Comparing the benefits and costs of combining CPAP and jaw-advancing devices for obstructive sleep apnea

This study will assess an intervention that could enhance the effectiveness of an existing NICE-recommended therapy for obstructive sleep apnoea (OSA). Both mandibular advancement devices (MAD) and continuous positive airway pressure (CPAP) therapy are part of NHS clinical services. Should the results prove effective and cost-effective the new ‘combination therapy’ could be rapidly introduced as an additional, efficient tool to help the significant number of patients who struggle with ‘monotherapy’.

OSA is caused by closure of the upper airway during sleep due to excessive muscle relaxation. Pauses in breathing caused by total (apnoeas) or partial (hypopnoeas) airway closure cause oxygen levels to drop and brief awakenings that result in excessive daytime sleepiness. OSA impacts all aspects of life, including social and work performance and driving safety, and impairs quality of life (QoL) [2]. OSA has other important health effects. It is linked to high blood pressure [3] and a 2.5 times higher risk of developing cardiovascular disease including heart attack and stroke [4]. OSA is associated with abnormal heart rhythms including atrial fibrillation, which itself undermines QoL and daytime functioning, and increases risks of stroke and road traffic accidents.

OSA is straightforward to diagnose and is treated with CPAP. A patient wears a mask connected to a small electric air pump that generates pressure to keep the throat open during sleep. CPAP is highly effective in treating OSA, improving daytime sleepiness, functional status and QoL. It is a treatment of choice for moderate-to-severe OSA [5, 6].

Unfortunately, not everyone tolerates CPAP and adherence rates range from 20 to 83%, so untreated OSA is prevalent [7]. Uncomfortably high pressure is a possible reason for CPAP intolerance. Patients start on a low pressure which is increased until OSA is controlled. For some patients the required pressure is higher than they can tolerate, making CPAP difficult to sleep with and defeating the purpose of treatment. In a nationwide survey of CPAP users, we found 68% had had difficulty using CPAP due to the pressure being delivered (details below). One method to try to improve CPAP tolerance is to use an auto-titrating CPAP machine. These aim to make CPAP more tolerable by continually monitoring for apnoeas and adjusting pressure according to whether they are occurring. Meta-analyses suggest auto-CPAP leads to a small improvement in CPAP adherence, although average pressure delivery is not much different to fixed pressure machines. Whilst auto-titration may help some patients to tolerate CPAP, it does not address all pressure-related issues and in some patients the pressure needed to control OSA exceeds CPAP's capacity, leading to ‘breakthrough’ OSA. More expensive machines deliver much higher pressures, but these are not available in every NHS service and there is no robust evidence that they are better tolerated [7].

MAD are worn in the mouth during sleep. They treat OSA by advancing the lower jaw and increasing airway space. MAD are not as good as CPAP at treating OSA. However, they are useful in milder diseases, are cost-effective [8] and sometimes help people with more severe OSA who cannot tolerate CPAP [5, 8].

Combining MAD with CPAP could potentially open the airway enough to allow CPAP pressure to be reduced. This could allow patients requiring higher CPAP pressures to tolerate treatment better and use CPAP more, increasing the chance of improving OSA symptoms and avoiding longer-term health impacts. If this study shows combination CPAP-MAD does reduce CPAP pressures, then it would also establish the utility of this intervention for improving the experiences of patients with ‘breakthrough’ OSA.

Up to 24% of middle-aged men and 9% of women have at least mild OSA (apnoea–hypopnoea index [AHI]) ≥ 5/h) (Young et al. 1993). Two per cent to 7% of adults have associated excessive daytime sleepiness, known as OSA syndrome (OSAS). Untreated OSA has a substantial economic burden in terms of reduced work productivity, increased road traffic accidents [9] and health care costs, especially for cardiovascular disease and stroke [10]. For example, older and middle-aged patients with OSA were shown to have roughly double the healthcare costs of aged-matched controls [11] which were reduced considerably with effective treatment [12]. However, the limited utility of other treatments for more severe OSA means that patients who cannot tolerate CPAP often remain untreated, so new ways to modify CPAP pressure delivery are still needed [13].

