Clinical Inquiries

Evidence summary

Yoga, through static physical postures (or asanas), uses stretching to improve muscular strength and flexibility, which could be beneficial for low-back-related pain management.¹ Hatha yoga, which incorporates breathing and movement, has provided limited benefit in musculoskeletal-related pain.² Hatha yoga is distinguished from other yoga practices in that it is based on the knowledge, development, and balance of psychophysical energies. A large systematic review of yoga used for various medical conditions found over 120 studies.³ Anecdotal reports were excluded. The authors reported no studies directly evaluating effect of yoga on back pain.

A randomized controlled trial studied a 6-week modified hatha yoga protocol with 22 patients.⁴ The yoga group spent an hour with a certified instructor twice weekly, while the control group received the same intervention delayed until the study phase was completed. This underpowered pilot study found trends in functional measurement scores for improved balance and flexibility, as well as decreased disability and depression in the yoga group, but the sample size was too small to detect significant changes.

Patients who practice hatha yoga say it is valuable for preventing and managing stress-related chronic health problems, including low back pain. In a survey of 3000 people receiving yoga for health ailments (1142 [38%] with back pain), 98% claimed that yoga benefited them.⁵

In a case series of 16 patients using various asanas for rehabilitation, 11 (69%) reported significant improvement, with near normal mobility and absence of pain.⁶ Those who reported recurring back pain also reported irregular practice of yoga. In another case series, 21 women aged ≥60 years (mean age, 75) with hyperkyphosis, participated in twice-weekly 1-hour sessions of hatha yoga for 12 weeks. Measured height increased by a mean of 0.52 cm, forward curvature diminished, patients were able to get out of chairs faster, and they had longer functional reach. Eleven patients (48%) reported increased postural awareness/improvement and improved well-being; 58% perceived improvement in their physical functioning.⁷

Clearly, more studies are required to determine the effects of yoga on lower back pain. Larger randomized sample sizes, group and individualized formats, and longer follow-up are needed. Control groups should involve both group and nongroup settings, to detect any benefit that may be derived from group support. No reports of harm from yoga in low-back pain therapy were reported in the few studies found.

Recommendations from others

The Philadelphia Panel formulated evidence-based guidelines for selected rehabilitation interventions in the management of low back pain for outpatient adults.⁸ Continuation of normal activity improves rate of return to work compared with enforced bed rest. Randomized controlled trials demonstrate no clinically important effect (15% improvement compared with control) with stretching or strengthening exercises, mechanical traction, or TENS. The panel found insufficient evidence to support the use of mechanical traction for patient global improvement and return to work. Therapeutic exercise—including stretching, strengthening, and mobility exercises—significantly reduces pain and improves function for chronic low back pain (longer than 12 weeks); but there was no clinical benefit in facilitating return to work. No specific comments on yoga appeared in their recommendations.

The US Preventive Services Task Force reports that evidence is insufficient to recommend for or against counseling patients to exercise to prevent low back pain; it makes no mention about yoga.⁹

Evidence summary

Yoga, through static physical postures (or asanas), uses stretching to improve muscular strength and flexibility, which could be beneficial for low-back-related pain management.¹ Hatha yoga, which incorporates breathing and movement, has provided limited benefit in musculoskeletal-related pain.² Hatha yoga is distinguished from other yoga practices in that it is based on the knowledge, development, and balance of psychophysical energies. A large systematic review of yoga used for various medical conditions found over 120 studies.³ Anecdotal reports were excluded. The authors reported no studies directly evaluating effect of yoga on back pain.

A randomized controlled trial studied a 6-week modified hatha yoga protocol with 22 patients.⁴ The yoga group spent an hour with a certified instructor twice weekly, while the control group received the same intervention delayed until the study phase was completed. This underpowered pilot study found trends in functional measurement scores for improved balance and flexibility, as well as decreased disability and depression in the yoga group, but the sample size was too small to detect significant changes.

Patients who practice hatha yoga say it is valuable for preventing and managing stress-related chronic health problems, including low back pain. In a survey of 3000 people receiving yoga for health ailments (1142 [38%] with back pain), 98% claimed that yoga benefited them.⁵

In a case series of 16 patients using various asanas for rehabilitation, 11 (69%) reported significant improvement, with near normal mobility and absence of pain.⁶ Those who reported recurring back pain also reported irregular practice of yoga. In another case series, 21 women aged ≥60 years (mean age, 75) with hyperkyphosis, participated in twice-weekly 1-hour sessions of hatha yoga for 12 weeks. Measured height increased by a mean of 0.52 cm, forward curvature diminished, patients were able to get out of chairs faster, and they had longer functional reach. Eleven patients (48%) reported increased postural awareness/improvement and improved well-being; 58% perceived improvement in their physical functioning.⁷

Clearly, more studies are required to determine the effects of yoga on lower back pain. Larger randomized sample sizes, group and individualized formats, and longer follow-up are needed. Control groups should involve both group and nongroup settings, to detect any benefit that may be derived from group support. No reports of harm from yoga in low-back pain therapy were reported in the few studies found.

Recommendations from others

The Philadelphia Panel formulated evidence-based guidelines for selected rehabilitation interventions in the management of low back pain for outpatient adults.⁸ Continuation of normal activity improves rate of return to work compared with enforced bed rest. Randomized controlled trials demonstrate no clinically important effect (15% improvement compared with control) with stretching or strengthening exercises, mechanical traction, or TENS. The panel found insufficient evidence to support the use of mechanical traction for patient global improvement and return to work. Therapeutic exercise—including stretching, strengthening, and mobility exercises—significantly reduces pain and improves function for chronic low back pain (longer than 12 weeks); but there was no clinical benefit in facilitating return to work. No specific comments on yoga appeared in their recommendations.

The US Preventive Services Task Force reports that evidence is insufficient to recommend for or against counseling patients to exercise to prevent low back pain; it makes no mention about yoga.⁹

Evidence summary

Yoga, through static physical postures (or asanas), uses stretching to improve muscular strength and flexibility, which could be beneficial for low-back-related pain management.¹ Hatha yoga, which incorporates breathing and movement, has provided limited benefit in musculoskeletal-related pain.² Hatha yoga is distinguished from other yoga practices in that it is based on the knowledge, development, and balance of psychophysical energies. A large systematic review of yoga used for various medical conditions found over 120 studies.³ Anecdotal reports were excluded. The authors reported no studies directly evaluating effect of yoga on back pain.

A randomized controlled trial studied a 6-week modified hatha yoga protocol with 22 patients.⁴ The yoga group spent an hour with a certified instructor twice weekly, while the control group received the same intervention delayed until the study phase was completed. This underpowered pilot study found trends in functional measurement scores for improved balance and flexibility, as well as decreased disability and depression in the yoga group, but the sample size was too small to detect significant changes.

Patients who practice hatha yoga say it is valuable for preventing and managing stress-related chronic health problems, including low back pain. In a survey of 3000 people receiving yoga for health ailments (1142 [38%] with back pain), 98% claimed that yoga benefited them.⁵

In a case series of 16 patients using various asanas for rehabilitation, 11 (69%) reported significant improvement, with near normal mobility and absence of pain.⁶ Those who reported recurring back pain also reported irregular practice of yoga. In another case series, 21 women aged ≥60 years (mean age, 75) with hyperkyphosis, participated in twice-weekly 1-hour sessions of hatha yoga for 12 weeks. Measured height increased by a mean of 0.52 cm, forward curvature diminished, patients were able to get out of chairs faster, and they had longer functional reach. Eleven patients (48%) reported increased postural awareness/improvement and improved well-being; 58% perceived improvement in their physical functioning.⁷

Clearly, more studies are required to determine the effects of yoga on lower back pain. Larger randomized sample sizes, group and individualized formats, and longer follow-up are needed. Control groups should involve both group and nongroup settings, to detect any benefit that may be derived from group support. No reports of harm from yoga in low-back pain therapy were reported in the few studies found.

Recommendations from others

The Philadelphia Panel formulated evidence-based guidelines for selected rehabilitation interventions in the management of low back pain for outpatient adults.⁸ Continuation of normal activity improves rate of return to work compared with enforced bed rest. Randomized controlled trials demonstrate no clinically important effect (15% improvement compared with control) with stretching or strengthening exercises, mechanical traction, or TENS. The panel found insufficient evidence to support the use of mechanical traction for patient global improvement and return to work. Therapeutic exercise—including stretching, strengthening, and mobility exercises—significantly reduces pain and improves function for chronic low back pain (longer than 12 weeks); but there was no clinical benefit in facilitating return to work. No specific comments on yoga appeared in their recommendations.

The US Preventive Services Task Force reports that evidence is insufficient to recommend for or against counseling patients to exercise to prevent low back pain; it makes no mention about yoga.⁹

Evidence Summary

The subjects of most ADHD medication studies have been school-age children. Most children with ADHD will have symptoms persisting into teenage years, and methylphenidate has been increasingly prescribed for them.^1,2 Various systematic reviews and meta-analyses have demonstrated the effectiveness of short-term methylphenidate in the treatment of adolescents with ADHD.^3-5 Most participants in these studies are males aged <13 years. Therefore, any conclusions about the effectiveness of methylphenidate in older adolescents must be inferred.

The most comprehensive systematic review found 8 well-controlled crossover trials with an average sample size of 24.8 (range, 9–48).⁶ The average duration of the studies was 6 weeks. The majority of the participants were white males with a mean age of 13 years. Each study showed statistically significant improvement from treatment with methylphenidate. Average effect sizes were calculated for 3 domains: ADHD symptoms (0.94), social behavior (1.06), and academic performance (1.25). Effect sizes were calculated using a modified Cohen’s d, which is the difference between the treated and untreated means divided by the standard deviation in the untreated condition. It is difficult to translate these changes in effect size into clinically meaningful outcome measures, although one rule of thumb estimates an effect size of 0.8 is moderate to large.

Of the 3 studies that reported the proportion of subjects with clinically significant improvement, the modal result was about one half showings improvement with methylphenidate. Of trials assessing dosing levels, <50% found significant differences between “low” and “high” doses. However, the researchers did not give their definition of low and high doses. Also, diminishing clinical improvement was noted with higher methylphenidate doses.

A single study on the once-daily stimulant preparation, extended-release methylphenidate, shows statistically significant improvement in adolescent ADHD.⁷ In this multicenter, randomized, double-blind, placebo-controlled trial of 177 adolescents, subjects were given placebo (n=87) or extended-release methylphenidate (n=90) at titrated doses from 18 mg/d to 72 mg/d. Following a subsequent 2-week randomization phase, clinical investigators found extended-release methylphenidate significantly superior to placebo (P=.001) on the ADHD scale. Subjects also rated it significantly superior to placebo (P=.001) on the Conners-Wells’ Self-Report Scale. Mean dose and average age were not reported. This study has been presented as an abstract and is not yet published.

Recommendations from others

The American Academy of Child and Adolescent Psychiatry (AACAP) supports the prescribing of methylphenidate in adolescents with ADHD.⁸ Given the unique psychosocial, environmental, and scheduling challenges of adolescence, the AACAP mentions extended-release methylphenidate as “well-suited for treatment of adolescents.”

Evidence Summary

The subjects of most ADHD medication studies have been school-age children. Most children with ADHD will have symptoms persisting into teenage years, and methylphenidate has been increasingly prescribed for them.^1,2 Various systematic reviews and meta-analyses have demonstrated the effectiveness of short-term methylphenidate in the treatment of adolescents with ADHD.^3-5 Most participants in these studies are males aged <13 years. Therefore, any conclusions about the effectiveness of methylphenidate in older adolescents must be inferred.

The most comprehensive systematic review found 8 well-controlled crossover trials with an average sample size of 24.8 (range, 9–48).⁶ The average duration of the studies was 6 weeks. The majority of the participants were white males with a mean age of 13 years. Each study showed statistically significant improvement from treatment with methylphenidate. Average effect sizes were calculated for 3 domains: ADHD symptoms (0.94), social behavior (1.06), and academic performance (1.25). Effect sizes were calculated using a modified Cohen’s d, which is the difference between the treated and untreated means divided by the standard deviation in the untreated condition. It is difficult to translate these changes in effect size into clinically meaningful outcome measures, although one rule of thumb estimates an effect size of 0.8 is moderate to large.

Of the 3 studies that reported the proportion of subjects with clinically significant improvement, the modal result was about one half showings improvement with methylphenidate. Of trials assessing dosing levels, <50% found significant differences between “low” and “high” doses. However, the researchers did not give their definition of low and high doses. Also, diminishing clinical improvement was noted with higher methylphenidate doses.

A single study on the once-daily stimulant preparation, extended-release methylphenidate, shows statistically significant improvement in adolescent ADHD.⁷ In this multicenter, randomized, double-blind, placebo-controlled trial of 177 adolescents, subjects were given placebo (n=87) or extended-release methylphenidate (n=90) at titrated doses from 18 mg/d to 72 mg/d. Following a subsequent 2-week randomization phase, clinical investigators found extended-release methylphenidate significantly superior to placebo (P=.001) on the ADHD scale. Subjects also rated it significantly superior to placebo (P=.001) on the Conners-Wells’ Self-Report Scale. Mean dose and average age were not reported. This study has been presented as an abstract and is not yet published.

