Skip to main content
. 2020 Apr 7;2020(4):CD007736. doi: 10.1002/14651858.CD007736.pub3

Bakker 2016.

Methods Open‐label, parallel‐arm, RCT.
Participants N = 83 participants with OSA
Inclusion criteria: AHI 4%, ≥ 10 or AHI 3%, ≥ 15; 45 to 75 years with established CVD or cardiometabolic disease (established coronary artery disease (≥ 70% stenosis in at least one major coronary artery), prior myocardial infarction, coronary artery revascularisation procedure, Ischaemic stroke, or diabetes) OR 55 to 75 years with at least three CVD risk factors (male sex, BMI ≥ 30, hypertension, dyslipidaemia, and ≥ 10 pack‐years of smoking).
Exclusion criteria: cardiovascular event < 4 months before enrolment, prior CPAP, ESS > 14 of 24, drowsy driving within 2 years, commercial driving, or an uncontrolled medical condition (including CSA, heart failure, uncontrolled hypertension, severe hypoxaemia, anaemia, and renal insufficiency).
Baseline characteristics: 33% female. Mean age 63.8 (NR). Mean AHI 22.8. Mean ESS NR. Mean BMI 31.1.
Country: USA
Interventions Eligible participants entered a run‐in phase before randomisation, consisting of 14 days wearing a nasal CPAP mask during sleep (without a CPAP device). Participants who reported using the mask during the majority of the run‐in and who were willing to continue using the mask were eligible for randomisation. Randomisation took place in a 1:1:1:1 ratio with a block size of 4, based on three stratification factors: diagnostic study (full night or split night with titration), site, CVD status (established or risk factors) to one of four study arms (two control conditions, two treatment conditions): conservative medical therapy (n = 44), sham CPAP (n = 42), active CPAP (n = 42), or active CPAP +ME (n = 41). Bakker 2016 reported only the active CPAP and CPAP + ME arms.
Intervention (Active CPAP + ME): overall goal of each ME session was to resolve the participants' ambivalence toward establishing consistent CPAP usage patterns and increase their confidence toward using CPAP regularly. Each participant was encourage to set concrete goals regarding their future CPAP use and identify rewards that they could provide themselves when those goals were achieved. ME was delivered during 1‐hour in‐person sessions at baseline and week 1, which included an educational video, and during phone calls of 10 to 30 minutes with the same psychologist at weeks 3, 4, 8, 12, 20, and 32.
Control (Active CPAP): CPAP
Study duration: 12 months.
Outcomes
  • CPAP usage (hours/night) at 6*, 12 months

Notes * Indicates primary outcome analysed in this Review.
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk "Randomization took place in a 1:1:1:1 ratio with a block size of 4, based on three stratification factors: diagnostic study (full night or split night with titration), site (BWH, BIDMC, or Joslin), CVD status (established or risk factors). Randomization was performed using a data‐entry system linked to an off‐site server holding the sequences."
Allocation concealment (selection bias) Unclear risk Insufficient information to determine sufficiency of sequence concealment: Sequences were concealed at off‐site location. Block sizes were fixed, so last assignment to each block could be predicted, but only likely to occur if investigators were aware of block sizes.
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 Low risk Data were available for all, or nearly all, participants randomised.
Selective reporting (reporting bias) Low risk Study protocol is not available, but thorough analysis of information presented in study's NCT archive does not suggest that outcomes of interest were chosen after data analysis already commenced.
Other bias Low risk No baseline imbalances or reported deviations from intended intervention.
Bias arising from the randomisation process (ROB2, primary outcome) Unclear risk Authors report, "Analyses in this paper compare the active CPAP and CPAP + ME arms only. Randomization took place in a 1:1:1:1 ratio with a block size of 4, based on three stratification factors: diagnostic study (full night or split night with titration), site (BWH, BIDMC, or Joslin), CVD status (established or risk factors). Randomization was performed using a data‐entry system linked to an off‐site server holding the sequences. Participants randomly assigned to a CPAP group immediately underwent secondary randomisation to use a device by one of two manufacturers (Philips Respironics or ResMed), using a randomisation sequence with a block size of 2." From Protocol (Yaggi 2016): 255 patients with TIA or stroke are randomly assigned using a 1:1:1 (control:standard PAP: enhanced PAP) randomisation scheme to either usual care or a home‐based diagnosis and treatment approach that includes ambulatory polysomnography and initiation of PAP for patients with sleep apnoea (Figure 1). Control patients could undergo diagnosis and treatment of sleep apnoea if suspected as part of usual clinical care. The randomisation is stratified by centre (Connecticut or Indiana) and neurologic event type (TIA or stroke)...."
 