If the results of this study show CPAP-MAD combination therapy to be both effective and cost-effective it would be straightforward to make this treatment widely available. Results of our national patient survey suggest there are a significant proportion of OSA patients around the UK who need help and could benefit. Successfully treating these patients would not only improve their symptoms but could also reduce risk of longer-term conditions associated with untreated OSA, such as stroke. At population level, there is strong evidence that CPAP adherence is cost-effective [14] and superior to no treatment after a minimum of 2 years’ treatment [15]. It has been estimated that effectively treating all OSA could save the NHS roughly £1000 per patient [16]. The British Lung Foundation commissioned a report in 2014 which estimated that a total of £55 million and 40,000 QALYs annually could be saved if all people with moderate to severe OSA were diagnosed and treated with CPAP [17].

This is a single-centre superiority trial comparing CPAP to CPAP + MAD (combination therapy), using an adaptive 2 × 2 randomised crossover design. The primary outcome is CPAP adherence. The treatment periods are 12 weeks each, with the first 2 weeks treated as an acclimatisation period.

Eligible patients will be adults referred to the Respiratory Support and Sleep Centre (RSSC) at Royal Papworth Hospital Foundation NHS Trust, UK, for CPAP titration.

Following initial screening for eligibility, potential participants are approached to introduce them to the trial by either a member of the research team or the CPAP Practitioners/Advanced Nurse Practitioners at the time of a patient’s CPAP initiation visit. The trial is explained, and a patient information sheet provided, and verbal agreement is gained to contact the patient to obtain their CPAP pressure reading once they have completed the initial CPAP titration. If the patient is still eligible and remains willing to take part in the trial, consent is taken by the Chief Investigator when the participant attends their six-week CPAP review.

Initial screening removes patients according to AHI/DI; ESS; quality of/out of date diagnostic study; presence of mixed/central sleep apnoea; previous use of CPAP/NIV; previous MAD use and other medical/sleep history related criteria. Once the pressure reading is confirmed to be ≥ 12 cm water the final eligibility criteria are checked by the Chief Investigator prior to consenting, including a dental and temporomandibular joint check for suitability to use a MAD.

There are no plans to use participant data in ancillary studies. If this is subsequently considered then the data would be anonymised. Further consent would be sought otherwise. No biological specimens will be taken.

CPAP and MAD are both existing NICE recommended therapies for OSA on their own as “monotherapy”. Providing CPAP and MAD as a combination therapy could improve CPAP adherence and so help treat OSA.

Participants with high CPAP pressure requirements after initial CPAP titration will undertake the 2–arm cross-over trial. They will be randomised to either standalone CPAP or CPAP + MAD therapy and will then cross over to the alternative arm. Each arm will last 12 weeks (including 2 weeks acclimatisation period). Auto-titrating CPAP machines that automatically adjust the pressure according to individual requirements will be used in the trial. This will enable evaluation of the impact of MAD on CPAP pressure requirements and telemonitoring functionality will allow remote data downloading.

Each participant has the right to withdraw from the study at any time. In addition, the Investigator may discontinue a participant from the study at any time if the Investigator considers it necessary for any reason including:

The reason for withdrawal will be recorded in the CRF.

Adherence is recorded automatically and is the primary outcome of the study.

Participants will receive a telephone call if they have not sent their dental impression to the device manufacturer within 2 weeks. Our experience with a previous MAD study identified this as an area where additional support reduces the chance of a participant dropping out at an early stage if they have difficulty with moulding a dental impression. There is also telephone support available to participants mid-treatment phase to “touch base” and deal with any concerns.

Any other medical care will be permitted as required. There are no prohibitions.

Participants will be reviewed at trial end and offered to continue with either treatment arm. Alternative options for OSA treatment will be discussed as required and appropriate. Participants will return to long term follow-up under the clinical service. Participants are not expected to come to significant harm from involvement in the trial. However, if they do then the NHS Indemnity scheme would apply.

See the “Objectives {7}” section above.

A sample size of 64 patients was selected based on power (90%) and type I error (5%) considerations for the primary endpoint of average hours of CPAP usage within the treatment window of 10 weeks.

Using data from the TOMADO study [5], we estimated a pooled variance of 5.51 within pairs of observations. Using this estimate as a reference (assuming that patient variability between CPAP and CPAP-MAD will be similar) we obtain a sample size of 58 patients to detect an hour increment in the average use of the treatment (combination over standard of care). The sample size formula used is Eq. (9) of Siyasinghe and Sooriyarachchi [18]. To allow for estimated loss to follow-up (informed from TOMADO experience) we intend to randomise an additional 10% of patients to give a final sample size of 64.