Recommendations from others

The American Academy of Child and Adolescent Psychiatry (AACAP) supports the prescribing of methylphenidate in adolescents with ADHD.⁸ Given the unique psychosocial, environmental, and scheduling challenges of adolescence, the AACAP mentions extended-release methylphenidate as “well-suited for treatment of adolescents.”

Evidence Summary

The subjects of most ADHD medication studies have been school-age children. Most children with ADHD will have symptoms persisting into teenage years, and methylphenidate has been increasingly prescribed for them.^1,2 Various systematic reviews and meta-analyses have demonstrated the effectiveness of short-term methylphenidate in the treatment of adolescents with ADHD.^3-5 Most participants in these studies are males aged <13 years. Therefore, any conclusions about the effectiveness of methylphenidate in older adolescents must be inferred.

The most comprehensive systematic review found 8 well-controlled crossover trials with an average sample size of 24.8 (range, 9–48).⁶ The average duration of the studies was 6 weeks. The majority of the participants were white males with a mean age of 13 years. Each study showed statistically significant improvement from treatment with methylphenidate. Average effect sizes were calculated for 3 domains: ADHD symptoms (0.94), social behavior (1.06), and academic performance (1.25). Effect sizes were calculated using a modified Cohen’s d, which is the difference between the treated and untreated means divided by the standard deviation in the untreated condition. It is difficult to translate these changes in effect size into clinically meaningful outcome measures, although one rule of thumb estimates an effect size of 0.8 is moderate to large.

Of the 3 studies that reported the proportion of subjects with clinically significant improvement, the modal result was about one half showings improvement with methylphenidate. Of trials assessing dosing levels, <50% found significant differences between “low” and “high” doses. However, the researchers did not give their definition of low and high doses. Also, diminishing clinical improvement was noted with higher methylphenidate doses.

A single study on the once-daily stimulant preparation, extended-release methylphenidate, shows statistically significant improvement in adolescent ADHD.⁷ In this multicenter, randomized, double-blind, placebo-controlled trial of 177 adolescents, subjects were given placebo (n=87) or extended-release methylphenidate (n=90) at titrated doses from 18 mg/d to 72 mg/d. Following a subsequent 2-week randomization phase, clinical investigators found extended-release methylphenidate significantly superior to placebo (P=.001) on the ADHD scale. Subjects also rated it significantly superior to placebo (P=.001) on the Conners-Wells’ Self-Report Scale. Mean dose and average age were not reported. This study has been presented as an abstract and is not yet published.

Recommendations from others

The American Academy of Child and Adolescent Psychiatry (AACAP) supports the prescribing of methylphenidate in adolescents with ADHD.⁸ Given the unique psychosocial, environmental, and scheduling challenges of adolescence, the AACAP mentions extended-release methylphenidate as “well-suited for treatment of adolescents.”

TABLE
Inclusion and exclusion criteria for using thrombolytics for patients with acute ischemic CVA

Evidence summary

The 2003 American Heart Association guidelines recommend rtPA for acute ischemic stroke “for carefully selected patients” who also need crucial “ancillary care.”³ The evidence for these guidelines comes primarily from large double-blind placebo-controlled studies using rtPA. However, these studies—including NINDS,⁴ ECASS,⁵ and ATLANTIS⁶ —differ in their dosing regimen, timing, and other exclusion criteria, and outcome measurements.

The NINDS study, often employed as a benchmark,^3,7,8 used a slightly lower dose of rtPA than other studies and “required that no anticoagulants or antiplatelet agents be given for 24 hours after treatment and that blood pressure be maintained within prespecified values.”⁴ Patients were evaluated for inclusion according to strict criteria, similar to those shown in the Table.

Patients in research studies who were treated outside protocol guidelines, and patients treated in community hospitals, have not fared as well as the patients in NINDS. In Connecticut,⁹ a review of thrombolysis in acute ischemic stroke revealed protocol deviations in 67% of the patients treated. The number needed to harm (NNH) for death was only 4 (in other words, there was an additional patient death for every 4 patients treated with rtPA), and significant extracranial hemorrhage had an NNH of 8. In Cleveland,¹⁰ 50% of patients treated had at least 1 major protocol violation, and the NNH for symptomatic intracranial hemorrhage was 6. A quality improvement program in the Cleveland area lowered protocol violations to 19% and the NNH rose to 15.¹¹

Improved outcomes similar to NINDS have been noted where there are stroke units or teams with personnel such as neurosurgeons, strict adherence to protocols, and facilities available to give accurate and expedient interventions and imaging (eg, neuroradiologic interpretations of CTs).¹ These limits restrict the practical and safe use of rtPA to few of the millions of stroke victims.

The net positive outcome found in the Cochrane review¹ results from subtracting the significant increase in symptomatic intracranial hemorrhage (NNH=16; 95% CI, 11–25) from the larger primary decrease in disability/death (NNT=10; 95% CI, 6–22).¹ The overlapping confidence intervals of the outcomes was bothersome to the Cochrane reviewers.

Recommendations from others

Recommendations from the American Heart Association,² the American Academy of Neurology,⁶ and the 6th American College of Chest Physicians Consensus Conference on Antithrombotic Therapy⁷ substantially agree. With minor variations, all recommend rtPA with inclusion/exclusion criteria similar to those outlined in the Table.

TABLE
Inclusion and exclusion criteria for using thrombolytics for patients with acute ischemic CVA

Evidence summary

The 2003 American Heart Association guidelines recommend rtPA for acute ischemic stroke “for carefully selected patients” who also need crucial “ancillary care.”³ The evidence for these guidelines comes primarily from large double-blind placebo-controlled studies using rtPA. However, these studies—including NINDS,⁴ ECASS,⁵ and ATLANTIS⁶ —differ in their dosing regimen, timing, and other exclusion criteria, and outcome measurements.

The NINDS study, often employed as a benchmark,^3,7,8 used a slightly lower dose of rtPA than other studies and “required that no anticoagulants or antiplatelet agents be given for 24 hours after treatment and that blood pressure be maintained within prespecified values.”⁴ Patients were evaluated for inclusion according to strict criteria, similar to those shown in the Table.

Patients in research studies who were treated outside protocol guidelines, and patients treated in community hospitals, have not fared as well as the patients in NINDS. In Connecticut,⁹ a review of thrombolysis in acute ischemic stroke revealed protocol deviations in 67% of the patients treated. The number needed to harm (NNH) for death was only 4 (in other words, there was an additional patient death for every 4 patients treated with rtPA), and significant extracranial hemorrhage had an NNH of 8. In Cleveland,¹⁰ 50% of patients treated had at least 1 major protocol violation, and the NNH for symptomatic intracranial hemorrhage was 6. A quality improvement program in the Cleveland area lowered protocol violations to 19% and the NNH rose to 15.¹¹

Improved outcomes similar to NINDS have been noted where there are stroke units or teams with personnel such as neurosurgeons, strict adherence to protocols, and facilities available to give accurate and expedient interventions and imaging (eg, neuroradiologic interpretations of CTs).¹ These limits restrict the practical and safe use of rtPA to few of the millions of stroke victims.

The net positive outcome found in the Cochrane review¹ results from subtracting the significant increase in symptomatic intracranial hemorrhage (NNH=16; 95% CI, 11–25) from the larger primary decrease in disability/death (NNT=10; 95% CI, 6–22).¹ The overlapping confidence intervals of the outcomes was bothersome to the Cochrane reviewers.

Recommendations from others

Recommendations from the American Heart Association,² the American Academy of Neurology,⁶ and the 6th American College of Chest Physicians Consensus Conference on Antithrombotic Therapy⁷ substantially agree. With minor variations, all recommend rtPA with inclusion/exclusion criteria similar to those outlined in the Table.

TABLE
Inclusion and exclusion criteria for using thrombolytics for patients with acute ischemic CVA

Evidence summary

The 2003 American Heart Association guidelines recommend rtPA for acute ischemic stroke “for carefully selected patients” who also need crucial “ancillary care.”³ The evidence for these guidelines comes primarily from large double-blind placebo-controlled studies using rtPA. However, these studies—including NINDS,⁴ ECASS,⁵ and ATLANTIS⁶ —differ in their dosing regimen, timing, and other exclusion criteria, and outcome measurements.

The NINDS study, often employed as a benchmark,^3,7,8 used a slightly lower dose of rtPA than other studies and “required that no anticoagulants or antiplatelet agents be given for 24 hours after treatment and that blood pressure be maintained within prespecified values.”⁴ Patients were evaluated for inclusion according to strict criteria, similar to those shown in the Table.

Patients in research studies who were treated outside protocol guidelines, and patients treated in community hospitals, have not fared as well as the patients in NINDS. In Connecticut,⁹ a review of thrombolysis in acute ischemic stroke revealed protocol deviations in 67% of the patients treated. The number needed to harm (NNH) for death was only 4 (in other words, there was an additional patient death for every 4 patients treated with rtPA), and significant extracranial hemorrhage had an NNH of 8. In Cleveland,¹⁰ 50% of patients treated had at least 1 major protocol violation, and the NNH for symptomatic intracranial hemorrhage was 6. A quality improvement program in the Cleveland area lowered protocol violations to 19% and the NNH rose to 15.¹¹

Improved outcomes similar to NINDS have been noted where there are stroke units or teams with personnel such as neurosurgeons, strict adherence to protocols, and facilities available to give accurate and expedient interventions and imaging (eg, neuroradiologic interpretations of CTs).¹ These limits restrict the practical and safe use of rtPA to few of the millions of stroke victims.

The net positive outcome found in the Cochrane review¹ results from subtracting the significant increase in symptomatic intracranial hemorrhage (NNH=16; 95% CI, 11–25) from the larger primary decrease in disability/death (NNT=10; 95% CI, 6–22).¹ The overlapping confidence intervals of the outcomes was bothersome to the Cochrane reviewers.

Recommendations from others

Recommendations from the American Heart Association,² the American Academy of Neurology,⁶ and the 6th American College of Chest Physicians Consensus Conference on Antithrombotic Therapy⁷ substantially agree. With minor variations, all recommend rtPA with inclusion/exclusion criteria similar to those outlined in the Table.

Evidence Summary

Screening for hypothyroidism is more likely to detect the elevated TSH and normal free thyroxine level (FT4) of subclinical hypothyroidism than it is to detect overt hypothyroidism with a high TSH and a low FT4. We reviewed the accuracy of detection, natural history, and benefits and harms of treating subclinical hypothyroidism.

Detection. Subclinical hypothyroidism is found in 7% to 26% of women (with increasing prevalence as women reach age 60 and 70 years); overt hypothyroidism occurs in approximately 5%.^1-4 Two studies assessed the ability of the history and physical to detect hypothyroidism.

The first study evaluated 1154 women (aged 50–72) in a primary care setting using both history and physical and TSH testing. TSH testing found 3 women with overt hypothyroidism not identified by history and physical. History and physical identified 286 women with indications for TSH testing, 2 of whom had mild hypothyroidism and 1 with mild hyperthyroidism.⁵

In the second study, 2000 adults from a primary care population underwent history and physical and TSH testing. The TSH screen identified 19 cases of hypothyroidism not found by history and physical, while the history and physical prompted evaluation of 35 patients without abnormal TSH.⁶

Natural history of subclinical hypothyroidism. Among 1210 primary care patients (700 women) aged >60 years, 73 women with subclinical hypothyroidism were identified and followed for 12 months. Of these, 13 (18%) progressed to overt disease.⁷ Another prospective study of 30 patients (6 women) with subclinical hypothyroidism found 16 (3 women) who progressed to overt disease within 24 months. The remaining patients maintained normal FT4 levels for at least 15 years despite persistently elevated TSH.⁸ A third study followed 2779 adults (1494 women) with all types of thyroid disease for 20 years and found that 55% of women with TSH >6 and a positive antibody test developed overt hypothyroidism. Ninety percent of patients with an initial TSH >10 eventually progressed to overt disease.²

Serum lipid reduction. A retrospective study of 709 women referred to an endocrine clinic for evaluation of abnormal lipoprotein levels identified 34 (4.8%) with undiagnosed hypothyroidism. Thyroid hormone treatment significantly reduced total cholesterol and low-density lipoprotein (LDL) cholesterol in patients with initial TSH >10, but not in those with a TSH <10.⁹

A randomized trial involving 42 women with subclinical hypothyroidism measured lipid levels before and after 6 months of levothyroxine treatment. Levothyroxine reduced total cholesterol and LDL significantly compared with placebo. Additionally, the subclinical hypothyroidism patients had higher baseline lipid levels when compared with 27 euthyroid controls.¹⁰

A meta-analysis combined 13 studies, involving 247 patients with subclinical hypothyroidism, all of whom were given thyroid replacement. All studies reported a decrease in total cholesterol (mean –7.9 mg/dL), and 9 reported a decrease in LDL (mean –10 mg/dL).¹¹ A second meta-analysis with 278 hypothyroid patients given thyroid replacement also found a reduction of total cholesterol (mean –15 mg/dL). LDL effects were not reported.¹² The clinical significance of lipid changes in these circumstances is unknown.