 Random component used for sequence generation not explicitly described. Insufficient information to determine sufficiency of sequence concealment: Sequences were concealed at off‐site location. Block sizes were fixed, so last assignment to each block could be predicted, but only likely to occur if investigators were aware of block sizes.
Authors provide a tabular summary of baseline characteristics for participants randomised to CPAP+ME and CPAP‐only intervention arms, and report in text: "The two groups were comparable at baseline in terms of age, sex, anthropometrics, race or ethnicity, and cardiovascular risk factors (Table 5)."
Bias due to deviations from intended interventions (ROB2, primary outcome) Low risk Yaggi 2016 report that participants were unblinded. (p. 5)
Deviation from protocol documented by authors: The original trial design included follow‐up for 12 months; however, participants randomly assigned after January 2013 were restricted to 6 months of follow‐up. Our primary analysis of CPAP adherence therefore took place for 6 months, rather than 12." This deviation is unlikely to have affected the outcome. Authors report, "All analyses were intention‐to‐treat." Additionally, all outcome measures appear to derive from all randomised participants in the two relevant treatment arms.
Protocol deviation was unlikely to have contributed to biased effect estimates as it was balanced across intervention arms and likely unrelated to outcome.
Bias due to missing outcome data (ROB2, primary outcome) 
 All outcomes Low risk Six‐month outcome data available for nearly all randomised participants. In addition to having primary outcome (CPAP usage) data for nearly all participants, authors documented reasons for missing adherence data and conducted sensitivity analyses, determining that the finding of no differences in adherence between groups was robust when a value of zero was assigned to missing usage data and when examining complete data only.
Bias in measurement of the outcome (ROB2, primary outcome) 
 All outcomes Low risk Outcome measured using objective CPAP usage data. Published report notes, "All participants were provided with either a REMstar
 
 Pro CPAP (Respironics Inc. Murrysville PA) device with a data‐storage Smart‐Card, or a Sandman Goodnight 420 Series Auto HH (Covidien, Mansfield, MA) which allowed usage data to be directly downloaded from the machine or via a memory key. Reviewers contacted authors to request information as to the distribution of CPAP device make across intervention arms. Authors responded (email) reporting, "Our CPAP machines were provided for the study by Respironics – REMstar Auto with C Flex or REMstar Pro M Series." Thus, the information provided by authors in personal correspondence suggests that no participants received the Sandman Goodnight 420 Series Auto HH and, therefore, that outcome measurement should not have differed between intervention groups.
Outcome "assessor" is CPAP device: no knowledge of allocation possible.
Bias in selection of the reported result (ROB2, primary outcome) 
 All outcomes Unclear risk NCT01261390: Original primary outcome measures of BestAir trial (submitted 15 Dec 2010) was "Effectiveness of continuous positive airway pressure therapy on cardiovascular disease, using mean 24 hour systolic blood pressure as the trial's primary endpoint. [ Time Frame: 3 years ]". Current primary outcome measures (submitted 18 Apr 2017) include those of interest in this review: "Difference in CPAP Adherence by Active Treatment Arm [ Time Frame: 6‐months ] Original secondary outcome measure (submitted 15dec2010): "Recruitment and retention rates of patients with moderate to severe obstructive sleep apnoea and cardiovascular disease risk factors or established CVD participating in a controlled trial. [ Time Frame: 3 years ]"
 
 "Adherence to CPAP therapy was tracked remotely by modem transmission. Outcome reported is mean hours of PAP use per night at the 6‐month time point. Comparison is between those with and without assignment to Motivational Enhancement as part of treatment randomisation." Current secondary outcome measures of relevance to our review (submitted 18 Apr 2017): Change in ESS, SAQLI, SF‐36, PHQ‐8 [6 & 12 months.
 
 Though it appears that CPAP adherence was a target of the intervention from the outset (i.e. it is repeatedly mentioned as a focus of attention in description of the active treatment arms), it does not appear to have been an outcome of interest at the time of design. Rather, as described in the protocol (described in Yaggi 2016), adherence was a primary process measure (and one likely to impact the primary physiological outcome of interest). Study primary completion date (March 2014, final collection date for primary outcome measure) occurred before current primary/secondary outcome measures posted to clinicaltrials.gov. However, the time stamps provided in NCT entry are merely the date that these outcomes were entered into the electronic system and likely do not represent the date that decisions about additional outcome measures were made. Thus, it is possible that all primary/secondary outcomes were finalised before unblinded outcome data were available for analysis.
Decision to change primary CPAP adherence outcome from 12‐month to 6‐month endpoint likely a consequence of a change in design/follow‐up necessitated by factors unrelated to the outcome that required follow‐up to shorten for participants enrolled after January, 2013. Moreover, for those participants with outcome data at 12 months, the results were similar.
Overall risk of bias (ROB2, primary outcome) 
 Machine usage Unclear risk