984–1995) and found 16 articles that met qualifying conditions for further review. Of these, 3 were methodologically adequate for final review. None of these 3 studies provided evidence showing the efficacy of steroid injections. The results of the major trials reviewed can be found in the Table .

TABLE
Major placebo-controlled trials of injectable steroids for shoulder pain

Recommendations from others

The American Academy of Orthopaedic Surgeons’ clinical guideline for shoulder pain⁹ recommends the following for rotator cuff disease: avoidance of irritating activity; anti-inflammatory medications if tolerated; exercises to recover and maintain passive range of motion; exercises to strengthen the rotator cuff once acute symptoms abated. If these are unsuccessful over several weeks, they recommend considering subacromial injection of local anesthetic and a short-acting corticosteroid. They gave their recommendation a “B” rating (some evidence exists to suggest benefit).

Evidence summary

A Cochrane Review of corticosteroid injections for shoulder pain found 7 randomized controlled trials comparing subacromial steroid injections with placebo.¹ The placebos were either injectable anesthetics alone or injectable anesthetics combined with oral placebo tablets. Six of the 7 studies used the anterolateral approach to inject under the acromion.

All studies used a clinical exam for diagnosis that showed pain with range of motion (especially abduction) or pain that was consistent with impingement syndrome. Most of the follow-up times were short, typically 4 to 12 weeks, and the longest study was 33 weeks. Meta-analyses often report the effect size using standard mean difference (SMD). A rule of thumb for interpretation of SMD is a value of 0.2 indicates a small effect, a value of 0.5 indicates a medium effect, and a value of 0.8 or larger indicates a large effect. If the 95% confidence interval [CI] does not include zero, then the SMD is statistically significant at the 5% level (P<.05).²

Two of the studies comparing steroid injection with placebo were methodologically suitable for meta-analysis; these studies showed thatsteroids provided a mild, short-term (4-week)benefit with respect to pain (SMD=0.83; 95% CI,0.39–1.26), function (SMD=0.63; 95% CI,0.20–1.06), and abductive range of motion(SMD=0.82; 95% CI, 0.39–1.25).^3,4

Results of the remaining, less rigorous trialswere conflicting and inconclusive. The reviewersalso found 3 randomized controlled trials comparing subacromial steroid injection with oralNSAIDs. The pooled results of these trials,encompassing 120 patients, found no differences in these 3 outcomes at 4 or 6 weeks. The review of an additional trial of 50 patients comparing subacromial steroid injection plus simultaneous oral NSAIDs with oral NSAIDs alone found no differences at 4 weeks. All 11 studies had small sample sizes, and suffered from variable methodological quality and heterogeneous results.

The reviewers concluded that steroids are probably better than placebo but provide little or no benefit in addition to NSAIDs, and that evidence is insufficient to guide treatment. Likewise, a Cochrane Review of multiple interventions for shoulder pain also found “little evidence to support or refute the efficacy of common interventions” and highlighted the need for new, well-designed trials.⁵

Another Cochrane Review examined 4 randomized controlled trials comparing physiotherapy interventions for shoulder pain.⁶ They found that steroid injections may be superior to physiotherapy for rotator cuff disease, but the type of physiotherapy and injection sites were not consistent across the studies, making creation of summary estimates inappropriate. The individual studies showed significant short-term benefits (3–7 weeks) of steroid injection over physiotherapy; however, long-term (6–52 weeks) benefits ranged from some benefit to no difference. These studies were consistent regarding age (mean age=53–55 years, SD ± 13–14 years) and complications reported, with the only side effect being postinjection soreness.

Hay et al⁷ conducted a multicenter, primary care–based randomized controlled trial with more than 200 patients, which was published too recently for inclusion in the Cochrane Review. They found no statistical difference in improvement between steroid injection without physiotherapy and physiotherapy alone at 6 weeks.