Symptom relief. Four small randomized controlled trials used symptom-rating scales to measure symptom relief with treatment of subclinical hypothyroidism. One study involved patients found by screening; the other 3 did not indicate means of diagnosis. Three studies found no significant improvement.^13-15 The most recent, involving 33 unblinded patients, found that those taking thyroid replacement had lower symptom scores (number needed to treat [NNT]=3.5).¹⁶

Cardiac manifestations. Subclinical hypothyroidism may be associated with ventricular dysfunction, myocardial infarction, and atherosclerosis.^18-20 A randomized controlled trial of 20 people with subclinical hypothyroidism found significantly improved left ventricular function assessed by echocardiography after 6 months of treatment with levothyroxine vs placebo.¹⁸ Whether treatment prevents myocardial infarction and atherosclerosis is unknown.^19,20 A cohort study, involving 2779 adults studied aged >20 years, did not find an association between subclinical hypothyroidism and ischemic heart disease.²

Risks of replacement. A meta-analysis of 41 controlled, cross-sectional studies involving 1250 women treated with thyroid replacement for all causes (ie, not specifically subclinical hypothyroidism) found that replacement therapy (mean duration of treatment, 7 to 9 years) was associated with bone loss in premenopausal women, but not in postmenopausal women.¹⁷

A randomized trial of 37 patients over 55 with subclinical hypothyroidism (28 of whom were women), found that thyroid hormone reduced bone mineral density, as assessed by dual-energy x-ray absorptiometry (DEXA) scans over a 10-month period.¹⁴ In several trials, patients withdrew due to adverse effects. Two of 37 patients receiving L-thyroxine in 1 study withdrew because of new atrial fibrillation and worsened angina, and 2 of 20 patients in another study withdrew because of nervousness and palpitations.^13,14

Recommendations from others

The US Preventive Services Task Force concluded the evidence is insufficient to recommend for or against routine screening for thyroid disease in adults. The yield of screening is greater in high-risk groups such as postpartum women, people with Down syndrome, and the elderly; however, there is poor evidence that screening these groups leads to clinically important benefits.²¹

The American Thyroid Association recommends screening men and women beginning at age 35 and every 5 years thereafter.²² The American Academy of Family Physicians recommends screening for men and women over age 60.²³ The American College of Physicians states screening may be indicated in women over age 50.²⁴

Evidence Summary

Screening for hypothyroidism is more likely to detect the elevated TSH and normal free thyroxine level (FT4) of subclinical hypothyroidism than it is to detect overt hypothyroidism with a high TSH and a low FT4. We reviewed the accuracy of detection, natural history, and benefits and harms of treating subclinical hypothyroidism.

Detection. Subclinical hypothyroidism is found in 7% to 26% of women (with increasing prevalence as women reach age 60 and 70 years); overt hypothyroidism occurs in approximately 5%.^1-4 Two studies assessed the ability of the history and physical to detect hypothyroidism.

The first study evaluated 1154 women (aged 50–72) in a primary care setting using both history and physical and TSH testing. TSH testing found 3 women with overt hypothyroidism not identified by history and physical. History and physical identified 286 women with indications for TSH testing, 2 of whom had mild hypothyroidism and 1 with mild hyperthyroidism.⁵

In the second study, 2000 adults from a primary care population underwent history and physical and TSH testing. The TSH screen identified 19 cases of hypothyroidism not found by history and physical, while the history and physical prompted evaluation of 35 patients without abnormal TSH.⁶

Natural history of subclinical hypothyroidism. Among 1210 primary care patients (700 women) aged >60 years, 73 women with subclinical hypothyroidism were identified and followed for 12 months. Of these, 13 (18%) progressed to overt disease.⁷ Another prospective study of 30 patients (6 women) with subclinical hypothyroidism found 16 (3 women) who progressed to overt disease within 24 months. The remaining patients maintained normal FT4 levels for at least 15 years despite persistently elevated TSH.⁸ A third study followed 2779 adults (1494 women) with all types of thyroid disease for 20 years and found that 55% of women with TSH >6 and a positive antibody test developed overt hypothyroidism. Ninety percent of patients with an initial TSH >10 eventually progressed to overt disease.²

Serum lipid reduction. A retrospective study of 709 women referred to an endocrine clinic for evaluation of abnormal lipoprotein levels identified 34 (4.8%) with undiagnosed hypothyroidism. Thyroid hormone treatment significantly reduced total cholesterol and low-density lipoprotein (LDL) cholesterol in patients with initial TSH >10, but not in those with a TSH <10.⁹

A randomized trial involving 42 women with subclinical hypothyroidism measured lipid levels before and after 6 months of levothyroxine treatment. Levothyroxine reduced total cholesterol and LDL significantly compared with placebo. Additionally, the subclinical hypothyroidism patients had higher baseline lipid levels when compared with 27 euthyroid controls.¹⁰

A meta-analysis combined 13 studies, involving 247 patients with subclinical hypothyroidism, all of whom were given thyroid replacement. All studies reported a decrease in total cholesterol (mean –7.9 mg/dL), and 9 reported a decrease in LDL (mean –10 mg/dL).¹¹ A second meta-analysis with 278 hypothyroid patients given thyroid replacement also found a reduction of total cholesterol (mean –15 mg/dL). LDL effects were not reported.¹² The clinical significance of lipid changes in these circumstances is unknown.

Symptom relief. Four small randomized controlled trials used symptom-rating scales to measure symptom relief with treatment of subclinical hypothyroidism. One study involved patients found by screening; the other 3 did not indicate means of diagnosis. Three studies found no significant improvement.^13-15 The most recent, involving 33 unblinded patients, found that those taking thyroid replacement had lower symptom scores (number needed to treat [NNT]=3.5).¹⁶

Cardiac manifestations. Subclinical hypothyroidism may be associated with ventricular dysfunction, myocardial infarction, and atherosclerosis.^18-20 A randomized controlled trial of 20 people with subclinical hypothyroidism found significantly improved left ventricular function assessed by echocardiography after 6 months of treatment with levothyroxine vs placebo.¹⁸ Whether treatment prevents myocardial infarction and atherosclerosis is unknown.^19,20 A cohort study, involving 2779 adults studied aged >20 years, did not find an association between subclinical hypothyroidism and ischemic heart disease.²

Risks of replacement. A meta-analysis of 41 controlled, cross-sectional studies involving 1250 women treated with thyroid replacement for all causes (ie, not specifically subclinical hypothyroidism) found that replacement therapy (mean duration of treatment, 7 to 9 years) was associated with bone loss in premenopausal women, but not in postmenopausal women.¹⁷

A randomized trial of 37 patients over 55 with subclinical hypothyroidism (28 of whom were women), found that thyroid hormone reduced bone mineral density, as assessed by dual-energy x-ray absorptiometry (DEXA) scans over a 10-month period.¹⁴ In several trials, patients withdrew due to adverse effects. Two of 37 patients receiving L-thyroxine in 1 study withdrew because of new atrial fibrillation and worsened angina, and 2 of 20 patients in another study withdrew because of nervousness and palpitations.^13,14

Recommendations from others

The US Preventive Services Task Force concluded the evidence is insufficient to recommend for or against routine screening for thyroid disease in adults. The yield of screening is greater in high-risk groups such as postpartum women, people with Down syndrome, and the elderly; however, there is poor evidence that screening these groups leads to clinically important benefits.²¹

The American Thyroid Association recommends screening men and women beginning at age 35 and every 5 years thereafter.²² The American Academy of Family Physicians recommends screening for men and women over age 60.²³ The American College of Physicians states screening may be indicated in women over age 50.²⁴

Evidence Summary

Screening for hypothyroidism is more likely to detect the elevated TSH and normal free thyroxine level (FT4) of subclinical hypothyroidism than it is to detect overt hypothyroidism with a high TSH and a low FT4. We reviewed the accuracy of detection, natural history, and benefits and harms of treating subclinical hypothyroidism.

Detection. Subclinical hypothyroidism is found in 7% to 26% of women (with increasing prevalence as women reach age 60 and 70 years); overt hypothyroidism occurs in approximately 5%.^1-4 Two studies assessed the ability of the history and physical to detect hypothyroidism.

The first study evaluated 1154 women (aged 50–72) in a primary care setting using both history and physical and TSH testing. TSH testing found 3 women with overt hypothyroidism not identified by history and physical. History and physical identified 286 women with indications for TSH testing, 2 of whom had mild hypothyroidism and 1 with mild hyperthyroidism.⁵

In the second study, 2000 adults from a primary care population underwent history and physical and TSH testing. The TSH screen identified 19 cases of hypothyroidism not found by history and physical, while the history and physical prompted evaluation of 35 patients without abnormal TSH.⁶

Natural history of subclinical hypothyroidism. Among 1210 primary care patients (700 women) aged >60 years, 73 women with subclinical hypothyroidism were identified and followed for 12 months. Of these, 13 (18%) progressed to overt disease.⁷ Another prospective study of 30 patients (6 women) with subclinical hypothyroidism found 16 (3 women) who progressed to overt disease within 24 months. The remaining patients maintained normal FT4 levels for at least 15 years despite persistently elevated TSH.⁸ A third study followed 2779 adults (1494 women) with all types of thyroid disease for 20 years and found that 55% of women with TSH >6 and a positive antibody test developed overt hypothyroidism. Ninety percent of patients with an initial TSH >10 eventually progressed to overt disease.²

Serum lipid reduction. A retrospective study of 709 women referred to an endocrine clinic for evaluation of abnormal lipoprotein levels identified 34 (4.8%) with undiagnosed hypothyroidism. Thyroid hormone treatment significantly reduced total cholesterol and low-density lipoprotein (LDL) cholesterol in patients with initial TSH >10, but not in those with a TSH <10.⁹

A randomized trial involving 42 women with subclinical hypothyroidism measured lipid levels before and after 6 months of levothyroxine treatment. Levothyroxine reduced total cholesterol and LDL significantly compared with placebo. Additionally, the subclinical hypothyroidism patients had higher baseline lipid levels when compared with 27 euthyroid controls.¹⁰

A meta-analysis combined 13 studies, involving 247 patients with subclinical hypothyroidism, all of whom were given thyroid replacement. All studies reported a decrease in total cholesterol (mean –7.9 mg/dL), and 9 reported a decrease in LDL (mean –10 mg/dL).¹¹ A second meta-analysis with 278 hypothyroid patients given thyroid replacement also found a reduction of total cholesterol (mean –15 mg/dL). LDL effects were not reported.¹² The clinical significance of lipid changes in these circumstances is unknown.

Symptom relief. Four small randomized controlled trials used symptom-rating scales to measure symptom relief with treatment of subclinical hypothyroidism. One study involved patients found by screening; the other 3 did not indicate means of diagnosis. Three studies found no significant improvement.^13-15 The most recent, involving 33 unblinded patients, found that those taking thyroid replacement had lower symptom scores (number needed to treat [NNT]=3.5).¹⁶

Cardiac manifestations. Subclinical hypothyroidism may be associated with ventricular dysfunction, myocardial infarction, and atherosclerosis.^18-20 A randomized controlled trial of 20 people with subclinical hypothyroidism found significantly improved left ventricular function assessed by echocardiography after 6 months of treatment with levothyroxine vs placebo.¹⁸ Whether treatment prevents myocardial infarction and atherosclerosis is unknown.^19,20 A cohort study, involving 2779 adults studied aged >20 years, did not find an association between subclinical hypothyroidism and ischemic heart disease.²

Risks of replacement. A meta-analysis of 41 controlled, cross-sectional studies involving 1250 women treated with thyroid replacement for all causes (ie, not specifically subclinical hypothyroidism) found that replacement therapy (mean duration of treatment, 7 to 9 years) was associated with bone loss in premenopausal women, but not in postmenopausal women.¹⁷

A randomized trial of 37 patients over 55 with subclinical hypothyroidism (28 of whom were women), found that thyroid hormone reduced bone mineral density, as assessed by dual-energy x-ray absorptiometry (DEXA) scans over a 10-month period.¹⁴ In several trials, patients withdrew due to adverse effects. Two of 37 patients receiving L-thyroxine in 1 study withdrew because of new atrial fibrillation and worsened angina, and 2 of 20 patients in another study withdrew because of nervousness and palpitations.^13,14

Recommendations from others

The US Preventive Services Task Force concluded the evidence is insufficient to recommend for or against routine screening for thyroid disease in adults. The yield of screening is greater in high-risk groups such as postpartum women, people with Down syndrome, and the elderly; however, there is poor evidence that screening these groups leads to clinically important benefits.²¹

The American Thyroid Association recommends screening men and women beginning at age 35 and every 5 years thereafter.²² The American Academy of Family Physicians recommends screening for men and women over age 60.²³ The American College of Physicians states screening may be indicated in women over age 50.²⁴

Evidence summary

No trials have tested or compared specific, comprehensive strategies for recurrent bacterial vaginosis. Given that bacterial vaginosis can also be asymptomatic, recurrence often cannot be differentiated from treatment failure. Accordingly, recurrent bacterial vaginosis may be prevented by using the most effective therapy for the initial episode. A 2002 meta-analysis by the Centers for Disease Control and Prevention’s (CDC) bacterial vaginosis working group reviewed the indications for therapy and best treatments for bacterial vaginosis.¹ The group found 25 trials evaluating oral metronidazole therapy involving 2742 women. Although cure rates using either 500 mg twice daily for 5 to 7 days or 2 g as a single dose were similar at 2 weeks post follow-up (85%; range 67%–98%), the single-dose regimen led to higher relapse rates 1 month after treatment (35%–50% vs 20%–33%).

