Abstract
OBJECTIVE: To evaluate the relationship between subclinical hypothyroidism (SCH) and the risk of miscarriage before 20 weeks of pregnancy. METHODS: Literature databases were searched, including the PubMed, Web of Science, Embase and Cochrane databases, from January 1, 1980, to December 31, 2015. The following search terms were used: subclinical hypothyroidism, hypothyroidism, thyroid dysfunction, thyroid hypofunction, subclinical thyroid disease, thyroid dysfunction, pregnancy loss, abortion and miscarriage. Studies comparing the prevalence of miscarriage in pregnant women with SCH with those who were euthyroid were selected. From the studies matched, the relative risk (RR) and corresponding 95% confidence interval (95% CI) were calculated to yield outcomes. All the statistical analyses were performed using Review Manager (Revman) Version 5.3 and Stata Version 12.0 software. The publication bias of the studies was assessed by forest plot and Begg's test, while the stability of the results was evaluated by sensitivity analysis. RESULTS: Nine articles satisfying the inclusion criteria were analysed. Compared to euthyroid pregnant women, patients with non-treated SCH had a higher prevalence of miscarriage (RR = 1.90, 95% CI1.59-2.27, P<0.01). Additionally, SCH patients in the international diagnostic criteria group were more likely to suffer miscarriages than those in the ATA diagnostic criteria group (χ2 = 11.493, P<0.01). Moreover, there was no difference between patients with treated SCH and euthyroid women (RR = 1.14, 95% CI0.82-1.58, P = 0.43). Compared to isolated SCH women, the miscarriage risk of SCH patients with thyroid autoimmunity (TAI) was obviously higher (RR = 2.47, 95% CI1.77-3.45, P<0.01), and isolated SCH patients also had a higher prevalence of miscarriages than euthyroid women (RR = 1.45, 95% CI1.07-1.96, P = 0.02).A heterogeneity test, forest plot and Begg's test suggested that there was no significant heterogeneity or publication bias among the included articles, while the result of sensitivity analysis showed that our study exhibited high stability. CONCLUSION: SCH is a risk factor for miscarriage in women before 20 weeks of pregnancy, and early treatments can reduce the miscarriage rate. Regardless of the diagnostic criteria used, the miscarriage rate increased as long as a pregnant woman was confirmed to have SCH. The results show that the omission diagnostic rate may increase when the ATA diagnostic criteria are used. In addition, SCH patients with TAI have a higher prevalence of miscarriage, while isolated SCH patients also have a higher miscarriage rate than euthyroid women. Thus, we recommend early treatments to avoid adverse pregnancy outcomes and complications.