Geographic variation in invasive cardiac procedure rates in New South Wales, Australia

Abstract

Objectives

This study explores the contribution of socio-demographic factors to the geographic variation in coronary artery bypass graft (CABG) and percutaneous coronary intervention (PCI) procedure rates in New South Wales, Australia.

Methods

With the utilisation of small area analysis and regression model techniques, the possible explanatory factors of the local government area (LGA) level variation in CABG and PCI rates in terms of coronary artery disease prevalence, supply and access to health-care services, socio-economic status and ethnic origin of the people were examined.

Results

Multivariate regression results show that distance to hospitals is negatively associated with LGA-specific CABG and PCI rates. The CABG rate is lower and PCI rate is higher in LGAs with higher percentages of European-born residents. Higher proportions of surgeries were recorded for relatively younger people in the lowest socio-economic LGAs.

Conclusions

The focus should be on educating people in the lowest socio-economic LGAs in lifestyle management in order to minimise surgical interventions at a younger age.

Appendix

Small area analysis formulae

Mean (prevalence) $\mu = \frac{{\Sigma {w_i}{x_i}}}{{\Sigma {w_i}}}$, where no weight w_i = 1 for i

Standard deviation $\sigma = \sqrt {\frac{1}{n}\sum\limits_{i = 1}^n {{w_i}{{\left( {{x_i} - \overline x } \right)}^2}} }$

Extremal quotient = Maximum/Minium

Coefficient of variation CV = (standard deviation/mean) * 100

Unweighted coefficient of variation (CVU)

CVU = CV where w_i = 1

Weighted coefficient of variation (CVW)

CVW = CV with specific w_i weights

Systematic component of variation (SCV)

$$ SCV = \left( {1/k} \right)\left[ {\sum\limits_{i = 1}^k {{{\left( {\left( {Oi - Ei} \right)/Ei} \right)}^2} - \sum\limits_{i = 1}^k {\left( {1/Ei} \right)} } } \right] * 1000 $$

where k is the number of LGAs, O_i is the observed number of admissions in LGA i, and E_i is the expected number.

Chi square ${\chi^2}{\text{ }} = {\text{ }}\frac{{{{\left( {{{\text{O}}_i} - {{\text{E}}_i}} \right)}^2}}}{{{{\text{E}}_{\text{i}}}}}$ O_i = observed value, E_i = expected value.

CVA = analysis of variance estimate of the ‘true’ coefficient of variation σ and the confidence limits

$$ {\text{CV}}{{\text{A}}^2} = \frac{{{\chi^2} - \left( {k - 1} \right)*{\text{MAF}}}}{{\mu *\left( {k - 1} \right)*{n_0}}} $$

• MAF: multiple adjustment factor, k: number of Local Government Areas, ${{\text{n}}_{\text{0}}}{\text{ = }}\overline {\text{u}} {\text{ - s}}_{\text{n}}^{\text{2}}{\text{/}}\left( {{\text{k*}}\overline {\text{u}} } \right)$, where $\overline {\text{u}}$ and ${\text{s}}_{\text{n}}^{\text{2}}$ are mean and variance of the number of people per area.

CVA confidence limits ${\text{CV}}{{\text{A}}^2}\pm \frac{{1.96*\sqrt {4\chi 2 - 2\left( {k - 1} \right)} }}{{\mu *{n_0}}}$

Source: (Diehr et al. 1993), p:YS50, YS53.