Tracking progress towards Sustainable Development Goal 3.2 in Kenya using time series models

Sustainable Development Goal (SDG) 3.2, which aims to reduce under-five mortality rate (UFMR) below 25 deaths per 1000 live births by 2030, is still a crucial target for improved child survival in sub-Saharan Africa because UFMRs are still high and progress has stalled in recent years.

This study aimed to model the possibility of reaching the SDG 3.2 target by 2030 and evaluate trends in under-five mortality in Kenya.

Three models: autoregressive integrated moving average (ARIMA), autoregressive fractionally integrated moving average (ARFIMA) and hybrid were fitted to annual national under-five mortality data from 1995 to 2022. Automated model selection showed ARIMA (0,2,1) as the best fitting model from information criteria, predictive accuracy and residual diagnostics. The model was tested with mean absolute error, root mean square error, mean absolute percentage error and tested against the 80/20 train-test split.

Kenya's UFMR has been slightly declining over the course of the study, but the ARIMA projection indicates that the rate of fall is slowing. By 2030, the UFMR is expected to be 27.8 deaths per 1000 live births (95% prediction interval (PI) 25.2 to 30.3), over the SDG 3.2 goal level (signifying an increase in predicted uncertainty). The upper bound of humanity's real 95% PI still far exceeds the aim, even as the lower bound has started to move closer. Kenya would require an accelerated annual decline in roughly 2.43 fatalities per 1000 starting in 2023 much higher than trends seen in the recent past to meet SDG 3.2.

Kenya's UFMR has significantly decreased; however, the SDG 3.2 target might not be met by 2030 without more initiatives. To accelerate progress, it will be essential to improve mother and child health services, increase community-level interventions, address social injustices and employ more focused county-specific strategies. Using additional high-quality data and improved modelling tools could enhance child mortality monitoring and prediction in the future.