TRANSITIONAL PROBABILITY AND TIMING OF FIRST SEXUAL INTERCOURSE AND FIRST BIRTH AMONG NIGERIAN WOMEN

Abstract

Age-at-first sex and age-at-first-birth have been identified as important markers of changes in social and health needs especially in the lives of women and an understanding of these important reproductive health events can provide clear understanding of how individual progress throughout these stages. Age-at-first-sex is an important indicator of risk of pregnancy as well as STls and HIV/ AIDS among young adults. The interval between these reproductive events has also called for consideration. The wider the interval between first sex and first marriage, for instance, exposes a woman to having multiple sexual partners. The interval between first sex and first birth also suggests possible use of contraceptives and intentional delay in childbearing among the couples given several personal and community level factors. These reproductive health events have survival times. Although many studies settle for the Cox proportional hazard model for estimating survival times but parametric methods have been described as having better estimates. Both parametric and non-parametric methods have been suggested for fitting these reproductive health events and even for the parametric methods, graphical and analytical methods have been used. The main objective of this study was to model the baseline hazard of age at first sex, age at first birth and the interval between them using the Weibull distribution. Data for the study was obtained from the Nigeria DHS 2013 for 38948 women of which 31075 (84.4%) ever had sex among whom 25370 (81.6%) ever gave birth. Exploratory data analysis, curve fining, and model fitting were employed for the analyses. Kaplan-Meier estimates of the survival function was estimated and substituted into the Log cumulative hazard plot to check the suitability of Weibull distribution to model the hazard function. Stgenreg procedure was used for analytical fitting of the hazard functions through maximum likelihood estimation. Fixed and random effect models for interval between first sex and first birth were fitted using streg and stmixed procedures respectively. The results showed that the age at first sex, age at first birth and their interval changed with birth cohort. Specifically, older women had wider interval of sex debut and first birth compared to younger women. Weibull distribution was appropriate for the baseline hazard of age at first sex, age at first birth and the interval between them. The median age at first sex obtained from the log cumulative hazard plot was 16.61 years (95% CI: 16.43-16.80) which was slightly higher than the Kaplan-Meier estimate of J 6 years (95% CI: 15.94-16.05). The estimate of median age at first sex from the MLE of Weibull parameters was 16.97 years (95% Cl: 14.97-19.23). The median age at first birth for both graphical and MLE estimates were 19.47 years (95% CI: 19.23-19.7 2) and 19.60 years (95% CI: 17.02-22.56) respectively. which are comparable to the KM estimate of 19 years (95% CI: 18. 94-19.06). The KM estimate for first sex-first birth interval was 2 years {95% CI: 1.97-2.03), which is also comparable to the median estimate from the Weibull estimation of 2.12 years (95% Cl: 2.06 -2.18). Random effects Weibull model showed that age at first sex, education. religion, women's birth cohort, contraceptive use, marital status, and ever terminated a pregnancy were significant factors of the interval between first sex and first birth.