The Times (6th September 2007) has reported a poster from the European Society of Cardiology Congress. The poster itself is available from the Wall Street Journal website. The headline read “switch to cheap statins raises risk of heart attack or stroke”. So is this something to worry about?
What happened in this study?
The authors (all Pfizer staff plus a statistician from biostatistics company) analysed the GP records of 2511 patients switched from atorvastatin to simvastatin and matched them with 9009 controls. Records were obtained from a UK primary care database (THIN). Importantly, the last data collection was June 2005. There were some key imbalances in baseline characteristics between switched patients and controls – for example, total cholesterol was 0.5mmol/L higher and there were more smokers in switched patients. There were also baseline differences in other cardiovascular drugs prescribed to the different groups.
What does this study claim?
Switching treatment was associated with a statistically significant increase in the risk of death or major cardiovascular events (myocardial infarction (MI), stroke or coronary revascularization) compared with patients who did not switch (hazard ratio (HR) 1.30, 95%CI 1.02 to 1.64; P = 0.030). Although this P value achieves the conventional level of statistical significance, this result could have occurred by chance with a marginally greater probability than the likelihood of throwing a double six in a board game – and we’ve all seen that happen!
Looking at these results in more detail, there was no statistically significant difference between the two groups in the rate of death. The set of outcomes labelled “major cardiovascular events” occurred more frequently in the switch group but looking at these outcomes individually, only stroke occurred statistically significantly more frequently in the switch group: there was no difference in rates of MI or revascularisation. Patients who switched were also more likely to stop taking their statin than those who did not switch (adjusted HR for discontinuation 2.15, 95%CI 1.96 to 2.36, P<0.001)
The reasons for switching statins in this cohort were not available, nor were the doses of atorvastatin or simvastatin prescribed. These are two very important factors which could affect the results. Since the time of this study (last data collected June 2005) pre-dates NHS switching programmes, it’s reasonable to assume that the switches here involved people who were non-responders, poor compliers, side effect sufferers etc, and so likely to discontinue treatment. These are not the sort of people who have been switched as a result of the Better Care Better Value indicators, the Moon & Bogle BMJ editorial and the work of medicines management teams in PCTs.
This type of study is inherently prone to bias and confounding – not least because of the identified imbalances in baseline characteristics. As the report itself indicates these results should be hypothesis generating. While the Times article does mention this, other reports have not, perhaps because, frankly, this wouldn’t then be newsworthy.
We hope somebody somewhere has already commissioned a high quality observational study of a potentially better validated UK database (e.g. GPRD) looking at event rates in switched patients over the last 12 months (although a longer time period would give a more statistically powerful result, when those data are available). In the meantime this study fails to provide adequate evidence that atorvastatin has advantages over generic alternatives which are less costly to the NHS and offer value for public money.
Readers may also wish to read the NPC blog article on the ASPEN study for another interesting study involving atorvastatin, but, curiously, one that has not received as much publicity as this one.