Fig. 3

Estimates of under-five mortality rate (U5MR) by sex and data source^a, Tonga, 2010–2018. ^a Estimates by international agencies and previous published reports are presented with present study empirical estimates for comparison purposes only and are not included in trend analysis. ⁽¹⁾ Estimates from the present study based on reconciled routinely collected data for 2010–2012, 2013–2015, and 2016–2018 are shown with 95% confidence intervals at years “2010.75”, “2013.75”, and “2016.75” respectively. ⁽²⁾ Source: [9] Range of estimates for 2005–2009 based on reconciled data is shown at year 2009. The lower and upper limits of the range represent estimates adjusted by two-source and three-source capture-recapture analysis. ⁽³⁾ Source: 2016 Annual Report [36], based on Reproductive Health Services community nursing reports. ⁽⁴⁾ Source: 2011 Census [28] and 2016 Census [22]. 2011 estimate (both sexes) is shown offset at year “2010.75”. ⁽⁵⁾ Source: 2012 Survey [38]. U5MR by sex refers to the 10-year period preceding the survey in 2012, confidence intervals are not available. U5MR (both sexes) refers to the five-year period preceding 2012 with 95% confidence interval. Estimates are shown offset at year “2009.75”. ⁽⁶⁾ Source: 2019 Survey [37]. U5MR calculated for the 10-year period preceding the survey in 2019. 95%CI are not available for U5MR by sex*. Estimates shown offset at year “2014.5”. ⁽⁷⁾ Source: United Nations Inter-agency Group for Child Mortality Estimation [39] Sex-specific U5MR estimates based on B-splines regression (BSR) model. Total U5MR estimate based on Bayesian B-splines bias-adjusted model. Annual estimates over 2010–2018 are shown with 90% uncertainty intervals offset at + 0.25 per year. 90% uncertainty intervals (UI) are not comparable to 95%UI and CI⁽⁸⁾ Source: [41]. Estimates modelled from spatiotemporal and Gaussian process regressions. *95% uncertainty intervals not available by sex for years 2011–2018. 95%UI are not comparable to 95%CI