Randomized trials are the optimal approach for evaluations of treatment efficacy but may not always be feasible. We study the adequacy of the case-control design in evaluating efficacy in a situation where the investigated
therapy, namely the administration of
magnesium sulfate for the prevention of
eclampsia in patients with
preeclampsia, has a suspected strong protective effect. A total of 66 cases of
eclampsia were ascertained from among deliveries occurring between 1977 and 1992 at two hospitals in Houston, Texas. Randomly selected preeclamptic controls were matched to cases based on hospital and month of delivery.
Magnesium sulfate administration prior to seizure occurrence had a strong protective effect against
eclampsia in patients with
preeclampsia (OR, 0.02; 95% CI, 0.01-0.05). This protective effect remained when controls were stratified by the degree of severity of
preeclampsia (mild-to-moderate OR, 0.03, 95% CI, 0.01-0.09 and severe OR, 0.005; 95% CI, 0.0005-0.04) and when cases were stratified by the timing of the first seizure (antepartum and intrapartum
seizures OR, 0.01; 95% CI, 0.003-0.05 and postpartum
seizures OR, 0.03; 95% CI, 0.005-0.15). The effect also remained after adjustment for other important predictors in a multivariate logistic regression model (OR, 0.11; 95% CI, 0.03-0.38). The results of this study are in support of a recent randomized trial on the efficacy of
magnesium sulfate as a prophylactic agent against
eclampsia. Although there are serious potential sources of bias in this study, the magnitude of the protective effect of
magnesium sulfate minimizes the likelihood that this effect can be explained by bias. Observational studies could be appropriate complements or alternatives to randomized trials in situations where a strong treatment effect is expected.