关键词: 孕产妇死亡率, 直接产科死因, 间接产科死因, 联合轨迹, 纵向聚类

Abstract:

Objective: To identify joint trajectory patterns of direct and indirect maternal mortality ratios (MMR) at the country level from 2000 to 2021, and to compare phase-specific changes during the millennium development goals (MDG, 2000 to 2015) and the sustainable development goals (SDG, 2015 to 2021) periods, as well as differences in health system and policy environments across trajectory groups. Methods: Data on maternal mortality among women aged 15-49 years in 204 countries and territories from 2000 to 2021 were obtained from the Global Burden of Disease (GBD) Study. Direct cause MMR and indirect cause MMR at five time points (2000, 2005, 2010, 2015, and 2021) were jointly analyzed using longitudinal K-means clustering (k=2-6). The optimal number of clusters was determined by the Calinski-Harabasz (CH) index. Based on the clustering results, a piecewise linear mixed effects model with random intercepts was fitted with a knot in 2015 to estimate the baseline intercept in 2000 and period-specific slopes for the MDG and SDG phases. For 2021, health system and policy-related indicators, including antenatal care coverage (≥4 visits, ANC4), proportion of women with a demand for contraception that are using a modern method, cesarean section rate, female human immunodeficiency virus (HIV) prevalence, in facility delivery rate, skilled birth attendance, and an abortion legality index were compared across clusters using the Kruskal-Wallis H test. All tests were two-sided, and P < 0.05 was considered statistically significant. Results: The CH index peaked at k=3 (CH=342.63), classifying the 204 countries and territories into three joint trajectory clusters: high burden (n= 66), medium burden (n=88), and low burden (n=50). In 2021, direct MMR was 334.6 (95%CI: 282.5- 386.7), 65.6 (95%CI: 50.4-80.7), and 5.4 (95%CI: 3.7-7.0) per 100 000 live births in the high, medium, and low burden clusters, respectively; indirect MMR was 33.6 (95%CI: 27.9-39.2), 18.2 (95%CI: 13.5-22.9), and 0.9 (95%CI: 0.6-1.3) per 100 000 live births, respectively. The piecewise mixed effects model showed significant declines in direct MMR during the MDG period in all clusters (slopes: -0.020, -0.016, and -0.036; P < 0.001), whereas declines slowed and became non-significant during the SDG period (-0.011, 0.011, and -0.006; P > 0.05). For indirect MMR, modest increases were observed during the MDG period in the high and medium burden clusters (0.029 and 0.015; P < 0.05), with no significant change in the low burden cluster (P > 0.05). During the SDG period, indirect MMR increased markedly in the medium burden cluster (slope: 0.121; 95%CI: 0.092-0.151; P < 0.001), while remaining broadly stable in the high and low burden clusters (P > 0.05). Health system and policy indicators differed significantly across the clusters (P < 0.001): the high burden cluster showed lower ANC4 coverage, lower in facility delivery and skilled birth attendance, lower demand for contraception satisfied by modern methods, and higher female HIV prevalence; the medium burden cluster achieved near universal in facility delivery and skilled birth attendance but had a higher cesarean section rate; the low-burden cluster generally showed more favorable indicator profiles and a higher abortion legality index. Conclusion: Distinct joint trajectories of direct and indirect maternal mortality were observed globally from 2000 to 2021. While reductions in direct maternal mortality were substantial during the MDG era, progress broadly slowed and plateaued during the SDG era. Meanwhile, the pronounced rise in indirect maternal mortality in medium-burden countries during the SDG period suggests potential structural risk accumulation even when overall MMR appears stable. Incorporating joint direct-indirect trajectories into routine monitoring may facilitate stage and cluster specific prioritization of maternal health interventions.

Key words: Maternal mortality ratio, Direct maternal deaths, Indirect maternal deaths, Joint trajectory, Longitudinal clustering