Prior event rate ratio adjustment for hidden confounding in observational studies of treatment effectiveness: a pairwise Cox likelihood approach

Observational studies provide a rich source of information for assessing effectiveness of treatment interventions in many situations where it is not ethical or practical to perform randomized controlled trials. However, such studies are prone to bias from hidden (unmeasured) confounding. A promising approach to identifying and reducing the impact of unmeasured confounding is prior event rate ratio (PERR) adjustment, a quasi‐experimental analytic method proposed in the context of electronic medical record database studies. In this paper, we present a statistical framework for using a pairwise approach to PERR adjustment that removes bias inherent in the original PERR method. A flexible pairwise Cox likelihood function is derived and used to demonstrate the consistency of the simple and convenient alternative PERR (PERR‐ALT) estimator. We show how to estimate standard errors and confidence intervals for treatment effect estimates based on the observed information and provide R code to illustrate how to implement the method. Assumptions required for the pairwise approach (as well as PERR) are clarified, and the consequences of model misspecification are explored. Our results confirm the need for researchers to consider carefully the suitability of the method in the context of each problem. Extensions of the pairwise likelihood to more complex designs involving time‐varying covariates or more than two periods are considered. We illustrate the application of the method using data from a longitudinal cohort study of enzyme replacement therapy for lysosomal storage disorders. © 2016 The Authors. Statistics in Medicine Published by John Wiley & Sons Ltd.

[1]  W. Henley,et al.  The effectiveness and cost-effectiveness of enzyme and substrate replacement therapies: a longitudinal cohort study of people with lysosomal storage disorders. , 2012, Health technology assessment.

[2]  Nils Lid Hjort,et al.  On inference in parametric survival data models , 1992 .

[3]  Gianluca Baio,et al.  Regression discontinuity designs: an approach to the evaluation of treatment efficacy in primary care using observational data , 2014, BMJ : British Medical Journal.

[4]  K. Nichol,et al.  Effectiveness of influenza vaccine in the community-dwelling elderly. , 2007, The New England journal of medicine.

[5]  W. Henley,et al.  Bias and Sensitivity Analysis When Estimating Treatment Effects from the Cox Model with Omitted Covariates , 2013, Biometrics.

[6]  William E. Henley,et al.  The NCS-LSD cohort study: a description of the methods and analyses used to assess the long-term effectiveness of enzyme replacement therapy and substrate reduction therapy in patients with lysosomal storage disorders , 2014, Journal of Inherited Metabolic Disease.

[7]  R. Willke,et al.  Statistical Issues with the Analysis of Nonrandomized Studies in Comparative Effectiveness Research , 2011, Journal of managed care pharmacy : JMCP.

[8]  N. Black.  Why we need observational studies to evaluate the effectiveness of health care , 1996, BMJ.

[9]  R. Tannen,et al.  Use of primary care electronic medical record database in drug efficacy research on cardiovascular outcomes: comparison of database and randomised controlled trial findings , 2009, BMJ : British Medical Journal.

[10]  R. Lyons,et al.  Incidence of Campylobacter and Salmonella Infections Following First Prescription for PPI: A Cohort Study Using Routine Data , 2013, The American Journal of Gastroenterology.

[11]  Til Stürmer,et al.  Performance of propensity score calibration--a simulation study. , 2007, American journal of epidemiology.

[12]  A. Hoes,et al.  Performance of prior event rate ratio adjustment method in pharmacoepidemiology: a simulation study , 2015, Pharmacoepidemiology and drug safety.

[13]  Per Capita,et al.  About the authors , 1995, Machine Vision and Applications.

[14]  C. Huber,et al.  Matched pair experiments: cox and maximum likelihood estimation , 1987 .

[15]  A. V. Peterson,et al.  On the regression analysis of multivariate failure time data , 1981 .

[16]  R. Prentice,et al.  Commentary on Andersen and Gill's "Cox's Regression Model for Counting Processes: A Large Sample Study" , 1982 .

[17]  Menggang Yu,et al.  A new “Comparative Effectiveness” assessment strategy using the THIN database: comparison of the cardiac complications of pioglitazone and rosiglitazone , 2013, Pharmacoepidemiology and drug safety.

[18]  L. J. Wei,et al.  Regression analysis of multivariate incomplete failure time data by modeling marginal distributions , 1989 .

[19]  Richard J. Cook,et al.  The Statistical Analysis of Recurrent Events , 2007 .

[20]  Joshua D. Angrist,et al.  Identification of Causal Effects Using Instrumental Variables , 1993 .

[21]  S. Johansen.  An Extension of Cox's Regression Model , 1983 .

[22]  L. J. Wei,et al.  The Robust Inference for the Cox Proportional Hazards Model , 1989 .

[23]  R. Tannen,et al.  Replication of the Scandinavian Simvastatin Survival Study using a primary care medical record database prompted exploration of a new method to address unmeasured confounding , 2008, Pharmacoepidemiology and drug safety.

[24]  Menggang Yu,et al.  Prior event rate ratio adjustment: numerical studies of a statistical method to address unrecognized confounding in observational studies , 2012, Pharmacoepidemiology and drug safety.

[25]  S Richardson,et al.  Simplified Bayesian Sensitivity Analysis for Mismeasured and Unobserved Confounders , 2010, Biometrics.

[26]  Dawei Xie,et al.  Replicated studies of two randomized trials of angiotensin‐ converting enzyme inhibitors: further empiric validation of the ‘prior event rate ratio’ to adjust for unmeasured confounding by indication , 2008, Pharmacoepidemiology and drug safety.