Six trials enrolling 946 women assessed the efficacy of various topical vaginal treatments. Metronidazole gel, clindamycin cream, and clindamycin ovules had a wide range of initial cure rates (50%–95%), but all had higher relapse rates at 4 weeks than did oral metronidazole for 1 week (34%–49%).¹ A more complete discussion of the effectiveness of antibiotics for bacterial vaginosis can be found in a recent Clinical Inquiry.²

The CDC reviewers identified causal relationships between bacterial vaginosis and plasmacell endometritis, postpartum fever, and posthysterectomy vaginal-cuff cellulitis. They therefore concluded it is reasonable to try to prevent post-procedure infections by treating women who have asymptomatic bacterial vaginosis before hysterectomy or pregnancy termination. Although bacterial vaginosis has been associated with preterm labor, trials evaluating treatment of bacterial vaginosis to prevent preterm delivery are conflicting. A Cochrane review of bacterial vaginosis and preterm labor suggests treating women at high risk for preterm birth may reduce the risk of low birthweight and preterm prelabor rupture of membranes.³

Patients frequently try to self-diagnose vaginal complaints and ask for treatments and retreatments by phone. However, a prospective study of 253 women who underwent a structured telephone interview and subsequent physical exam found a poor correlation between telephone diagnosis and final clinical diagnosis (kappa coefficient of 0.12—very poor agreement).⁴ Accordingly, clinical and laboratory evaluation of vaginal discharge and especially recurrent symptoms is essential for diagnostic accuracy and treatment for bacterial vaginosis (Table 1).

For recurrent symptomatic bacterial vaginosis, 1 option is suppressive therapy with metronidazole gel 0.75%. After initial daily retreatment for 10 days, this can be used twice weekly for 4 to 6 months to decrease symptoms. This strategy is based on expert opinion but is currently undergoing clinical trial.

One small crossover randomized controlled trial of 46 women with bacterial vaginosis studied the consumption of live L acidophilus cultures.⁵ Only 20 of the women had recurrent bacterial vaginosis. The groups were randomized to eat yogurt with and without live L acidophilus cultures. While the results were encouraging (50% reduction in episodes of bacterial vaginosis and increase in detectable vaginal Lactobacillus), only 7 women actually completed the study protocol.

Douching is the best-studied risk factor for bacterial vaginosis. A recent multicenter cross-sectional study of 1200 women assessed douching practices and found that recent douching increased the risk of bacterial vaginosis twofold (odds ratio=2.1; 95% confidence interval, 1.3–3.1).⁶ Evidence for the other risk factors listed in Table 2 is based on smaller studies or expert opinion.^7,8

For women who continue to have recurrent or unresolved vaginal symptoms not explained by candidiasis or sexually transmitted infections such as trichomoniasis, consider less common causes such as atrophic vaginitis, chemical/irritant vaginitis, allergic vaginitis, Behçets disease, desquamative interstitial vaginitis, or erosive lichen planus vaginitis.⁹

TABLE 1
Amsel criteria for diagnosis of bacterial vaginosis

TABLE 2
Risk factors for bacterial vaginosis

Recommendations from others

No organizations have developed guidelines for treating recurrent bacterial vaginosis. In 2002, the Association for Genitourinary Medicine and the Medical Society for the Study of Venereal Diseases released national guidelines on the management of bacterial vaginosis,¹⁰ which generally agrees with the previously described CDC recommendations.

Evidence summary

No trials have tested or compared specific, comprehensive strategies for recurrent bacterial vaginosis. Given that bacterial vaginosis can also be asymptomatic, recurrence often cannot be differentiated from treatment failure. Accordingly, recurrent bacterial vaginosis may be prevented by using the most effective therapy for the initial episode. A 2002 meta-analysis by the Centers for Disease Control and Prevention’s (CDC) bacterial vaginosis working group reviewed the indications for therapy and best treatments for bacterial vaginosis.¹ The group found 25 trials evaluating oral metronidazole therapy involving 2742 women. Although cure rates using either 500 mg twice daily for 5 to 7 days or 2 g as a single dose were similar at 2 weeks post follow-up (85%; range 67%–98%), the single-dose regimen led to higher relapse rates 1 month after treatment (35%–50% vs 20%–33%).

Six trials enrolling 946 women assessed the efficacy of various topical vaginal treatments. Metronidazole gel, clindamycin cream, and clindamycin ovules had a wide range of initial cure rates (50%–95%), but all had higher relapse rates at 4 weeks than did oral metronidazole for 1 week (34%–49%).¹ A more complete discussion of the effectiveness of antibiotics for bacterial vaginosis can be found in a recent Clinical Inquiry.²

The CDC reviewers identified causal relationships between bacterial vaginosis and plasmacell endometritis, postpartum fever, and posthysterectomy vaginal-cuff cellulitis. They therefore concluded it is reasonable to try to prevent post-procedure infections by treating women who have asymptomatic bacterial vaginosis before hysterectomy or pregnancy termination. Although bacterial vaginosis has been associated with preterm labor, trials evaluating treatment of bacterial vaginosis to prevent preterm delivery are conflicting. A Cochrane review of bacterial vaginosis and preterm labor suggests treating women at high risk for preterm birth may reduce the risk of low birthweight and preterm prelabor rupture of membranes.³

Patients frequently try to self-diagnose vaginal complaints and ask for treatments and retreatments by phone. However, a prospective study of 253 women who underwent a structured telephone interview and subsequent physical exam found a poor correlation between telephone diagnosis and final clinical diagnosis (kappa coefficient of 0.12—very poor agreement).⁴ Accordingly, clinical and laboratory evaluation of vaginal discharge and especially recurrent symptoms is essential for diagnostic accuracy and treatment for bacterial vaginosis (Table 1).

For recurrent symptomatic bacterial vaginosis, 1 option is suppressive therapy with metronidazole gel 0.75%. After initial daily retreatment for 10 days, this can be used twice weekly for 4 to 6 months to decrease symptoms. This strategy is based on expert opinion but is currently undergoing clinical trial.

One small crossover randomized controlled trial of 46 women with bacterial vaginosis studied the consumption of live L acidophilus cultures.⁵ Only 20 of the women had recurrent bacterial vaginosis. The groups were randomized to eat yogurt with and without live L acidophilus cultures. While the results were encouraging (50% reduction in episodes of bacterial vaginosis and increase in detectable vaginal Lactobacillus), only 7 women actually completed the study protocol.

Douching is the best-studied risk factor for bacterial vaginosis. A recent multicenter cross-sectional study of 1200 women assessed douching practices and found that recent douching increased the risk of bacterial vaginosis twofold (odds ratio=2.1; 95% confidence interval, 1.3–3.1).⁶ Evidence for the other risk factors listed in Table 2 is based on smaller studies or expert opinion.^7,8

For women who continue to have recurrent or unresolved vaginal symptoms not explained by candidiasis or sexually transmitted infections such as trichomoniasis, consider less common causes such as atrophic vaginitis, chemical/irritant vaginitis, allergic vaginitis, Behçets disease, desquamative interstitial vaginitis, or erosive lichen planus vaginitis.⁹

TABLE 1
Amsel criteria for diagnosis of bacterial vaginosis

TABLE 2
Risk factors for bacterial vaginosis

Recommendations from others

No organizations have developed guidelines for treating recurrent bacterial vaginosis. In 2002, the Association for Genitourinary Medicine and the Medical Society for the Study of Venereal Diseases released national guidelines on the management of bacterial vaginosis,¹⁰ which generally agrees with the previously described CDC recommendations.

Evidence summary

No trials have tested or compared specific, comprehensive strategies for recurrent bacterial vaginosis. Given that bacterial vaginosis can also be asymptomatic, recurrence often cannot be differentiated from treatment failure. Accordingly, recurrent bacterial vaginosis may be prevented by using the most effective therapy for the initial episode. A 2002 meta-analysis by the Centers for Disease Control and Prevention’s (CDC) bacterial vaginosis working group reviewed the indications for therapy and best treatments for bacterial vaginosis.¹ The group found 25 trials evaluating oral metronidazole therapy involving 2742 women. Although cure rates using either 500 mg twice daily for 5 to 7 days or 2 g as a single dose were similar at 2 weeks post follow-up (85%; range 67%–98%), the single-dose regimen led to higher relapse rates 1 month after treatment (35%–50% vs 20%–33%).

Six trials enrolling 946 women assessed the efficacy of various topical vaginal treatments. Metronidazole gel, clindamycin cream, and clindamycin ovules had a wide range of initial cure rates (50%–95%), but all had higher relapse rates at 4 weeks than did oral metronidazole for 1 week (34%–49%).¹ A more complete discussion of the effectiveness of antibiotics for bacterial vaginosis can be found in a recent Clinical Inquiry.²

The CDC reviewers identified causal relationships between bacterial vaginosis and plasmacell endometritis, postpartum fever, and posthysterectomy vaginal-cuff cellulitis. They therefore concluded it is reasonable to try to prevent post-procedure infections by treating women who have asymptomatic bacterial vaginosis before hysterectomy or pregnancy termination. Although bacterial vaginosis has been associated with preterm labor, trials evaluating treatment of bacterial vaginosis to prevent preterm delivery are conflicting. A Cochrane review of bacterial vaginosis and preterm labor suggests treating women at high risk for preterm birth may reduce the risk of low birthweight and preterm prelabor rupture of membranes.³

Patients frequently try to self-diagnose vaginal complaints and ask for treatments and retreatments by phone. However, a prospective study of 253 women who underwent a structured telephone interview and subsequent physical exam found a poor correlation between telephone diagnosis and final clinical diagnosis (kappa coefficient of 0.12—very poor agreement).⁴ Accordingly, clinical and laboratory evaluation of vaginal discharge and especially recurrent symptoms is essential for diagnostic accuracy and treatment for bacterial vaginosis (Table 1).

For recurrent symptomatic bacterial vaginosis, 1 option is suppressive therapy with metronidazole gel 0.75%. After initial daily retreatment for 10 days, this can be used twice weekly for 4 to 6 months to decrease symptoms. This strategy is based on expert opinion but is currently undergoing clinical trial.

One small crossover randomized controlled trial of 46 women with bacterial vaginosis studied the consumption of live L acidophilus cultures.⁵ Only 20 of the women had recurrent bacterial vaginosis. The groups were randomized to eat yogurt with and without live L acidophilus cultures. While the results were encouraging (50% reduction in episodes of bacterial vaginosis and increase in detectable vaginal Lactobacillus), only 7 women actually completed the study protocol.

Douching is the best-studied risk factor for bacterial vaginosis. A recent multicenter cross-sectional study of 1200 women assessed douching practices and found that recent douching increased the risk of bacterial vaginosis twofold (odds ratio=2.1; 95% confidence interval, 1.3–3.1).⁶ Evidence for the other risk factors listed in Table 2 is based on smaller studies or expert opinion.^7,8

For women who continue to have recurrent or unresolved vaginal symptoms not explained by candidiasis or sexually transmitted infections such as trichomoniasis, consider less common causes such as atrophic vaginitis, chemical/irritant vaginitis, allergic vaginitis, Behçets disease, desquamative interstitial vaginitis, or erosive lichen planus vaginitis.⁹

TABLE 1
Amsel criteria for diagnosis of bacterial vaginosis

TABLE 2
Risk factors for bacterial vaginosis

Recommendations from others

No organizations have developed guidelines for treating recurrent bacterial vaginosis. In 2002, the Association for Genitourinary Medicine and the Medical Society for the Study of Venereal Diseases released national guidelines on the management of bacterial vaginosis,¹⁰ which generally agrees with the previously described CDC recommendations.

Evidence summary

A systematic review of cardiac rehabilitation programs evaluated 14 randomized controlled trials with exercise-based interventions.¹ An updated review added 5 more for a total of 2984 patients with coronary heart disease.² Patients with coronary heart disease comprised those with prior MI, prior coronary artery bypass graft surgery, or PTCA, and those with angina pectoris and angiographically confirmed coronary heart disease.

Exercise-based cardiac rehabilitation significantly reduced all-cause mortality (relative risk [RR]=0.76; 95% confidence interval [CI], 0.59–0.98) compared with usual care (NNT=66; 95% CI, 35–273). Cardiac mortality also decreased significantly with exercise (RR=0.73; 95% CI, 0.56–0.96) compared with usual care (NNT=49; 95% CI, 26–120).

Six studies showed particularly significant improvement in total cardiac mortality.^3-8 Exercise was variably defined. Training sessions lasted 30 minutes and occurred on 2 to 5 days per week. Intensity was typically 75% to 85% of a maximum work capacity determined on an exercise test before initiating the training sessions. The type of exercise ranged from cycling alone to circuit training with 6 stationary devices. Patients were trained with supervision 1 to 36 months and followed for a mean of 24 months (range, 6–60 months).

A trend was observed toward decreased recurrence of nonfatal MI with exercise-based cardiac rehabilitation, which did not reach significance (RR=0.78; 95% CI, 0.59–1.03). An inadequate number of subjects is the most likely reason; however, other possibilities include an increase in the frequency of nonfatal MI after rehabilitation, or an increased rate of survival after MI for patients undergoing exercise-based rehabilitation.

The studies included in these reviews had several limitations. The population appears skewed in age (mean=54 years, with patients aged >65 years excluded from most studies) and gender (4.9% female); ethnicity was rarely reported. The adequacy of randomization was poor or unclear in 71% of studies, and only 4 trials reported blind assessment of outcomes. Finally, in 34% of studies the loss of participants to follow-up was more than 20%.

