Correlates of antenatal care visits among expectant mothers in Malawi: application of generalised estimating equations and likelihood-based poisson and negative binomial mixed models.

Despite recent improvements in antenatal care (ANC) coverage in sub-Saharan Africa, the timing and frequency of ANC visits remain inconsistent. This contributes to increased pregnancy-related complications among mothers in the region. This study utilised Poisson and Negative Binomial mixed regression with maximum likelihood estimation and generalized estimating equations (GEE) to analyse the factors associated with ANC visits among mothers in Malawi. Secondary data for 7,125 mothers from the individual recode file of the 2024 Malawi Demographic and Health Survey were used. All the analyses were carried out using R software version 4.5.2. The results indicated substantial between-cluster variation of ANC visits, indicating higher chances of observing similar ANC outcomes for mothers staying in the same cluster. The likelihood-based estimation methods yielded efficient estimates with smaller standard errors, p-values, and narrower confidence intervals than the GEE. However, both methods produced estimates with similar directions and magnitudes. On average, mothers in Malawi attend four ANC visits during their pregnancies. Higher ANC attendance rates were observed among mothers aged 20 years and above, those with tertiary education, and those with educated husbands. Conversely, lower rates were found in mothers with at least five children, those from middle wealth households, and those exposed to newspapers. Using alternative statistical estimation methods can enhance the reliability of findings in public health research. The health policy interventions that can utilise shared community cultural values, norms and networks at cluster level, as well as spouse involvement may improve ANC programs in sub-Saharan Africa.