Q&A

CLINICAL QUESTION: How accurate is noninvasive glucose monitoring?

BACKGROUND: Tight glycemic control has been shown to improve diabetic outcomes, but current methods for monitoring blood glucose levels are painful and invasive. This trial compares traditional finger-stick glucose monitoring with the non- invasive technique of transdermal iontophoresis.

POPULATION STUDIED: A total of 92 adults with either type 1 or type 2 diabetes were enrolled from 2 diabetes centers and 3 contract research organizations. The mean age was 42 years, and 60% were women. The study population seems similar to that of a typical family practice with respect to the accuracy of noninvasive glucose monitoring.

STUDY DESIGN AND VALIDITY: The noninvasive Glucowatch biographer was compared with finger-stick capillary glucose measurements. After calibration by a single finger-stick measurement, participants wore up to 2 biographers for 12 to 15 hours while getting 2 finger-sticks per hour. Diet and insulin regimens were altered to provide a wide range of glucose levels (40-400 mg/dL).

OUTCOMES MEASURED: The major outcomes assessed were the correlation between biographer and finger-stick measures and the clinical significance of errors. Adverse effects of the biographer were also noted. Other patient-oriented outcomes, such as cost, patient satisfaction, reduction of symptoms, complications of hypoglycemia and hyperglycemia, and ease of use were not addressed.

RESULTS: Biographer readings lagged behind blood glucose measurements by a mean of 18 minutes. There was close tracking of blood glucose levels over a range of 40 to 400 mg/dL for up to 12 hours after a single calibration, and 97% of the biographer readings were associated with a finger-stick reading that was therapeutically equivalent. The average difference between measurements was 15.6%. The highest frequency of clinically significant errors was seen at blood glucose levels below 70 mg/dL; nearly one half of finger-stick measures below 70 mg/dL were read by the biographer as higher. Mild skin irritation occurred at the iontophoresis site.

CLINICAL QUESTION: How accurate is noninvasive glucose monitoring?

BACKGROUND: Tight glycemic control has been shown to improve diabetic outcomes, but current methods for monitoring blood glucose levels are painful and invasive. This trial compares traditional finger-stick glucose monitoring with the non- invasive technique of transdermal iontophoresis.

POPULATION STUDIED: A total of 92 adults with either type 1 or type 2 diabetes were enrolled from 2 diabetes centers and 3 contract research organizations. The mean age was 42 years, and 60% were women. The study population seems similar to that of a typical family practice with respect to the accuracy of noninvasive glucose monitoring.

STUDY DESIGN AND VALIDITY: The noninvasive Glucowatch biographer was compared with finger-stick capillary glucose measurements. After calibration by a single finger-stick measurement, participants wore up to 2 biographers for 12 to 15 hours while getting 2 finger-sticks per hour. Diet and insulin regimens were altered to provide a wide range of glucose levels (40-400 mg/dL).

OUTCOMES MEASURED: The major outcomes assessed were the correlation between biographer and finger-stick measures and the clinical significance of errors. Adverse effects of the biographer were also noted. Other patient-oriented outcomes, such as cost, patient satisfaction, reduction of symptoms, complications of hypoglycemia and hyperglycemia, and ease of use were not addressed.

RESULTS: Biographer readings lagged behind blood glucose measurements by a mean of 18 minutes. There was close tracking of blood glucose levels over a range of 40 to 400 mg/dL for up to 12 hours after a single calibration, and 97% of the biographer readings were associated with a finger-stick reading that was therapeutically equivalent. The average difference between measurements was 15.6%. The highest frequency of clinically significant errors was seen at blood glucose levels below 70 mg/dL; nearly one half of finger-stick measures below 70 mg/dL were read by the biographer as higher. Mild skin irritation occurred at the iontophoresis site.

CLINICAL QUESTION: How accurate is noninvasive glucose monitoring?

BACKGROUND: Tight glycemic control has been shown to improve diabetic outcomes, but current methods for monitoring blood glucose levels are painful and invasive. This trial compares traditional finger-stick glucose monitoring with the non- invasive technique of transdermal iontophoresis.

POPULATION STUDIED: A total of 92 adults with either type 1 or type 2 diabetes were enrolled from 2 diabetes centers and 3 contract research organizations. The mean age was 42 years, and 60% were women. The study population seems similar to that of a typical family practice with respect to the accuracy of noninvasive glucose monitoring.

STUDY DESIGN AND VALIDITY: The noninvasive Glucowatch biographer was compared with finger-stick capillary glucose measurements. After calibration by a single finger-stick measurement, participants wore up to 2 biographers for 12 to 15 hours while getting 2 finger-sticks per hour. Diet and insulin regimens were altered to provide a wide range of glucose levels (40-400 mg/dL).

OUTCOMES MEASURED: The major outcomes assessed were the correlation between biographer and finger-stick measures and the clinical significance of errors. Adverse effects of the biographer were also noted. Other patient-oriented outcomes, such as cost, patient satisfaction, reduction of symptoms, complications of hypoglycemia and hyperglycemia, and ease of use were not addressed.

RESULTS: Biographer readings lagged behind blood glucose measurements by a mean of 18 minutes. There was close tracking of blood glucose levels over a range of 40 to 400 mg/dL for up to 12 hours after a single calibration, and 97% of the biographer readings were associated with a finger-stick reading that was therapeutically equivalent. The average difference between measurements was 15.6%. The highest frequency of clinically significant errors was seen at blood glucose levels below 70 mg/dL; nearly one half of finger-stick measures below 70 mg/dL were read by the biographer as higher. Mild skin irritation occurred at the iontophoresis site.

CLINICAL QUESTION: Is a supervised group exercise program effective in treating depression in older adults?

BACKGROUND: Major depressive disorder (MDD) occurs in up to 18% of older adults¹ and is a major cause of morbidity, decreased quality of life, and mortality in this age group. Effective treatment options include psychotherapy and the use of antidepressants.² Aerobic exercise is an effective therapeutic option in younger adults with MDD, but this option has not been well studied in the elderly.

POPULATION STUDIED: Older adults aged 50 to 77 years were recruited from the community who met Diagnostic and Statistical Manual of Mental Disorders, 4th edition criteria for MDD. Of the 604 subjects initially screened, 156 met the inclusion criteria and were randomly assigned to 1 of the 3 treatment groups. The groups were similar in mean age (57 years), race, sex, marital status, history of recurrent depression, and severity of depression.

STUDY DESIGN AND VALIDITY: This was a randomized controlled trial. After an initial assessment by clinical psychologists, eligible subjects were randomized to exercise (supervised group aerobics consisting of walking or jogging for 30 minutes 3 times a week at an intensity to achieve 70% to 85% of maximum heart rate); antidepressant therapy (sertraline 50 mg titrated up to 200 mg a day as needed to achieve effectiveness); or a combination of exercise and antidepressants. Randomization was stratified on the basis of severity of depression as determined by the Hamilton Rating Scale for Depression (HAM-D). These patients were assessed periodically for 16 weeks by investigators who were blinded to the treatment allocation.

OUTCOMES MEASURED: Treatment response was measured by the HAM-D and the Beck Depression Inventory at weeks 1, 2, 3, 4, 6, 8, and 12. Aerobic capacity was measured at the beginning and end of the trial using a symptom-limited graded exercise treadmill test.

RESULTS: At the end of the 4-month trial, 60% to 69% of the subjects were no longer depressed; there was no statistical difference among the 3 groups. Patients receiving drug therapy responded faster, though those in the exercise group caught up by the end of the trial. Of note, 15% of the medication-only group and 26% of the exercise group did not complete the study because of adverse effects or inability to follow therapy. The patients in the exercise programs also showed significant improvements in their aerobic capacity (mean = 11% increase) compared with those in the medication-only group.

