Abstract
BACKGROUND: Tonga is a South Pacific Island country with a population of 100,651 (2016 Census). This study examines Tongan infant mortality rates (IMR), under-five mortality rates (U5MR), adult mortality and life expectancy (LE) at birth from 2010 to 2018 using a recent collation of empirical mortality data over the past decade for comparison with other previously published mortality estimates. METHODS: Routinely collected mortality data for 2010-2018 from the Ministry of Health, national (Vaiola) hospital, community nursing reports, and the Civil Registry, were consolidated by deterministic and probabilistic linkage of individual death records. Completeness of empirical mortality reporting was assessed by capture-recapture analysis. The reconciled data were aggregated into triennia to reduce stochastic variation, and used to estimate IMR and U5MR (per 1000 live births), adult mortality (15-59, 15-34, 35-59, and 15-64 years), and LE at birth, employing the hypothetical cohort method (with statistical testing). Mortality trends and differences were assessed by Poisson regression. Mortality findings were compared with published national and international agency estimates. RESULTS: Over the three triennia in 2010-2018, levels varied minimally for IMR (12-14) and U5MR (15-19) per 1000 births (both ns, p > 0.05), and also for male LE at birth of 64-65 years, and female LE at birth 69-70 years. Cumulated risks of adult mortality were significantly higher in men than women; period mortality increases in 15-59-year women from 18 to 21% were significant (p < 0.05). Estimated completeness of the reconciled data was > 95%. International agencies reported generally comparable estimates of IMR and U5MR, with varying uncertainty intervals; but they reported significantly lower adult mortality and higher LE than the empirical estimates from this study. CONCLUSIONS: Life expectancy in Tonga over 2010-2018 has remained relatively low and static, with low IMR and U5MR, indicating the substantial impact from premature adult mortality. This analysis of empirical data (> 95% complete) indicates lower LE and higher premature adult mortality than previously reported by international agencies using indirect and modelled methods. Continued integration of mortality recording and data systems in Tonga is important for improving the completeness and accuracy of mortality estimation for local health monitoring and planning.