Henry S. Sacks

Meta-Analyses of Randomized Controlled Trials

A new type of research, termed meta-analysis, attempts to analyze and combine the results of previous reports. We found 86 meta-analyses of reports of randomized controlled trials in the English-language literature. We evaluated the quality of these meta-analyses, using a scoring method that considered 23 items in six major areas--study design, combinability, control of bias, statistical analysis, sensitivity analysis, and application of results. Only 24 meta-analyses (28 percent) addressed all six areas, 31 (36 percent) addressed five, 25 (29 percent) addressed four, 5 (6 percent) addressed three, and 1 (1 percent) addressed two. Of the 23 individual items, between 1 and 14 were addressed satisfactorily (mean +/- SD, 7.7 +/- 2.7). We conclude that an urgent need exists for improved methods in literature searching, quality evaluation of trials, and synthesizing of the results.

Bias in treatment assignment in controlled clinical trials.

Controlled clinical trials of the treatment of acute myocardial infarction offer a unique opportunity for the study of the potential influence on outcome of bias in treatment assignment. A group of 145 papers was divided into those in which the randomization process was blinded (57 papers), those in which it may have been unblinded (45 papers), and those in which the controls were selected by a nonrandom process (43 papers). At least one prognostic variable was maldistributed (P less than 0.05) in 14.0 per cent of the blinded-randomization studies, in 26.7 per cent of the unblinded-randomization studies, and in 58.1 per cent of the nonrandomized studies. Differences in case-fatality rates between treatment and control groups (P less than 0.05) were found in 8.8 per cent of the blinded-randomization studies, 24.4 per cent of the unblinded-randomization studies, and 58.1 per cent of the nonrandomized studies. These data emphasize the importance of keeping those who recruit patients for clinical trials from suspecting which treatment will be assigned to the patient under consideration.

The Women's Interagency HIV Study

The Womens Interagency HIV Study comprises the largest U.S. cohort to date of human immunodeficiency virus (HIV)-seropositive women (N = 2,058) with a comparison cohort of seronegative women (N = 568). The methodology, training, and quality assurance activities employed are described. The study population, enrolled between October 1994 and November 1995 through six clinical consortia throughout the United States (totaling 23 sites) represents a typically hard-to-reach study population. More than half of the women in each cohort were living below the federally defined levels of poverty. The women ranged in age from 16 to 73 years; approximately one-quarter self-identified as Latina or Hispanic, over one-half as African-American not of Hispanic origin, and less than 20% as white, non-Hispanic origin. Self-reporting of HIV exposure risk included injection drug use by 34% of the seropositive women and 28% of the seronegative women, heterosexual contact (42% vs 26%), transfusion risk (4% vs 3%) and no identified risk (20% vs 43%). Demographic and HIV exposure risk characteristics of the seropositive cohort were comparable with characteristics of nationally reported AIDS cases in U.S. women. This well characterized cohort of HIV-seropositive and high-risk seronegative women represents a rich opportunity for future studies of HIV disease progression and pathogenesis.

Randomized versus historical controls for clinical trials

To compare the use of randomized controls (RCTs) and historical controls (HCTs) for clinical trials, we searched the literature for therapies studied by both methods. We found six therapies for which 50 RCTs and 56 HCTs were reported. Forty-four of 56 HCTs (79 percent) found the therapy better than the control regimen, but only 10 of 50 RCTs (20 percent) agreed. For each therapy, the treated patients in RCTs and HCTs of the same therapy was largely due to differences in outcome for the control groups, with HCT control patients generally doing worse than the RCT control groups. Adjustment of the outcomes of the HCTs for prognostic factors, when possible, did not appreciably change the results. The data suggest that biases in patient selection may irretrievably weight the outcome of HCts in favor of new therapies. RCTs may miss clinically important benefits because of inadequate attention to sample size. The predictive value of each might be improved by reconsidering the use of p less than 0.05 as the significance level for all types of clinical trials, and by the use of confidence intervals around estimates of treatment effects.

