The Value of Stratified Economic Analysis in Informing Decision-Making: A Case Study of Interventions for Obstructive Sleep Apnea
Presenting author: Dr. Bernice Tsoi, Health Economist, CADTH
Co-author: Karen Lee, Director, Health Economics, CADTH
Background: When patient characteristics influence the clinical effectiveness of interventions, stratified analyses should be conducted. We illustrate through an example of treatments for obstructive sleep apnea (OSA) how cost-effectiveness results may differ depending on whether stratification is undertaken, and the potential implications to policy-makers.
Methods: Based on an economic model developed for OSA, the impact of stratified analyses was explored. The clinical evidence suggests subgroup-specific treatment effects exist according to baseline apnea-hypopnea index (AHI; a measure of disease severity based on the number of apnea and hypopnea events per hour of sleep). Relative efficacy was incorporated into the model, estimated from two different data sets:
- stratified measures (from pairwise estimates of individual meta-analyses)
- a single aggregated measure (from a network meta-analysis of the entire studied population).
Results: The interpretation on the cost-effectiveness of treatment strategies for OSA was found to be dependent on a patient’s baseline disease severity. At a willingness-to-pay threshold of $50,000 per quality-adjusted life-year, the interventions considered most likely cost-effective differed by disease severity, with different interpretations depending on whether aggregate or stratified data were used for the clinical treatment estimates.
Conclusion: When heterogeneity exists in the clinical estimates, these need to be accounted for in economic models. By using treatment estimates from the population as a whole compared with stratified populations, different findings can be reached. This study highlights the value of stratified economic analysis to facilitate meaningful policy recommendations, and may have implications toward more efficient use of resources.
Economic Evaluation of a Portable, Home-Testing Device (BresoDx) for Obstructive Sleep Apnea
Presenting author: Dr. Petros Pechlivanoglou, Scientist, Hospital for Sick Children Research Institute
Background: Five per cent of adults have obstructive sleep apnea (OSA), but most are undiagnosed. Untreated patients have daytime sleepiness, impaired function, and an increased risk of motor vehicle accidents and death. Diagnosis of OSA is costly; in Ontario in 2008, it cost more than $40 million. Polysomnography (PSG) is the gold standard for diagnosing OSA, but requires overnight testing in a dedicated sleep lab. BresoDx is a locally developed single-channel (sound) mask-like portable monitor (PM) that can be used at home.
Aim: To assess the cost-utility of the BresoDx PM in guiding the diagnosis of OSA in patients with suspected OSA.
Methods: As part of a comprehensive technology assessment, we conducted a cost-utility analysis comparing three strategies: diagnosis guided by i) BresoDx at home; ii) BresoDx at home followed by PSG for patients with negative or indeterminate home test results; iii) and PSG for all patients. We relied on a micro-simulation to model the diagnosis and natural history of OSA. The model was partially populated by our companion randomized trial.
Results: BresoDx-guided diagnosis resulted in marginally lower quality-adjusted life-years QALYs (-0.04) and less treatment- and diagnosis-related costs (-$32) compared to PSG-guided diagnosis. BresoDx followed by PSG had similar QALYs but higher total costs ($62) as all negatively diagnosed patients undergo a second PSG-guided diagnosis.
Conclusions: Given the model assumptions, BresoDx-based diagnosis followed by PSG is similarly effective but marginally more costly versus PSG-guided diagnosis. For scenarios of price reduction for BresoDx or more conservative follow-up after OSA diagnosis, BresoDx can be a cost-saving option.
Diagnostic Accuracy of Level IV Portable Sleep Monitors Versus Polysomnography for Obstructive Sleep Apnea: A Systematic Review and Meta-Analysis
Presenting author: Dr. Lusine Abrahamyan, Clinical Epidemiologist and Assistant Professor, Toronto Health Economics and Technology Assessment (THETA) Collaborative and University of Toronto
Co-authors: Dr. Yeva Sahakyan, Institute of Health Policy, Management and Evaluation, and Suzanne Chung, University of Toronto; Dr. Petros Pechlivanoglou, Hospital for Sick Children Research Institute; Steven Carcone, University of Toronto; and Dr. Valeria Rac, Dr. Murray Krahn, and Joanna Bielecki, Toronto Health Economics and Technology Assessment (THETA) Collaborative, University Health Network
Purpose: Obstructive sleep apnea (OSA) is an under-diagnosed disorder affecting 3% to 4% of women and 6% to 9% of men. In-laboratory, overnight type I polysomnography (PSG) is the current gold standard for diagnosing OSA. Diagnostic sleep studies can be also conducted at home with type IV portable monitors (PM) that use fewer channels but offer better comfort and lower costs. We aimed to systematically review the evidence on the diagnostic ability of type IV PMs compared to PSG in diagnosing patients with suspected OSA.
