Article type
Year
Abstract
Background: When determining the dose-related response of a drug treatment on a continuous objective outcome, a lot of potentially valuable data is lost if trials without a parallel placebo (baseline-controlled trials) are excluded. Insufficient trials could limit the ability to construct a comprehensive dose-response relationship.
Objective: In an attempt to determine the comprehensive dose-response effect of fluvastatin on total cholesterol and LDL cholesterol, we tested whether it was possible and appropriate to include baseline-controlled trials. We did this by: 1) comparing of the effect size in placebo-controlled trials with the effect size in baseline-controlled trials; 2) comparing the placebo group effect size with zero.
Methods: 1) After stratification by dose (20mg, 40mg, 80mg), the effect-size of pooled placebo-controlled treatment groups (N=1841) was compared with that of pooled baseline-controlled treatment groups (N=3782). This comparison was made using a two-tailed unpaired t-test. 2) The effect-sizes of placebo groups were pooled after stratifying by dose (2.5 mg, 5 mg, 10 mg, 20 mg, 40 mg, and 80 mg doses). The total number of patients was 1109 for total cholesterol measurements and 1144 for LDL cholesterol measurements. A two-tailed one-sample t-test was used to determine if the mean value of pooled placebo groups was significantly different from zero.
Results: The treatment effect for the baseline-controlled trials was not significantly different (p>0.1) from the treatment effect for placebo-controlled trials. For each of the dose stratification group comparisons, placebo effect-size was not significantly different from zero (p>0.1).
Conclusions: In this example, meta-analysis of trials with an objective laboratory measurement showed no significant placebo effect. Under these circumstances, baseline-controlled trials can be included in REVMAN by setting the placebo effect as zero and imputing the standard deviation. This increases the precision of the effect size estimation for each dose and thus allows a better estimate of the dose response relationship.
Objective: In an attempt to determine the comprehensive dose-response effect of fluvastatin on total cholesterol and LDL cholesterol, we tested whether it was possible and appropriate to include baseline-controlled trials. We did this by: 1) comparing of the effect size in placebo-controlled trials with the effect size in baseline-controlled trials; 2) comparing the placebo group effect size with zero.
Methods: 1) After stratification by dose (20mg, 40mg, 80mg), the effect-size of pooled placebo-controlled treatment groups (N=1841) was compared with that of pooled baseline-controlled treatment groups (N=3782). This comparison was made using a two-tailed unpaired t-test. 2) The effect-sizes of placebo groups were pooled after stratifying by dose (2.5 mg, 5 mg, 10 mg, 20 mg, 40 mg, and 80 mg doses). The total number of patients was 1109 for total cholesterol measurements and 1144 for LDL cholesterol measurements. A two-tailed one-sample t-test was used to determine if the mean value of pooled placebo groups was significantly different from zero.
Results: The treatment effect for the baseline-controlled trials was not significantly different (p>0.1) from the treatment effect for placebo-controlled trials. For each of the dose stratification group comparisons, placebo effect-size was not significantly different from zero (p>0.1).
Conclusions: In this example, meta-analysis of trials with an objective laboratory measurement showed no significant placebo effect. Under these circumstances, baseline-controlled trials can be included in REVMAN by setting the placebo effect as zero and imputing the standard deviation. This increases the precision of the effect size estimation for each dose and thus allows a better estimate of the dose response relationship.