In 1996, van der Heijden et al⁸ systematically reviewed randomized clinical trials of steroid injections for shoulder disorders, including rota-tor cuff disease, adhesive capsulitis, rheumatoid conditions, and periarthritis. They screened more than 200 articles from searches in Medline (1966–1995) and EMBASE (1984–1995) and found 16 articles that met qualifying conditions for further review. Of these, 3 were methodologically adequate for final review. None of these 3 studies provided evidence showing the efficacy of steroid injections. The results of the major trials reviewed can be found in the Table .

TABLE
Major placebo-controlled trials of injectable steroids for shoulder pain

Recommendations from others

The American Academy of Orthopaedic Surgeons’ clinical guideline for shoulder pain⁹ recommends the following for rotator cuff disease: avoidance of irritating activity; anti-inflammatory medications if tolerated; exercises to recover and maintain passive range of motion; exercises to strengthen the rotator cuff once acute symptoms abated. If these are unsuccessful over several weeks, they recommend considering subacromial injection of local anesthetic and a short-acting corticosteroid. They gave their recommendation a “B” rating (some evidence exists to suggest benefit).

Evidence summary

Multiple case-control studies have identified risk factors for SIDS, which are presented along (with odds ratios) in Table 1.^1-6 None of these case-control studies found apnea of prematurity to be a risk factor for SIDS.

A prospective cohort study of 1079 infants monitored for cardiorespiratory events, the Collaborative Home Infant Monitor Evaluation (CHIME) study, demonstrated that prior to 43 weeks postconceptional age, preterm infants had a statistically significant greater risk of extreme events (apnea or bradycardia longer than 30 seconds) compared with healthy term infants (Table 2). After 43 weeks postconceptional age, there were no differences in incidence of apnea or bradycardia, comparing preterm and term infants. Neither preterm infants nor infants with apnea, bradycardia, or apparent life-threatening events had increased incidences of SIDS.⁷

Significant financial costs are associated with home monitoring. The average monthly cost is $300 to $400, not including physician fees. This would lead to an estimated annual cost of $24 million dollars if every infant <1500 grams in the United States were monitored.⁸

The psychological costs of home apnea monitoring have also been studied. One hundred and four parents of monitored and unmonitored infants were enrolled in a questionnaire study to determine emotional distress and family functioning. As is common among families in the postpartum period, all experienced increased stress. But parents of monitored infants, compared with parents of unmonitored infants, had an increased incidence of subjective depression (number needed to harm [NNH]=7) and hostility (NNH=12) at 2 weeks postpartum. Interestingly, at 1-year follow-up interviews, 83% of parents who had consistently used the monitor reported feeling more secure for having used it and 69% believed that monitor use had been helpful.⁹

Recommendations from others

The American Academy of Pediatrics (AAP) acknowledges that no established predictive or precursor relationship exists between prolonged apnea and SIDS, stating that the “prevention of SIDS is not an acceptable indication for home cardiorespiratory monitoring.” They issue a weak recommendation that home cardiorespiratory monitoring may be necessary for recurrent apnea, recurrent bradycardia, hypoxemia, chronic lung disease, and technology-dependent infants. Finally, they state that monitoring should be discontinued at 43 weeks postconceptional age or after cessation of extreme cardiorespiratory events, whichever occurs last. The AAP recommends proven practices such as supine sleeping position, a safe sleeping environment, and elimination of prenatal and postnatal exposure to tobacco smoke to decrease the risk of SIDS.⁸

TABLE 1
Risk factors for SIDS

Evidence summary

Multiple case-control studies have identified risk factors for SIDS, which are presented along (with odds ratios) in Table 1.^1-6 None of these case-control studies found apnea of prematurity to be a risk factor for SIDS.

