Overlap Analysis of the West of Scotland Coronary Prevention Study (WOSCOPS)
To the Editor:
The results of the above article are being used for pharmaceutical promotion for prescribing pravastatin.1 The overlap analysis in this article is interpreted as suggesting that pravastatin reduces mortality over and above its effect on serum cholesterol. We believe that this assertion may not be justified.
The case comparison might not be valid in this analysis. The sample selection was based on on-treatment rather than baseline LDL cholesterol for both placebo and pravastatin groups. The baseline LDL cholesterol levels were clearly different between the 2 groups. The pravastatin group started with higher LDL cholesterol levels, and therefore those in this group were of a greater cardiovascular risk and had more room for risk reduction with active treatment. In contrast, the placebo group had low baseline LDL cholesterol levels, suggesting inherently lower baseline cardiovascular risk. It is not difficult to appreciate that the magnitude of any given favorable effect is far greater when interventions are directed at high-risk groups. The overlap analysis compared 2 groups with very different baseline risks.
To conclude that pravastatin has additional effect over and above its effect in lowering cholesterol, the analysis would need to be controlled for the change in lipids rather than for the on-treatment lipids values. Because the study failed to find a direct correlation between the reduction in LDL cholesterol and mortality, this confounder perhaps cannot be accounted for statistically.
- Copyright © 1999 by American Heart Association
West of Scotland Coronary Prevention Study Group. Influence of pravastatin and plasma lipids on clinical events in the West of Scotland Coronary Prevention Study (WOSCOPS). Circulation. 1998;97:1440–1445.
We thank Drs Lim and Yee for their comments on our overlap analyses. In this comparison of the 2 treatment arms of the study, we asked the question whether subjects who lowered their LDL cholesterol with pravastatin achieved the risk usually associated with that lower level. We agree that those treated with the drug would initially have been at a high baseline risk due to their high LDL cholesterol level. It was anticipated that over 5 years, the risk in the pravastatin-treated subjects would approach that seen in individuals who habitually had the lower LDL level. The fact that treatment reduced the risk to a value significantly lower than that of matched controls is both remarkable and unexpected.
The analysis proposed by the correspondents, examining the change in LDL, would be confounded fundamentally by the drug-treatment effect and is therefore inappropriate.