Modeling Age Patterns of Under-5 Mortality: Results From a Log-Quadratic Model Applied to High-Quality Vital Registration Data

Abstract

Information about how the risk of death varies with age within the 0-5 age range represents critical evidence for guiding health policy. This study proposes a new model for summarizing regularities about how under-5 mortality is distributed by detailed age. The model is based on a newly compiled database that contains under-5 mortality information by detailed age in countries with high-quality vital registration systems, covering a wide array of mortality levels and patterns. It uses a log-quadratic approach in predicting a full mortality schedule between ages 0 and 5 on the basis of only one or two parameters. With its larger number of age-groups, the proposed model offers greater flexibility than existing models in terms of both entry parameters and model outcomes. We present applications of this model for evaluating and correcting under-5 mortality information by detailed age in countries with problematic mortality data.

Keywords: Age patterns of mortality; Indirect methods; Model life tables; Neonatal mortality; Under-5 mortality.

Fig. 1

Relationship between age-specific mortality rates (_nM_x) and the probability of dying between age 28 days and 5 years (q(28d,5y)) for each of the first four weeks of life (₇M_0(d), ₇M_7(d), ₇M_14(d), and ₇M_21(d)) in the harmonized Under-5 Mortality Database (U5MD), for both sexes combined

Fig. 2

Relationship between q(x) and q(5y) for x = 7d, 28d, and 12m, with observed values in the final U5MD versus values predicted using the log-quadratic model with k = 0, −1, or +1, for both sexes combined

Fig. 3

Effect of varying q(5y) versus k on q(x) and _nM_x in the log-quadratic model, for both sexes combined

Fig. 4

Observed and predicted values of q(x) and _nM_x for Finland, 1933, for both sexes combined

Fig. 5

Observed and predicted values of q(x) and _nM_x for Belgium, 1949, for both sexes combined, with 95% confidence intervals

Fig. 6

Observed and predicted values of q(x) and _nM_x for selected Demographic and Health Surveys, for both sexes combined. In panel c, k* is estimated on the basis of observed q(x) points, excluding ages 9 months to 21 months.

Fig. 7

Levels of neonatal (q(28d)), infant (q(12m)), and under-5 (q(5y)) mortality rates for Jordan in 2015, for both sexes combined, with 95% confidence intervals. VR = vital registration; DHS = 2017–2018 Demographic and Health Survey, with mortality values centered in 2015; obs. = unadjusted VR values; adj. = adjusted VR values using the log-quadratic model with the observed VR-based value of q(28d,5y) and k = 0 as inputs.