Association of Adrenocorticosteroid Therapy and Peptic-Ulcer Disease

We reexamined the association between corticosteroid therapy and subsequent peptic ulceration or gastrointestinal hemorrhage by pooling data from 71 controlled clinical trials in which patients were randomized to systemic corticosteroids (or ACTH) or to nonsteroid therapy. Of 3064 steroid-treated patients evaluated for peptic ulcer, 55 (1.8 per cent) had ulcers, as compared with 23 of 2897 controls (0.8 per cent) (relative risk, 2.3; 95 per cent confidence interval, 1.4 to 3.7). Of 3135 steroid-treated patients evaluated for gastrointestinal hemorrhage, 78 (2.5 per cent) had bleeding, as compared with 48 of 2976 controls (1.6 per cent) (relative risk, 1.5; 95 per cent confidence interval, 1.1 to 2.2). The incidence of ulcers varied directly with the dosage of steroids. When separate analyses were performed for studies that were double-blind, used only oral steroids, used only parenteral steroids, or excluded patients with a history of ulcer, the trend remained consistent but did not always reach statistical significance. This study strongly suggests that corticosteroids do increase the risk of peptic ulcers and gastrointestinal hemorrhage.

A Controlled Trial Comparing Continued Zidovudine with Didanosine in Human Immunodeficiency Virus Infection

BACKGROUND Although zidovudine is effective in patients with human immunodeficiency virus (HIV) infection, its efficacy may decline with prolonged use. Didanosine is another inhibitor of HIV reverse transcriptase. We evaluated the effectiveness of changing anti-HIV treatment from zidovudine to didanosine. METHODS This multicenter, double-blind study involved 913 patients who had tolerated zidovudine for at least 16 weeks. The patients had the acquired immunodeficiency syndrome (AIDS), AIDS-related complex with less than or equal to 300 CD4 cells per cubic milliliter, or asymptomatic HIV infection with less than or equal to 200 CD4 cells per cubic milliliter. They were randomly assigned to receive 600 mg per day of zidovudine, 750 mg per day of didanosine, or 500 mg per day of didanosine. RESULTS There were significantly fewer new AIDS-defining events and deaths among the 298 subjects assigned to 500 mg per day of didanosine than among the subjects who continued to receive zidovudine (relative risk, 1.39; 95 percent confidence interval, 1.06 to 1.82; P = 0.015). With 750 mg of didanosine, there was no clear benefit over zidovudine (relative risk, 1.10; 95 percent confidence interval, 0.86 to 1.42). The efficacy of didanosine was unrelated to the duration of previous zidovudine treatment. In the two didanosine groups, there were improvements in the number of CD4 cells (P less than 0.001 for both groups) and in p24 antigen levels (P = 0.03 in the 500-mg group; P = 0.005 in the 750-mg group), as compared with the zidovudine group. CONCLUSIONS Changing treatment from zidovudine to 500 mg per day of didanosine appears to slow the progression of HIV disease.

A Survey of Clinical Trials of Antibiotic Prophylaxis in Colon Surgery: Evidence against Further Use of No-Treatment Controls

To evaluate the use of antibiotics given prophylactically of colon surgery, we examined 26 trials published from 1965 to 1980 in which patients given various antibiotic regiments were compared with controls given no antibiotic treatment. In 22 (85 per cent of these trials) antibiotics reduced postoperative wound infection (p less than 0.05 in 14). Combining the results of the trials published from 1965 to 1975 reveals a 95 per cent confidence interval from the true difference in infection rates of 14 +/- 6 per cent (36 per cent for control group vs. 22 per cent for treatment group) and the true difference in death rates of 6.7 +/- 4.4 per cent (11.2 per cent for control group vs 4.5 per cent for treatment group). Yet trials employing control groups given no treatment continue to be reported. Since the use of such controls is justified only when no effective alternative therapy exists, we believe that any further trials of antibiotic prophylaxis in colon surgery should employ a previously proved standard. However, steadily increasing efficacy of treatment means that comparisons of new therapies with standard therapies will become prohibitively expensive because of the large number of patients required.