CLINICAL QUESTION: Is a supervised group exercise program effective in treating depression in older adults?

BACKGROUND: Major depressive disorder (MDD) occurs in up to 18% of older adults¹ and is a major cause of morbidity, decreased quality of life, and mortality in this age group. Effective treatment options include psychotherapy and the use of antidepressants.² Aerobic exercise is an effective therapeutic option in younger adults with MDD, but this option has not been well studied in the elderly.

POPULATION STUDIED: Older adults aged 50 to 77 years were recruited from the community who met Diagnostic and Statistical Manual of Mental Disorders, 4th edition criteria for MDD. Of the 604 subjects initially screened, 156 met the inclusion criteria and were randomly assigned to 1 of the 3 treatment groups. The groups were similar in mean age (57 years), race, sex, marital status, history of recurrent depression, and severity of depression.

STUDY DESIGN AND VALIDITY: This was a randomized controlled trial. After an initial assessment by clinical psychologists, eligible subjects were randomized to exercise (supervised group aerobics consisting of walking or jogging for 30 minutes 3 times a week at an intensity to achieve 70% to 85% of maximum heart rate); antidepressant therapy (sertraline 50 mg titrated up to 200 mg a day as needed to achieve effectiveness); or a combination of exercise and antidepressants. Randomization was stratified on the basis of severity of depression as determined by the Hamilton Rating Scale for Depression (HAM-D). These patients were assessed periodically for 16 weeks by investigators who were blinded to the treatment allocation.

OUTCOMES MEASURED: Treatment response was measured by the HAM-D and the Beck Depression Inventory at weeks 1, 2, 3, 4, 6, 8, and 12. Aerobic capacity was measured at the beginning and end of the trial using a symptom-limited graded exercise treadmill test.

RESULTS: At the end of the 4-month trial, 60% to 69% of the subjects were no longer depressed; there was no statistical difference among the 3 groups. Patients receiving drug therapy responded faster, though those in the exercise group caught up by the end of the trial. Of note, 15% of the medication-only group and 26% of the exercise group did not complete the study because of adverse effects or inability to follow therapy. The patients in the exercise programs also showed significant improvements in their aerobic capacity (mean = 11% increase) compared with those in the medication-only group.

CLINICAL QUESTION: Is a supervised group exercise program effective in treating depression in older adults?

BACKGROUND: Major depressive disorder (MDD) occurs in up to 18% of older adults¹ and is a major cause of morbidity, decreased quality of life, and mortality in this age group. Effective treatment options include psychotherapy and the use of antidepressants.² Aerobic exercise is an effective therapeutic option in younger adults with MDD, but this option has not been well studied in the elderly.

POPULATION STUDIED: Older adults aged 50 to 77 years were recruited from the community who met Diagnostic and Statistical Manual of Mental Disorders, 4th edition criteria for MDD. Of the 604 subjects initially screened, 156 met the inclusion criteria and were randomly assigned to 1 of the 3 treatment groups. The groups were similar in mean age (57 years), race, sex, marital status, history of recurrent depression, and severity of depression.

STUDY DESIGN AND VALIDITY: This was a randomized controlled trial. After an initial assessment by clinical psychologists, eligible subjects were randomized to exercise (supervised group aerobics consisting of walking or jogging for 30 minutes 3 times a week at an intensity to achieve 70% to 85% of maximum heart rate); antidepressant therapy (sertraline 50 mg titrated up to 200 mg a day as needed to achieve effectiveness); or a combination of exercise and antidepressants. Randomization was stratified on the basis of severity of depression as determined by the Hamilton Rating Scale for Depression (HAM-D). These patients were assessed periodically for 16 weeks by investigators who were blinded to the treatment allocation.

OUTCOMES MEASURED: Treatment response was measured by the HAM-D and the Beck Depression Inventory at weeks 1, 2, 3, 4, 6, 8, and 12. Aerobic capacity was measured at the beginning and end of the trial using a symptom-limited graded exercise treadmill test.

RESULTS: At the end of the 4-month trial, 60% to 69% of the subjects were no longer depressed; there was no statistical difference among the 3 groups. Patients receiving drug therapy responded faster, though those in the exercise group caught up by the end of the trial. Of note, 15% of the medication-only group and 26% of the exercise group did not complete the study because of adverse effects or inability to follow therapy. The patients in the exercise programs also showed significant improvements in their aerobic capacity (mean = 11% increase) compared with those in the medication-only group.

CLINICAL QUESTION: Does celecoxib relieve pain and inflammation in rheumatoid arthritis without the adverse gastrointestinal (GI) effects of traditional non-steroidal anti-inflammatory drugs (NSAIDs)?

BACKGROUND: NSAIDs are widely prescribed to control the pain and inflammation of rheumatoid arthritis. Their use can be limited by the risk of GI toxicity, an adverse event that is thought to be mediated by nonspecific cyclooxygenase (COX) inhibition. Celecoxib, a COX-2-specific inhibitor, offers the possibility of relieving pain and inflammation in rheumatoid arthritis without GI injury.

POPULATION STUDIED: This study included 1149 adult patients with a mean age of 54 years, 73% of whom were women. All met the American College of Rheumatology diagnostic criteria for rheumatoid arthritis for at least 3 months and were in a functional class of I, II, or III. Stable doses of glucocorticoids or disease-modifying antirheumatic drugs were permitted. Concomitant NSAIDs, injectable corticosteroids, anticoagulants, and antiulcer drugs were prohibited.

STUDY DESIGN AND VALIDITY: This was a randomized double-blind placebo-controlled multicenter 12-week trial funded by the manufacturer of celecoxib. Before enrollment, appropriate baseline clinical and laboratory assessments (including Helicobacter pylori antibody testing and upper endoscopy) were conducted. There were 5 treatment arms: patients received placebo, celecoxib (100 mg, 200 mg, or 400 mg twice daily) or naproxen 500 mg twice daily. All patients underwent a second upper GI endoscopy at treatment termination.

OUTCOMES MEASURED: The primary outcome was improvement in signs and symptoms of rheumatoid arthritis using American College of Rheumatology criteria. The incidence of gastroduodenal ulceration (defined as mucosal breaks Ž3 mm in diameter with unequivocal depth) was a secondary outcome.

RESULTS: Forty percent of the patients withdrew from the study; the majority of withdrawals were because of treatment failure. The rate of treatment failure was similar for all 4 active treatment groups (21%-29%) and higher in the placebo group (45%). In general, all celecoxib doses provided efficacy similar to naproxen and superior to placebo. Compared with placebo recipients (29%), there were significantly more responders in the celecoxib groups (39%-44%; P <.05; number needed to treat [NNT] = 8) and the naproxen group (36%; P <.05). Maximal treatment effects with celecoxib were apparent and significant by week 2 and sustained through week 12.

CLINICAL QUESTION: Does celecoxib relieve pain and inflammation in rheumatoid arthritis without the adverse gastrointestinal (GI) effects of traditional non-steroidal anti-inflammatory drugs (NSAIDs)?

BACKGROUND: NSAIDs are widely prescribed to control the pain and inflammation of rheumatoid arthritis. Their use can be limited by the risk of GI toxicity, an adverse event that is thought to be mediated by nonspecific cyclooxygenase (COX) inhibition. Celecoxib, a COX-2-specific inhibitor, offers the possibility of relieving pain and inflammation in rheumatoid arthritis without GI injury.

POPULATION STUDIED: This study included 1149 adult patients with a mean age of 54 years, 73% of whom were women. All met the American College of Rheumatology diagnostic criteria for rheumatoid arthritis for at least 3 months and were in a functional class of I, II, or III. Stable doses of glucocorticoids or disease-modifying antirheumatic drugs were permitted. Concomitant NSAIDs, injectable corticosteroids, anticoagulants, and antiulcer drugs were prohibited.