The recruitment rate will be closely monitored by the project team and the Trial Steering Committee. The aim is to have 50% recruited by months 14–16 from when the trial opens for recruitment and an assessment of feasibility for recruitment will be undertaken at this time if recruitment is below 35%.

One potential area for modification if recruitment is low is the eligibility threshold for CPAP pressure. The chosen inclusion criterion of ≥ 14 cm water was informed by the limited published data and a survey of our CPAP patients’ pressure ranges. Pressure readings for all patients screened for the trial will be recorded and if necessary we will examine the recruitment uplift potential of reducing the threshold to, e.g. 13 or 12 cm water.

From the initial review of screening results during the project team meeting of September 28th 2022, it was evident there was a significant chance of missing potentially suitable participants through pressure threshold requirements. Of the 22 not eligible due to pressure, 18 had reading < 10 cm, one had reading 10–11 cm; two had reading 12–13 cm and one had reading 13–14 cm. Following communication with TSC, it was agreed to lower the eligibility threshold to ≥ 12 cm water and amendment was completed and granted Trust approval on 16^th November 2022.

An important component of this trial is for the research team to work closely with the clinical service, specifically the CPAP practitioners (a representative will be included in the TSC). It will provide a direct line of communication to the staff conducting recruitment to get real-time feedback from patients approached (and should hopefully aid with participant retention). The CPAP team’s understanding of the patient population at Papworth will also be extremely important when it comes to adapting the logistics of the recruitment process in the event of low recruitment; and for the future rollout of any changes to clinical practice that the trial results lead to.

Randomisation company Sealed Envelope will generate the allocation sequence. Permuted block randomisation will be used to randomise patients with random permuted block sizes. The allocation ratio of the combination therapy to CPAP is 1:1. There will be no stratification.

Any investigator seeking to randomise a patient to the trial has to answer screening questions and confirm consent on the online website. Once this is complete the system will release the randomisation allocation. Sealed Envelope provides a system and generates sequences independent of the trial team.

The randomisation service will be hosted by Sealed Envelope, who will generate the allocation sequence.

Due to the nature of the intervention, patients and clinical staff cannot be blinded whilst the patient is receiving randomised therapy. However, a team of research staff will collect data on outcomes and these staff will be blinded. The interim analysis and sample size re-estimation will be done by an independent unblinded statistician so that the trial statisticians can remain blinded.

As above: due to the nature of the intervention, patients and clinical staff cannot be blinded whilst the patient is receiving randomised therapy. There is no need to unblind the trial statisticians until the trial analysis is complete.

Papworth Trials Unit Collaboration (PTUC) Data Management team will provide data management oversight for the study and will coordinate with the Statistical and Health Economics teams to design the trial case CRF and to ensure data quality.

All data will be collated into the bespoke trial database within OpenClinica, including the sleep diary for the combination treatment arm. Participants will be given the option to access a patient portal (Participate) within OpenClinica to complete the daily sleep diary and questionnaires for follow-up visits conducted remotely. If paper sleep diaries are used the relevant data will be transcribed into the trial data base.

Treatment preference will be recorded in OC.

As recruitment: the research team will work closely with the clinical service, specifically the CPAP practitioners (a representative will be included in the TSC). It will provide a direct line of communication to the staff conducting recruitment to get real-time feedback from patients approached, which we hope will aid participant retention. The availability of telephone support aims to deal with any concerns that could impact on participant retention in a timely fashion. We have anticipated there might be reticence to attend appointments in person at Papworth and therefore retained the option to conduct the research visits 2 and 3 remotely. To improve data collection for follow-ups, we have put in place processes for participants to record data on either paper or by using the personal electronic CRF system available to the study so they can complete the study questionnaires at a time that suits them, rather than during core working hours. In addition, there is an option for participants to send electronic images of the sleep diary entries to Papworth.

The investigator/clinical research team must maintain source documents (patient’s medical record) for each patient in the study, consisting of all demographic and medical information. A copy of the consent form and patient information sheet will also be uploaded into the patient’s medical record. All information in the CRFs, apart from the questionnaires, must be traceable to and consistent with the source documents in the patient’s hospital medical notes (ICH/GCP 4.9.2).

Full CPAP data downloads will occur at the end of each treatment arm. Bespoke reports of specific datasets will be generated as required and then entered directly into the eCRF.

On its return to Papworth, data from the WatchPAT device will be transferred to a secure server by research staff. The sleep study reports will be downloaded from this server as required and then directly entered into the eCRF.