A prospective cohort study of 1079 infants monitored for cardiorespiratory events, the Collaborative Home Infant Monitor Evaluation (CHIME) study, demonstrated that prior to 43 weeks postconceptional age, preterm infants had a statistically significant greater risk of extreme events (apnea or bradycardia longer than 30 seconds) compared with healthy term infants (Table 2). After 43 weeks postconceptional age, there were no differences in incidence of apnea or bradycardia, comparing preterm and term infants. Neither preterm infants nor infants with apnea, bradycardia, or apparent life-threatening events had increased incidences of SIDS.⁷

Significant financial costs are associated with home monitoring. The average monthly cost is $300 to $400, not including physician fees. This would lead to an estimated annual cost of $24 million dollars if every infant <1500 grams in the United States were monitored.⁸

The psychological costs of home apnea monitoring have also been studied. One hundred and four parents of monitored and unmonitored infants were enrolled in a questionnaire study to determine emotional distress and family functioning. As is common among families in the postpartum period, all experienced increased stress. But parents of monitored infants, compared with parents of unmonitored infants, had an increased incidence of subjective depression (number needed to harm [NNH]=7) and hostility (NNH=12) at 2 weeks postpartum. Interestingly, at 1-year follow-up interviews, 83% of parents who had consistently used the monitor reported feeling more secure for having used it and 69% believed that monitor use had been helpful.⁹

Recommendations from others

The American Academy of Pediatrics (AAP) acknowledges that no established predictive or precursor relationship exists between prolonged apnea and SIDS, stating that the “prevention of SIDS is not an acceptable indication for home cardiorespiratory monitoring.” They issue a weak recommendation that home cardiorespiratory monitoring may be necessary for recurrent apnea, recurrent bradycardia, hypoxemia, chronic lung disease, and technology-dependent infants. Finally, they state that monitoring should be discontinued at 43 weeks postconceptional age or after cessation of extreme cardiorespiratory events, whichever occurs last. The AAP recommends proven practices such as supine sleeping position, a safe sleeping environment, and elimination of prenatal and postnatal exposure to tobacco smoke to decrease the risk of SIDS.⁸

TABLE 1
Risk factors for SIDS

Evidence summary

Multiple case-control studies have identified risk factors for SIDS, which are presented along (with odds ratios) in Table 1.^1-6 None of these case-control studies found apnea of prematurity to be a risk factor for SIDS.

A prospective cohort study of 1079 infants monitored for cardiorespiratory events, the Collaborative Home Infant Monitor Evaluation (CHIME) study, demonstrated that prior to 43 weeks postconceptional age, preterm infants had a statistically significant greater risk of extreme events (apnea or bradycardia longer than 30 seconds) compared with healthy term infants (Table 2). After 43 weeks postconceptional age, there were no differences in incidence of apnea or bradycardia, comparing preterm and term infants. Neither preterm infants nor infants with apnea, bradycardia, or apparent life-threatening events had increased incidences of SIDS.⁷

Significant financial costs are associated with home monitoring. The average monthly cost is $300 to $400, not including physician fees. This would lead to an estimated annual cost of $24 million dollars if every infant <1500 grams in the United States were monitored.⁸

The psychological costs of home apnea monitoring have also been studied. One hundred and four parents of monitored and unmonitored infants were enrolled in a questionnaire study to determine emotional distress and family functioning. As is common among families in the postpartum period, all experienced increased stress. But parents of monitored infants, compared with parents of unmonitored infants, had an increased incidence of subjective depression (number needed to harm [NNH]=7) and hostility (NNH=12) at 2 weeks postpartum. Interestingly, at 1-year follow-up interviews, 83% of parents who had consistently used the monitor reported feeling more secure for having used it and 69% believed that monitor use had been helpful.⁹

Recommendations from others

The American Academy of Pediatrics (AAP) acknowledges that no established predictive or precursor relationship exists between prolonged apnea and SIDS, stating that the “prevention of SIDS is not an acceptable indication for home cardiorespiratory monitoring.” They issue a weak recommendation that home cardiorespiratory monitoring may be necessary for recurrent apnea, recurrent bradycardia, hypoxemia, chronic lung disease, and technology-dependent infants. Finally, they state that monitoring should be discontinued at 43 weeks postconceptional age or after cessation of extreme cardiorespiratory events, whichever occurs last. The AAP recommends proven practices such as supine sleeping position, a safe sleeping environment, and elimination of prenatal and postnatal exposure to tobacco smoke to decrease the risk of SIDS.⁸

TABLE 1
Risk factors for SIDS

Evidence summary

A variety of brief ADHD-specific rating scales are used for both parent and teacher assessment of child behavior. Rating scales are generally evaluated to establish mean scores for affected and unaffected children. Many scales publish such normative data in commercially available manuals. Some scales have been evaluated by 1 or more independent studies to compare children with and without ADHD. Rating scales have not been evaluated as a sole tool for the diagnosis of ADHD.