STUDY DESIGN AND VALIDITY: This was a randomized double-blind placebo-controlled multicenter 12-week trial funded by the manufacturer of celecoxib. Before enrollment, appropriate baseline clinical and laboratory assessments (including Helicobacter pylori antibody testing and upper endoscopy) were conducted. There were 5 treatment arms: patients received placebo, celecoxib (100 mg, 200 mg, or 400 mg twice daily) or naproxen 500 mg twice daily. All patients underwent a second upper GI endoscopy at treatment termination.

OUTCOMES MEASURED: The primary outcome was improvement in signs and symptoms of rheumatoid arthritis using American College of Rheumatology criteria. The incidence of gastroduodenal ulceration (defined as mucosal breaks Ž3 mm in diameter with unequivocal depth) was a secondary outcome.

RESULTS: Forty percent of the patients withdrew from the study; the majority of withdrawals were because of treatment failure. The rate of treatment failure was similar for all 4 active treatment groups (21%-29%) and higher in the placebo group (45%). In general, all celecoxib doses provided efficacy similar to naproxen and superior to placebo. Compared with placebo recipients (29%), there were significantly more responders in the celecoxib groups (39%-44%; P <.05; number needed to treat [NNT] = 8) and the naproxen group (36%; P <.05). Maximal treatment effects with celecoxib were apparent and significant by week 2 and sustained through week 12.

CLINICAL QUESTION: Does celecoxib relieve pain and inflammation in rheumatoid arthritis without the adverse gastrointestinal (GI) effects of traditional non-steroidal anti-inflammatory drugs (NSAIDs)?

BACKGROUND: NSAIDs are widely prescribed to control the pain and inflammation of rheumatoid arthritis. Their use can be limited by the risk of GI toxicity, an adverse event that is thought to be mediated by nonspecific cyclooxygenase (COX) inhibition. Celecoxib, a COX-2-specific inhibitor, offers the possibility of relieving pain and inflammation in rheumatoid arthritis without GI injury.

POPULATION STUDIED: This study included 1149 adult patients with a mean age of 54 years, 73% of whom were women. All met the American College of Rheumatology diagnostic criteria for rheumatoid arthritis for at least 3 months and were in a functional class of I, II, or III. Stable doses of glucocorticoids or disease-modifying antirheumatic drugs were permitted. Concomitant NSAIDs, injectable corticosteroids, anticoagulants, and antiulcer drugs were prohibited.

STUDY DESIGN AND VALIDITY: This was a randomized double-blind placebo-controlled multicenter 12-week trial funded by the manufacturer of celecoxib. Before enrollment, appropriate baseline clinical and laboratory assessments (including Helicobacter pylori antibody testing and upper endoscopy) were conducted. There were 5 treatment arms: patients received placebo, celecoxib (100 mg, 200 mg, or 400 mg twice daily) or naproxen 500 mg twice daily. All patients underwent a second upper GI endoscopy at treatment termination.

OUTCOMES MEASURED: The primary outcome was improvement in signs and symptoms of rheumatoid arthritis using American College of Rheumatology criteria. The incidence of gastroduodenal ulceration (defined as mucosal breaks Ž3 mm in diameter with unequivocal depth) was a secondary outcome.

RESULTS: Forty percent of the patients withdrew from the study; the majority of withdrawals were because of treatment failure. The rate of treatment failure was similar for all 4 active treatment groups (21%-29%) and higher in the placebo group (45%). In general, all celecoxib doses provided efficacy similar to naproxen and superior to placebo. Compared with placebo recipients (29%), there were significantly more responders in the celecoxib groups (39%-44%; P <.05; number needed to treat [NNT] = 8) and the naproxen group (36%; P <.05). Maximal treatment effects with celecoxib were apparent and significant by week 2 and sustained through week 12.

CLINICAL QUESTION: What is the preferred treatment for a primary basal cell carcinoma?

BACKGROUND: Basal cell carcinoma (BCC) is the most common nonmelanoma skin cancer in the world, with incidences varying between 146 and 317 per 100,000 population in the United States. This slow-growing, locally invasive tumor is most often successfully treated with outpatient office procedures. The authors performed a systematic literature review comparing the effectiveness of 7 treatment modalities for primary BCC: Mohs micrographic surgery (MMS), surgical excision (SE), cryosurgery (CS), curettage and electrodesiccation (CE), radiotherapy (RT), immunotherapy (IM) with interferon or fluorouracil, and photodynamic therapy (PDT) (the latter 2 are investigational).

POPULATION STUDIED: The authors selected and summarized 18 studies, with a total number of 9930 patients treated for primary BCC: 2660 treated with MMS in 3 studies, 1303 with SE in 3 studies, 798 with CS in 4 studies, 4212 with CE in 6 studies, 862 with RT in 1 study, 95 with IM in 1 study, and 0 with PDT (no studies met the inclusion criteria). The broad source base and variety of languages of the primary studies suggests a wide range of patient characteristics such as age, geography, race, and skin type. However, details are not offered. Also undefined are the clinical settings. Exclusion criteria included nonprimary BCC (recurrences), retrospective studies, studies with less than 5 years of follow-up or less than 50 patients, and studies of treatment modality combinations.

STUDY DESIGN AND VALIDITY: This was an extensive review of the literature from 1970 through 1997, including original research published in English, French, German, Dutch, Italian, and Spanish. The search included MEDLINE, EMBASE and CANCERLIT; dermatology yearbooks from 1978 through 1996; and other textbooks, reviews, editorials, existing guidelines, and study references. Of the 298 prospective studies identified, 51% were found using MEDLINE, and 18 met the inclusion criteria. The authors do not describe how they evaluated the quality of studies.

OUTCOMES MEASURED: The studies were quite heterogeneous. The primary outcome measure was the 5-year recurrence rates of treated primary BCCs. The recurrence rates were tabulated from studies as either the raw recurrence rate (absolute number of patients with recurrence divided by number of patients with primary BCC at the start of the study), the strict recurrence rate (absolute number of patients with recurrence divided by number of patients with primary BCC observed for at least 5 years), or the life-table cumulative 5-year recurrence rate. Other outcomes such as cost, cosmetic results, and patient satisfaction were not measured.

RESULTS: The heterogeneity of study designs, reporting methods, and outcomes measured precluded extensive comparisons between studies. Raw recurrence rates may be artificially low, strict recurrence rates too high, and cumulative 5-year recurrence rates were not consistently available. In studies reporting life-table cumulative 5-year rates, the rate of recurrence ranged from 0% to 18.8% across the 5 most common therapies, but lacked otherwise conclusive trends. On the basis of these data, evidence-based guidelines for the treatment of primary BCC cannot yet be developed. Within some treatment modalities, variations in recurrence rates were consistent with risk factors previously identified in the literature. Higher rates of recurrence were associated with certain histologic BCC subtypes (micronodular, adenoid and morphea), location in the H-zone of the face or on ear or eyelid, and size larger than 2 centimeters. The authors conclude that surgical excision remains the treatment of choice for primary BCC, and that MMS is recommended for larger BCCs in the H-zone of the face and those with more aggressive growth patterns; however, both of these conclusions are not definitively supported by their data.

CLINICAL QUESTION: What is the preferred treatment for a primary basal cell carcinoma?

BACKGROUND: Basal cell carcinoma (BCC) is the most common nonmelanoma skin cancer in the world, with incidences varying between 146 and 317 per 100,000 population in the United States. This slow-growing, locally invasive tumor is most often successfully treated with outpatient office procedures. The authors performed a systematic literature review comparing the effectiveness of 7 treatment modalities for primary BCC: Mohs micrographic surgery (MMS), surgical excision (SE), cryosurgery (CS), curettage and electrodesiccation (CE), radiotherapy (RT), immunotherapy (IM) with interferon or fluorouracil, and photodynamic therapy (PDT) (the latter 2 are investigational).

