Do Statins Protect Against Pneumonia?

Statins are used to treat high cholesterol.  Recently, researchers have observed a reduction in infections such as pneumonia among patients taking statins. While some of these findings have prompted investigators to propose several biological mechanisms for the observed protective effects of statins, others have speculated that this effect may be the product of a healthy person bias, in which healthier patients, who are less likely to develop infections, are more likely to be prescribed a statin. The objective of this study is to determine if statins protect patients against pneumonia.

We obtained all Medicare claims files, enrollment information, and Part D prescription drug events for patients hospitalized with an acute myocardial infarction (AMI) in 2007 and 2008.

We estimated ordinary least squares models (OLS) for the full sample (124,695) and for the subset of patients who survived 1 year post-discharge (N=104,870). We also estimated Cox proportional hazards (Cox) models for the full sample to examine the time to pneumonia. All models included a variety of covariates including patient demographics, medications, diagnoses, length of hospital stay and out-of-pocket drug costs. The outcome variable is a diagnosis of pneumonia during the 1 year post-AMI. Our independent variable of interest is whether or not the patient filled a statin prescription within the first 30 days post-discharge.

For our instrument, we defined local treatment areas around patient ZIP codes. We create local area practice style measures for alternative size definitions based on threshold numbers of patients (50, 100, 150, 200) living within a specified driving time of each ZIP code. For each size our instrument was the ratio of the actual to the predicted statin treatment rate for all subjects in the area.

For our full sample, 8% (N=9,975) developed pneumonia within 1 year post-discharge, and 54% (N=67,335) filled a statin prescription within the first 30 days post-discharge. Those prescribed statins were less likely to develop pneumonia in the following year. For the full sample, the statin coefficient was -0.01 (p=0.004), and for the one-year survivors, -0.004 (p = 0.012). For the Cox model the hazard ratio was 0.919 (p = 0.0004), indicating that the time to pneumonia is greater for those taking a statin.

Using 2-stage least squares and our 4 different treatment areas, our first-stage F-statistics ranged from 981.46 to 3072.47 for both samples.  None of the statin coefficients were significantly different than zero. For the full sample, statin coefficients ranged from -0.003 to 0.002 (p>0.9).  For the survivors, the coefficients ranged from -0.004 to -0.008 (p>0.5). For the Cox model, the hazard ratios ranged from 0.93 to 0.98 (p>0.9).

We demonstrate that in one population eligible for statin use -- AMI patients -- the protective effect of statins against pneumonia may reflect a healthy-user bias. For those patients whose statin treatment is affected by the rate of stain treatment in their area, increasing statin treatment will not decrease infections.  Given that statins have many known side effects, statins should not be prescribed to patients solely to prevent infections.