A well-done study randomized 101 male patients (age <70 years) with stable angina to either a daily exercise program or standard PTCA with stenting.⁹ After 12 months, event-free survival was significantly greater among patients randomized to exercise than in those randomized to PTCA with stenting (88% vs 70%; P=.023; NNT=5.5). Cardiovascular events were defined as percutaneous interventions, hospitalizations, acute MI, cerebrovascular accidents, coronary artery bypass graft operation, and death.

Recommendation from others

The American Heart Association (AHA) supports aggressive risk factor management for patients with coronary heart disease, and recommends a minimum of 30 minutes of exercise 3 to 4 days per week as well as an increase in daily lifestyle activities.¹⁰ The American College of Cardiology endorses the position of the AHA.

Evidence summary

A systematic review of cardiac rehabilitation programs evaluated 14 randomized controlled trials with exercise-based interventions.¹ An updated review added 5 more for a total of 2984 patients with coronary heart disease.² Patients with coronary heart disease comprised those with prior MI, prior coronary artery bypass graft surgery, or PTCA, and those with angina pectoris and angiographically confirmed coronary heart disease.

Exercise-based cardiac rehabilitation significantly reduced all-cause mortality (relative risk [RR]=0.76; 95% confidence interval [CI], 0.59–0.98) compared with usual care (NNT=66; 95% CI, 35–273). Cardiac mortality also decreased significantly with exercise (RR=0.73; 95% CI, 0.56–0.96) compared with usual care (NNT=49; 95% CI, 26–120).

Six studies showed particularly significant improvement in total cardiac mortality.^3-8 Exercise was variably defined. Training sessions lasted 30 minutes and occurred on 2 to 5 days per week. Intensity was typically 75% to 85% of a maximum work capacity determined on an exercise test before initiating the training sessions. The type of exercise ranged from cycling alone to circuit training with 6 stationary devices. Patients were trained with supervision 1 to 36 months and followed for a mean of 24 months (range, 6–60 months).

A trend was observed toward decreased recurrence of nonfatal MI with exercise-based cardiac rehabilitation, which did not reach significance (RR=0.78; 95% CI, 0.59–1.03). An inadequate number of subjects is the most likely reason; however, other possibilities include an increase in the frequency of nonfatal MI after rehabilitation, or an increased rate of survival after MI for patients undergoing exercise-based rehabilitation.

The studies included in these reviews had several limitations. The population appears skewed in age (mean=54 years, with patients aged >65 years excluded from most studies) and gender (4.9% female); ethnicity was rarely reported. The adequacy of randomization was poor or unclear in 71% of studies, and only 4 trials reported blind assessment of outcomes. Finally, in 34% of studies the loss of participants to follow-up was more than 20%.

A well-done study randomized 101 male patients (age <70 years) with stable angina to either a daily exercise program or standard PTCA with stenting.⁹ After 12 months, event-free survival was significantly greater among patients randomized to exercise than in those randomized to PTCA with stenting (88% vs 70%; P=.023; NNT=5.5). Cardiovascular events were defined as percutaneous interventions, hospitalizations, acute MI, cerebrovascular accidents, coronary artery bypass graft operation, and death.

Recommendation from others

The American Heart Association (AHA) supports aggressive risk factor management for patients with coronary heart disease, and recommends a minimum of 30 minutes of exercise 3 to 4 days per week as well as an increase in daily lifestyle activities.¹⁰ The American College of Cardiology endorses the position of the AHA.

Evidence summary

A systematic review of cardiac rehabilitation programs evaluated 14 randomized controlled trials with exercise-based interventions.¹ An updated review added 5 more for a total of 2984 patients with coronary heart disease.² Patients with coronary heart disease comprised those with prior MI, prior coronary artery bypass graft surgery, or PTCA, and those with angina pectoris and angiographically confirmed coronary heart disease.

Exercise-based cardiac rehabilitation significantly reduced all-cause mortality (relative risk [RR]=0.76; 95% confidence interval [CI], 0.59–0.98) compared with usual care (NNT=66; 95% CI, 35–273). Cardiac mortality also decreased significantly with exercise (RR=0.73; 95% CI, 0.56–0.96) compared with usual care (NNT=49; 95% CI, 26–120).

Six studies showed particularly significant improvement in total cardiac mortality.^3-8 Exercise was variably defined. Training sessions lasted 30 minutes and occurred on 2 to 5 days per week. Intensity was typically 75% to 85% of a maximum work capacity determined on an exercise test before initiating the training sessions. The type of exercise ranged from cycling alone to circuit training with 6 stationary devices. Patients were trained with supervision 1 to 36 months and followed for a mean of 24 months (range, 6–60 months).

A trend was observed toward decreased recurrence of nonfatal MI with exercise-based cardiac rehabilitation, which did not reach significance (RR=0.78; 95% CI, 0.59–1.03). An inadequate number of subjects is the most likely reason; however, other possibilities include an increase in the frequency of nonfatal MI after rehabilitation, or an increased rate of survival after MI for patients undergoing exercise-based rehabilitation.

The studies included in these reviews had several limitations. The population appears skewed in age (mean=54 years, with patients aged >65 years excluded from most studies) and gender (4.9% female); ethnicity was rarely reported. The adequacy of randomization was poor or unclear in 71% of studies, and only 4 trials reported blind assessment of outcomes. Finally, in 34% of studies the loss of participants to follow-up was more than 20%.

A well-done study randomized 101 male patients (age <70 years) with stable angina to either a daily exercise program or standard PTCA with stenting.⁹ After 12 months, event-free survival was significantly greater among patients randomized to exercise than in those randomized to PTCA with stenting (88% vs 70%; P=.023; NNT=5.5). Cardiovascular events were defined as percutaneous interventions, hospitalizations, acute MI, cerebrovascular accidents, coronary artery bypass graft operation, and death.

Recommendation from others

The American Heart Association (AHA) supports aggressive risk factor management for patients with coronary heart disease, and recommends a minimum of 30 minutes of exercise 3 to 4 days per week as well as an increase in daily lifestyle activities.¹⁰ The American College of Cardiology endorses the position of the AHA.

Evidence summary

Superficial thrombophlebitis refers to erythema, pain, induration, and other findings of inflammation in superficial veins, usually due to infection or thrombosis. Typically, SVTP is localized problem, but some lower-extremity SVTP is associated with increased risk of DVT and PE, particularly the long saphenous vein. This review will not address thrombosis in the superficial femoral vein, a portion of the deep venous system, which requires full DVT therapy.¹

Since saphenous vein thrombosis above the knee is associated with DVT and PE, 1 systematic review looked at papers comparing anticoagulation (IV heparin followed by 6 weeks to 6 months of warfarin) with surgical ligation of the saphenous vein (either alone or combined with vein stripping or with vein stripping and perforator ligation).¹ The review included primarily case series with widely varying protocols. According to the authors, the data “suggests that medical management with anticoagulants is somewhat superior” to surgery for preventing DVT and PE. However, the fewest extensions of SVTP occurred when vein ligation was combined with stripping of the thrombosed vein and interruption of perforators.

In a more recent trial, patients randomized to subcutaneous heparin at 12,500 units twice daily for a week followed by 10,000 units twice daily had fewer vascular complications of proximal saphenous vein thrombosis than those receiving heparin at 5000 units twice daily (6/30 in the low-dose group and 1/30 in the high-dose group; P<.05; number needed to treat [NNT]=6).² There were no bleeding complications in either group.

One large double-blind randomized controlled trial compared tenoxicam (an NSAID available in Canada, similar to piroxicam), enoxaparin (Lovenox), and placebo for 8 to 12 days in 427 patients with SVTP of the leg measuring 5 cm or more.³ Patients were also treated with compression hose. Patients who required immediate anticoagulation or venous ligation were excluded. Within 3 months, 35% of patients taking placebo developed an extension or recurrence of their SVTP or a DVT, compared with 16% to 17% of treated patients (NNT=6). There was no significant difference in outcome between subcutaneous enoxaparin at fixed (40 mg/d) or adjusted doses (1.5 mg/kg), or 20 mg/d oral tenoxicam. In a small randomized trial (n=40), intramuscular defibrotide provided better symptom resolution than low-dose heparin for patients with uncomplicated SVTP of the leg.⁴

For infusion-related SVTP, a randomized controlled trial of 120 patients found both oral and topical diclofenac effective in reducing symptoms (NNT=3), although oral diclofenac had significantly more gastrointestinal side effects (number needed to harm=3 for dyspepsia).⁵ Two double-blind trials of topical heparin showed it to be superior to placebo in reducing symptoms and speeding healing.^6,7

In the larger study (n=126), 44% of patients treated with 1000 IU/g heparin gel 3 times a day were symptom-free at 1 week, compared with 26% on placebo (NNT=6).⁷ A randomized trial of infusion-related SVTP (n=100) found that 2% nitroglycerin gel eliminated pain in 50 hours vs 72 hours with topical heparin (P<.05).⁸ A smaller, underpowered double-blind trial of topical heparin, piroxicam gel, and placebo (22 to 24 patients in each treatment arm) failed to find efficacy with either therapy.⁹

Recommendation from others

For SVTP of the leg that does not include the proximal saphenous vein, Up To Date recommends compression and oral NSAIDs, noting that NSAIDs are inexpensive, help with symptom control, and appear comparable to low-molecular-weight heparin in limiting complications.¹⁰

Evidence summary

Superficial thrombophlebitis refers to erythema, pain, induration, and other findings of inflammation in superficial veins, usually due to infection or thrombosis. Typically, SVTP is localized problem, but some lower-extremity SVTP is associated with increased risk of DVT and PE, particularly the long saphenous vein. This review will not address thrombosis in the superficial femoral vein, a portion of the deep venous system, which requires full DVT therapy.¹

Since saphenous vein thrombosis above the knee is associated with DVT and PE, 1 systematic review looked at papers comparing anticoagulation (IV heparin followed by 6 weeks to 6 months of warfarin) with surgical ligation of the saphenous vein (either alone or combined with vein stripping or with vein stripping and perforator ligation).¹ The review included primarily case series with widely varying protocols. According to the authors, the data “suggests that medical management with anticoagulants is somewhat superior” to surgery for preventing DVT and PE. However, the fewest extensions of SVTP occurred when vein ligation was combined with stripping of the thrombosed vein and interruption of perforators.

In a more recent trial, patients randomized to subcutaneous heparin at 12,500 units twice daily for a week followed by 10,000 units twice daily had fewer vascular complications of proximal saphenous vein thrombosis than those receiving heparin at 5000 units twice daily (6/30 in the low-dose group and 1/30 in the high-dose group; P<.05; number needed to treat [NNT]=6).² There were no bleeding complications in either group.

One large double-blind randomized controlled trial compared tenoxicam (an NSAID available in Canada, similar to piroxicam), enoxaparin (Lovenox), and placebo for 8 to 12 days in 427 patients with SVTP of the leg measuring 5 cm or more.³ Patients were also treated with compression hose. Patients who required immediate anticoagulation or venous ligation were excluded. Within 3 months, 35% of patients taking placebo developed an extension or recurrence of their SVTP or a DVT, compared with 16% to 17% of treated patients (NNT=6). There was no significant difference in outcome between subcutaneous enoxaparin at fixed (40 mg/d) or adjusted doses (1.5 mg/kg), or 20 mg/d oral tenoxicam. In a small randomized trial (n=40), intramuscular defibrotide provided better symptom resolution than low-dose heparin for patients with uncomplicated SVTP of the leg.⁴

For infusion-related SVTP, a randomized controlled trial of 120 patients found both oral and topical diclofenac effective in reducing symptoms (NNT=3), although oral diclofenac had significantly more gastrointestinal side effects (number needed to harm=3 for dyspepsia).⁵ Two double-blind trials of topical heparin showed it to be superior to placebo in reducing symptoms and speeding healing.^6,7

In the larger study (n=126), 44% of patients treated with 1000 IU/g heparin gel 3 times a day were symptom-free at 1 week, compared with 26% on placebo (NNT=6).⁷ A randomized trial of infusion-related SVTP (n=100) found that 2% nitroglycerin gel eliminated pain in 50 hours vs 72 hours with topical heparin (P<.05).⁸ A smaller, underpowered double-blind trial of topical heparin, piroxicam gel, and placebo (22 to 24 patients in each treatment arm) failed to find efficacy with either therapy.⁹

Recommendation from others

For SVTP of the leg that does not include the proximal saphenous vein, Up To Date recommends compression and oral NSAIDs, noting that NSAIDs are inexpensive, help with symptom control, and appear comparable to low-molecular-weight heparin in limiting complications.¹⁰

Evidence summary

Superficial thrombophlebitis refers to erythema, pain, induration, and other findings of inflammation in superficial veins, usually due to infection or thrombosis. Typically, SVTP is localized problem, but some lower-extremity SVTP is associated with increased risk of DVT and PE, particularly the long saphenous vein. This review will not address thrombosis in the superficial femoral vein, a portion of the deep venous system, which requires full DVT therapy.¹

Since saphenous vein thrombosis above the knee is associated with DVT and PE, 1 systematic review looked at papers comparing anticoagulation (IV heparin followed by 6 weeks to 6 months of warfarin) with surgical ligation of the saphenous vein (either alone or combined with vein stripping or with vein stripping and perforator ligation).¹ The review included primarily case series with widely varying protocols. According to the authors, the data “suggests that medical management with anticoagulants is somewhat superior” to surgery for preventing DVT and PE. However, the fewest extensions of SVTP occurred when vein ligation was combined with stripping of the thrombosed vein and interruption of perforators.

