Circulating 25-hydroxyvitamin D levels and hypertension risk after adjusting for publication bias

Background Previous systematic reviews reported that serum vitamin D deficiency was associated with risk of hypertension. The aim was to conduct a meta-epidemiological analysis for evaluating the potential effects of publication bias. Methods The selection criterion was defined as a follow-up study for evaluating the association between circulating 25-hydroxyvitam D level and hypertension risk in adults. A funnel plot and Egger's test were used to detect a publication bias. If a publication bias was identified, trim-and-fill analysis (TFA) with linear estimator was performed to estimate a summary relative risk (sRR). Results The meta-analysis of 13 cohorts resulted in the lower the vitamin D, the higher the risk of hypertension statistically significant (sRR, 1.22; 95% confidence interval [CI], 1.05 to 1.41). But The P-value of Egger's test (=0.015) and asymmetry of the funnel plot showed that there was a publication bias. TFA resulted in that statistical significance disappeared in the association between vitamin D level and hypertension risk in total cohorts (filled sRR, 1.03; 95% CI, 0.89 to 1.18) as well as men and women cohorts. Conclusions The publication bias-adjusted results by TFA had no statistically significant association between vitamin D levels and the risk of hypertension. The significant results in previous systematic reviews might be interpreted as due to publication bias.