Patient Outcomes, Health Care Resource Use, and Costs Associated with High Versus Low HEDIS Asthma Medication Ratio.

BACKGROUND: The Healthcare Effectiveness Data and Information Set (HEDIS) quality measures for asthma include the asthma medication ratio (AMR) as a marker of quality of care for patients with asthma. Few data are available to describe the association between health care use and costs in patients with high versus low AMR.

OBJECTIVE: To characterize health care use and costs associated with high versus low AMR in patients participating in commercial health plans.

METHODS: In a commercial claims database, this study retrospectively identified patients aged 5 to 64 years on December 31, 2011, who met the HEDIS definition of asthma in the premeasurement year (January 1, 2010-December 31, 2010) and the measurement year (January 1, 2011-December 31, 2011). Each patient was classified as having either high or low AMR based on the HEDIS definition. AMR was calculated as the ratio of controller to total asthma medications; high AMR was defined as ≥ 0.5. Annual per-patient health care use and costs were compared in patients with high versus low AMR using (a) multivariable linear regression models to estimate mean annual number of office visits, oral corticosteroids (OCS) bursts (≤ 15-day supply), and costs and (b) negative binomial models to estimate mean annual hospitalization and emergency department (ED) visits. All estimates were adjusted for age, sex, region, and Charlson Comorbidity Index score to control for differences between patients with high versus low AMR.

RESULTS: Patients were identified with high (30,575) and low (6,479) AMR. An average patient with high AMR had more all-cause office visits (14.1 vs. 11.0; P < 0.001), fewer all-cause hospitalizations (0.109 vs. 0.215; P < 0.001), fewer all-cause ED visits (0.321 vs. 0.768; P < 0.001), and fewer OCS bursts (0.83 vs. 1.33; P < 0.001) than an average patient with low AMR. An average patient with high AMR had fewer asthma-related hospitalizations (0.024 vs. 0.088; P < 0.001) and ED visits (0.060 vs. 0.304; P < 0.001) than an average patient with low AMR. Numbers of asthma-related annual office visits were similar between the high and low AMR groups (high 2.2 vs. low 2.2; not significant). The rate of poor asthma control events (≥ 6 short-acting beta-agonist dispensing events or ≥ 2 OCS bursts, asthma-related ED visits, or hospitalizations) was greater in patients with low AMR than in patients with high AMR (74.3% vs. 26.9%). An average patient with high AMR had lower annual nonmedication costs than an average patient with low AMR ($5,733 vs. $6,295; P = 0.011). Similar trends emerged for asthma-related costs. A patient with high AMR had higher average total annual health care costs than a patient with low AMR ($9,811 vs. $8,398; P < 0.001). These increased costs primarily resulted from increased medication costs for patients with high versus low AMR ($4,077 vs. $2,103; P < 0.001).

CONCLUSIONS: Although patients with high AMR had more office visits and higher medication (which resulted in higher overall health care) costs, their care was marked by fewer OCS bursts (indicating fewer instances of poor asthma control), fewer ED visits, and fewer hospitalizations and lower non-medication costs than those patients with low AMR. These findings support the use of AMR as a care quality measure for patients with persistent asthma.

DISCLOSURES: This study was funded by Genentech. Luskin has received consulting and lecture fees, research and travel support, and payment for developing educational presentations from Genentech and has received lecture fees from Merck. Raimundo and Solari are employees of Genentech. Antonova was employed by Genentech at the time of this study. Broder and Chang are employees of Partnership for Health Analytic Research, which received funding from Genentech to conduct this research. Study concept and design were contributed by all authors. Broder and Chang conducted analyses. All authors interpreted the data. Antonova wrote the manuscript with assistance from the other authors. All authors participated in manuscript review and revisions.