In a more recent trial, patients randomized to subcutaneous heparin at 12,500 units twice daily for a week followed by 10,000 units twice daily had fewer vascular complications of proximal saphenous vein thrombosis than those receiving heparin at 5000 units twice daily (6/30 in the low-dose group and 1/30 in the high-dose group; P<.05; number needed to treat [NNT]=6).² There were no bleeding complications in either group.

One large double-blind randomized controlled trial compared tenoxicam (an NSAID available in Canada, similar to piroxicam), enoxaparin (Lovenox), and placebo for 8 to 12 days in 427 patients with SVTP of the leg measuring 5 cm or more.³ Patients were also treated with compression hose. Patients who required immediate anticoagulation or venous ligation were excluded. Within 3 months, 35% of patients taking placebo developed an extension or recurrence of their SVTP or a DVT, compared with 16% to 17% of treated patients (NNT=6). There was no significant difference in outcome between subcutaneous enoxaparin at fixed (40 mg/d) or adjusted doses (1.5 mg/kg), or 20 mg/d oral tenoxicam. In a small randomized trial (n=40), intramuscular defibrotide provided better symptom resolution than low-dose heparin for patients with uncomplicated SVTP of the leg.⁴

For infusion-related SVTP, a randomized controlled trial of 120 patients found both oral and topical diclofenac effective in reducing symptoms (NNT=3), although oral diclofenac had significantly more gastrointestinal side effects (number needed to harm=3 for dyspepsia).⁵ Two double-blind trials of topical heparin showed it to be superior to placebo in reducing symptoms and speeding healing.^6,7

In the larger study (n=126), 44% of patients treated with 1000 IU/g heparin gel 3 times a day were symptom-free at 1 week, compared with 26% on placebo (NNT=6).⁷ A randomized trial of infusion-related SVTP (n=100) found that 2% nitroglycerin gel eliminated pain in 50 hours vs 72 hours with topical heparin (P<.05).⁸ A smaller, underpowered double-blind trial of topical heparin, piroxicam gel, and placebo (22 to 24 patients in each treatment arm) failed to find efficacy with either therapy.⁹

Recommendation from others

For SVTP of the leg that does not include the proximal saphenous vein, Up To Date recommends compression and oral NSAIDs, noting that NSAIDs are inexpensive, help with symptom control, and appear comparable to low-molecular-weight heparin in limiting complications.¹⁰

Evidence summary

Five recent RCTs have demonstrated similar mortality and cardiovascular morbidity in atrial fibrillation patients treated with either a rate-control or rhythm-control strategy.^1-5

The AFFIRM trial, the largest (n=4060), was a nonblinded, randomized, multicenter study with an average follow-up of 3.5 years.¹ The patients were aged 65 years or older and had at least 1 other risk factor for stroke. The rhythm-control group was given an antiarrhythmic medication chosen by the treating physician, while the rate-control group was given either a beta-blocker, a calcium-channel blocker, digoxin, or a combination of these as needed. Heart-rate goals were a resting pulse under 80 beats per minute, and a pulse after a 6-minute walk under 110 beats per minute. An intention-totreat analysis was followed.

There was no difference between the 2 groups for the composite endpoints of death, disabling stroke, disabling anoxic encephalopathy, major bleeding, or cardiac arrest. A nonsignificant trend was observed for mortality favoring the rate-control group (relative risk [RR]=1.15; 95% confidence interval [CI], 0.99–1.34). Quality-of-life measures were equivalent in the 2 groups at all points in the study.¹

More patients in the rhythm-control group required hospitalization (number needed to harm [NNH]=12.3; P<.001) and had adverse drug effects (P.001 for each of pulmonary events [NNH=18], gastrointestinal events [NNH=17], bradycardia [NNH=56], and prolonged QT [NNH=63]). This trial did not include younger patients without stroke risk factors, or those with paroxysmal atrial fibrillation.¹

The 4 other RCTs also found no greater benefit with a rhythm-control strategy vs rate-control for most patients with atrial fibrillation.^2-5

Two systematic reviews have looked at the efficacy of medications for ventricular rate control in atrial fibrillation.^6,7 The first analyzed 54 trials involving 17 agents and focused on digoxin calcium-channel blockers and beta-blockers. The second systematic review evaluated 45 trials with similar agents. Both reviews were unable to perform mathematical pooling due to the heterogeneity of the studies. However, both showed strong evidence for superior ventricular rate control at both exercise and rest with verapamil and diltiazem compared with placebo.^6,7

All beta-blockers tested were effective in rate-control during exercise and most (excluding labetalol and celiprolol) were effective at rest.^6,7 Digoxin was ineffective during exercise and less effective than beta-blockers or calcium-channel blockers at rest.^6-8 The combination of digoxin plus a calcium-channel blocker or beta-blocker may have increased benefit compared with either drug alone.⁶ Evidence was insufficient to recommend propafenone, clonidine, or amiodarone for rate control.⁷

In select patients, a rhythm-control approach may be desirable. A meta-analysis of 60 RCTs evaluated 8 drugs for acute cardioversion. Ibutilide, flecainide, dofetilide, propafenone, and amiodarone were found to have the strongest evidence of efficacy.⁶ There was moderate evidence for quinidine and insufficient evidence for disopyramide and sotalol.⁶ Studies of pharmacologic conversion suffer from small ample size, short follow-up, and variable duration of atrial fibrillation.⁶ A review of limited research reveals an 80% to 85% immediate success rate for DC cardioversion, with rare side-effects of ventricular tachycardia, transient AV node dysfunction, and significant skin blistering.⁶

For patients who elect a rhythm-control approach, RCTs demonstrate the need for continued long-term anticoagulation in high-risk patients even if they are maintained in sinus rhythm.^1,4,5 (High-risk patients are defined as those aged >65 years, or those <65 years with 1 or more stroke risk factors: diabetes, hypertension, heart failure, prior transient ischemic attack or stroke or systemic embolism, or echocardiographic evidence of a left atrium >50 mm, a shortening fraction <25%, or an ejection fraction <40%.)

Recommendation from others

The American Academy of Family Practice/American College of Physicians’ clinical guidelines support a rate-control strategy for most patients with atrial fibrillation and recommend atenolol, metoprolol, diltiazem, or verapamil as the first-choice drugs.⁸ Digoxin is recommended as a second-line agent. DC cardioversion and pharmacologic conversion for patients who desire a rhythm-control strategy are described as “appropriate options.”⁸

Evidence summary

Five recent RCTs have demonstrated similar mortality and cardiovascular morbidity in atrial fibrillation patients treated with either a rate-control or rhythm-control strategy.^1-5

The AFFIRM trial, the largest (n=4060), was a nonblinded, randomized, multicenter study with an average follow-up of 3.5 years.¹ The patients were aged 65 years or older and had at least 1 other risk factor for stroke. The rhythm-control group was given an antiarrhythmic medication chosen by the treating physician, while the rate-control group was given either a beta-blocker, a calcium-channel blocker, digoxin, or a combination of these as needed. Heart-rate goals were a resting pulse under 80 beats per minute, and a pulse after a 6-minute walk under 110 beats per minute. An intention-totreat analysis was followed.

There was no difference between the 2 groups for the composite endpoints of death, disabling stroke, disabling anoxic encephalopathy, major bleeding, or cardiac arrest. A nonsignificant trend was observed for mortality favoring the rate-control group (relative risk [RR]=1.15; 95% confidence interval [CI], 0.99–1.34). Quality-of-life measures were equivalent in the 2 groups at all points in the study.¹

More patients in the rhythm-control group required hospitalization (number needed to harm [NNH]=12.3; P<.001) and had adverse drug effects (P.001 for each of pulmonary events [NNH=18], gastrointestinal events [NNH=17], bradycardia [NNH=56], and prolonged QT [NNH=63]). This trial did not include younger patients without stroke risk factors, or those with paroxysmal atrial fibrillation.¹

The 4 other RCTs also found no greater benefit with a rhythm-control strategy vs rate-control for most patients with atrial fibrillation.^2-5

Two systematic reviews have looked at the efficacy of medications for ventricular rate control in atrial fibrillation.^6,7 The first analyzed 54 trials involving 17 agents and focused on digoxin calcium-channel blockers and beta-blockers. The second systematic review evaluated 45 trials with similar agents. Both reviews were unable to perform mathematical pooling due to the heterogeneity of the studies. However, both showed strong evidence for superior ventricular rate control at both exercise and rest with verapamil and diltiazem compared with placebo.^6,7

All beta-blockers tested were effective in rate-control during exercise and most (excluding labetalol and celiprolol) were effective at rest.^6,7 Digoxin was ineffective during exercise and less effective than beta-blockers or calcium-channel blockers at rest.^6-8 The combination of digoxin plus a calcium-channel blocker or beta-blocker may have increased benefit compared with either drug alone.⁶ Evidence was insufficient to recommend propafenone, clonidine, or amiodarone for rate control.⁷

In select patients, a rhythm-control approach may be desirable. A meta-analysis of 60 RCTs evaluated 8 drugs for acute cardioversion. Ibutilide, flecainide, dofetilide, propafenone, and amiodarone were found to have the strongest evidence of efficacy.⁶ There was moderate evidence for quinidine and insufficient evidence for disopyramide and sotalol.⁶ Studies of pharmacologic conversion suffer from small ample size, short follow-up, and variable duration of atrial fibrillation.⁶ A review of limited research reveals an 80% to 85% immediate success rate for DC cardioversion, with rare side-effects of ventricular tachycardia, transient AV node dysfunction, and significant skin blistering.⁶

For patients who elect a rhythm-control approach, RCTs demonstrate the need for continued long-term anticoagulation in high-risk patients even if they are maintained in sinus rhythm.^1,4,5 (High-risk patients are defined as those aged >65 years, or those <65 years with 1 or more stroke risk factors: diabetes, hypertension, heart failure, prior transient ischemic attack or stroke or systemic embolism, or echocardiographic evidence of a left atrium >50 mm, a shortening fraction <25%, or an ejection fraction <40%.)

Recommendation from others

The American Academy of Family Practice/American College of Physicians’ clinical guidelines support a rate-control strategy for most patients with atrial fibrillation and recommend atenolol, metoprolol, diltiazem, or verapamil as the first-choice drugs.⁸ Digoxin is recommended as a second-line agent. DC cardioversion and pharmacologic conversion for patients who desire a rhythm-control strategy are described as “appropriate options.”⁸

Evidence summary

Five recent RCTs have demonstrated similar mortality and cardiovascular morbidity in atrial fibrillation patients treated with either a rate-control or rhythm-control strategy.^1-5

The AFFIRM trial, the largest (n=4060), was a nonblinded, randomized, multicenter study with an average follow-up of 3.5 years.¹ The patients were aged 65 years or older and had at least 1 other risk factor for stroke. The rhythm-control group was given an antiarrhythmic medication chosen by the treating physician, while the rate-control group was given either a beta-blocker, a calcium-channel blocker, digoxin, or a combination of these as needed. Heart-rate goals were a resting pulse under 80 beats per minute, and a pulse after a 6-minute walk under 110 beats per minute. An intention-totreat analysis was followed.

There was no difference between the 2 groups for the composite endpoints of death, disabling stroke, disabling anoxic encephalopathy, major bleeding, or cardiac arrest. A nonsignificant trend was observed for mortality favoring the rate-control group (relative risk [RR]=1.15; 95% confidence interval [CI], 0.99–1.34). Quality-of-life measures were equivalent in the 2 groups at all points in the study.¹

More patients in the rhythm-control group required hospitalization (number needed to harm [NNH]=12.3; P<.001) and had adverse drug effects (P.001 for each of pulmonary events [NNH=18], gastrointestinal events [NNH=17], bradycardia [NNH=56], and prolonged QT [NNH=63]). This trial did not include younger patients without stroke risk factors, or those with paroxysmal atrial fibrillation.¹

The 4 other RCTs also found no greater benefit with a rhythm-control strategy vs rate-control for most patients with atrial fibrillation.^2-5

Two systematic reviews have looked at the efficacy of medications for ventricular rate control in atrial fibrillation.^6,7 The first analyzed 54 trials involving 17 agents and focused on digoxin calcium-channel blockers and beta-blockers. The second systematic review evaluated 45 trials with similar agents. Both reviews were unable to perform mathematical pooling due to the heterogeneity of the studies. However, both showed strong evidence for superior ventricular rate control at both exercise and rest with verapamil and diltiazem compared with placebo.^6,7

All beta-blockers tested were effective in rate-control during exercise and most (excluding labetalol and celiprolol) were effective at rest.^6,7 Digoxin was ineffective during exercise and less effective than beta-blockers or calcium-channel blockers at rest.^6-8 The combination of digoxin plus a calcium-channel blocker or beta-blocker may have increased benefit compared with either drug alone.⁶ Evidence was insufficient to recommend propafenone, clonidine, or amiodarone for rate control.⁷

In select patients, a rhythm-control approach may be desirable. A meta-analysis of 60 RCTs evaluated 8 drugs for acute cardioversion. Ibutilide, flecainide, dofetilide, propafenone, and amiodarone were found to have the strongest evidence of efficacy.⁶ There was moderate evidence for quinidine and insufficient evidence for disopyramide and sotalol.⁶ Studies of pharmacologic conversion suffer from small ample size, short follow-up, and variable duration of atrial fibrillation.⁶ A review of limited research reveals an 80% to 85% immediate success rate for DC cardioversion, with rare side-effects of ventricular tachycardia, transient AV node dysfunction, and significant skin blistering.⁶

For patients who elect a rhythm-control approach, RCTs demonstrate the need for continued long-term anticoagulation in high-risk patients even if they are maintained in sinus rhythm.^1,4,5 (High-risk patients are defined as those aged >65 years, or those <65 years with 1 or more stroke risk factors: diabetes, hypertension, heart failure, prior transient ischemic attack or stroke or systemic embolism, or echocardiographic evidence of a left atrium >50 mm, a shortening fraction <25%, or an ejection fraction <40%.)

