Geographical variation in coronary revascularisation rates.

Geographical variation in coronary revascularisation rates Sir-I found the paper by Black et all on the geographical variation in coronary re-vascularisation rates extremely interesting. This is a vital area if the NHS is to try to achieve a more equitable service. I am impressed by the tremendous size of the study embarked upon to investigate these variations and the wealth of information available for analysis. The authors report a negative correlation between SMRs for coronary heart disease and revascularisation indicating inequities in service provision. However, this conflicts with the positive correlation between social deprivation indices and revascularisation. This implies that higher intervention rates are associated with districts with lower SMRs and with more deprived districts. The authors argue that this could be confounded by the close proximity of specialist centres to more socially deprived districts. I believe there are important possible explanations for this relationship that have not been discussed. The first is related to the statistical analyses. Pearson correlation coefficients are quoted for relationships between rates for coronary artery bypass grafting (CABG) and percutaneous transluminal coronary angioplasty (PTCA) and both the Jarman social deprivation index and the Department of the Environment social index (DoE). These show CABG to have a significant correlation with both the Jarman and DoE indices, and when all revascularisations are considered together there is a significant correlation with the DoE index. The re-vascularisation rates and DoE index are illustrated in figure 2 of the paper. On closer inspection, however, it is evident that these relationships could be due to one outlying district. I have entered the data for CABG rates obtained from figure 2 and reanalysed the data removing this district. This shows no significant correlation between CABG and the DoE index. It would also lead to non-significant correlation between rates of all revascularisations and the DoE index. Evidently this unusual district requires further investigation. A more appropriate analysis would be a binomial or Poisson regression model.2 This could adjust for differences in district population size which is not allowed for in the correlation analysis. A further issue relates to the indices used as indicators of coronary heart disease morbidity. The Jarman and DoE indices combine both direct measures of material deprivation, for example, unemployment and overcrowding , and indirect measures of material deprivation, for example, lone pensioners, single parents, and ethnicity. Studies34 have shown that these two indices correlate less well with measures of …