Hwang 2017.

Methods	Cluster‐randomised parallel‐group study
Participants	N = 1455 patients with suspected OSA were randomised to four study arms, by class‐based (cluster) randomised design. This study used the existing home‐based testing triage structure at the trialists institution. As they report, "Most patients are referred by primary care physicians, and a sleep medicine physician triages appropriate patients to home sleep apnoea testing after review of the referral information and electronic health record chart. HSAT classes (up to 13 people) are led by a sleep‐trained respiratory therapist and sleep technologist and provide interactive OSA education and individualised HSAT setup. After a one‐night test, each patient returns for an individual appointment with a respiratory therapist to review the results. Those with OSA are recommended to undergo a 1‐week CPAP trial followed by an individual return appointment with a respiratory therapist to review CPAP data and patient experience. Patients willing to commit to CPAP therapy are immediately dispensed a device; otherwise CPAP troubleshooting or alternative treatments are discussed." This trial enrolled Consecutive patients referred to the Kaiser Permanente Fontana Sleep Disorders Center (Fontana, CA) for evaluation of suspected OSA and triaged to HSAT between November 2014 and August 2015.To conform to the sleep centre's usual care procedures, groups of patients were randomised, with all participants in each HSAT class following the same treatment arm. Inclusion criteria: at least 18 years of age, no previous sleep testing or trial of OSA therapy, eligible for HSAT. Exclusion criteria: at risk of other sleep disorders (e.g. severe insomnia), significant cardiopulmonary disease (e.g. heart failure, chronic respiratory failure), or English not preferred language. Baseline characteristics: 51% female. Mean age 49.1 (±12.5). Mean AHI 22.7. Mean ESS 9.1. Mean BMI 34. Country: USA
Interventions	Classes (and all participants in each class) were randomised (1:1:1:1) to one of four arms: 1) web‐based OSA education (Tel‐Ed, n=380), 2) telemonitoring and automated feedback (Tel‐TM, n = 375), 3) Tel‐Ed + Tel‐TM (Tel‐Both, n = 346), and 4) usual care (UC, n = 354) using a four‐arm, randomised, factorial design. Usual care: all patients attended a 1‐hour, small‐group education class with HSAT set‐up. After the trial, those willing to continue CPAP were prescribed therapy and scheduled for a 3‐month follow‐up appointment. Tel‐Ed: education about the pathophysiology of OSA, health‐related risks, impact on daytime vigilance, introduction to CPAP therapy. For patients eventually determined to have OSA, a link to a second education program was emailed. This focused on how to use CPAP, potential benefits, methods of acclimating, and equipment care instructions. Education sessions were interactive and self‐paced. Tel‐TM: Intervention based on automatic processing of device data. During the 3‐month study period, if CPAP usage thresholds were met, a message was automatically sent to the patient providing encouragement to improve use or positively reinforcing successful adherence. Tel‐both: patients received Tel‐Ed and Tel‐TM Study duration: 90 days
Outcomes	CPAP usage (hours/night) at 90 days Sleepiness (ESS) Residual AHI N of adherent participants (Medicare definition, usage ≥ 4 hours per night) QoL (FOSQ) Withdrawals
Notes	Trialists included three intervention arms. One arm was educational (Tel‐Ed), one was Supportive (Tel‐TM) and the third was Mixed (Tel‐Both). These were compared to control in respective meta‐analyses (i.e. Educational, Supportive, Mixed).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Author's 'group randomisation' does not conform to individual or block randomisation schemes.
Allocation concealment (selection bias)	Unclear risk	No reference to allocation concealment method.
Blinding (performance bias and detection bias) All outcomes	High risk	No evidence of blinding of participants, personnel or outcome assessors. Study had subjective outcomes, so knowing group assignment could influence these outcomes.
Incomplete outcome data (attrition bias) All outcomes	High risk	Authors report, "Overall, 66% showed up for the HSAT class and were tested, and 81.1% of those tested were diagnosed with OSA (AHI>5). Five hundred and fifty‐six patients (38.2% of all randomised patients; 71.7% of all patients with OSA) were eventually prescribed CPAP, all of whom had 90‐day usage data to be included for analysis."