The test characteristics of a particular scale depend on the cut points for a positive or negative test. The usefulness of psychological tests in discriminating normal from abnormal behavior is often reported as “effect size.” The effect size is the difference in mean scores between 2 populations divided by an estimate of the individual standard deviation.¹ An effect size of 4.0 means that abnormal subjects and normal controls are separated 4 standard deviations and thus almost completely separated. An effect size of 1.0 shows significant overlap between the 2 populations. An effect size of 4.0 is roughly equivalent to a sensitivity and specificity of 97%. An effect size of 1.0 is roughly equal to a sensitivity and specificity of 71%.

Table 1 outlines the characteristics and effect size of several available brief ADHD-specific checklists.^2-4,6,11-13 Typically, the gold standard was a clinical diagnostic interview, usually conducted by a clinical psychologist, as well as supporting data from schools and parents.

TABLE
Descriptive characteristics of abbreviated symptom checklists for ADHD

Recommendations from others

The American Academy of Pediatrics states that the use of ADHD-specific checklists is a clinical option when evaluating children for ADHD. They caution that the ADHD scales may function less well in clinicians’ offices than suggested by reported effect size and, in addition, rating scales are subject to bias and may convey a false sense of validity. They also state that it is not known if these scales provide additional information beyond a careful clinical assessment.⁷

The Institute for Clinical Systems Improvement recommends use of at least 1 ADHD-specific rating scale to be administered to parents and teachers. This information should be used as part of the overall historical database for the child and should not be used as the sole criteria for diagnosis of ADHD.⁸

Many sources agree that ADHD-specific rating scales allow a rapid and consistent collection of information from multiple sources. However, the information they provide is necessary, but not sufficient, to make a definitive diagnosis of ADHD. In addition to assisting in diagnosis, checklists can be helpful in monitoring treatment changes once a diagnosis has been established.

Evidence summary

A variety of brief ADHD-specific rating scales are used for both parent and teacher assessment of child behavior. Rating scales are generally evaluated to establish mean scores for affected and unaffected children. Many scales publish such normative data in commercially available manuals. Some scales have been evaluated by 1 or more independent studies to compare children with and without ADHD. Rating scales have not been evaluated as a sole tool for the diagnosis of ADHD.

The test characteristics of a particular scale depend on the cut points for a positive or negative test. The usefulness of psychological tests in discriminating normal from abnormal behavior is often reported as “effect size.” The effect size is the difference in mean scores between 2 populations divided by an estimate of the individual standard deviation.¹ An effect size of 4.0 means that abnormal subjects and normal controls are separated 4 standard deviations and thus almost completely separated. An effect size of 1.0 shows significant overlap between the 2 populations. An effect size of 4.0 is roughly equivalent to a sensitivity and specificity of 97%. An effect size of 1.0 is roughly equal to a sensitivity and specificity of 71%.

Table 1 outlines the characteristics and effect size of several available brief ADHD-specific checklists.^2-4,6,11-13 Typically, the gold standard was a clinical diagnostic interview, usually conducted by a clinical psychologist, as well as supporting data from schools and parents.

TABLE
Descriptive characteristics of abbreviated symptom checklists for ADHD

Recommendations from others

The American Academy of Pediatrics states that the use of ADHD-specific checklists is a clinical option when evaluating children for ADHD. They caution that the ADHD scales may function less well in clinicians’ offices than suggested by reported effect size and, in addition, rating scales are subject to bias and may convey a false sense of validity. They also state that it is not known if these scales provide additional information beyond a careful clinical assessment.⁷

The Institute for Clinical Systems Improvement recommends use of at least 1 ADHD-specific rating scale to be administered to parents and teachers. This information should be used as part of the overall historical database for the child and should not be used as the sole criteria for diagnosis of ADHD.⁸

Many sources agree that ADHD-specific rating scales allow a rapid and consistent collection of information from multiple sources. However, the information they provide is necessary, but not sufficient, to make a definitive diagnosis of ADHD. In addition to assisting in diagnosis, checklists can be helpful in monitoring treatment changes once a diagnosis has been established.

