Objective: To evaluate whether the V Leiden mutation is consistently associated with hypertension in pregnancy across populations of pregnant women.
Design: Meta-analysis of studies comparing women with and without hypertension in pregnancy for the V Leiden mutation.
Methods: Studies were identified with MEDLINE and EMBASE searches complemented with perusal of bibliographies of retrieved articles and communication with investigators. Data were evaluated with random effects models and between-study heterogeneity was estimated. Sensitivity analyses examined the effect of population and study characteristics. Bias diagnostics evaluated the evolution of the postulated effect over time and the potential for publication bias. RESULTS Across 19 studies (2742 hypertensive women, 2403 controls), V Leiden mutation increased the odds of hypertensive disease of pregnancy by 2.25-fold [95% confidence interval (CI), 1.50-3.38], but there was large between-study heterogeneity (P = 0.002). The results were similar and heterogeneity persisted when sensitivity analyses were limited to studies with Caucasians, proteinuria, diastolic hypertension threshold > 110 mmHg, specified selection of cases, and matching. While studies published up to 2000 showed an odds ratio of 3.16 (95% CI, 2.04-4.92), no association was seen in studies published in 2001-2002 (odds ratio 0.97; 95% CI, 0.61-1.54). There was also evidence of potential publication bias: the five largest studies showed no association (odds ratio 1.21; 95% CI, 0.84-1.74).
Conclusions: Although modest effects of V Leiden mutation on the risk of hypertension in pregnancy cannot be excluded, the association observed in early and small studies may be typical of bias, in particular time-lag bias and publication bias.