Reliability of self-reported data from the 2011 Ja-Kids birth cohort study: demographic, social, obstetric history, medical and outcome indicators.

Abstract

The Ja-Kids Longitudinal Study (JA-Kids) aims to improve the health and development of Jamaican children by identifying social, demographic, environmental and clinical factors that help or hinder these processes. As clinical indicators relied on maternal reporting, we aim to evaluate the quality of the self-reported data. Women were recruited across Jamaica during pregnancy or at delivery from July 1-30 September 2011. Indicators were compared between women recruited while pregnant and at delivery to understand possible differences between the sub-populations. Variables reported more than once between pregnancy and delivery were assessed to evaluate level of agreement (reliability). Clinical indicators from the literature were contrasted with study findings to determine how maternal reporting align with published prevalence (validity). Intra-class correlation and the kappa (κ) statistic were used to assess reliability while chi-squared, Fisher's-exact or students-t were used to compare differences over time; p values ≤0.05 were considered statistically significant. Women recruited during pregnancy (n = 3970) were younger, less parous and possibly more socially disadvantaged than those recruited at delivery (n = 5803). Socio-demographic and selected clinical indicators showed good to moderate (0.421 < κ < 0.681) reporting consistency between pregnancy and delivery for previous C-section (κ = 0.681), pre-existing diabetes mellitus (κ = 0.616) and prior twin gestations (0.580). Most conditions however showed only fair agreement (0.21 < κ < 0.40) including previous gestational hypertension (κ = 0.387), asthma (κ = 0.365), premature rupture of membranes (κ = 0.324), eclampsia (κ = 0.257) and essential hypertension (κ = 0.213). Infectious conditions had poor reliability. Prevalence rates for most conditions, except sickle cell disease, were lower than the published literature. Complications and outcomes which were well defined for women were better reported than those requiring clinical judgment (e.g. prior C-section versus specific hypertensive disorders of pregnancy). NCDs with only episodic acute effects were not well reported, e.g. asthma, hypertension and sickle cell disease. Maternal reporting of pregnancy complications needs to be interpreted with caution.