All Investigators and research staff must comply with the requirements of the Data Protection Act 2018 with regards to the collection, storage, processing, and disclosure of personal information and will uphold the Act’s core principles.

All data used in the formulation of trial reports will only contain anonymised data. The Data Management lead will ensure confidentiality of data is preserved when the data is transmitted to the Sponsor and Co-Investigators.

Patient identifiable information will be stored for a maximum of 12 months after the end of the study: the data will remain stored for 15 years as per the Trust policy. The study data will be exported from OpenClinica and archived locally on Royal Papworth Hospital NHS Foundation Trust servers.

n/a: There will be no collection of biological specimens.

The economic analysis will account for short-term cost-effectiveness, using a within-trial analysis, and long-term cost-effectiveness using a decision model. The within-trial cost-effectiveness and cost-utility analyses will be conducted from the viewpoint of the NHS, personal social services, and patients. Patient-specific resource use (device cost, device fitting and maintenance, delivery of kits and devices, use of GP and hospital services, treatment for adverse events) will be measured using data extracted from patient records and patient-reported data. Shared resource use (training of nurses and patients to use devices) will be accounted for using a mix of administrative logs and staff interviews. Valuation of resources used will be based on national unit costs or, in their absence, literature or local unit cost data. Outcome measures for quality of life (EQ-5D-5L, SF36) will be patient-reported and use NICE-recommended valuation methods, whereas effectiveness (adherence) will be device reported with no further valuation. A mixed effects model will be used to estimate differences in patient costs and outcomes (i.e. adherence, QALYs). We will investigate the importance of controlling for the following variables at baseline; quality of life, health service use, time period, oxygen desaturation, ESS at baseline, age, gender and BMI. We will undertake deterministic sensitivity analysis to consider, for example, varying the price and length of life of the MAD, and the impact of missing value imputation on findings. Results will be reported as total costs and effects for each arm, incremental cost-effectiveness ratios, cost-effectiveness acceptability curves and incremental net benefit.

The structure of the long-term cost-effectiveness model will be based on McDaid et al. [14], and account for the long-term impacts, e.g. on hypertension, stroke, and road traffic accidents. The model will be parameterised using; data from the PAPMAT trial (on treatment effects, costs, health utilities, the relationship between ESS and quality of life), updated national data (for mortality), estimates from McDaid et al. where a previous update did not find better data despite reviews of hundreds of abstracts (i.e. CVD risk, OSAS, road traffic accident risk), and a new evidence review on compliance. Deterministic sensitivity analysis will include; alternative estimates from the McDaid et al. model and Sharples et al. 2014 HTA report, alternative appropriately inflated costs from Sharples 2014 report. Probabilistic sensitivity analysis will be undertaken to assess uncertainty that will be illustrated using cost-effectiveness acceptability curves.

The proposed interim analysis will be conducted when 29 patients have completed the primary endpoint, which is expected to happen at Month 21. The primary rationale behind the timing of the interim analysis is based on the minimum sample size needed to make precise estimates of the variance parameters in each arm of the trial. A range of published literature exists on the topic of minimum sample size for pilot studies, which is akin to the sample size requirements for an interim sample size re-estimation or internal pilot.

Traditionally a “rule of thumb” approach was used to set the sample size for pilot studies at around 30, though there have also been a number of papers published giving a more scientific basis for the selection of sample size for pilot studies.

For two-arm studies with a continuous outcome, Julious [19] recommends a total sample size of 24 and Keiser [20] recommends a total sample size of 20–40, whilst Sim [21] recommends a total sample size of at least 50 and Teare [22] recommends a total sample size of at least 70. Our recommendation for performing the interim sample size re-estimation after 29 patients have completed their primary endpoint sits towards the lower end of these figures. This must be carefully balanced with the duration of treatment and follow-up, particularly in the case of trials using a cross-over design, which will naturally delay the point at which the interim analysis occurs.

Sample size re-estimation is the primary task of the interim analysis. The trial statistician will prepare the analysis code to calculate the standard deviations (SD) of the crossover difference in the primary endpoint between the CPAP/MAD combination therapy and the CPAP therapy where the treatment group is labelled using a dummy randomisation list. An independent statistician will then run the code with the real randomisation list provided by the data manager and output the updated estimates of the SD for the crossover difference. The original sample size calculation will be repeated with the updated estimates to provide an updated sample size estimate which will be reported to the DMEC.