Recommendation from others

The American Academy of Family Practice/American College of Physicians’ clinical guidelines support a rate-control strategy for most patients with atrial fibrillation and recommend atenolol, metoprolol, diltiazem, or verapamil as the first-choice drugs.⁸ Digoxin is recommended as a second-line agent. DC cardioversion and pharmacologic conversion for patients who desire a rhythm-control strategy are described as “appropriate options.”⁸

Evidence summary

We found no studies evaluating statins for the primary prevention of stroke. An observational study of 433 patients with ischemic stroke found that patients who were taking statins before hospital admission more often had better outcomes (51%) than those who were not taking statins (38%). However, the groups differed in many respects.¹ Many coronary event prevention and treatment trials using statins include the risk of primary and recurrent ischemic stroke as secondary endpoints for patients with high cardiac risk.

Primary prevention of stroke in vascular disease. The Heart Protection Study followed 20,536 patients in the United Kingdom (aged 40–80 years), 3280 with a history of cerebrovascular disease (defined as nondisabling stroke, transient cerebral ischemic attack, or carotid endarterectomy or angioplasty) and 17,256 with other occlusive arterial disease, coronary artery disease, or diabetes. Patients were randomized to receive either simvastatin 40 mg or placebo for an average of 5 years. The endpoint was major vascular events: myocardial infarction, stroke of any type, and revascularization procedure.

Simvastatin reduced the combined risk of non-fatal or fatal ischemic stroke for patients with no history of cerebrovascular disease (3.2% for simvastatin vs 4.8% with placebo; relative risk reduction=33%, number needed to treat [NNT]=63; P=.0001).² As noted in other well-done studies, the Heart Protection Study showed no difference in the number of hemorrhagic strokes between treatment and placebo groups. There were 3500 subjects with pretreatment low-density lipoprotein (LDL) cholesterol <100 mg/dL; lowering LDL to 65 mg/dL reduced major vascular event risk by about 25%.³

Hypertension with multiple cardiovascular risk factors and cholesterol <250 mg/dL. The ASCOT-LLA study compared atorvastatin with placebo in 10,305 hypertensive Caucasian patients with multiple cardiovascular risk factors and a total nonfasting cholesterol of 250 mg/dL (6.5 mmol/L) or less. Patients were aged 40 to 79 years and had at least 3 other cardiovascular risk factors (left ventricular hypertrophy, abnormal electrocardiogram, type 2 diabetes, peripheral artery disease, stroke or transient ischemic attack, male sex, age >55 years, proteinuria or microalbuminuria, smoking, family history of premature coronary heart disease). The study was stopped early at a median of 3.3 years because atorvastatin significantly reduced cardiac events. Atorvastatin also significantly reduced ischemic strokes when compared with placebo (relative risk [RR]=0.73, 95% confidence interval [CI], 0.56–0.96; P=.024). This study did not differentiate between first or second stroke. The NNT was 155.⁴

Ischemic stroke and coronary disease. The LIPID trial randomized 9014 patients with a history of acute coronary syndromes and total cholesterol of 150 to 270 mg/dL (4 to 7 mmol/L) to either pravastatin or placebo and followed them for 6 years. Among the 350 patients with prior ischemic stroke, there were 388 new ischemic stokes over the course of the study. When adjusted for risk factors (atrial fibrillation, history of cerebrovascular accident, diabetes, hypertension, cigarette smoking, body mass index, and male sex), pravastatin reduced recurrent ischemic stroke by 21% relative to placebo (P=.024). The reduction was not modified by baseline lipid level.⁵

A meta-analysis of 15 randomized placebo-controlled trials using various statins (32,684 participants) assessed the risk of strokes for patients with a history of coronary disease. Among patients who had cerebrovascular disease, statins significantly reduced recurrent ischemic stroke (RR=0.74; 95% CI, 0.64–0.86). One recurrence of ischemic stroke would be prevented for every 110 coronary disease patients treated with a statin. Achieving final total cholesterol <232 mg/dL correlated with reduced risk of recurrent stroke.⁶ Three of the studies evaluated primary prevention of stroke and did not show a significant risk reduction (RR=0.85; P=.4). Statins did not reduce the rate of hemorrhagic stroke or fatal strokes.

Risks of statins. In 1 study involving 35,000 participants and 158,000 person-years of observation, there were 8 cases of rhabdomyolysis in the treatment groups vs 5 in the placebo groups.⁷ Forty-three deaths attributed to statin therapy have been reported to the Food and Drug Administration from 1987 to 2001, or 1 per million person-years of use. The Heart Protection Study found simvastatin and placebo users reported myopathy or muscle pain at the same annual rate of 0.01%.

Recommendations from others

We found no recommendations specifically regarding the use of statins to prevent stroke. However, the Third Report of the National Cholesterol Education Program, Adult Treatment Panel III (NCEP-ATP III) describes symptomatic carotid artery disease as a coronary heart disease risk equivalent and recommends therapy to reduce the LDL below 100 mg/dL.⁸

Evidence summary

We found no studies evaluating statins for the primary prevention of stroke. An observational study of 433 patients with ischemic stroke found that patients who were taking statins before hospital admission more often had better outcomes (51%) than those who were not taking statins (38%). However, the groups differed in many respects.¹ Many coronary event prevention and treatment trials using statins include the risk of primary and recurrent ischemic stroke as secondary endpoints for patients with high cardiac risk.

Primary prevention of stroke in vascular disease. The Heart Protection Study followed 20,536 patients in the United Kingdom (aged 40–80 years), 3280 with a history of cerebrovascular disease (defined as nondisabling stroke, transient cerebral ischemic attack, or carotid endarterectomy or angioplasty) and 17,256 with other occlusive arterial disease, coronary artery disease, or diabetes. Patients were randomized to receive either simvastatin 40 mg or placebo for an average of 5 years. The endpoint was major vascular events: myocardial infarction, stroke of any type, and revascularization procedure.

Simvastatin reduced the combined risk of non-fatal or fatal ischemic stroke for patients with no history of cerebrovascular disease (3.2% for simvastatin vs 4.8% with placebo; relative risk reduction=33%, number needed to treat [NNT]=63; P=.0001).² As noted in other well-done studies, the Heart Protection Study showed no difference in the number of hemorrhagic strokes between treatment and placebo groups. There were 3500 subjects with pretreatment low-density lipoprotein (LDL) cholesterol <100 mg/dL; lowering LDL to 65 mg/dL reduced major vascular event risk by about 25%.³

Hypertension with multiple cardiovascular risk factors and cholesterol <250 mg/dL. The ASCOT-LLA study compared atorvastatin with placebo in 10,305 hypertensive Caucasian patients with multiple cardiovascular risk factors and a total nonfasting cholesterol of 250 mg/dL (6.5 mmol/L) or less. Patients were aged 40 to 79 years and had at least 3 other cardiovascular risk factors (left ventricular hypertrophy, abnormal electrocardiogram, type 2 diabetes, peripheral artery disease, stroke or transient ischemic attack, male sex, age >55 years, proteinuria or microalbuminuria, smoking, family history of premature coronary heart disease). The study was stopped early at a median of 3.3 years because atorvastatin significantly reduced cardiac events. Atorvastatin also significantly reduced ischemic strokes when compared with placebo (relative risk [RR]=0.73, 95% confidence interval [CI], 0.56–0.96; P=.024). This study did not differentiate between first or second stroke. The NNT was 155.⁴

Ischemic stroke and coronary disease. The LIPID trial randomized 9014 patients with a history of acute coronary syndromes and total cholesterol of 150 to 270 mg/dL (4 to 7 mmol/L) to either pravastatin or placebo and followed them for 6 years. Among the 350 patients with prior ischemic stroke, there were 388 new ischemic stokes over the course of the study. When adjusted for risk factors (atrial fibrillation, history of cerebrovascular accident, diabetes, hypertension, cigarette smoking, body mass index, and male sex), pravastatin reduced recurrent ischemic stroke by 21% relative to placebo (P=.024). The reduction was not modified by baseline lipid level.⁵

A meta-analysis of 15 randomized placebo-controlled trials using various statins (32,684 participants) assessed the risk of strokes for patients with a history of coronary disease. Among patients who had cerebrovascular disease, statins significantly reduced recurrent ischemic stroke (RR=0.74; 95% CI, 0.64–0.86). One recurrence of ischemic stroke would be prevented for every 110 coronary disease patients treated with a statin. Achieving final total cholesterol <232 mg/dL correlated with reduced risk of recurrent stroke.⁶ Three of the studies evaluated primary prevention of stroke and did not show a significant risk reduction (RR=0.85; P=.4). Statins did not reduce the rate of hemorrhagic stroke or fatal strokes.

Risks of statins. In 1 study involving 35,000 participants and 158,000 person-years of observation, there were 8 cases of rhabdomyolysis in the treatment groups vs 5 in the placebo groups.⁷ Forty-three deaths attributed to statin therapy have been reported to the Food and Drug Administration from 1987 to 2001, or 1 per million person-years of use. The Heart Protection Study found simvastatin and placebo users reported myopathy or muscle pain at the same annual rate of 0.01%.

Recommendations from others

We found no recommendations specifically regarding the use of statins to prevent stroke. However, the Third Report of the National Cholesterol Education Program, Adult Treatment Panel III (NCEP-ATP III) describes symptomatic carotid artery disease as a coronary heart disease risk equivalent and recommends therapy to reduce the LDL below 100 mg/dL.⁸

Evidence summary

We found no studies evaluating statins for the primary prevention of stroke. An observational study of 433 patients with ischemic stroke found that patients who were taking statins before hospital admission more often had better outcomes (51%) than those who were not taking statins (38%). However, the groups differed in many respects.¹ Many coronary event prevention and treatment trials using statins include the risk of primary and recurrent ischemic stroke as secondary endpoints for patients with high cardiac risk.

Primary prevention of stroke in vascular disease. The Heart Protection Study followed 20,536 patients in the United Kingdom (aged 40–80 years), 3280 with a history of cerebrovascular disease (defined as nondisabling stroke, transient cerebral ischemic attack, or carotid endarterectomy or angioplasty) and 17,256 with other occlusive arterial disease, coronary artery disease, or diabetes. Patients were randomized to receive either simvastatin 40 mg or placebo for an average of 5 years. The endpoint was major vascular events: myocardial infarction, stroke of any type, and revascularization procedure.

Simvastatin reduced the combined risk of non-fatal or fatal ischemic stroke for patients with no history of cerebrovascular disease (3.2% for simvastatin vs 4.8% with placebo; relative risk reduction=33%, number needed to treat [NNT]=63; P=.0001).² As noted in other well-done studies, the Heart Protection Study showed no difference in the number of hemorrhagic strokes between treatment and placebo groups. There were 3500 subjects with pretreatment low-density lipoprotein (LDL) cholesterol <100 mg/dL; lowering LDL to 65 mg/dL reduced major vascular event risk by about 25%.³

Hypertension with multiple cardiovascular risk factors and cholesterol <250 mg/dL. The ASCOT-LLA study compared atorvastatin with placebo in 10,305 hypertensive Caucasian patients with multiple cardiovascular risk factors and a total nonfasting cholesterol of 250 mg/dL (6.5 mmol/L) or less. Patients were aged 40 to 79 years and had at least 3 other cardiovascular risk factors (left ventricular hypertrophy, abnormal electrocardiogram, type 2 diabetes, peripheral artery disease, stroke or transient ischemic attack, male sex, age >55 years, proteinuria or microalbuminuria, smoking, family history of premature coronary heart disease). The study was stopped early at a median of 3.3 years because atorvastatin significantly reduced cardiac events. Atorvastatin also significantly reduced ischemic strokes when compared with placebo (relative risk [RR]=0.73, 95% confidence interval [CI], 0.56–0.96; P=.024). This study did not differentiate between first or second stroke. The NNT was 155.⁴

Ischemic stroke and coronary disease. The LIPID trial randomized 9014 patients with a history of acute coronary syndromes and total cholesterol of 150 to 270 mg/dL (4 to 7 mmol/L) to either pravastatin or placebo and followed them for 6 years. Among the 350 patients with prior ischemic stroke, there were 388 new ischemic stokes over the course of the study. When adjusted for risk factors (atrial fibrillation, history of cerebrovascular accident, diabetes, hypertension, cigarette smoking, body mass index, and male sex), pravastatin reduced recurrent ischemic stroke by 21% relative to placebo (P=.024). The reduction was not modified by baseline lipid level.⁵

A meta-analysis of 15 randomized placebo-controlled trials using various statins (32,684 participants) assessed the risk of strokes for patients with a history of coronary disease. Among patients who had cerebrovascular disease, statins significantly reduced recurrent ischemic stroke (RR=0.74; 95% CI, 0.64–0.86). One recurrence of ischemic stroke would be prevented for every 110 coronary disease patients treated with a statin. Achieving final total cholesterol <232 mg/dL correlated with reduced risk of recurrent stroke.⁶ Three of the studies evaluated primary prevention of stroke and did not show a significant risk reduction (RR=0.85; P=.4). Statins did not reduce the rate of hemorrhagic stroke or fatal strokes.