Evidence summary

A variety of brief ADHD-specific rating scales are used for both parent and teacher assessment of child behavior. Rating scales are generally evaluated to establish mean scores for affected and unaffected children. Many scales publish such normative data in commercially available manuals. Some scales have been evaluated by 1 or more independent studies to compare children with and without ADHD. Rating scales have not been evaluated as a sole tool for the diagnosis of ADHD.

The test characteristics of a particular scale depend on the cut points for a positive or negative test. The usefulness of psychological tests in discriminating normal from abnormal behavior is often reported as “effect size.” The effect size is the difference in mean scores between 2 populations divided by an estimate of the individual standard deviation.¹ An effect size of 4.0 means that abnormal subjects and normal controls are separated 4 standard deviations and thus almost completely separated. An effect size of 1.0 shows significant overlap between the 2 populations. An effect size of 4.0 is roughly equivalent to a sensitivity and specificity of 97%. An effect size of 1.0 is roughly equal to a sensitivity and specificity of 71%.

Table 1 outlines the characteristics and effect size of several available brief ADHD-specific checklists.^2-4,6,11-13 Typically, the gold standard was a clinical diagnostic interview, usually conducted by a clinical psychologist, as well as supporting data from schools and parents.

TABLE
Descriptive characteristics of abbreviated symptom checklists for ADHD

Recommendations from others

The American Academy of Pediatrics states that the use of ADHD-specific checklists is a clinical option when evaluating children for ADHD. They caution that the ADHD scales may function less well in clinicians’ offices than suggested by reported effect size and, in addition, rating scales are subject to bias and may convey a false sense of validity. They also state that it is not known if these scales provide additional information beyond a careful clinical assessment.⁷

The Institute for Clinical Systems Improvement recommends use of at least 1 ADHD-specific rating scale to be administered to parents and teachers. This information should be used as part of the overall historical database for the child and should not be used as the sole criteria for diagnosis of ADHD.⁸

Many sources agree that ADHD-specific rating scales allow a rapid and consistent collection of information from multiple sources. However, the information they provide is necessary, but not sufficient, to make a definitive diagnosis of ADHD. In addition to assisting in diagnosis, checklists can be helpful in monitoring treatment changes once a diagnosis has been established.

Evidence summary

A prospective cohort study compared cardiovascular events among patients with white-coat hypertension vs those with sustained hypertension. The study evaluated 479 patients with persistently elevated clinic systolic blood pressures of 140 to 180 mm Hg. Using 24-hour intraarterial ambulatory blood pressure monitoring (ABPM), they found that 126 patients had ambulatory blood pressures below 140/90 mm Hg (white-coat hypertension) while 353 patients maintained pressures above 140/90 mm Hg (sustained hypertension). On average, white-coat hypertension patients were younger than sustained hypertension patients (44 vs 52 years) but were otherwise similar. Over the next 9 years, patients with white-coat hypertension had significantly fewer cardiovascular events than patients with sustained hypertension (Table).¹

Another prospective cohort study compared fatal and nonfatal cardiovascular event rates among patients who had white-coat hypertension, sustained hypertension, or were normotensive. Investigators performed 24-hour ABPM on 1187 patients who had clinic blood pressures over 140/90 on three visits. They found that 228 patients had white-coat hypertension, defined as mean ambulatory blood pressures below the 90th percentile of a normotensive population, and 959 patients had sustained hypertension. They followed these patients, along with 205 normotensive controls, for a mean of 3.2 years. Cardiovascular event rates did not differ significantly between normotensive and white-coat hypertension patients (P=.83; see Table), but the difference in event-free survival between the sustained hypertension group and both the white-coat hypertension and normotensive groups was highly significant (P=.002).²