Drug Treatment of Hypertension in the Elderly: A Meta-Analysis

Estimates of hypertension in patients older than 65 years vary according to population factors and the cut-off value used [1]. Despite estimated prevalences of as high as 64% [2], conservative estimates are 20% for systolic-diastolic hypertension and 10% to 20% for isolated systolic hypertension [1, 3]. Hypertension increases the risk for cardiovascular death [4] and is one of the most modifiable risk factors [5]. Drug treatment has been shown to benefit the young, but the elderly have typically been under-represented in randomized, controlled trials [6, 7]. As people age, hypertension becomes less and less of a risk factor, presumably because those most susceptible have already died [8]. Indeed, blood pressure and mortality rates have been shown to vary inversely in very elderly men [8-10] and women [11, 12]. Because of the risk of drug interactions and side effects, questions about the riskbenefit of treatment in very old and frail patients have been raised [1, 8]. We did a meta-analysis to determine the effect of treatment on all-cause and cause-specific mortality, stroke, and coronary heart disease in the elderly. We also tried to evaluate the effects of disease severity and age of the trial populations on these end points to determine whether treatment should be applied to all elderly subgroups. Methods Data Sources We identified randomized, controlled trials of antihypertensive drug treatment in the elderly through a MEDLINE search of articles published from January 1980 to February 1992. The following key words were used: hypertension, elderly or aged, and randomized, controlled trials. Additional studies were identified from pertinent review articles, a review of Current Contents (Clinical Medicine), and references of the retrieved articles. Study Selection and Data Extraction The following were the study selection criteria: 1) random assignment to treatment (control patients were not required to have received placebo); 2) hypertension defined as diastolic (>89 mm Hg), systolic (>159 mm Hg), or isolated systolic (>159 mm Hg systolic/<90 mm Hg diastolic) on at least one measurement; 3) interventions that included any antihypertensive drug treatments administered in any form (single, combined, or stepped); and 4) end points numerically identifiable at 5 years for elderly patients. Because some trials reported ages by decade (50 to 59 years, 60 to 69 years, and so forth), and some reported ages as younger or older than 65 years, we extracted data from any groups that included patients with a minimum age of 60 years. We used for our analyses the protocol definitions of morbidity end points in the primary studies that were not uniform for either coronary events or strokes. Transient ischemic attacks were excluded. In some instances, subtypes of strokes were combined. In one case, the number of deaths at 5 years had to be approximated from the mortality rate [13]. One trial [14] provided only morbidity end points. Data were extracted by two readers who were blinded to treatment assignment, and the final result was reached by consensus. Analysis and Data Synthesis Included trials were graded blindly by two readers according to a previously described quality-grading system [15] that assigns a score between 0 and 1 using more than 30 elements of study design, randomization, blinding, statistical analysis, and reporting. The outcomes were mortality and other events in the treatment and control groups using intention-to-treat analysis [16]. We used the Yusuf-Peto method [17, 18] to obtain a combined estimate of the odds ratio under the assumption of homogeneity of the odds ratio across the strata; it provided a test of significance of the combined odds ratio. There is also a test of the homogeneity of effect (that is, that any difference between the study results can be attributed to random variation). Other methods used included the DerSimonian and Laird random-effects model [19], the Robins, Greenland, and Breslow model [20], the Mantel-Haenszel model [21], and the pooled relative risk model [22]. Because no major differences in results were found, results are presented only for the Yusuf-Peto method. We used approximate chi-square tests for statistical heterogeneity [18]. All P values are two-sided. We considered blood pressure reduction an intermediate effect [16] rather than the primary end point of interest. We defined blood pressure reduction as the mean systolic or diastolic value of treated patients minus the mean of the controls at 5 years of follow-up. We ranked the primary study populations according to severity of illness (from 1 to 7, with 1 being the least ill) using two main available criteria: the health care setting from which the patients were recruited and the amount of end-organ damage at study entry. We assumed that patients in long-term care facilities were the most severely ill, followed by those in acute care hospitals and then by those in primary care, and