Evaluating Different Physician’s Prescribing Preference Based Instrumental Variables in the Study of Beta2-Agonist Use and the Risk of Acute Myocardial Infarction Md Jamal Uddin, RHH Groenwold, A de Boer, ASM Afonso, P Primatesta, C Becker, SV Belitser, AW Hoes, KCB Roes, OH Klungel Division of Pharmacoepidemiology and Clinical Pharmacology University of Utrecht, Netherlands
Disclosures Acknowledgements: The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) for the Innovative Medicine Initiative (www.imi.europa.eu) under Grant Agreement n° 115004. The research leading to these results was conducted as part of the PROTECT consortium (Pharmacoepidemiological Research on Outcomes of Therapeutics by a European ConsorTium, www.imi-protect.eu) which is a public-private partnership coordinated by the European Medicines Agency. Conflicts of interest: “Olaf Klungel had received unrestricted funding for pharmacoepidemiological research from the Dutch private-public funded Top Institute Pharma.”
Background Instrumental variable (IV) analysis with physician’s prescribing preference (PPP) as an IV has been used to control for unmeasured confounding PPP can be defined in several ways, but it is unclear how different definitions of PPPs perform across databases We aimed to assess the performance of PPPs in a study of long-acting beta2-agonist (LABA) use and myocardial infarction (MI)
Methods Database Mondriaan, Netherlands (N=1.4 M) CPRD, UK (N=5 M) Study Population Adult patients with a diagnosis of asthma and/or COPD At least one prescription of inhaled beta2-agonists or inhaled muscarinic antagonists (new users) Cohorts COPD, Asthma, COPD & Asthma, Combined
Exposure: Time-fixed: LABA vs. Non-LABA (first prescription) Time-varying: Current LABA users vs. all other users (Non-LABA, Past and recent LABA) Outcome: Myocardial Infarction (MI) Data were analysed using conventional method and the IV methods
Characteristics of study population CPRD Mondriaan Cohort size 490,499 27,459 Number of MI cases 5,739 447 LABA users 15% 29% Follow-up period 2002-2009 Median follow-up time (years) 4.58 3.50 Mean age of patients (years) 51 52
Exposure Time-varying Definition of Instrumental Variables Exposure Time-fixed Exposure Time-varying
Definition of Instrumental Variables Exposure: Time-fixed Physician’s prescribing preference with previous single prescription (PPP1) 5 prescriptions (PPP5) 10 prescriptions (PPP10) Proportion of LABA prescriptions (PLP) Exposure: Time-Varying Proportions of time for current LABA use (PTL) in a practice
MI Assessment of IV assumptions Instrumental variables LABA vs. Non-LABA Correlation, Odds Ratio MI Standardized Difference Ref: Brookhart et al. 2006, Epidemiology
Results
Violation (X) of Assumptions of Instrumental Variables Assumption-I IV is associated with exposure Assumption-II is independent of confounders
Neither of the IVs we considered appears to be valid, hence, the HRs were invalid
Key Findings For time-fixed LABA exposure, the IV PPP10 outperformed the other IVs regarding strength of the IV and balance of confounders between IV categories None of the IVs we considered appeared valid for time- varying LABA exposure, hence estimates were not valid CPRD included approximately 18 times more subjects than Mondriaan, and thus IVs estimates appeared more precise
Conclusions The apparent validity of IV analysis using PPP strongly depends on how this IV is defined and in which database it is applied We recommend to consider several plausible IVs, assess their validity, and only proceed with the IV analysis with apparently valid IV(s)
Thank You!!!