Selective reporting (reporting bias)	Low risk	NCT02279901: Original/current primary outcome measures (unchanged from original submission on 30 Oct 2014) was: 3‐month average CPAP use/night (experimental pathway vs. traditional pathway). Secondary measures (30 Oct 2014, unchanged from original submission): Difference in 3‐month average CPAP use/night (experimental vs. experimental), ESS, FOSQ‐10, adherence to provider encounters (no‐show), healthcare utilisation.
Other bias	Unclear risk	Baseline characteristics for all randomised participants presented in tabular form by authors but no statistical comparison reported. No deviations from intended intervention reported or suspected.
Bias arising from the randomisation process (ROB2, primary outcome)	Unclear risk	Only information provided as to the randomisation procedures used was within Methods: "To conform to the sleep centre's usual care procedures, groups of patients were randomised, with all participants in each HSAT class following the same treatment arm. Classes were randomised (1:1:1:1)...both). Randomization was performed for each HSAT class (not in blocks), using a computerized random number generator." In their discussion, authors further note, "...group rather than individual randomisation was performed; nevertheless, treatment arm baseline characteristics were similar and within‐group correlations were taken into account in analyses." Author's 'group randomisation' does not conform to individual or block randomisation schemes. No reference to allocation concealment method. Baseline characteristics for all randomised participants presented in tabular form by authors but no statistical comparison reported. Authors do report that "baseline characteristics for those prescribed CPAP were similar between the four arms."
Bias due to deviations from intended interventions (ROB2, primary outcome)	Low risk	Authors report, "Clinicians providing routine care and study analysts were blinded to study arm assignment." No deviations documented; none suspected based upon review. Probable mITT. There is one discrepancy in Tel‐Ed group n = 164 in Study flowchart and baseline characteristics table 2 and the outcome data table n = 163.
Bias due to missing outcome data (ROB2, primary outcome) All outcomes	Unclear risk	Authors report, "use), 162 HSAT classes—accounting for 1,455 patients—were randomised (Figure 1). Overall, 66% showed up for the HSAT class and were tested, and 81.1% of those tested were diagnosed with OSA (AHI>5). Five hundred and fifty‐six patients (38.2% of all randomised patients; 71.7% of all patients with OSA) were eventually prescribed CPAP, all of whom had 90‐day usage data to be included for analysis." Neither analyses to correct for bias nor sensitivity analyses were conducted. Missing outcome data occurred for documented reasons (declined CPAP) that could be related to the outcome. Across intervention arms, the proportion of participants diagnosed with OSA (in evaluation period), who accepted CPAP therapy during the evaluation phase and, therefore, for whom 90 day CPAP usage data were available/analysed ranged from 68.3% (Tel‐TM) to 76.3% (Tel‐Ed) groups. UC and Tel‐Both proportions were 71%. Thus, as noted by study authors, 'drop‐out' rates were similar across groups. Additionally, there is no evidence that reasons for dropout differed across intervention arms and cited reasons for all dropouts in author's study flowchart is 'declined CPAP' and is noted in text to be 'elected for non‐CPAP therapies'. Thus, due to similar proportions and cited reasons across intervention arms, missing outcome data are unlikely to lead to bias in the estimated effects of the intervention.
Bias in measurement of the outcome (ROB2, primary outcome) All outcomes	Low risk	Outcome measured using objective CPAP usage data. Each intervention group outcome data ascertained via automated CPAP device monitoring; devices identical or sufficiently similar (i.e. similar distributions of CPAP device make) across groups. Outcome "assessor" is CPAP device: no knowledge of allocation possible.
Bias in selection of the reported result (ROB2, primary outcome) All outcomes	Low risk	NCT02279901: Original/current primary outcome measures (unchanged from original submission on 30 Oct 2014) was: 3‐month average CPAP use/night (experimental pathway vs. traditional pathway). Secondary measures (30 Oct 2014, unchanged from original submission): Difference in 3‐month average CPAP use/night (experimental vs. experimental), ESS, FOSQ‐10, adherence to provider encounters (no‐show), healthcare utilisation. Primary outcome identical in protocol and in published report. Some secondary outcomes not noted in protocol but are in published report. One outcome time point (for primary adherence outcome) planned, one outcome time point reported. No evidence that multiple analyses (e.g. variable adherence 'thresholds') were conducted.
Overall risk of bias (ROB2, primary outcome) Machine usage	Unclear risk	‐