Statins and outcomes in patients with pneumonia

We read with interest the article on the use of statins for patients with pneumonia.¹ Although an important addition to the literature, several issues limit this article's usefulness.

The choice of an outcome measure combining in-hospital mortality and admission to intensive care is curious for a prospective study with such a rich clinical database. Previous research shows that 30 day mortality is largely pneumonia related,² and from our recent study,³ 33% of deaths were after discharge and before 30 days. A preferred way to examine the impact of statins on sepsis would be to examine sepsis-related outcomes (vasopressor use, incidence of severe sepsis, or mortality alone).

The findings that age >65 years, ischaemic heart disease, and using levofloxacin are protective, or that PSI ⁴ class III has an odds ratio of 2.45, have not been previously reported and seem implausible. Inclusion of younger subjects who are less often prescribed statins, and are at much lower risk for mortality, reduces the ability to see an effect. A specific list of variables entered into the final model would be informative to assess potential multicollinearity.

We also believe it is inappropriate to label the odds ratios as “potential harm” or “potential benefit” as all of the 95% confidence intervals include 1.0. These odds ratios and 95% confidence intervals show no association, not potential “harm” or “benefit.”

Overall, the study suffers from faults in the study analyses, notably a failure to assess interactions and multicollinearity in the face of counterintuitive results, undermining the contention that previous findings may be attributable to healthy user bias. Future research needs to adjust for factors associated with healthy user bias, patient frailty, and other forms of potential confounding. Only well designed randomised controlled trials will be able to determine finally whether statins have a role in the management of serious infectious diseases.