Methods in adult health and mortality

This paper proposes a new method to estimate disability transition rates from national cross-sectional health surveys. The proposed method estimates age-specific transition rates from cross-sectional data according to well-documented longitudinal age-specific health transition rates of other populations, used as standards, and the proportion of health and unhealthy individuals by age, reported in cross-sectional datasets. In order to estimate healthy life expectancy, this paper makes use of most recent Brazilian health survey data The preliminary results indicate that the estimated disability transition rates are consistent with the current literature. Moreover, the estimated parameters for the simple model specification seem to produce very reliable results. In 1998, 2003 and 2008 the estimated life expectancy – with and without any disability – do not show significant statistical differences from other estimates, produced by other methods.
A second exercise will be conducted by estimating the parameters including covariates: sex, race and education and to estimate differentials in healthy life expectancy.

The individual rate of aging is defined as the relative derivative of one's risk of death with respect to one's age. The b-hypothesis, formulated by Vaupel (2010), postulates that all humans share the same rate of aging. In order to check this hypothesis given the existing aggregate data on human mortality, we present several statistical approaches, their advantages and shortcomings, as well as some preliminary conclusions.

Event history analyses make the explicit assumption of independence between censoring and event. Under this hypothesis right censoring due to survey time does not create a selection bias. However, when censoring is not independent from the event of interest (e.g. migration in relation to death) then results suffer from potential bias. This paper presents a model to deal with non-independent right- as well as left-censoring, when the same determinants may cause in-migration, out-migration and mortality. The model follows the rationale of two-stage regression models controlling for selection biases to control for both observed and unobserved heterogeneity in migration. The method is applied on longitudinal adult mortality data collected by the APHRC Nairobi Urban Health and Demographic Surveillance System (NUHDSS), situated in two Nairobi slums – Korogocho and Viwandani – where circular migration is high. Results confirm selection for both out- and in-migration. The method produces higher adult mortality rates than raw estimates. After controlling for migration effect, median age at death from age 15 would be 46 year old (against gross estimate of 70) for females, and 52 (against 70) for males. Migration is most likely an important strategy adopted by slum-dwellers to avoid health risks attached to slum environment.

This paper is addressing the modelling of old age mortality and its
dependence of factors earlier in life. We argue for alternatives
to the widely used proportional hazards (PH) model, especially Cox
regression. There are several reasons for this. First, it is well known
that old age mortality very often is well described by the Gompertz
distribution. Second, accelerated failure time (AFT) models can be
expressed as linear models, which is important when interest lies in the analysis of mediating effects in the analysis of the impact of early-life factors on old-age mortality. Third, the results of an AFT model fit is easier and more intuitive to interpret in tems of years lost or gained,
compared to the PH model fit which reports relative risks. Fourth, contrary to "common knowledge", the family of Gompertz distributions is both a collection of PH families and a collection of AFT families, which we demonstrate in the paper. For instance, Kleinbaum and Klein (2005), in their text book on survival analysis, writes: "The Gompertz model is a parametric PH model but not an AFT model". This mistake is reiterated by other authors.