Abstract

The birth interval data in sub-Saharan Africa is often collected using complex survey designs. The data also tend to be skewed toward one end of the durations between first and second childbirth. Care must therefore be taken when analyzing such data to avoid biased conclusions. This study aimed to compare the performance of the Cox, exponential, and Weibull mixed-effects survival regression models with normal and gamma random effects in analyzing women’s birth interval data in Malawi. The study used a nationally representative sample of 18,988 mothers from 850 clusters collected as part of the 2015–2016 Malawi Demographic and Health Survey data who had at least a child. The results indicated that the majority of mothers had a second child and that the average birth interval was 37.4 months, with a standard deviation of 29.3 months. Additionally, all the models reported statistically significant variations in birth intervals between clusters. However, the Weibull mixed-effects survival model with gamma-distributed random effects provided the best fit compared to the Cox proportional hazard mixed model with normal or gamma random effects, the exponential mixed model with normal or gamma random effects, and the Weibull mixed model with normal random effects. The findings also revealed that mothers from rural areas, the central region, and those who used traditional or modern contraceptive methods had a higher likelihood of having a second child, while women from middle or rich families, those with secondary or tertiary education, and those who had their first child alive had a lower likelihood of giving birth to a second child. These findings suggest that it’s important to consider the clustering effects of the data when analyzing the risk factors of women’s birth intervals. The maternal health policies aiming to regulate women’s child spacing should consider interventions that improve women’s economic livelihoods and educational attainment and those that reduce child mortality.

Original language	en
Pages (from-to)	195-224
Publication status	Published - 2025