POPULATION STUDIED: The authors selected and summarized 18 studies, with a total number of 9930 patients treated for primary BCC: 2660 treated with MMS in 3 studies, 1303 with SE in 3 studies, 798 with CS in 4 studies, 4212 with CE in 6 studies, 862 with RT in 1 study, 95 with IM in 1 study, and 0 with PDT (no studies met the inclusion criteria). The broad source base and variety of languages of the primary studies suggests a wide range of patient characteristics such as age, geography, race, and skin type. However, details are not offered. Also undefined are the clinical settings. Exclusion criteria included nonprimary BCC (recurrences), retrospective studies, studies with less than 5 years of follow-up or less than 50 patients, and studies of treatment modality combinations.

STUDY DESIGN AND VALIDITY: This was an extensive review of the literature from 1970 through 1997, including original research published in English, French, German, Dutch, Italian, and Spanish. The search included MEDLINE, EMBASE and CANCERLIT; dermatology yearbooks from 1978 through 1996; and other textbooks, reviews, editorials, existing guidelines, and study references. Of the 298 prospective studies identified, 51% were found using MEDLINE, and 18 met the inclusion criteria. The authors do not describe how they evaluated the quality of studies.

OUTCOMES MEASURED: The studies were quite heterogeneous. The primary outcome measure was the 5-year recurrence rates of treated primary BCCs. The recurrence rates were tabulated from studies as either the raw recurrence rate (absolute number of patients with recurrence divided by number of patients with primary BCC at the start of the study), the strict recurrence rate (absolute number of patients with recurrence divided by number of patients with primary BCC observed for at least 5 years), or the life-table cumulative 5-year recurrence rate. Other outcomes such as cost, cosmetic results, and patient satisfaction were not measured.

RESULTS: The heterogeneity of study designs, reporting methods, and outcomes measured precluded extensive comparisons between studies. Raw recurrence rates may be artificially low, strict recurrence rates too high, and cumulative 5-year recurrence rates were not consistently available. In studies reporting life-table cumulative 5-year rates, the rate of recurrence ranged from 0% to 18.8% across the 5 most common therapies, but lacked otherwise conclusive trends. On the basis of these data, evidence-based guidelines for the treatment of primary BCC cannot yet be developed. Within some treatment modalities, variations in recurrence rates were consistent with risk factors previously identified in the literature. Higher rates of recurrence were associated with certain histologic BCC subtypes (micronodular, adenoid and morphea), location in the H-zone of the face or on ear or eyelid, and size larger than 2 centimeters. The authors conclude that surgical excision remains the treatment of choice for primary BCC, and that MMS is recommended for larger BCCs in the H-zone of the face and those with more aggressive growth patterns; however, both of these conclusions are not definitively supported by their data.

CLINICAL QUESTION: What is the preferred treatment for a primary basal cell carcinoma?

BACKGROUND: Basal cell carcinoma (BCC) is the most common nonmelanoma skin cancer in the world, with incidences varying between 146 and 317 per 100,000 population in the United States. This slow-growing, locally invasive tumor is most often successfully treated with outpatient office procedures. The authors performed a systematic literature review comparing the effectiveness of 7 treatment modalities for primary BCC: Mohs micrographic surgery (MMS), surgical excision (SE), cryosurgery (CS), curettage and electrodesiccation (CE), radiotherapy (RT), immunotherapy (IM) with interferon or fluorouracil, and photodynamic therapy (PDT) (the latter 2 are investigational).

POPULATION STUDIED: The authors selected and summarized 18 studies, with a total number of 9930 patients treated for primary BCC: 2660 treated with MMS in 3 studies, 1303 with SE in 3 studies, 798 with CS in 4 studies, 4212 with CE in 6 studies, 862 with RT in 1 study, 95 with IM in 1 study, and 0 with PDT (no studies met the inclusion criteria). The broad source base and variety of languages of the primary studies suggests a wide range of patient characteristics such as age, geography, race, and skin type. However, details are not offered. Also undefined are the clinical settings. Exclusion criteria included nonprimary BCC (recurrences), retrospective studies, studies with less than 5 years of follow-up or less than 50 patients, and studies of treatment modality combinations.

STUDY DESIGN AND VALIDITY: This was an extensive review of the literature from 1970 through 1997, including original research published in English, French, German, Dutch, Italian, and Spanish. The search included MEDLINE, EMBASE and CANCERLIT; dermatology yearbooks from 1978 through 1996; and other textbooks, reviews, editorials, existing guidelines, and study references. Of the 298 prospective studies identified, 51% were found using MEDLINE, and 18 met the inclusion criteria. The authors do not describe how they evaluated the quality of studies.

OUTCOMES MEASURED: The studies were quite heterogeneous. The primary outcome measure was the 5-year recurrence rates of treated primary BCCs. The recurrence rates were tabulated from studies as either the raw recurrence rate (absolute number of patients with recurrence divided by number of patients with primary BCC at the start of the study), the strict recurrence rate (absolute number of patients with recurrence divided by number of patients with primary BCC observed for at least 5 years), or the life-table cumulative 5-year recurrence rate. Other outcomes such as cost, cosmetic results, and patient satisfaction were not measured.

RESULTS: The heterogeneity of study designs, reporting methods, and outcomes measured precluded extensive comparisons between studies. Raw recurrence rates may be artificially low, strict recurrence rates too high, and cumulative 5-year recurrence rates were not consistently available. In studies reporting life-table cumulative 5-year rates, the rate of recurrence ranged from 0% to 18.8% across the 5 most common therapies, but lacked otherwise conclusive trends. On the basis of these data, evidence-based guidelines for the treatment of primary BCC cannot yet be developed. Within some treatment modalities, variations in recurrence rates were consistent with risk factors previously identified in the literature. Higher rates of recurrence were associated with certain histologic BCC subtypes (micronodular, adenoid and morphea), location in the H-zone of the face or on ear or eyelid, and size larger than 2 centimeters. The authors conclude that surgical excision remains the treatment of choice for primary BCC, and that MMS is recommended for larger BCCs in the H-zone of the face and those with more aggressive growth patterns; however, both of these conclusions are not definitively supported by their data.

CLINICAL QUESTION: Is a methylprednisolone injection proximal to the carpal tunnel an effective treatment for carpal tunnel syndrome (CTS)?

BACKGROUND: CTS is a common problem caused by compression of the median nerve at the wrist resulting in hand numbness, loss of dexterity, muscle wasting, and decreased functional ability at work. This study investigated the efficacy of a corticosteroid injection just proximal (not into) the carpal tunnel for CTS.

POPULATION STUDIED: Study participants included 60 patients referred to a neurology clinic in Amsterdam, Netherlands, with CTS symptoms for longer than 3 months’ duration and confirmed with electrophysiological tests. Patients in the intervention and control groups had symptoms for an average of 32 months and 25 months, respectively. In patients with bilateral symptoms, the arm with the most severe symptoms was chosen for randomization. Patients aged younger than 18 years and those who had previous treatment for CTS were excluded.

