Tennis PS, Johannes CB, Camargo C, Sturmer T, Lanes S, Brown J, Andrews EB, Davis K, Schatz M. Feasibility of ruling out small treatment-associated increase in asthma mortality risk. Poster presented at the 29th International Conference on Pharmacoepidemiology and Therapeutic Risk Management; August 2013. Montreal, Canada. [abstract] Pharmacoepidemiol Drug Saf. 2013 Aug; 22(Suppl 1):S34-5. doi: 10.1002/pds.3512

BACKGROUND: Asthma mortality is rare. Published stud-ies evaluating possible increased mortality risk withuse of long-acting beta-agonists (LABA) in fixed-dosecombination with inhaled corticosteroids (ICS) hadlimited sizes. Ongoing safety trials are evaluating acomposite endpoint including asthma hospitalizations.

OBJECTIVES: Using a new user design, assess feasibilityof establishing with sufficient precision an asthma mor-tality ratio (MR) of≤1.40 for LABA+ICS relative toselected non-LABA maintenance therapy.

METHODS: We established a distributed network toassemble a cohort of asthma patients aged 4–100 yearsfrom 10 US health plans–total>70 million enrolleesbetween 2001–2010. We identified a persistent asthmacohort (PAC):≥4 asthma medication dispensings and≥1 asthma diagnosis in 12 months. New use was thefirst exposure of interest after the cohort entry andwithout any prior exposure of interest. Exposures ofinterest included Advair (salmeterol+fluticasone propi-onate), ICS monotherapy, and leukotriene receptorantagonist+ICS. To identify asthma deaths, data part-ners linked PAC data to the National Death Index.From person-years (PY) of exposure and asthma mor-tality rate in the PAC over all follow-up regardless ofexposure, we derived the MR threshold that could beruled out with probability of 0.80 based on the upper95% confidence limit if the true MR=1.00.

RESULTS: Of 5,881,438 patients with an asthma diagno-sis from 10 data partners 17% fulfilled the PAC defini-tion (2,399,564 PY). The most common exclusion(65%) was having<4 asthma medication dispensingsin 12 months. There were 11,531 PY of Advair newuse and 17,231 PY of comparator exposures new use.Across the PAC, there were 278 asthma deaths; overallasthma mortality rate was 1.16 per 10,000 PY. If thetrue asthma MR=1.00, the study could rule out anMR>29.

CONCLUSIONS: Even with 10 data sources, use ofguideline definitions for persistent asthma and strict exposure definitions yielded a study size too small to address the objective. We are reconsidering PAC inclusion criteria, exposure definitions, and MR threshold as part of an ongoing feasibility assessment.