that patients recruited from their communities were the least ill. An approximation of baseline end-organ damage was provided by simple addition of cardiovascular disease markers at trial entry and by the severity of the exclusion criteria. Trials with less strict inclusion criteria usually had more end-organ damage burden and vice-versa. One trial could not be ranked because of the lack of end-organ damage description [23-25] and another because of the lack of mortality data [14]. Further analysis of the trial populations according to ranking was done by using 5-year incidence rates of various end points. Analysis for trend according to ranking was done using weighted regression [26]. The ranking obtained was entered as an effect modifier measure into a meta-regression [16] in which each trial was a unit of analysis. The dependent variables for the meta-regression were log odds ratio and log relative risk. Statistical methods for the effect of severity-of-illness ranking on outcome are provided in the Appendix. Results Trial Characteristics We identified 31 trials but excluded 22 for the following reasons: Four gave no age criteria [27-30]; 11 did not give results separately for elderly patients [31-41]; 2 included only patients with previous strokes [42, 43]; 1 used multiple interventions [44]; 3 used historical controls [45-47]; and 1 had fewer than 5 years of follow-up [48]. The 9 identified studies that fulfilled the inclusion criteria are shown in Tables 1 and 2 [1, 3, 14, 23-25,44-74], but only 7 could be ranked according to severity of illness. These 9 studies included multiple publications that were perused to abstract the required data. Table 1. Descriptive Data on Trial Population and Study Design* Table 2. Hypertension Results* The 15 559 patients older than 59 years studied in the trials (7750 treated patients and 7809 controls) had an average follow-up of 4.13 years, which yields more than 64 000 patient-years of follow-up. The trials had a mean quality score of 0.67 (range, 0.39 to 0.91), which is higher than that of trials of mild hypertension we reviewed in a previous study [75]. Some trials did not provide systolic blood pressure reduction [50, 62, 73], and another [13] provided only blood pressure at 6 months of follow-up (Tables 1 and 2). The average diastolic and systolic blood pressure reduction at 5 years was 8.72 mm Hg and 17.13 mm Hg, respectively. All trials included stepped care (defined as sequential use of one or more drugs) except for one [13] that used only methyldopa. All of the stepped-care regimens included treatment with diuretic agents, and five of them [25, 62, 65, 72, 74] included -blockers. In three trials, no placebo was used in the control group [13, 65, 73]. Three trials [13, 51, 66] did not provide drop-out data. Between 10% [74] and 35% [65] of the treated groups in the remaining trials had premature termination of drug treatment. It was not possible to pool treatment side effects because they were recorded differently. In one trial [74], 44% of the patients in the control group received active drug treatment at 5 years of follow-up. Meta-analysis All-cause mortality was approximately 12% lower in the treated patients (odds ratio, 0.88; 95% CI, 0.80 to 0.97; 953 events compared with 1069 events). Treated patients had 36% fewer fatal strokes (odds ratio, 0.64; CI, 0.49 to 0.82; 94 events compared with 149 events). Fatal coronary events in the treated group were 25% lower (odds ratio, 0.75; CI, 0.64 to 0.88; 263 events compared with 350 events) (Figure 1). However, nonvascular deaths were 8% higher in the treated group (odds ratio, 1.08; CI, 0.94 to 1.25; 423 events compared with 396 events) (data not shown). Nonfatal strokes were 35% lower in the treated patients (odds ratio, 0.65; CI, 0.55 to 0.76; 247 events compared with 382 events), and coronary morbidity was 15% lower (odds ratio, 0.85; CI, 0.73 to 0.99; 325 events compared with 379 events) (Figure 2). When the clinical history definition of coronary events in the Hypertension Detection and Follow-up Program [76] was used, the observed reduction in coronary morbidity was 23% (odds ratio, 0.77; CI, 0.64 to 0.92). We observed no consistent pattern in a separate analysis of most end points in those trials with placebo control compared with those without. Figure 1. Results of meta-analysis of mortality end points. Left. Right. Because the Systolic Hypertension in the Elderly Program (SHEP) trial [74] only considered systolic hypertension, and patients with isolated systolic hypertension may not be similar to those with diastolic hypertension, analyses were done that included and excluded the SHEP trial. The SHEP trial results were not consistently different from the results of the other trials in the direction and magnitude of effect (Figures 1 and 2). Figure 2. Results of meta-analysis of morbidity end points. Left. Right. However, the SHEP trial was not the only study that included patients with isolated systolic hypertension. In one trial

