Objective:To systematically evaluate the relationship between maternal arsenic exposure and neonatal mortality (NM) and infant mortality (IM).Methods:Literature searches were conducted through PubMed, Web of Science, Embase, Cochrane Library, CNKI database, Wanfang Data Knowledge Service Platform, VIP Chinese Journal Service Platform, and the Chinese Biomedical Literature Database to include case-control, cohort, and cross-sectional studies on the relationship between maternal arsenic exposure and NM, IM published domestically and internationally. The search period was from database establishment to June 4, 2023. Data analysis was conducted using Stata MP 16.0 software, and heterogeneity tests were performed using I2 statistics and Q-test. Fixed effect model (no significant heterogeneity, I2≤50%, P≥0.100) or random effect model (significant heterogeneity, I2 > 50%, P < 0.100) was selected according to heterogeneity among study results for meta-analysis. The OR value (95% CI) was used as the effect value, and subgroup analysis was performed based on different exposure index, arsenic exposure levels in drinking water and study types. At the same time, the dose-response relationship between maternal arsenic exposure and NM, IM was analyzed using generalized least square method. Results:Finally, 9 English literature articles (including 3 053 women and 74 172 maternal and infant pairs) were included, including 6 articles on NM outcomes and 8 articles on IM outcomes. After heterogeneity testing, there was significant heterogeneity in NM ( I2 = 75.20%, P = 0.001) and IM ( I2 = 62.50%, P = 0.009) among all studies. Random effect model was used for meta-analysis, and the combined OR values (95% CI) of NM and IM were 1.38 (1.11 - 1.73) and 1.51 (1.21 - 1.89), respectively. According to the exposure index grouping, in the NM outcome, all studies used drinking water arsenic as the exposure index, and the combined OR value (95% CI) of drinking water arsenic was 1.38 (1.11 - 1.73). In the IM outcome, the combined OR values (95% CI) for urinary arsenic and drinking water arsenic were 3.42 (1.38 - 8.47) and 1.44 (1.16 - 1.79), respectively. According to the grouping of arsenic exposure levels in drinking water, the combined OR values (95% CI) for high and low exposure levels ( > 50 and > 10 - 50 μg/L) in NM and IM outcomes were 1.18 (0.97 - 1.44), 1.54 (1.41 - 1.67), and 1.22 (1.03 - 1.43), 1.55 (1.18 - 2.03), respectively. According to the study types grouping, the combined OR values (95% CI) for retrospective, prospective, and cross-sectional studies in NM and IM outcomes were 1.54 (1.41 - 1.67), 1.11 (0.96 - 1.28), 1.90 (1.01 - 3.55), and 1.55 (1.18 - 2.03), 2.01 (0.82 - 4.94), 1.58 (0.87 - 2.88), respectively. The dose-response relationship analysis showed that the dose-response relationship between maternal arsenic exposure and IM exhibited a non-linear trend (χ 2 = 5.75, P = 0.017). Conclusion:Maternal arsenic exposure is correlated with NM and IM, and there is a non-linear dose-response relationship with IM.

Objective:To systematically evaluate the relationship between maternal arsenic exposure and neonatal mortality (NM) and infant mortality (IM).Methods:Literature searches were conducted through PubMed, Web of Science, Embase, Cochrane Library, CNKI database, Wanfang Data Knowledge Service Platform, VIP Chinese Journal Service Platform, and the Chinese Biomedical Literature Database to include case-control, cohort, and cross-sectional studies on the relationship between maternal arsenic exposure and NM, IM published domestically and internationally. The search period was from database establishment to June 4, 2023. Data analysis was conducted using Stata MP 16.0 software, and heterogeneity tests were performed using I2 statistics and Q-test. Fixed effect model (no significant heterogeneity, I2≤50%, P≥0.100) or random effect model (significant heterogeneity, I2 > 50%, P < 0.100) was selected according to heterogeneity among study results for meta-analysis. The OR value (95% CI) was used as the effect value, and subgroup analysis was performed based on different exposure index, arsenic exposure levels in drinking water and study types. At the same time, the dose-response relationship between maternal arsenic exposure and NM, IM was analyzed using generalized least square method. Results:Finally, 9 English literature articles (including 3 053 women and 74 172 maternal and infant pairs) were included, including 6 articles on NM outcomes and 8 articles on IM outcomes. After heterogeneity testing, there was significant heterogeneity in NM ( I2 = 75.20%, P = 0.001) and IM ( I2 = 62.50%, P = 0.009) among all studies. Random effect model was used for meta-analysis, and the combined OR values (95% CI) of NM and IM were 1.38 (1.11 - 1.73) and 1.51 (1.21 - 1.89), respectively. According to the exposure index grouping, in the NM outcome, all studies used drinking water arsenic as the exposure index, and the combined OR value (95% CI) of drinking water arsenic was 1.38 (1.11 - 1.73). In the IM outcome, the combined OR values (95% CI) for urinary arsenic and drinking water arsenic were 3.42 (1.38 - 8.47) and 1.44 (1.16 - 1.79), respectively. According to the grouping of arsenic exposure levels in drinking water, the combined OR values (95% CI) for high and low exposure levels ( > 50 and > 10 - 50 μg/L) in NM and IM outcomes were 1.18 (0.97 - 1.44), 1.54 (1.41 - 1.67), and 1.22 (1.03 - 1.43), 1.55 (1.18 - 2.03), respectively. According to the study types grouping, the combined OR values (95% CI) for retrospective, prospective, and cross-sectional studies in NM and IM outcomes were 1.54 (1.41 - 1.67), 1.11 (0.96 - 1.28), 1.90 (1.01 - 3.55), and 1.55 (1.18 - 2.03), 2.01 (0.82 - 4.94), 1.58 (0.87 - 2.88), respectively. The dose-response relationship analysis showed that the dose-response relationship between maternal arsenic exposure and IM exhibited a non-linear trend (χ 2 = 5.75, P = 0.017). Conclusion:Maternal arsenic exposure is correlated with NM and IM, and there is a non-linear dose-response relationship with IM.