STUDY DESIGN AND VALIDITY: Patients were randomized to receive an injection of either lignocaine (Lidocaine 10 mg) and methylprednisolone 40 mg or a lignocaine 10-mg injection only. The site of injection was proximal to the carpal tunnel on the volar side of the forearm 4 cm proximal to the wrist crease, between the tendon of the radial flexor muscle and the long palmar muscle. Injections were given at a 10Þ to 20Þ angle with a 3-cm needle. At baseline, there were no significant differences between the control group and the intervention group. The study was performed at one clinic where one neurologist performed all injections. Thus, we do not know if the results of this technique can be consistently reproduced. No patients were reported lost to follow-up at 1 year. To ensure blinding of the treatment assignment, a pharmacist wrapped the syringes in paper and a second neurologist performed outcomes assessment interviews. One month after the initial injection, patients were asked whether they had no symptoms or only minor symptoms that they considered so much improved that they felt no further treatment was necessary. Investigators broke the trial code at follow-up assessment visits to offer nonresponders an injection with methylprednisolone or surgery.

OUTCOMES MEASURED: Patients were considered improved if they self-reported no symptoms or only minor symptoms needing no additional treatment. Other symptoms (weakness, nighttime pain) or impact on lifestyle and occupation were not reported.

RESULTS: At 1 month, 20% of the patients in the control group had improved compared with 77% of patients in the intervention group (P <.001; number needed to treat = 1.8). After 1 year, 8 of the 23 patients (35%) who initially responded to methylprednisolone required a second injection. A total of 86% of nonresponders in the control group improved after receiving a methylprednisolone injection, but 50% of these patients went on to need surgical treatment within 1 year. The investigators reported no side effects to the injection.

CLINICAL QUESTION: Is a methylprednisolone injection proximal to the carpal tunnel an effective treatment for carpal tunnel syndrome (CTS)?

BACKGROUND: CTS is a common problem caused by compression of the median nerve at the wrist resulting in hand numbness, loss of dexterity, muscle wasting, and decreased functional ability at work. This study investigated the efficacy of a corticosteroid injection just proximal (not into) the carpal tunnel for CTS.

POPULATION STUDIED: Study participants included 60 patients referred to a neurology clinic in Amsterdam, Netherlands, with CTS symptoms for longer than 3 months’ duration and confirmed with electrophysiological tests. Patients in the intervention and control groups had symptoms for an average of 32 months and 25 months, respectively. In patients with bilateral symptoms, the arm with the most severe symptoms was chosen for randomization. Patients aged younger than 18 years and those who had previous treatment for CTS were excluded.

STUDY DESIGN AND VALIDITY: Patients were randomized to receive an injection of either lignocaine (Lidocaine 10 mg) and methylprednisolone 40 mg or a lignocaine 10-mg injection only. The site of injection was proximal to the carpal tunnel on the volar side of the forearm 4 cm proximal to the wrist crease, between the tendon of the radial flexor muscle and the long palmar muscle. Injections were given at a 10Þ to 20Þ angle with a 3-cm needle. At baseline, there were no significant differences between the control group and the intervention group. The study was performed at one clinic where one neurologist performed all injections. Thus, we do not know if the results of this technique can be consistently reproduced. No patients were reported lost to follow-up at 1 year. To ensure blinding of the treatment assignment, a pharmacist wrapped the syringes in paper and a second neurologist performed outcomes assessment interviews. One month after the initial injection, patients were asked whether they had no symptoms or only minor symptoms that they considered so much improved that they felt no further treatment was necessary. Investigators broke the trial code at follow-up assessment visits to offer nonresponders an injection with methylprednisolone or surgery.

OUTCOMES MEASURED: Patients were considered improved if they self-reported no symptoms or only minor symptoms needing no additional treatment. Other symptoms (weakness, nighttime pain) or impact on lifestyle and occupation were not reported.

RESULTS: At 1 month, 20% of the patients in the control group had improved compared with 77% of patients in the intervention group (P <.001; number needed to treat = 1.8). After 1 year, 8 of the 23 patients (35%) who initially responded to methylprednisolone required a second injection. A total of 86% of nonresponders in the control group improved after receiving a methylprednisolone injection, but 50% of these patients went on to need surgical treatment within 1 year. The investigators reported no side effects to the injection.

CLINICAL QUESTION: Is a methylprednisolone injection proximal to the carpal tunnel an effective treatment for carpal tunnel syndrome (CTS)?

BACKGROUND: CTS is a common problem caused by compression of the median nerve at the wrist resulting in hand numbness, loss of dexterity, muscle wasting, and decreased functional ability at work. This study investigated the efficacy of a corticosteroid injection just proximal (not into) the carpal tunnel for CTS.

POPULATION STUDIED: Study participants included 60 patients referred to a neurology clinic in Amsterdam, Netherlands, with CTS symptoms for longer than 3 months’ duration and confirmed with electrophysiological tests. Patients in the intervention and control groups had symptoms for an average of 32 months and 25 months, respectively. In patients with bilateral symptoms, the arm with the most severe symptoms was chosen for randomization. Patients aged younger than 18 years and those who had previous treatment for CTS were excluded.

STUDY DESIGN AND VALIDITY: Patients were randomized to receive an injection of either lignocaine (Lidocaine 10 mg) and methylprednisolone 40 mg or a lignocaine 10-mg injection only. The site of injection was proximal to the carpal tunnel on the volar side of the forearm 4 cm proximal to the wrist crease, between the tendon of the radial flexor muscle and the long palmar muscle. Injections were given at a 10Þ to 20Þ angle with a 3-cm needle. At baseline, there were no significant differences between the control group and the intervention group. The study was performed at one clinic where one neurologist performed all injections. Thus, we do not know if the results of this technique can be consistently reproduced. No patients were reported lost to follow-up at 1 year. To ensure blinding of the treatment assignment, a pharmacist wrapped the syringes in paper and a second neurologist performed outcomes assessment interviews. One month after the initial injection, patients were asked whether they had no symptoms or only minor symptoms that they considered so much improved that they felt no further treatment was necessary. Investigators broke the trial code at follow-up assessment visits to offer nonresponders an injection with methylprednisolone or surgery.

OUTCOMES MEASURED: Patients were considered improved if they self-reported no symptoms or only minor symptoms needing no additional treatment. Other symptoms (weakness, nighttime pain) or impact on lifestyle and occupation were not reported.

RESULTS: At 1 month, 20% of the patients in the control group had improved compared with 77% of patients in the intervention group (P <.001; number needed to treat = 1.8). After 1 year, 8 of the 23 patients (35%) who initially responded to methylprednisolone required a second injection. A total of 86% of nonresponders in the control group improved after receiving a methylprednisolone injection, but 50% of these patients went on to need surgical treatment within 1 year. The investigators reported no side effects to the injection.

CLINICAL QUESTION: Will eradication of Helicobacter pylori improve the symptoms of nonulcer dyspepsia?

BACKGROUND: Although H pylori is associated with peptic ulcer disease, chronic gastritis, and other syndromes, its role in nonulcer dyspepsia is much more controversial. A recent meta-analysis suggested that there may be some benefit to H pylori eradication in nonulcer dyspepsia, but it did not include 2 recent studies with negative results.¹ The authors of this study performed the first large multicentered randomized trial in the United States to evaluate the efficacy of H pylori eradication in nonulcer dyspepsia.

POPULATION STUDIED: Consecutive adult patients were invited to participate if they had at least moderate pain or discomfort in the upper abdomen for at least 3 months. Subjects all had a normal esophagogastroduodenal endoscopy (EGD) and a positive result on a carbon-13 urea breath test for H pylori. The average age was 46 years, and the other demographic characteristics were reasonably representative of the general population. Of the 640 patients screened, 337 were enrolled. Most of the exclusions were because patients did not have H pylori or had other underlying conditions.

STUDY DESIGN AND VALIDITY: This was a randomized multicenter trial. Patients were randomly assigned to receive either omeprazole 20 mg, amoxicillin 1000 mg, or clarithromycin 500 mg twice daily for 2 weeks, or identical placebos. In their power analysis, the authors calculated that 270 patients would provide an 88% chance of detecting a 20% reduction in symptoms. Overall, the methods of this study were appropriate to answer this particular question.

