Relationship between environmental estrogens and hypospadias: a meta-analysis

Abstract

Objective To evaluate the correlation between environmental estrogen and incidence of hypospadias. Methods Medline, Web of Science (ISI), EMBASE, the Cochrane Library, EBSCO Host, Elsevier Science Direct, Chinese Biomedicine (CBMdisc), and China National Knowledge Infrastructure(CNKI) databases, had been searched and secondary manual search for library was also conducted. Case-control and cohort studies, reporting odds ratio (OR) or relative risk (RR) of environmental estrogen exposure before or during pregnancy and their 95% confidence interval (CI) for the occurrence of hypospadias and published from January 1990 to July 2011, were included. Review Manager 5.2 software were used for data analysis. Results Altogether, 18 studies were finally recruited including two cohort studies (only diethylstilbestrol was involved) and 16 case-control studies. While the 16 case-control studies showed significant heterogeneity (P=0.002), and the pooled OR (95%CI) calculated by random effect model was 1.35(1.08-1.70). For the three case-control studies on diethylstilbestrol exposured during pregnancy, no heterogeneity was found (P=0.250) and the pooled OR (95%CI) with fixed effect model was 2.12(1.18-3.80). For the three case-control studies on oral contraceptives and the other three on exposure to dichlorodiphenyltrichloroethane (DDT) or DDT-derived materials, the pooled OR (95%CI) were 1.01 (0.81-1.27) and 1.04 (0.61-1.74) respectively with fixed effect model as no heterogeneity was found (P=0.450 and 0.980). There were two case-control studies on phytoestrogens and two on phthalates, and both showed significant heterogeneity (both P=0.020). The pooled OR (95%CI) with random effect model were 1.88 (0.52-6.79) and 1.68 (0.47-5.94), respectively. The pooled OR (95%CI) of case-control studies in Europe (11 studies) and North America(five studies) were 1.51(1.13-2.03) and 1.13(0.86-1.53) with the random and fixed effect model, separately (heterogeneity test P values were 0.002 and 0.110, respectively). Among those 11 studies in Europe, there were three from Holland and three from England with the pooled OR(95%CI) of 1.47(0.90-2.38) and 2.14(0.74-6.18) with fixed and random effect model (heterogeneity test P values were 0.400 and 0.000), respectively. For the two cohort studies, the pooled RR (95%CI) was 6.19 (0.39-99.55) (heterogeneity test P value was 0.002). Analysis of sensitivity and bias of publication found that publications with positive results were more than those with negative results. Conclusions Environmental estrogens may have effect on hypospadias with geographical differences. Maternal exposure to diethylstilbestrol during pregnancy increases the risk of hypospadias in the offspring. Key words: Hypospadias; Estrogens; Meta-analysis