In contrast, a recent 10-year longitudinal study of 146 normotensive people, 76 people with white-coat hypertension, and 344 with sustained hypertension showed that cardiovascular event rates were similar for patients with white-coat and sustained hypertension, and were significantly higher than in the normotensive group (P=.03 overall, P=.03 between white-coat hypertension and normotension and P=.01 between sustained hypertension and normotension).³

One randomized trial evaluated outcomes of antihypertensive therapy for white-coat hypertension for patients aged >60 years. Ninety-nine patients with white-coat hypertension were identified on the basis of systolic blood pressure greater than 160 mm Hg in clinic and normal 24-hour ABPM and were randomized to either place-bo or drug therapy. Active treatment did not significantly lower ambulatory blood pressure in white-coat hypertension, but it did reduce blood pressure measured in clinic. After a year, medication produced an absolute reduction in cardiovascular events of 8.6%, and in stroke of 4.2%. Neither result was statistically significant due to the small sample size.⁴

TABLE
Cohort studies of patients with white-coat hypertension

Recommendations from others

The American College of Cardiology and American Academy of Family Physicians have made no specific recommendations about white-coat hypertension. The Blood Pressure Monitoring Task Force V concluded that a significant number of white-coat hypertension patients become truly hypertensive over years of follow-up.⁵

Experts agree that patients with white-coat hypertension should be indefinitely monitored for the development of sustained hypertension.⁶ Treatment is not needed unless the patient has sustained hypertension, evidence of cardiovascular disease, or signs of target organ injury.^7,8 Typically, expert opinion recommends confirming the diagnosis of white-coat hypertension with home blood pressure records or ambulatory blood pressure monitoring.

Evidence summary

A prospective cohort study compared cardiovascular events among patients with white-coat hypertension vs those with sustained hypertension. The study evaluated 479 patients with persistently elevated clinic systolic blood pressures of 140 to 180 mm Hg. Using 24-hour intraarterial ambulatory blood pressure monitoring (ABPM), they found that 126 patients had ambulatory blood pressures below 140/90 mm Hg (white-coat hypertension) while 353 patients maintained pressures above 140/90 mm Hg (sustained hypertension). On average, white-coat hypertension patients were younger than sustained hypertension patients (44 vs 52 years) but were otherwise similar. Over the next 9 years, patients with white-coat hypertension had significantly fewer cardiovascular events than patients with sustained hypertension (Table).¹

Another prospective cohort study compared fatal and nonfatal cardiovascular event rates among patients who had white-coat hypertension, sustained hypertension, or were normotensive. Investigators performed 24-hour ABPM on 1187 patients who had clinic blood pressures over 140/90 on three visits. They found that 228 patients had white-coat hypertension, defined as mean ambulatory blood pressures below the 90th percentile of a normotensive population, and 959 patients had sustained hypertension. They followed these patients, along with 205 normotensive controls, for a mean of 3.2 years. Cardiovascular event rates did not differ significantly between normotensive and white-coat hypertension patients (P=.83; see Table), but the difference in event-free survival between the sustained hypertension group and both the white-coat hypertension and normotensive groups was highly significant (P=.002).²

In contrast, a recent 10-year longitudinal study of 146 normotensive people, 76 people with white-coat hypertension, and 344 with sustained hypertension showed that cardiovascular event rates were similar for patients with white-coat and sustained hypertension, and were significantly higher than in the normotensive group (P=.03 overall, P=.03 between white-coat hypertension and normotension and P=.01 between sustained hypertension and normotension).³

One randomized trial evaluated outcomes of antihypertensive therapy for white-coat hypertension for patients aged >60 years. Ninety-nine patients with white-coat hypertension were identified on the basis of systolic blood pressure greater than 160 mm Hg in clinic and normal 24-hour ABPM and were randomized to either place-bo or drug therapy. Active treatment did not significantly lower ambulatory blood pressure in white-coat hypertension, but it did reduce blood pressure measured in clinic. After a year, medication produced an absolute reduction in cardiovascular events of 8.6%, and in stroke of 4.2%. Neither result was statistically significant due to the small sample size.⁴

