Just a Good Deal or Truly a Steal? Medical Cost Savings and the Impact on the Cost-Effectiveness of Treating Sleep Apnea

INCREASINGLY, DECISION MAKERS SEEKING TO IDENTIFY INTERVENTIONS THAT OFFER THE MAXIMUM HEALTH BENEFIT ARE USING SYSTEMATIC economic evaluation.¹ A type of economic evaluation, cost-effective analysis (CEA) compares interventions based on the differences in costs needed to provide an additional unit of a specific health outcome. This outcome is usually summarized as quality adjusted life year (QALY) gained, to allow comparisons across different diseases and health states.² As knowledge about the prevalence, treatments, and consequences of obstructive sleep apnea (OSA) has accrued, the challenge has been to convince decision makers to allocate scarce resources to the diagnosis and treatment of OSA.

The results of CEA support the value of the diagnosis and treatment of OSA.^3–5 Studies have provided estimates, from a payer perspective, ranging from $2,000-11,000 per QALY over 5 years for treating moderate to severe OSA in a sleep clinic population. These values could be considered a very good buy for payers when compared to other generally accepted medical interventions such as the use of blood pressure lowering drugs in hypertensive patients.⁶

The treatment of OSA, which has harmful and costly multi-organ consequences, may result in future medical cost savings by preventing those consequences. If future cost savings result, the cost-effectiveness ratio may be even more favorable than reported or treatment may dominate over no intervention (lower costs and better outcomes). The reason given by authors of some CEAs to not include the future medical cost savings in models is the lack of empiric data to support this savings.^3,5

What is the empiric evidence that untreated OSA results in increased medical costs? In this issue Banno and colleagues present data from an observational study that shows physician fees and visits are higher prior to diagnosis in 223 clinically diagnosed obese women with OSA than in obese and normal weight population controls.⁷ This result is consistent with findings of previous studies examining health care costs in patients identified clinically with OSA compared to controls without OSA.^8,9 These studies documented an approximately two-fold increase in medical costs prior to diagnosis relative to control groups matched for age, gender, residence, and in some cases family physician. The current study adds to the literature by controlling for obesity-related differences in healthcare use by matching on obesity categories.

The current study⁷ did not demonstrate that the higher healthcare use seen before diagnosis is attributable to untreated OSA. Increased healthcare use may increase the likelihood of diagnosis of OSA. Subjects who see healthcare providers more frequently for medical issues (including those not caused by OSA) may more likely be diagnosed with OSA than subjects who have less illness. Also individuals who are more aggressive in seeking care for symptoms may be more likely to seek care for their symptoms of OSA, leading to a higher likelihood of diagnosis particularly where access to these services is limited. To attribute higher health care use to untreated OSA, these threats to the validity of the comparison need to be removed by statistical adjustment or by careful choice of a control group.

Guidance on factors other than clinical characteristics that can affect healthcare utilization is provided by a model proposed by Andersen and Newman.¹⁰ These include predisposing factors (i.e. demographic, social structure and beliefs), enabling factors (i.e., income, insurance type, family characteristics), and illness level (i.e., perceived health state, symptoms, diagnosis).¹⁰ A study design that accounts for these potential confounders may provide stronger comparisons across groups of patients with and without OSA. Population based studies in which the identification of OSA in cases and controls is not biased by healthcare utilization characteristics are a promising approach to providing a more valid comparison. The Sleep Heart Health Study showed that OSA was associated with an 18% increase in predicted health care use based on medication use, but these analyses were limited by the lack of direct assessment of healthcare costs.¹¹

Does treatment of OSA reduce future healthcare use? Investigators have used a pre-post quasi-experimental design (subjects not randomized with respect to treatment) to answer this question.^12–14 Since regression to the mean is a significant threat to validity in a situation where high healthcare use may predispose to the identification of OSA, it is important to compare the intervention group to an untreated OSA group identified in a similar manner. Peker and colleagues¹⁴ studied patients with cardiopulmonary disease and OSA (N=54). Hospitalizations were decreased in the 2 years following treatment in patients using CPAP versus an increase in patients who did not use CPAP.¹⁴ A reduction in healthcare costs after treatment in a broader OSA group (without requirement for comorbid illness) in comparison to untreated OSA control group has not been demonstrated in adults, though it has in a pediatric population.¹⁵

An acceptable approach to account for future medical cost savings in CEA is to use results from intervention studies that demonstrate a reduction in adverse consequences and data on the medical costs of treating those consequences to estimate medical cost savings. A recently published CEA that included medical cost savings from preventing cardiovascular events, strokes and motor vehicle accidents indicated that the cost savings exceeded the costs of treating OSA after 13 years.¹⁶ The validity of this approach is dependent on the strength of evidence from intervention studies showing reduced adverse outcomes.

The study of medical costs in OSA has practical implications for our field since the validation and estimation of future medical cost savings will make the cost-effectiveness of treating OSA even more compelling to decision makers. Future economic studies with more rigorous study designs can assist in the formulation of more comprehensive CEA. The addition of healthcare utilization and cost measures to prospective clinical outcome studies can also advance the assessment of the economic value of interventions for OSA.

DISCLOSURE STATEMENT

The authors have indicated no financial conflicts of interest.