Assessing factors influencing adolescent sexual debut in Nigeria: a multi-cluster survival analysis approach

Abstract

Introduction: Early sexual debut is an area of concern in Nigeria with implications for reproductive health. Methods: This study addresses this by proposing a more effective survival model-one that incorporates both independent and identically distributed (IID) and Besag intrinsically conditional auto-regressive (ICAR) random effect priors, using a generalised additive model that accounts for both individual and spatial influences on age at first sex. We analyse data from the 2018 Nigerian NDHS survey and compare our model with existing models: a model without the cluster frailty effect, a model that ignores the Besag ICAR and includes the IID, and a model that ignores the IID and includes only the Besag ICAR. Results: Our approach, which combines independent and spatial random effects, outperforms others based on statistical criteria (Deviation Information Criterion and the Widely Applicable Information Criterion). Discussion: As shown in this study, the proposed model effectively captures the complexity of age at first sex. In addition, visualisations reveal significant geographic and social clusters with high rates of early sexual debut in Nigeria. These findings highlight the importance of considering multi-level clustering to better understand and promote healthy sexual behaviour among young Nigerians through targeted interventions.