Another important task is to present a summary of the safety data so that the DEMC can evaluate if there is evidence of treatment safety or treatment harm. The frequencies of the AEs and SAEs will be summarised by the treatment group. Note that AEs that are OSA symptoms will not be reported.

In addition, information about patient recruitment and non-compliance will be reported at the interim analysis. The summary of recruitment data will be presented as well as the treatment group and month where appropriate. Non-compliance will be summarised by treatment group and the reasons for non-compliance will be reported where appropriate.

The trial statisticians and data management team will assess data quality to ensure collecting high-quality data. The completeness of study data and data completeness will be monitored regularly.

Treatment efficacy will not be assessed at the interim analysis. In the event that the DMEC request additional data analyses at the interim stage the trial statistician will be responsible for providing these (via the independent statistician if unblinding is required).

The primary analysis is an adjusted analysis, intention to treat. The only additional analysis planned beyond the primary and secondary analyses is a subgroup analysis for ODI categories, comparing treatment compliance.

Missing data will be quantified per variable (%). All essential variables for the outcomes are expected to be complete or at least low missing percentage before starting analysis.

For participants who discontinue the allocated treatment CPAP + MAD but continue to use the CPAP machine, we will use their adherence data as if they were still continuing their allocated treatment (i.e. intention to treat). For participants who discontinue the allocated treatment CPAP (and thus no longer provide adherence data), we will impute their adherence as zero hours for each remaining day. This pragmatic approach reflects the primary endpoint, which is the number of hours of CPAP adherence, and the wider research question, which is to examine if CPAP adherence is greater for participants allocated to CPAP + MAD compared to those allocated to CPAP only. More details with regards to the treatment of missing data will be given in the SAP, and in the health economics analysis plan (HEAP) in terms of costs.

The protocol and statistical analysis plan (SAP) will be published open access, before the completion of the trial. The statistical code used will be available online (through the software website GitHub) within 12 months of publication of results.

The DMEC will be led by an Independent Chair who is an expert in the field. As per NIHR guidelines, the DMEC will include an independent expert Statistician and a Clinician.

Annual DMEC meetings will review progress against the agreed milestones, recruitment and safety. The independent DMEC will: (1) review the assumptions underlying the sample size calculations and determine whether additional interim analyses of trial data should be undertaken; (2) develop clear, robust safety stopping rules based on regular (at least yearly) adverse event monitoring; (3) consider results of other interim analyses and relevant information arising elsewhere; (4) consider any requests for the release of interim trial data and advise the trial steering committee on this; and (5) make recommendations to the trial steering committee about continuation of recruitment. The trial statistician will provide the interim reports for the DMEC.

A monitoring plan for the trial is designed prior to the trial opening and is agreed by the TSC. The initial monitoring takes place soon after recruitment begins and then at six monthly intervals (unless there is a reason to increase the frequency). The trial is audited to determine whether trial-related activities are in accordance with protocol, SOPs and GCP as part of the R&D Department’s ongoing program of audits. The monitoring and audits are undertaken by appropriately trained staff from the Royal Papworth R&D department, independent of the Trials Unit and trial investigators, Monitoring and audit findings are reported to the R&D Quality & Audit Committee, which in turn reports to the R&D Directorate Committee.

Before the start of the study, or implementation of any amendment, approval of the trial protocol, protocol amendments, informed consent forms and other relevant documents will be obtained from the Regional Ethics Committee (REC) and Health Research Authority (HRA). All correspondence with the REC and RfPB will be retained in the Trial Master File (Sponsor File/Investigator Site File).

We will make the findings of the study publicly available through publication or other dissemination tools without any unnecessary delay and an honest accurate and transparent account of the study will be given. Any discrepancies from the study as planned in this protocol will be explained.

Before the start of the study, or implementation of any amendment, approval of the trial protocol, protocol amendments, informed consent forms and other relevant documents will be obtained from the Regional Ethics Committee (REC) and Health Research Authority (HRA). All correspondence with the REC and RfPB will be retained in the Trial Master File (Sponsor File/Investigator Site File).

Annual reports will be submitted to the REC in accordance with national requirements. It is the Chief Investigator’s responsibility to produce the annual reports as required.

A model consent form can be provided on request.

The authors declare that they have no competing interests.

The Principal Investigator will have access to the final dataset. The trial statisticians will have access to a blinded full dataset for analysis, whilst the health economists will have access to an unblinded full dataset for analysis. Upon request, we will provide access to a pseudo-anonymised final dataset.