Is the fetus safe when spermicides fail?

Extrapolation of findings from reproductive studies in animals to humans

March, 1997

Studies have produced conflicting results. One statistical method for obtaining an overall quantitative estimate of the effect of a drug on reproduction, especially when conflict exists, is meta-analysis.³ This technique has been widely used in such cases and is gaining increased popularity in medicine. We recently studied the relationship between mother's use of spermicides and subsequent adverse fetal outcomes using meta-analysis.⁴ Nine studies investigating teratogenicity met the inclusion criteria; the Mantel-Haenszel summary odds ratio was 1.02 (95% confidence interval [CI] 0.78 to 1.32). The analyses were 0.10 for significance from unity ( P = 0.748 ) and 8.73 for homogeneity of effects (P = 0.365). Studies comparing specific abnormalities with other abnormalities also indicated no association (odds ratio = 0.96; 95% CI 0.72 to 1.28). Studies investigating other adverse events (spontaneous abortion, stillbirth, low fetal weight, prematurity, increased incidence of female births) also had negative results. Cohen's D, the overall effect size as determined by Tukey's method, was -0.001 (95% CI-0.018 to 0.017). These results indicate that mothers' use of spermicides is not associated with adverse fetal outcomes.

Because rates of spontaneous adverse reproductive outcomes are very low, most studies that attempt to establish relationships between potential teratogens and outcome cannot recruit enough cases to reach an adequate level of statistical power. Physicians and scientists, trying to determine whether a drug or chemical is a potential human teratogen, must often decide which of the studies that present opposing results is more credible.

Meta-analysis allows date from different studies to be combined, so that appropriate statistical power can be achieved. The main criticism of the method is that "good" studies are combined with "bad" studies with each receiving equal weight. However, when meta-analysis is performed appropriately, the problem dissipates. If a blinded reviewer includes or excludes studies according to preset criteria, such methodologic issues are minimized.

In the litigious atmosphere around health care today, we must develop and crystallize scientific tools powerful enough to disqualify the devastating effects of invalid cases presented to juries. Classically, studies presented by plaintiffs (generally parents of malformed children) must stand against conflicting studies presented by defendants (generally pharmaceutical manufacturers, physicians, or both). Lawyers, judges, and physicians cannot be expected to be able to evaluate single studies in isolation from the statistical context of all available evidence. Meta-analysis is the only approach that combines such data and derives an overall estimate of risk.

Beyond its inherent advantages, meta-analysis also forces reviewers to evaluate the methods and results of each study closely to determine acceptability for inclusion. The value of using blinded reviewers is self-evident.

The meta-analysis in this study showed that mothers' use of spermicides is not associated with fetal malformations or any other reported abnormal fetal outcomes. We hope that these results, on the basis of all available studies, will help stop the flood of unnecessary and unfounded litigation. We also hope that meta-analysis will become the standard for evaluating reproductive outcome after exposure to drugs, chemicals, radiation, and infections during pregnancy.

References

1. Huggins G, Vessy M, Flavel R, Yeates D, McPherson K. Vaginal spermicides and outcome of pregnancy : finding in a large cohort study. Contraception 1982;25:219-30.
2. Bracken MB, Vita K. Frequency of non-hormonal contraception around conception and association with congenital malformations in offspring. Am J Epidemiol 1983;117 : 281-91.
3. Einarson TR, Leeder JS, Koren G.A method for meta-analysis of epidemiologic studies. Drug Intell Clin Pharm 1988;22: 813-24.
4. Einarson TR, Koren G, Mattice D. Maternal spermicide use and adverse reproductive outcome. A meta-analysis. Am J Obstet Gynecol 1990;162:655-60.