Statins in pneumonia - magic versus science?

Several studies have investigated the use of statins as an adjunct in the treatment of pneumonia, some with conflicting conclusions. The most recent of these large observational studies again suggests statin use is associated with an improved outcome for patients with pneumonia. How best to incorporate these findings into current practice is of great interest. Hidden confounders plague database interrogation and so cast doubt on the real or causal nature of observed associations. Data from large, observational studies must be complemented by smaller biological studies and randomised controlled trials in an effort to complete missing pieces in the biological puzzle of the use of statins in patients with pneumonia.


Introduction
Th e latest study on the therapeutic value of statins in pneumonia adds further support to the association between statin use and improved patient outcomes [1]. Th e role of statins in the management of sepsis has been increasingly studied but remains contentious. Animal studies of statin use in sepsis have shown a signifi cant benefi t with respect to mortality [2] but a plethora of obser vational studies have come to confl icting conclusions [3,4]. Many mechanisms have been postulated to explain why statins may infl uence sepsis [5], and it is possible some of these may be of particular benefi t in pneumonia [6][7][8].
Statins are taken by millions of people around the world and pneumonia represents one of the most common infections seen in clinical practice. Th is provides an opportunity to review a large number of cases of pneumonia that involve coincidental statin use. As would be expected for a drug targeting vascular disease and hyper cholesterolaemia, statin users have diff erent comorbidities and risk profi les to those patients that do not take statins. Th erefore, observational and retro spective studies are potentially confounded by both an unhealthy user eff ect (a cohort often older with more coexisting cardiovascular disease) and the 'healthy user' eff ect [9].

Current studies
Th e recent study by Nielsen and colleagues [1] is a well conducted, large, observational study. Th e authors provide extensive detail on coexisting diseases in each of the cohorts in an eff ort to reveal hidden confounders that plague database interrogation. Th is paper adds another observational study to a large bank that already exists addressing the outcome from pneumonia in statin users [10][11][12][13]. It serves to remind us that while again suggesting a possible therapeutic role for these agents in pneumonia, it does not prove a clear cause and eff ect relationship.
Th e utility of the study is weakened by a number of factors. Firstly, it defi ned 'current use' as one prescription fi lled in the past 125 days. While the results suggest the improvement in pneumonia outcome was associated with current and not former statin users, a huge variation in timing of statin administration in relation to onset of infection is possible within these defi nitions. In addition to the timing of administration, the impact of one statin over another or drug dosage remains unclear. Finally, large registry databases pose other challenges. As was highlighted recently, both the consistent use of a diagnostic category over time and the motivations for recording specifi c diagnosis can further confound the validity of results from large national or administrative databases [14].
It is fortuitous that prevalence has provided the opportunity to observe statin use in pneumonia patients. Th ese observations have provided the impetus to investigate statins as a potential adjuvant therapy for severe lung injury. Two randomised controlled trials are currently underway: Statins for Acutely Injured Lungs from Sepsis (SAILS) and Hydroxymethylglutaryl-CoA reductase in hibi tion with simvastatin in Acute lung injury to Reduce Pulmonary dysfunction (HARP 2).
SAILS will compare 20 mg rosuvastatin to placebo. In contrast, HARP-2 will compare 80 mg simvastatin to

Abstract
Several studies have investigated the use of statins as an adjunct in the treatment of pneumonia, some with confl icting conclusions. The most recent of these large observational studies again suggests statin use is associated with an improved outcome for patients with pneumonia. How best to incorporate these fi ndings into current practice is of great interest. Hidden confounders plague database interrogation and so cast doubt on the real or causal nature of observed associations. Data from large, observational studies must be complemented by smaller biological studies and randomised controlled trials in an eff ort to complete missing pieces in the biological puzzle of the use of statins in patients with pneumonia.
placebo. Whilst these may be equipotent doses in the management of hypercholesterolaemia, it remains unknown what eff ects these agents at these doses will have in acute lung injury. Th ese two drugs have pharmacokinetic diff erences, including lipid solubility, volume of distribution, metabolism and potency. Without an explanation of how statins may benefi t patients with lung injury, predicting the impact of diff ering pharmacology is impossible. Compared to the general population, the pharmacokinetics of atorvastatin diff ers in critical illness, and perhaps with sepsis in particular [15]. Exploring commonly used statin pharmacology in critically ill patients with pneumonia would better defi ne doseresponse relationships on varied biological endpoints and this may provide some valuable insights into mechanisms that could explain an association with improved patient outcomes.
In addition to statin use possibly being associated with a reduction in the severity of infection, a second intriguing possibility is that statins may reduce the risk of infection. A post hoc analysis of the JUPITER trial suggested a modest reduction in the incidence of pneumonia in patients randomised to receive 20 mg of rosuvastatin [16]. A nested case control study has also suggested a reduced incidence of pneumonia in statin users [17]. Th e recent study by Nielsen and colleagues [1] again suggests a decrease in the risk of hospitalization for pneumonia with statin use, but only after adjustment for potential confounding variables. Pneumonia represents an illness with known risk factors and the administration of statins as preventative therapy could be prospectively explored in this patient population in an eff ort to avoid confusion from uncontrolled or unknown factors.

Conclusion
Rather than question what can be learned from observational data while the results of randomised trials are awaited, perhaps we can use all these research methods to better inform our clinical practice. Whilst observational studies to date are far from conclusive, it is hard to see how randomised controlled trials will resolve a complex debate that is still in its infancy. It is more likely that a multipronged approach with retrospective large databases, smaller biological studies and randomised trials will help complete the biological puzzle of statin use and its role in patients with pneumonia.