Chris R. Fernandez1 • Mark Kaiser1 • Jen McClurg1 • Sam Rusk1 • Nick Glattard1 • Fred Turkington1 • Yoav N. Nygate1 • Maggie Richard2 • Ian Duncan2 • Nathaniel Watson3
Research studying the economics of Obstructive Sleep Apnea (OSA) therapy faces confounds including the prevalence of undiagnosed OSA, rate of diagnosed patients not starting continuous positive airway pressure (CPAP) therapy, spectrum of CPAP treatment adherence, and effects of concurrent co-morbidity.
We provide an actuarial analysis to study the economic impact of OSA therapy using CPAP, accounting for these confounds, using the 2016-2018 Medicare 5% Limited Data Set (LDS) Analytical File, a random sample of Medicare Claims containing approximately 2.9 million patients/year, resulting in N=2,001,538 eligible Fee-For-Service patients excluding managed care patients and incomplete data.
We segmented a qualified study population of 52,087 members with OSA into three cohorts and three 12-month time-periods, including a total of 1,351,838 patient months.
The cohorts analyzed were A) patients with OSA diagnosis and ≥12 months treatment, B) patients with OSA and <12 months treatment, and C) patients with OSA diagnosis who never received treatment, requiring members to have at least 3 months of membership in each 12-month timeframe to qualify.
We analyzed the healthcare costs in each cohort in the year before treatment (BTY), the first year of treatment (DTY), and following treatment year (PTY).
Treatment adherence qualifications were based on therapy related claims (e.g. CPAP therapy initiation, CPAP supply claims) within any 6-month time window following OSA testing and diagnosis.
We applied actuarial risk adjustment within each cohort and time-period to provide a risk-adjusted cost comparison and applied trend rates to analyze costs in terms of 2016 dollars.
Results were analyzed cross- sectionally given zero-to-seven co-morbidities among obesity, hypertension, type-II diabetes, depression, COPD, CHF, and/or prior stroke, facility-vs-home testing, and with-or-without surgical procedures.
The average per-member-per-month (PMPM) total medical spend for beneficiaries with OSA was 209% higher than average PMPM of Medicare beneficiaries.
The average PMPM total medical spend was highest in the diagnosed-but- never-treated cohort-C ($1,375), second highest in <12-months treatment cohort-B ($1,005), and lowest in ≥12-months treatment cohort-A ($983). In both cohorts that started therapy, average costs decreased from before-treatment year (BTY) to post-treatment year (PTY), and from the first to second year on therapy. Compared to no-therapy cohort-C, costs were 29% lower in cohort-A and 27% lower in cohort-B. Among co-morbid members in the 75th spend percentile, we observed BTY to DTY PMPM reduction of 17.7% and 15.6% in Cohort-A and –B, respectively.
In the same cohorts, we observed 20% and 21% PMPM reductions between BTY and PTY, or after the 2nd treatment year. Patients undergoing surgical procedures had higher costs but lower spend reduction in initial and following year of therapy (22% and 5%).
We compared prevalence rates of major comorbidities for each cohort of members and each time period of interest. We also compared these rates to the baseline population (FFS Medicare without OSA).
In the OSA population, patients with 2-4 comorbidities account for 76% of the population whereas in the baseline population members with 2-4 comorbidities only account for 48%. In the figures below it is apparent how prevalent some of the most expensive conditions are and the degree of multiple morbidity within the Medicare OSA population.
To view the data presented above, please download the poster.
We observed significant differences in cost between OSA patients that started treatment versus those that did not, and those differences further increased the year following therapy onset.
These findings imply that receiving treatment for OSA reduces a patients total medical spend. In terms of mean cost, the ≥12-month and <12-month cohort costs decreased in both follow-up treatment years.