OUTCOMES MEASURED: The primary outcome reported was successful treatment (defined by the presence of no more than mild symptoms of dyspepsia during the previous week). Patients completed symptom diaries and the validated Gastrointestinal Symptom Rating Scale at baseline and 1, 3, 6, 9, and 12 months after enrollment. The urea breath test was repeated at the first return visit, and all patients had a repeat EGD and urea breath test at the end of the study (1 year). Quality of life was measured at enrollment and the end of the study using the Medical Outcomes Study Short Form-36 survey instrument.

RESULTS: With respect to the primary outcome, 46% of the patients in the intervention group had successful treatment compared with 50% in the control group (absolute difference = -4%; 95% confidence interval [CI], -15 to 8). Treatment was well tolerated and compliance was excellent in both groups. The absolute difference between groups for complete resolution of dyspeptic symptoms was 5% (95% CI, -5 to 15). There was no significant difference in quality-of-life scores. Not surprisingly, subjects in the intervention group did have a substantial reduction in the presence of H pylori.

CLINICAL QUESTION: Will eradication of Helicobacter pylori improve the symptoms of nonulcer dyspepsia?

BACKGROUND: Although H pylori is associated with peptic ulcer disease, chronic gastritis, and other syndromes, its role in nonulcer dyspepsia is much more controversial. A recent meta-analysis suggested that there may be some benefit to H pylori eradication in nonulcer dyspepsia, but it did not include 2 recent studies with negative results.¹ The authors of this study performed the first large multicentered randomized trial in the United States to evaluate the efficacy of H pylori eradication in nonulcer dyspepsia.

POPULATION STUDIED: Consecutive adult patients were invited to participate if they had at least moderate pain or discomfort in the upper abdomen for at least 3 months. Subjects all had a normal esophagogastroduodenal endoscopy (EGD) and a positive result on a carbon-13 urea breath test for H pylori. The average age was 46 years, and the other demographic characteristics were reasonably representative of the general population. Of the 640 patients screened, 337 were enrolled. Most of the exclusions were because patients did not have H pylori or had other underlying conditions.

STUDY DESIGN AND VALIDITY: This was a randomized multicenter trial. Patients were randomly assigned to receive either omeprazole 20 mg, amoxicillin 1000 mg, or clarithromycin 500 mg twice daily for 2 weeks, or identical placebos. In their power analysis, the authors calculated that 270 patients would provide an 88% chance of detecting a 20% reduction in symptoms. Overall, the methods of this study were appropriate to answer this particular question.

OUTCOMES MEASURED: The primary outcome reported was successful treatment (defined by the presence of no more than mild symptoms of dyspepsia during the previous week). Patients completed symptom diaries and the validated Gastrointestinal Symptom Rating Scale at baseline and 1, 3, 6, 9, and 12 months after enrollment. The urea breath test was repeated at the first return visit, and all patients had a repeat EGD and urea breath test at the end of the study (1 year). Quality of life was measured at enrollment and the end of the study using the Medical Outcomes Study Short Form-36 survey instrument.

RESULTS: With respect to the primary outcome, 46% of the patients in the intervention group had successful treatment compared with 50% in the control group (absolute difference = -4%; 95% confidence interval [CI], -15 to 8). Treatment was well tolerated and compliance was excellent in both groups. The absolute difference between groups for complete resolution of dyspeptic symptoms was 5% (95% CI, -5 to 15). There was no significant difference in quality-of-life scores. Not surprisingly, subjects in the intervention group did have a substantial reduction in the presence of H pylori.

CLINICAL QUESTION: Will eradication of Helicobacter pylori improve the symptoms of nonulcer dyspepsia?

BACKGROUND: Although H pylori is associated with peptic ulcer disease, chronic gastritis, and other syndromes, its role in nonulcer dyspepsia is much more controversial. A recent meta-analysis suggested that there may be some benefit to H pylori eradication in nonulcer dyspepsia, but it did not include 2 recent studies with negative results.¹ The authors of this study performed the first large multicentered randomized trial in the United States to evaluate the efficacy of H pylori eradication in nonulcer dyspepsia.

POPULATION STUDIED: Consecutive adult patients were invited to participate if they had at least moderate pain or discomfort in the upper abdomen for at least 3 months. Subjects all had a normal esophagogastroduodenal endoscopy (EGD) and a positive result on a carbon-13 urea breath test for H pylori. The average age was 46 years, and the other demographic characteristics were reasonably representative of the general population. Of the 640 patients screened, 337 were enrolled. Most of the exclusions were because patients did not have H pylori or had other underlying conditions.

STUDY DESIGN AND VALIDITY: This was a randomized multicenter trial. Patients were randomly assigned to receive either omeprazole 20 mg, amoxicillin 1000 mg, or clarithromycin 500 mg twice daily for 2 weeks, or identical placebos. In their power analysis, the authors calculated that 270 patients would provide an 88% chance of detecting a 20% reduction in symptoms. Overall, the methods of this study were appropriate to answer this particular question.

OUTCOMES MEASURED: The primary outcome reported was successful treatment (defined by the presence of no more than mild symptoms of dyspepsia during the previous week). Patients completed symptom diaries and the validated Gastrointestinal Symptom Rating Scale at baseline and 1, 3, 6, 9, and 12 months after enrollment. The urea breath test was repeated at the first return visit, and all patients had a repeat EGD and urea breath test at the end of the study (1 year). Quality of life was measured at enrollment and the end of the study using the Medical Outcomes Study Short Form-36 survey instrument.

RESULTS: With respect to the primary outcome, 46% of the patients in the intervention group had successful treatment compared with 50% in the control group (absolute difference = -4%; 95% confidence interval [CI], -15 to 8). Treatment was well tolerated and compliance was excellent in both groups. The absolute difference between groups for complete resolution of dyspeptic symptoms was 5% (95% CI, -5 to 15). There was no significant difference in quality-of-life scores. Not surprisingly, subjects in the intervention group did have a substantial reduction in the presence of H pylori.

CLINICAL QUESTION: Is H pylori associated with nonulcer dyspepsia, and if so, will eradication improve symptoms?

BACKGROUND: H pylori is associated with peptic ulcer disease, chronic gastritis, and other syndromes. The role of H pylori in nonulcer dyspepsia is less well defined. Many people have symptoms of dyspepsia, accounting for 3% to 5% of all primary care visits. Acid suppression therapy and motility agents have been the mainstay of treatment for empiric management of nonulcer dyspepsia. If H pylori is a cause of nonulcer dyspepsia, its eradication may reduce or remove symptoms. Since no study has been definitive, the authors undertook a pair of meta-analyses to address this question.

POPULATION STUDIED: The first meta-analysis was to investigate the possible association between H pylori and nonulcer dyspepsia. The authors identified 23 articles, involving more than 7000 patients. These studies were completed in Europe, Africa, Asia, and North America. The second part of this meta-analysis included only randomized trials testing whether eradication of H pylori would improve symptoms of nonulcer dyspepsia. The authors found 5 randomized trials of 778 patients from North America and Europe. Two of those studies had 1-year follow-up and used 3 agents to eradicate the H pylori. The other 3 studies followed patients for only 8 weeks and used a single agent to attempt eradication of H pylori. The latter 3 studies had lower overall quality scores.

STUDY DESIGN AND VALIDITY: These meta-analyses were reasonably well done, although the reported search strategy was somewhat limited. The authors did not include studies published in languages other than English or published only in abstract format. Their reported search strategy only included MEDLINE searches, although they may have searched references of selected papers. No assessment of potential publication biases was undertaken. The authors developed quality scores for each of the articles included, and they performed appropriate sensitivity analysis, testing whether study methods or quality affected the results. Overall, the quality scores of the included studies were not very high; the median score was only 41% of the best possible score. Most studies had reduced scores because of inadequate definition of dyspepsia and poor (or no) control for confounding.

