The lesser known effects of statinsBMJ 2006; 333 doi: http://dx.doi.org/10.1136/bmj.39024.513218.BE (Published 09 November 2006) Cite this as: BMJ 2006;333:980
Benefits on infectious outcomes may be explained by “healthy user” effect
- Reimar W Thomsen, senior research fellow ()
Statins reduce the risk of vascular events and are also cost effective.1 In recent years, non-randomised studies have linked statins with a list of biologically diverse actions, indicating that statins have pleiotropic effects.2 3 For severe infections—such as pneumonia, sepsis, or bacteraemia—at least six studies have linked statin use with decreased (to one third or less in some reports) risks of severe sepsis or death; this has caused great excitement among infectious disease and intensive care physicians.4 5
In this issue of the BMJ, Majumdar and colleagues report outcomes in 3415 Canadian patients with community acquired pneumonia.6 They found that statins slightly reduced the risk of in-hospital mortality or admission to intensive care (crude relative risk 0.80). However, statin users may be “healthy users,” because younger, healthier, better educated, and socioeconomically more privileged people may be more likely to receive preventive treatments than less privileged frail people.7 After controlling for measures of the “healthy user” effect (including up to date immunisations, quitting smoking, and walking unaided) and a propensity score for statin use, the relative risk among statin users increased to 1.10 (95% confidence interval 0.76 to 1.60).
Why were the findings by Majumdar and colleagues different from earlier studies? Their study was population based and thus likely to represent statin users in daily clinical practice, rather than those followed by university hospital specialists. Furthermore, unmeasured socioeconomic confounding is less likely to occur within the Canadian state sponsored healthcare system than, say, in the United States. Their results corroborated the findings of unaltered short term mortality among statin users with bacteraemia reported in Denmark, which has a similar healthcare system.8 Finally, the investigators measured and controlled for a wide range of healthy user markers, which further reduced the likelihood of residual confounding.
Does this mean that we can write off any additional benefits of statins? Not entirely, as the Canadian study does have limitations. Of note, statin users had a lower risk of admission to intensive care or death (crude relative risk 0.80), despite being older, having more comorbidities, and using more drugs. It is odd that controlling for these variables lowered the apparent protective effect instead of augmenting it. This could be an artefact of using the composite endpoint of risk of admission to intensive care or death, as older age and more comorbidity may “protect” against admission to intensive care.6 Controlling for severity of pneumonia among statin users decreased the protective effect further. Pneumonia may have been less severe in statin users because they are more likely to come to the attention of the healthcare system than non-users. Alternatively, statin use itself may affect severity, in which case severity should not be controlled for. In addition, the analysis included only in-hospital mortality. In patients with severe infections, the beneficial effect of statins may become apparent after the first few weeks of use, when the protective effect against sustained vascular events triggered by inflammation may be the greatest.8
Nevertheless, in an area susceptible to publication bias, Majumdar and colleagues should be congratulated for publishing an important “negative” study. The current abundance of “positive” observational studies might tempt drug companies to promote statins on the basis of their presumed diverse beneficial effects, without supporting evidence. This would be unwise, as is discussed in another paper published online in the BMJ (10.1136/bmj.39006.531146.BE).9 We have learnt from the lessons of hormone replacement therapy and antioxidants (when some doctors and drug companies prematurely recommended drugs on the basis of positive observational studies), that therapeutic recommendations should not be based solely on animal studies, plausible biological mechanisms, and findings from observational studies. Large meta-analyses of trials have shown that serious adverse events with statin therapy are rare.10 Currently, for patients admitted to hospital with severe infections, it seems sensible to continue statins in those already receiving them11 until convincing results from randomised trials prove otherwise.
Hospital admissions for sepsis and pneumonia are on the rise in ageing Western populations,12 and positive results from statin therapy trials would have important public health implications, given the availability and relatively low cost of these drugs.5 Such trials should enrol not only selected groups of young and otherwise advantaged patients (as has often been the case for sepsis, for example) but also those who bear most of the disease burden—elderly patients with multiple morbidities. Well designed observational studies may also contribute to our understanding of the issue if they eliminate or quantify biases by collecting data on confounding factors and applying modern epidemiological techniques, including propensity scores and sensitivity analyses.7 Long term observational studies of statin users examining cardiovascular outcomes after infection would be especially useful.
Meanwhile, clinicians might be wondering what protects “healthy” patients with pneumonia who take statins from adverse outcomes, if it is not statins. Because most frail patients will not be admitted to high tech intensive care units in specialised institutions or enrolled in costly trials,13 observational studies will remain essential to shape strategies to improve the outcome of these severe infections.
Competing interests: None declared.