- Participants: Patients ≥ 16 years old with symptoms suggestive of OSA
- Intervention: Type IV PM for diagnosing OSA
- Comparator: In-laboratory PSG
- Outcomes: Diagnostic accuracy measures
- Studies: Cross-sectional, prospective observational, experimental, and quasi-experimental studies
- Information sources: MEDLINE and the Cochrane Library from January 1, 2010 to May 10, 2016. All review stages were conducted independently by two investigators.
Results: We screened 6,054 abstracts and 115 full-text articles to select 24 full-text articles for final review. These 24 studies enrolled a total of 2,068 patients with suspected OSA and evaluated 11 different PMs with 1 to 6 channels. Only 7 (29%) studies tested PMs in a home setting. The mean difference (bias) between the PSG-measured and PM-measured apnea-hypopnea index (AHI) ranged from −6.5 to 13.5 events per hour. At AHI ≥ 5 events per hour, the sensitivity of type IV PMs varied from 67.5% to 100%, and specificity from 25% to 100%.
Conclusion: Level IV PMs offer the potential to widen access to treatment for this under-diagnosed condition. Policy recommendations regarding PM use should consider the health and societal implications of false-positive and false-negative diagnoses, and its cost-effectiveness.
SIESTA Home Sleep Study With BresoDx Portable Monitor for the Diagnosis of Obstructive Sleep Apnea: A Pragmatic Randomized Controlled Trial
Presenting authors: Dr. Valeria Rac, Associate Program Director, Toronto Health Economics and Technology Assessment (THETA) Collaborative and Assistant Professor, University of Toronto; and Dr. Nicholas Mitsakakis, Biostatistician, Toronto Health Economics and Technology Assessment (THETA) Collaborative and Assistant Professor, Institute of Health Policy Management and Evaluation, University of Toronto
Co-authors: Dr. Lusine Abrahamyan, Clinical Epidemiologist and Assistant Professor, Toronto Health Economics and Technology Assessment (THETA) Collaborative and University of Toronto; Suzanne Chung, Research Associate, Toronto Health Economics and Technology Assessment (THETA) Collaborative; Dr. Petros Pechlivanoglou, Scientist, Hospital for Sick Children; Dr. Michael Fitzpatrick, Chief of Staff & Chief of Medical & Academic Affairs, Hotel Dieu Hospital, Kingston; Dr. Murray Krahn, Program Director, Toronto Health Economics and Technology Assessment (THETA) Collaborative and Professor, Department of Medicine, University of Toronto
The objectives of the randomized controlled trials were to evaluate:
to evaluate 1) the accuracy of the clinical diagnosis of obstructive sleep apnea (OSA) guided by the home sleep study with a BresoDx portable monitor (PM), BresoDx against the clinical diagnosis guided by the polysomnography (PSG) as the reference standard; and
2) the agreement between the apnea-hypopnea index (AHI) measured with BresoDx and PSG.
SIESTA was a pragmatic, multi-centre randomized single-blinded trial that enrolled adult patients referred to three Ontario sleep clinics with symptoms suggestive of OSA. Patients were randomized into BresoDx/PSG or PSG/BresoDx sequence arms. In each sleep clinic, two sleep specialists were randomized to conduct patient clinical assessment and diagnosis. A total of 243 patients completed both sleep studies: 126 in the BresoDx/PSG arm, and 117 in the PSG/BresoDx arm. A total of 213 patients completed both consultation visits: 111 in BresoDx/PSG and 102 in PSG/BresoDx. The sensitivity of the diagnostic accuracy of the BresoDx-informed clinical diagnosis against the PSG was 0.867 to 0.897, and specificity was 0.396 to 0.458. At the AHI cut-off of ≥ 5 events per hour, specificity of the BresoDx PM was 0.47 (95% confidence interval [CI], 0.35 to 0.60), sensitivity was 0.85 (95% CI, 0.78 to 0.90) and the area under the curve (AUC) was 0.794. For the cut-off of ≥ 15 events per hour, specificity of the BresoDx PM was 0.95 (95% CI, 0.90 to 0.98), sensitivity was 0.60 (95% CI, 0.51 to 0.69) and the AUC was 0.856. A combined strategy of initial screening with BresoDx followed by PSG for patients with negative test results could be used with reasonable clinical confidence as a screening/diagnostic device for OSA in a population with a high pretest probability for moderate or severe OSA.