OUTCOMES MEASURED: The authors created summary scores for studies and then combined them to form summary estimates. In the meta-analysis of association, the authors created an odds ratio (OR) for an association between H pylori and nonulcer dyspepsia. In the treatment meta-analysis, the OR represents the odds of improvement of nonulcer dyspepsia symptoms after H pylori treatment.

RESULTS: For the association between H pylori and nonulcer dyspepsia, a summary OR of 1.6 (95% confidence interval [CI], 1.4 - 1.8) was found. Of note, studies of lower quality showed a stronger association. In addition, studies using a weaker methodology (case-control) had a higher OR (OR = 2.0; 95% CI, 1.7 - 2.5) than the studies that used prevalence measures in the general population (OR = 1.3; 95% CI, 1.0 - 1.7). In the meta-analysis investigating potential benefit of H pylori eradication, the summary OR was 1.9 (95% CI, 1.3 - 2.6). Studies with 8-week follow-up and using a single agent found a much stronger association (OR = 15.4) than those that used a 1-year follow-up and 3 agents to eradicate H pylori (OR = 1.4; 95% CI, 1.0 - 2.3).

CLINICAL QUESTION: Is H pylori associated with nonulcer dyspepsia, and if so, will eradication improve symptoms?

BACKGROUND: H pylori is associated with peptic ulcer disease, chronic gastritis, and other syndromes. The role of H pylori in nonulcer dyspepsia is less well defined. Many people have symptoms of dyspepsia, accounting for 3% to 5% of all primary care visits. Acid suppression therapy and motility agents have been the mainstay of treatment for empiric management of nonulcer dyspepsia. If H pylori is a cause of nonulcer dyspepsia, its eradication may reduce or remove symptoms. Since no study has been definitive, the authors undertook a pair of meta-analyses to address this question.

POPULATION STUDIED: The first meta-analysis was to investigate the possible association between H pylori and nonulcer dyspepsia. The authors identified 23 articles, involving more than 7000 patients. These studies were completed in Europe, Africa, Asia, and North America. The second part of this meta-analysis included only randomized trials testing whether eradication of H pylori would improve symptoms of nonulcer dyspepsia. The authors found 5 randomized trials of 778 patients from North America and Europe. Two of those studies had 1-year follow-up and used 3 agents to eradicate the H pylori. The other 3 studies followed patients for only 8 weeks and used a single agent to attempt eradication of H pylori. The latter 3 studies had lower overall quality scores.

STUDY DESIGN AND VALIDITY: These meta-analyses were reasonably well done, although the reported search strategy was somewhat limited. The authors did not include studies published in languages other than English or published only in abstract format. Their reported search strategy only included MEDLINE searches, although they may have searched references of selected papers. No assessment of potential publication biases was undertaken. The authors developed quality scores for each of the articles included, and they performed appropriate sensitivity analysis, testing whether study methods or quality affected the results. Overall, the quality scores of the included studies were not very high; the median score was only 41% of the best possible score. Most studies had reduced scores because of inadequate definition of dyspepsia and poor (or no) control for confounding.

OUTCOMES MEASURED: The authors created summary scores for studies and then combined them to form summary estimates. In the meta-analysis of association, the authors created an odds ratio (OR) for an association between H pylori and nonulcer dyspepsia. In the treatment meta-analysis, the OR represents the odds of improvement of nonulcer dyspepsia symptoms after H pylori treatment.

RESULTS: For the association between H pylori and nonulcer dyspepsia, a summary OR of 1.6 (95% confidence interval [CI], 1.4 - 1.8) was found. Of note, studies of lower quality showed a stronger association. In addition, studies using a weaker methodology (case-control) had a higher OR (OR = 2.0; 95% CI, 1.7 - 2.5) than the studies that used prevalence measures in the general population (OR = 1.3; 95% CI, 1.0 - 1.7). In the meta-analysis investigating potential benefit of H pylori eradication, the summary OR was 1.9 (95% CI, 1.3 - 2.6). Studies with 8-week follow-up and using a single agent found a much stronger association (OR = 15.4) than those that used a 1-year follow-up and 3 agents to eradicate H pylori (OR = 1.4; 95% CI, 1.0 - 2.3).

CLINICAL QUESTION: Is H pylori associated with nonulcer dyspepsia, and if so, will eradication improve symptoms?

BACKGROUND: H pylori is associated with peptic ulcer disease, chronic gastritis, and other syndromes. The role of H pylori in nonulcer dyspepsia is less well defined. Many people have symptoms of dyspepsia, accounting for 3% to 5% of all primary care visits. Acid suppression therapy and motility agents have been the mainstay of treatment for empiric management of nonulcer dyspepsia. If H pylori is a cause of nonulcer dyspepsia, its eradication may reduce or remove symptoms. Since no study has been definitive, the authors undertook a pair of meta-analyses to address this question.

POPULATION STUDIED: The first meta-analysis was to investigate the possible association between H pylori and nonulcer dyspepsia. The authors identified 23 articles, involving more than 7000 patients. These studies were completed in Europe, Africa, Asia, and North America. The second part of this meta-analysis included only randomized trials testing whether eradication of H pylori would improve symptoms of nonulcer dyspepsia. The authors found 5 randomized trials of 778 patients from North America and Europe. Two of those studies had 1-year follow-up and used 3 agents to eradicate the H pylori. The other 3 studies followed patients for only 8 weeks and used a single agent to attempt eradication of H pylori. The latter 3 studies had lower overall quality scores.

STUDY DESIGN AND VALIDITY: These meta-analyses were reasonably well done, although the reported search strategy was somewhat limited. The authors did not include studies published in languages other than English or published only in abstract format. Their reported search strategy only included MEDLINE searches, although they may have searched references of selected papers. No assessment of potential publication biases was undertaken. The authors developed quality scores for each of the articles included, and they performed appropriate sensitivity analysis, testing whether study methods or quality affected the results. Overall, the quality scores of the included studies were not very high; the median score was only 41% of the best possible score. Most studies had reduced scores because of inadequate definition of dyspepsia and poor (or no) control for confounding.

OUTCOMES MEASURED: The authors created summary scores for studies and then combined them to form summary estimates. In the meta-analysis of association, the authors created an odds ratio (OR) for an association between H pylori and nonulcer dyspepsia. In the treatment meta-analysis, the OR represents the odds of improvement of nonulcer dyspepsia symptoms after H pylori treatment.

RESULTS: For the association between H pylori and nonulcer dyspepsia, a summary OR of 1.6 (95% confidence interval [CI], 1.4 - 1.8) was found. Of note, studies of lower quality showed a stronger association. In addition, studies using a weaker methodology (case-control) had a higher OR (OR = 2.0; 95% CI, 1.7 - 2.5) than the studies that used prevalence measures in the general population (OR = 1.3; 95% CI, 1.0 - 1.7). In the meta-analysis investigating potential benefit of H pylori eradication, the summary OR was 1.9 (95% CI, 1.3 - 2.6). Studies with 8-week follow-up and using a single agent found a much stronger association (OR = 15.4) than those that used a 1-year follow-up and 3 agents to eradicate H pylori (OR = 1.4; 95% CI, 1.0 - 2.3).

CLINICAL QUESTION: Should we use warfarin or aspirin to prevent stroke in patients with nonvalvular atrial fibrillation?

BACKGROUND: Patients with atrial fibrillation have a substantial risk of stroke, but the best medication for anticoagulation remains unclear. This meta-analysis compared the safety and effectiveness of warfarin and aspirin in preventing stroke in patients with nonvalvular atrial fibrillation.

