Affiliations: Vanderbilt University School of Medicine, Nashville, TN,
Lovelace Respiratory Research Unit, Albuquerque, NM
Correspondence to: Carlos G. Grijalva, MD, MPH, 1500 Twenty-First Ave, S Suite #2600, The Village at Vanderbilt, Vanderbilt University Medical Center, Nashville, TN 37212; e-mail: email@example.com
We read with great interest the article by Frost et al1(April 2007), and although we concur in the need for measures to attenuate the effect of a potential influenza pandemic, we believe there might be other explanations for the apparent benefit of statins in the prevention of influenza/pneumonia and COPD-associated mortality. Study cohort members were classified as either statin exposed (moderate and low dose) or unexposed for the entire follow-up (ie, exposure status was fixed). Moreover, for statin users, the time from enrollment to initiation of statin therapy (phase 1) was necessarily free of events and contributed follow-up time in which statins were not taken. The inclusion of this “immortal time” would result in exposure misclassification and may artificially reduce the event rate in the statin-exposed group, producing an apparent benefit of statins.2 The classification of patients who used but did not accumulate 90 days of statin use is unclear. Their inclusion in the unexposed group would result in exposure misclassification, and their exclusion from the statin-exposed group would result in selection of a subgroup of patients (ie, with good tolerance to the drug). Allowing statin use to change after its initiation (time-varying exposure) would reduce exposure misclassification.
Another consideration is the “healthy user” effect, in which statins use could be related to a constellation of healthy behaviors that were not included in the analysis.3Different designs using the same exposure definitions and covariates (eg, nested case control) will not account for this either. As the authors stated, unmeasured confounders could explain the observed benefits. Finally, logistic regression does not account for censoring.4
Data from clinical trials have failed to show benefits of statins in preventing mortality due to nonvascular respiratory causes.5 Meanwhile, the assessment of additional benefits of statins using observational studies requires the careful consideration of known sources of bias.
The authors have no conflicts of interest to disclose.
The author has no conflict of interest to disclose.
We thank Drs. Grijalva, Arbogast, and Griffin for their observation that our study1may have included an immortal time period such that participants were not permitted to die. Immortal time has been clearly demonstrated in the literature2 to cause substantial bias in observational studies when there is a difference in the application of immortal time between the exposed and unexposed groups. In our study, both the exposed and the individually matched unexposed participants were equally subject to immortal periods prior to the initiation of statins (12 months, phase 1) in order to estimate mortality risk differences. A second period (the 90-day follow-up period after the exposed subject initiated treatment with a statin) was included to avoid misclassifying noncompliant users as statin exposed. We believe that immortal time did not bias our results substantially because it was nondifferential.
We do not believe that confounding by a healthy user effect can completely account for such a strong protective effect, especially in preventing deaths associated with COPD. In our proportional hazards analysis, patients not receiving statins were 7.7 times more likely to die from COPD than moderate daily dose users of statins.
A “healthy user” effect may arise because patients who receive a medication may differ when compared to those that do not with respect to comorbidities and other risk factors that may be associated with disease severity and health-care quality, utilization, and access.3–4 In our study, like many other observational studies of the effects of statin use, statin users compared to nonusers were somewhat older and had more comorbidities, such as heart failure and diabetes. The number of comorbidities in addition to number of medications and number of influenza vaccinations (proxy measures for health-care utilization) were adjusted for in our final statistical regression models.
Become a CHEST member and receive a FREE subscription as a benefit of membership.
Individuals can purchase this article on ScienceDirect.
Individuals can purchase a subscription to the journal.
Individuals can purchase a subscription to the journal or buy individual articles.
Learn more about membership or Purchase a Full Subscription.
Institutional access is now available through ScienceDirect and can be purchased at myelsevier.com.
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 3
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a reminder to the email address on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.