TABLE
Cohort studies of patients with white-coat hypertension

Recommendations from others

The American College of Cardiology and American Academy of Family Physicians have made no specific recommendations about white-coat hypertension. The Blood Pressure Monitoring Task Force V concluded that a significant number of white-coat hypertension patients become truly hypertensive over years of follow-up.⁵

Experts agree that patients with white-coat hypertension should be indefinitely monitored for the development of sustained hypertension.⁶ Treatment is not needed unless the patient has sustained hypertension, evidence of cardiovascular disease, or signs of target organ injury.^7,8 Typically, expert opinion recommends confirming the diagnosis of white-coat hypertension with home blood pressure records or ambulatory blood pressure monitoring.

Evidence summary

A prospective cohort study compared cardiovascular events among patients with white-coat hypertension vs those with sustained hypertension. The study evaluated 479 patients with persistently elevated clinic systolic blood pressures of 140 to 180 mm Hg. Using 24-hour intraarterial ambulatory blood pressure monitoring (ABPM), they found that 126 patients had ambulatory blood pressures below 140/90 mm Hg (white-coat hypertension) while 353 patients maintained pressures above 140/90 mm Hg (sustained hypertension). On average, white-coat hypertension patients were younger than sustained hypertension patients (44 vs 52 years) but were otherwise similar. Over the next 9 years, patients with white-coat hypertension had significantly fewer cardiovascular events than patients with sustained hypertension (Table).¹

Another prospective cohort study compared fatal and nonfatal cardiovascular event rates among patients who had white-coat hypertension, sustained hypertension, or were normotensive. Investigators performed 24-hour ABPM on 1187 patients who had clinic blood pressures over 140/90 on three visits. They found that 228 patients had white-coat hypertension, defined as mean ambulatory blood pressures below the 90th percentile of a normotensive population, and 959 patients had sustained hypertension. They followed these patients, along with 205 normotensive controls, for a mean of 3.2 years. Cardiovascular event rates did not differ significantly between normotensive and white-coat hypertension patients (P=.83; see Table), but the difference in event-free survival between the sustained hypertension group and both the white-coat hypertension and normotensive groups was highly significant (P=.002).²

In contrast, a recent 10-year longitudinal study of 146 normotensive people, 76 people with white-coat hypertension, and 344 with sustained hypertension showed that cardiovascular event rates were similar for patients with white-coat and sustained hypertension, and were significantly higher than in the normotensive group (P=.03 overall, P=.03 between white-coat hypertension and normotension and P=.01 between sustained hypertension and normotension).³

One randomized trial evaluated outcomes of antihypertensive therapy for white-coat hypertension for patients aged >60 years. Ninety-nine patients with white-coat hypertension were identified on the basis of systolic blood pressure greater than 160 mm Hg in clinic and normal 24-hour ABPM and were randomized to either place-bo or drug therapy. Active treatment did not significantly lower ambulatory blood pressure in white-coat hypertension, but it did reduce blood pressure measured in clinic. After a year, medication produced an absolute reduction in cardiovascular events of 8.6%, and in stroke of 4.2%. Neither result was statistically significant due to the small sample size.⁴

TABLE
Cohort studies of patients with white-coat hypertension

Recommendations from others

The American College of Cardiology and American Academy of Family Physicians have made no specific recommendations about white-coat hypertension. The Blood Pressure Monitoring Task Force V concluded that a significant number of white-coat hypertension patients become truly hypertensive over years of follow-up.⁵

Experts agree that patients with white-coat hypertension should be indefinitely monitored for the development of sustained hypertension.⁶ Treatment is not needed unless the patient has sustained hypertension, evidence of cardiovascular disease, or signs of target organ injury.^7,8 Typically, expert opinion recommends confirming the diagnosis of white-coat hypertension with home blood pressure records or ambulatory blood pressure monitoring.