POPULATION STUDIED: This paper combined the results of 16 randomized-controlled trials from Europe and North America of 9874 patients with nonvalvular atrial fibrillation. The mean age of subjects was 69 to 71 years, 29% to 38% were women, 45% had hypertension, and 20% to 40% had a previous stroke or transient ischemic attack (TIA). International normalized ratio targets in most studies were 2 to 3 or 2 to 3.5; aspirin dosages ranged from 50 mg per day to 1300 mg per day.

STUDY DESIGN AND VALIDITY: The authors of this meta-analysis reviewed published randomized trials found through OVID and MEDLINE (1966-1999) and by inquiries to the Cochrane Collaboration Stroke Review Group and the Antithrombotic Trialists Collaboration. Studies of valvular atrial fibrillation were excluded. Primary (those that included patients without previous stroke) and secondary (those that included patients with previous stroke or TIA) prevention trials were included. Trials reporting results for subgroups of patients with atrial fibrillation were included, as were nonblinded trials. Two reviewers extracted information on exposure and outcomes by intention-to-treat analysis; homogeneity was tested for each comparison. This was a well-done meta-analysis. Its strengths included the limitation to randomized-controlled trials and the use of independent reviewers; weaknesses included the lack of assessment of study quality, inconsistent reporting of tests for homogeneity, and the lack of analysis of confounders such as risks for hemorrhage, other risks for strokes, or amount of medication. Important outcomes not addressed include cost, patient acceptability, functional status, and quality of life.

OUTCOME MEASURED: The primary outcome reported was the occurrence of all strokes (hemorrhagic and ischemic) presented as relative risk reduction (RRD), absolute risk reduction (ARR) and number needed to treat (NNT) for primary and secondary prevention. Other outcomes included were occurrence of ischemic stroke, intracranial hemorrhage, all-cause mortality, and major extracranial bleeding.

RESULTS: The mean follow-up of patients was 1.7 years. Treatment with adjusted-dose warfarin was superior to placebo for all trials, yielding a 62% RRD (95% confidence interval [CI], 48% - 72%). In patients without a history of stroke or TIA, warfarin produced an ARR of 2.7% (NNT = 37). For secondary prevention, 12 patients needed to be treated to prevent another significant event from occurring (ARR = 2.7%). Of those patients, 60 had to be treated to prevent one death of any cause (ARR = 2.7%).

CLINICAL QUESTION: Should we use warfarin or aspirin to prevent stroke in patients with nonvalvular atrial fibrillation?

BACKGROUND: Patients with atrial fibrillation have a substantial risk of stroke, but the best medication for anticoagulation remains unclear. This meta-analysis compared the safety and effectiveness of warfarin and aspirin in preventing stroke in patients with nonvalvular atrial fibrillation.

POPULATION STUDIED: This paper combined the results of 16 randomized-controlled trials from Europe and North America of 9874 patients with nonvalvular atrial fibrillation. The mean age of subjects was 69 to 71 years, 29% to 38% were women, 45% had hypertension, and 20% to 40% had a previous stroke or transient ischemic attack (TIA). International normalized ratio targets in most studies were 2 to 3 or 2 to 3.5; aspirin dosages ranged from 50 mg per day to 1300 mg per day.

STUDY DESIGN AND VALIDITY: The authors of this meta-analysis reviewed published randomized trials found through OVID and MEDLINE (1966-1999) and by inquiries to the Cochrane Collaboration Stroke Review Group and the Antithrombotic Trialists Collaboration. Studies of valvular atrial fibrillation were excluded. Primary (those that included patients without previous stroke) and secondary (those that included patients with previous stroke or TIA) prevention trials were included. Trials reporting results for subgroups of patients with atrial fibrillation were included, as were nonblinded trials. Two reviewers extracted information on exposure and outcomes by intention-to-treat analysis; homogeneity was tested for each comparison. This was a well-done meta-analysis. Its strengths included the limitation to randomized-controlled trials and the use of independent reviewers; weaknesses included the lack of assessment of study quality, inconsistent reporting of tests for homogeneity, and the lack of analysis of confounders such as risks for hemorrhage, other risks for strokes, or amount of medication. Important outcomes not addressed include cost, patient acceptability, functional status, and quality of life.

OUTCOME MEASURED: The primary outcome reported was the occurrence of all strokes (hemorrhagic and ischemic) presented as relative risk reduction (RRD), absolute risk reduction (ARR) and number needed to treat (NNT) for primary and secondary prevention. Other outcomes included were occurrence of ischemic stroke, intracranial hemorrhage, all-cause mortality, and major extracranial bleeding.

RESULTS: The mean follow-up of patients was 1.7 years. Treatment with adjusted-dose warfarin was superior to placebo for all trials, yielding a 62% RRD (95% confidence interval [CI], 48% - 72%). In patients without a history of stroke or TIA, warfarin produced an ARR of 2.7% (NNT = 37). For secondary prevention, 12 patients needed to be treated to prevent another significant event from occurring (ARR = 2.7%). Of those patients, 60 had to be treated to prevent one death of any cause (ARR = 2.7%).

CLINICAL QUESTION: Should we use warfarin or aspirin to prevent stroke in patients with nonvalvular atrial fibrillation?

BACKGROUND: Patients with atrial fibrillation have a substantial risk of stroke, but the best medication for anticoagulation remains unclear. This meta-analysis compared the safety and effectiveness of warfarin and aspirin in preventing stroke in patients with nonvalvular atrial fibrillation.

POPULATION STUDIED: This paper combined the results of 16 randomized-controlled trials from Europe and North America of 9874 patients with nonvalvular atrial fibrillation. The mean age of subjects was 69 to 71 years, 29% to 38% were women, 45% had hypertension, and 20% to 40% had a previous stroke or transient ischemic attack (TIA). International normalized ratio targets in most studies were 2 to 3 or 2 to 3.5; aspirin dosages ranged from 50 mg per day to 1300 mg per day.

STUDY DESIGN AND VALIDITY: The authors of this meta-analysis reviewed published randomized trials found through OVID and MEDLINE (1966-1999) and by inquiries to the Cochrane Collaboration Stroke Review Group and the Antithrombotic Trialists Collaboration. Studies of valvular atrial fibrillation were excluded. Primary (those that included patients without previous stroke) and secondary (those that included patients with previous stroke or TIA) prevention trials were included. Trials reporting results for subgroups of patients with atrial fibrillation were included, as were nonblinded trials. Two reviewers extracted information on exposure and outcomes by intention-to-treat analysis; homogeneity was tested for each comparison. This was a well-done meta-analysis. Its strengths included the limitation to randomized-controlled trials and the use of independent reviewers; weaknesses included the lack of assessment of study quality, inconsistent reporting of tests for homogeneity, and the lack of analysis of confounders such as risks for hemorrhage, other risks for strokes, or amount of medication. Important outcomes not addressed include cost, patient acceptability, functional status, and quality of life.

OUTCOME MEASURED: The primary outcome reported was the occurrence of all strokes (hemorrhagic and ischemic) presented as relative risk reduction (RRD), absolute risk reduction (ARR) and number needed to treat (NNT) for primary and secondary prevention. Other outcomes included were occurrence of ischemic stroke, intracranial hemorrhage, all-cause mortality, and major extracranial bleeding.

RESULTS: The mean follow-up of patients was 1.7 years. Treatment with adjusted-dose warfarin was superior to placebo for all trials, yielding a 62% RRD (95% confidence interval [CI], 48% - 72%). In patients without a history of stroke or TIA, warfarin produced an ARR of 2.7% (NNT = 37). For secondary prevention, 12 patients needed to be treated to prevent another significant event from occurring (ARR = 2.7%). Of those patients, 60 had to be treated to prevent one death of any cause (ARR = 2.7%).