Infection with Human Immunodeficiency Virus Type 1 (HIV-1) among Recipients of Antibody-Positive Blood Donations

OBJECTIVE To assess the incidence of human immunodeficiency virus type 1(HIV-1) transmission by antibody (anti-HIV-1)-positive blood components, and to determine the immunologic and clinical course in HIV-1-infected recipients. DESIGN AND SUBJECTS We retrospectively tested approximately 200,000 donor blood component specimens stored in late 1984 and 1985 for anti-HIV-1, and we contacted recipients of positive specimens to determine their serologic status. They were compared with both recipients of HIV-1-negative transfusions and healthy (untransfused) controls. Subjects were seen at 3- to 6-month intervals for up to 4 years for clinical and immunologic evaluations. MEASUREMENTS AND MAIN RESULTS Of 133 recipients, 9 had other possible exposures. Excluding these cases, 111 of 124 (89.5%) were anti-HIV-1-positive (95% CI, 84.1% to 94.5%). The recipients sex, age, underlying condition, and type of component did not influence infection rates. The cumulative risk for developing the acquired immunodeficiency syndrome (AIDS) within 38 months after transfusion was 13% (CI, 7.5% to 21.6%). At 36 +/- 3 months after the index transfusion, seropositive recipients had lower counts of CD2+CDw26+, CD4+, CD4+CD29+, and CD4+CD45RA+subsets and more CD8+I2+ lymphocytes than did recipients of anti-HIV-1-negative transfusions. The CD4+ and CD2+CDw26+subsets changed the most rapidly. The absolute CD8+ count remained normal. CONCLUSIONS Transfusion of anti-HIV-1-positive blood infected 90% of recipients. The rate of progression to AIDS within the first 38 months after infection was similar to that reported for homosexual men and hemophiliacs. Although most lymphocyte subset counts changed over time, CD8+ counts were constant.

A meta-analysis of randomized control trials of progestational agents in pregnancy

The continued use of progestational agents in attempts to achieve a normal outcome of pregnancy in women with a ‘high‐risk’ pregnancy (previous miscarriage, stillbirth or present preterm labour) prompted this meta‐analysis of randomized control trials of such therapy. Of 20 trials of a progestogen 15 had combinable data. Combined comparisons, using odds ratios with confidence intervals, were made of the rates of livebirths at term or preterm and the sum of term and preterm deliveries, miscarriages, stillbirths and neonatal deaths. All but one comparison failed to show a significant benefit. Only the preterm delivery versus the term delivery comparison approached statistical significance. There were average deficiencies of quality apparent in the studies, and a test for heterogeneity among the studies was positive, but these caveats do not diminish the conclusion that progestogens should not be used outside of randomized trials at present. If trials are done, they should include only women with demonstrated hormonal abnormalities who are carrying a live fetus as shown by ultrasonography.