Table 2

Bayesian multivariable local municipality mortality risk factor analysis, South Africa, 2007
Indicator Bivariate analyses (unadjusted) Multivariable model (adjusted)
Standard Poisson model i BYM CAR ii spatial Poisson model
RR (95% CI) p-value RR (95% BCI iii) p-value
Composite service delivery index score iv 2.90 (1.99,4.20) <0.001 1.84 (1.43,2.34) <0.001
High local municipality income inequality 1.23 (1.04,1.46) 0.015 1.14 (1.02,1.29) 0.024
Low-medium district population density 1
 High density vi – non-metropolitan 1.13 (1.00,1.28) 0.050 0.97 (0.87,1.09) 0.648
 High density - metropolitan municipality 0.66 (0.52,0.83) <0.001 0.73 (0.65,0.82) <0.001
District antenatal HIV sero-prevalence 1.02 (1.01,1.03) <0.001 1.02 (1.02,1.03) <0.001

i: robust standard errors to adjust for local municipality “cluster”.

ii: incorporated an unstructured local municipality random effect and a structured normal CAR spatial random effect.

iii: Bayesian credibility interval.

iv: increasing score indicates increasingly poor service delivery (square transformation used due to violation of linear assumption in Poisson framework).

v: lower tertiale.

vi: upper quartile and adjusting for metropolitan.

Sartorius and Sartorius

Sartorius and Sartorius International Journal of Health Geographics 2013 12:8   doi:10.1186/1476-072X-12-8

Open Data