Risks of statins. In 1 study involving 35,000 participants and 158,000 person-years of observation, there were 8 cases of rhabdomyolysis in the treatment groups vs 5 in the placebo groups.⁷ Forty-three deaths attributed to statin therapy have been reported to the Food and Drug Administration from 1987 to 2001, or 1 per million person-years of use. The Heart Protection Study found simvastatin and placebo users reported myopathy or muscle pain at the same annual rate of 0.01%.

Recommendations from others

We found no recommendations specifically regarding the use of statins to prevent stroke. However, the Third Report of the National Cholesterol Education Program, Adult Treatment Panel III (NCEP-ATP III) describes symptomatic carotid artery disease as a coronary heart disease risk equivalent and recommends therapy to reduce the LDL below 100 mg/dL.⁸

Evidence summary

Thromboprophylaxis with warfarin for patients with NVAF has been studied in 5 major clinical trials.^1-5 Pooled analysis with more than 2900 patients revealed an annual stroke risk of 4.5% for control patients and 1.4% for patients receiving adjusted-dose warfarin (number needed to treat [NNT] for 1 year=32).⁶ Studies comparing aspirin with placebo for treatment of NVAF are less robust and have heterogeneous results. Combined data from the Atrial Fibrillation Aspirin Anticoagulation Study (AFASAK-1),¹ the European Atrial Fibrillation Trial,⁷ and the Stroke Prevention in Atrial Fibrillation (SPAF) I studies² revealed a small but statistically significant reduction in stroke rates (relative risk reduction [RRR]=21%; 8.1% vs 6.3% annual stroke rate; NNT=55), with no increase in major bleeding risk.⁸

The SPAF III investigators further compared adjusted-dose warfarin with low-intensity, fixed-dose warfarin plus aspirin in high-risk patients with NVAF.⁹ An interim analysis at 1.1 years revealed superiority in the reduction of ischemic strokes and systemic embolisms with adjusted-dose warfarin (7.9% vs 1.9% per year, respectively; NNT=16), which led to the trial’s termination. Rates of major hemorrhage did not differ between treatment groups (2.4% per year with combination vs 2.1% per year with warfarin).

Two similar studies published in 1998 were terminated early, in light of the SPAF III results. The Second Copenhagen Atrial Fibrillation, Aspirin, and Anticoagulation Study¹⁰ (AFASAK-2) completed 3 of the scheduled 6 years; it compared warfarin 1.25 mg/d, warfarin 1.25 mg/d plus aspirin 300 mg, aspirin 300 mg alone, and adjusted-dose warfarin (INR=2.0–3.0) to treat NVAF for patients with a median age of 74 years (range, 44–89). The cumulative stroke event rate after 1 year was 5.8% on fixed-dose warfarin, 7.2% on combination, 3.6% on aspirin, and 2.8% on adjusted-dose warfarin. The researchers concluded that while the difference was not statistically significant, adjusted-dose warfarin seemed superior to other treatments after 1 year.

In a similar fashion, Pengo et al¹¹ randomized patients with NVAF aged >60 years to fixed-dose (1.25 mg/d) or adjusted-dose warfarin (INR=2.0–3.0) to evaluate ischemic stroke rates and major bleeding. This trial enrolled 303 patients who were followed up for 14.5 months before discontinuation of the trial. The rate of ischemic stroke was significantly higher in the fixed-dose warfarin group compared with the adjusted warfarin group (3.7% vs 0% per year; NNT=27). Major bleeds were more frequent in the adjusted warfarin group (2.6% vs 1% per year, number needed to harm=63). While the combined primary endpoint did not show a significant benefit for adjusted-dose warfarin, this study suggests that fixed-dose warfarin does not protect against ischemic stroke in NVAF patients.

The intensity of warfarin therapy and stroke severity has recently been studied for patients with NVAF.¹² A subtherapeutic INR (<2.0) on the day of admission was independently associated with severe stroke (odds ratio=1.9; 95% confidence interval [CI], 1.1–3.4), and risk of death at 30 days (hazard ratio, 3.4; 95% CI, 1.1–10.1) compared with an INR of 2.0 or greater. Furthermore, an admission INR of 1.5–1.9 had a similar mortality rate (18%) as an INR of <1.5 (15%), and for those patients on aspirin (15%).

These findings further support the importance of achieving therapeutic INR goals for patients with NVAF.

Recommendations from others

The American Heart Association, the American College of Cardiology,¹³ and the American College of Chest Physicians (ACCP)¹⁴ recommend adjusted-dose warfarin for nonvalvular atrial fibrillation patients at high risk for ischemic stroke. Risk stratification is a key component in order to maximize efficacy while minimizing bleeding risk.

The Table summarizes the ACCP guidelines for prevention of ischemic stroke based on patient risk factors.

TABLE
ACCP Stroke Prevention Guidelines 2001

Evidence summary

Thromboprophylaxis with warfarin for patients with NVAF has been studied in 5 major clinical trials.^1-5 Pooled analysis with more than 2900 patients revealed an annual stroke risk of 4.5% for control patients and 1.4% for patients receiving adjusted-dose warfarin (number needed to treat [NNT] for 1 year=32).⁶ Studies comparing aspirin with placebo for treatment of NVAF are less robust and have heterogeneous results. Combined data from the Atrial Fibrillation Aspirin Anticoagulation Study (AFASAK-1),¹ the European Atrial Fibrillation Trial,⁷ and the Stroke Prevention in Atrial Fibrillation (SPAF) I studies² revealed a small but statistically significant reduction in stroke rates (relative risk reduction [RRR]=21%; 8.1% vs 6.3% annual stroke rate; NNT=55), with no increase in major bleeding risk.⁸

The SPAF III investigators further compared adjusted-dose warfarin with low-intensity, fixed-dose warfarin plus aspirin in high-risk patients with NVAF.⁹ An interim analysis at 1.1 years revealed superiority in the reduction of ischemic strokes and systemic embolisms with adjusted-dose warfarin (7.9% vs 1.9% per year, respectively; NNT=16), which led to the trial’s termination. Rates of major hemorrhage did not differ between treatment groups (2.4% per year with combination vs 2.1% per year with warfarin).

Two similar studies published in 1998 were terminated early, in light of the SPAF III results. The Second Copenhagen Atrial Fibrillation, Aspirin, and Anticoagulation Study¹⁰ (AFASAK-2) completed 3 of the scheduled 6 years; it compared warfarin 1.25 mg/d, warfarin 1.25 mg/d plus aspirin 300 mg, aspirin 300 mg alone, and adjusted-dose warfarin (INR=2.0–3.0) to treat NVAF for patients with a median age of 74 years (range, 44–89). The cumulative stroke event rate after 1 year was 5.8% on fixed-dose warfarin, 7.2% on combination, 3.6% on aspirin, and 2.8% on adjusted-dose warfarin. The researchers concluded that while the difference was not statistically significant, adjusted-dose warfarin seemed superior to other treatments after 1 year.

In a similar fashion, Pengo et al¹¹ randomized patients with NVAF aged >60 years to fixed-dose (1.25 mg/d) or adjusted-dose warfarin (INR=2.0–3.0) to evaluate ischemic stroke rates and major bleeding. This trial enrolled 303 patients who were followed up for 14.5 months before discontinuation of the trial. The rate of ischemic stroke was significantly higher in the fixed-dose warfarin group compared with the adjusted warfarin group (3.7% vs 0% per year; NNT=27). Major bleeds were more frequent in the adjusted warfarin group (2.6% vs 1% per year, number needed to harm=63). While the combined primary endpoint did not show a significant benefit for adjusted-dose warfarin, this study suggests that fixed-dose warfarin does not protect against ischemic stroke in NVAF patients.

The intensity of warfarin therapy and stroke severity has recently been studied for patients with NVAF.¹² A subtherapeutic INR (<2.0) on the day of admission was independently associated with severe stroke (odds ratio=1.9; 95% confidence interval [CI], 1.1–3.4), and risk of death at 30 days (hazard ratio, 3.4; 95% CI, 1.1–10.1) compared with an INR of 2.0 or greater. Furthermore, an admission INR of 1.5–1.9 had a similar mortality rate (18%) as an INR of <1.5 (15%), and for those patients on aspirin (15%).

These findings further support the importance of achieving therapeutic INR goals for patients with NVAF.

Recommendations from others

The American Heart Association, the American College of Cardiology,¹³ and the American College of Chest Physicians (ACCP)¹⁴ recommend adjusted-dose warfarin for nonvalvular atrial fibrillation patients at high risk for ischemic stroke. Risk stratification is a key component in order to maximize efficacy while minimizing bleeding risk.

The Table summarizes the ACCP guidelines for prevention of ischemic stroke based on patient risk factors.

TABLE
ACCP Stroke Prevention Guidelines 2001

Evidence summary

Thromboprophylaxis with warfarin for patients with NVAF has been studied in 5 major clinical trials.^1-5 Pooled analysis with more than 2900 patients revealed an annual stroke risk of 4.5% for control patients and 1.4% for patients receiving adjusted-dose warfarin (number needed to treat [NNT] for 1 year=32).⁶ Studies comparing aspirin with placebo for treatment of NVAF are less robust and have heterogeneous results. Combined data from the Atrial Fibrillation Aspirin Anticoagulation Study (AFASAK-1),¹ the European Atrial Fibrillation Trial,⁷ and the Stroke Prevention in Atrial Fibrillation (SPAF) I studies² revealed a small but statistically significant reduction in stroke rates (relative risk reduction [RRR]=21%; 8.1% vs 6.3% annual stroke rate; NNT=55), with no increase in major bleeding risk.⁸

The SPAF III investigators further compared adjusted-dose warfarin with low-intensity, fixed-dose warfarin plus aspirin in high-risk patients with NVAF.⁹ An interim analysis at 1.1 years revealed superiority in the reduction of ischemic strokes and systemic embolisms with adjusted-dose warfarin (7.9% vs 1.9% per year, respectively; NNT=16), which led to the trial’s termination. Rates of major hemorrhage did not differ between treatment groups (2.4% per year with combination vs 2.1% per year with warfarin).

Two similar studies published in 1998 were terminated early, in light of the SPAF III results. The Second Copenhagen Atrial Fibrillation, Aspirin, and Anticoagulation Study¹⁰ (AFASAK-2) completed 3 of the scheduled 6 years; it compared warfarin 1.25 mg/d, warfarin 1.25 mg/d plus aspirin 300 mg, aspirin 300 mg alone, and adjusted-dose warfarin (INR=2.0–3.0) to treat NVAF for patients with a median age of 74 years (range, 44–89). The cumulative stroke event rate after 1 year was 5.8% on fixed-dose warfarin, 7.2% on combination, 3.6% on aspirin, and 2.8% on adjusted-dose warfarin. The researchers concluded that while the difference was not statistically significant, adjusted-dose warfarin seemed superior to other treatments after 1 year.

In a similar fashion, Pengo et al¹¹ randomized patients with NVAF aged >60 years to fixed-dose (1.25 mg/d) or adjusted-dose warfarin (INR=2.0–3.0) to evaluate ischemic stroke rates and major bleeding. This trial enrolled 303 patients who were followed up for 14.5 months before discontinuation of the trial. The rate of ischemic stroke was significantly higher in the fixed-dose warfarin group compared with the adjusted warfarin group (3.7% vs 0% per year; NNT=27). Major bleeds were more frequent in the adjusted warfarin group (2.6% vs 1% per year, number needed to harm=63). While the combined primary endpoint did not show a significant benefit for adjusted-dose warfarin, this study suggests that fixed-dose warfarin does not protect against ischemic stroke in NVAF patients.

The intensity of warfarin therapy and stroke severity has recently been studied for patients with NVAF.¹² A subtherapeutic INR (<2.0) on the day of admission was independently associated with severe stroke (odds ratio=1.9; 95% confidence interval [CI], 1.1–3.4), and risk of death at 30 days (hazard ratio, 3.4; 95% CI, 1.1–10.1) compared with an INR of 2.0 or greater. Furthermore, an admission INR of 1.5–1.9 had a similar mortality rate (18%) as an INR of <1.5 (15%), and for those patients on aspirin (15%).

These findings further support the importance of achieving therapeutic INR goals for patients with NVAF.

Recommendations from others

The American Heart Association, the American College of Cardiology,¹³ and the American College of Chest Physicians (ACCP)¹⁴ recommend adjusted-dose warfarin for nonvalvular atrial fibrillation patients at high risk for ischemic stroke. Risk stratification is a key component in order to maximize efficacy while minimizing bleeding risk.

The Table summarizes the ACCP guidelines for prevention of ischemic stroke based on patient risk factors.

TABLE
ACCP Stroke Prevention Guidelines 2001