Understanding the clinical management of obstructive sleep apnoea in tetraplegia: a qualitative study using the theoretical domains framework

People with tetraplegia experience a range of complications from their injury, affecting almost every system of their body. Obstructive sleep apnoea (OSA) is one such complication, with prevalence estimates of up to 83% in the acute phase, and up to 97% in the community dwelling chronic population [1, 2]. The quality of life of people with tetraplegia and OSA is up to five times the minimally important clinical difference worse than their peers without OSA [3]. OSA has been associated with daytime sleepiness, poor memory, attention and information processing in both the acute and chronic populations, and is therefore likely to impact on rehabilitation and vocational outcomes [4, 5]. Improving the management of OSA has the potential to prevent these undesirable consequences of spinal cord injury (SCI). A recent multicentre randomised controlled trial of treating CPAP following acute, traumatic tetraplegia found that while CPAP did not improve neurocognitive function, it did improve subjective daytime sleepiness [6].

Guidelines developed by the Consortium of Spinal Cord Medicine recommend diagnostic testing with polysomnography for all people with SCI with excessive daytime sleepiness or other symptoms of sleep disordered breathing [7]. These guidelines also recommend the prescription of positive airway pressure (PAP) therapy for those with a positive diagnosis of OSA. Similar recommendations have been published by the Spinal Cord Injury Rehabilitation Evidence (SCIRE) project, a Canadian research collaboration that produces evidence-based practice recommendations for health professionals working in SCI rehabilitation [8]. The SCIRE recommendations include vigilance for suggestive signs and symptoms and further testing with oximetry or polysomnography when these signs are present. Management adherent to these recommendations therefore requires routine screening for the signs and symptoms of OSA, and subsequent investigation.

Both guidelines are not explicit in fully detailing the recommended clinical practices, potentially hampering efforts by clinicians aiming to practice according to evidence-based guidelines [9]. In particular, screening practices are recommended with little indication of how, when or where screening for signs and symptoms of OSA should be undertaken. Diagnosis of OSA is recommended with polysomnography in one guideline, and polysomnography or oximetry in the other, with no indication of who would perform these tests and what the clinical criteria for diagnosis should be. Furthermore, only one guideline recommends a specific type of treatment; initially with continuous positive airway pressure (CPAP), and with bi-level PAP as a second option for those unable to tolerate CPAP.

The lack of actionable recommendations in the OSA in SCI guidelines reflects a lack of robust clinical evidence. While the guidelines are based on evidence from non-randomised studies and the expert panel consensus was reported to be strong, there is little randomised trial evidence in this setting. SCI is a relatively small and specialised clinical area, and as such, there are significant challenges for the conduct of clinical trials [10]. Thus, few guidelines in SCI are based on strong evidence. A review of knowledge translation research in SCI revealed almost all interventions were based on the findings of individual studies and expert opinion, with only one citing evidence from a randomised control trial [11]. Given the high prevalence and significant morbidity of OSA in tetraplegia, practice concordant with the best available evidence in the form of the current guidelines is important, and will contribute to reducing variation in practice and improving the clinical management of OSA.

Very little is known about the current management of OSA in chronic tetraplegia. An older study investigating OSA treatment in people with chronic SCI found that in a service providing care to approximately 600 veterans with chronic SCI, approximately 15% of people with tetraplegia had received a diagnoses of OSA [12]. Given the high prevalence estimates in this population, this is likely to reflect low screening and subsequent testing for OSA. More current research is required to determine the extent of OSA under-diagnosis in the present clinical environment.

To our knowledge, there have been no studies that systematically describe the current management of OSA in SCI, nor what influences the clinical behaviours of health professionals involved in the care of people with SCI and OSA. Anecdotally, practice is highly varied. A systematic review of barriers to physician adherence to clinical practice guidelines generally (not specifically in SCI) identified many factors that may influence practice, including lack of awareness, familiarity and dis/agreement with the guidelines, poor physician self-efficacy, low outcome expectancy, inertia of previous practice and external barriers such as lack of time, environmental factors and staff shortages [13]. Understanding the prevailing and contextual influences on clinical practice is essential for the development of any intervention aiming to improve the management of OSA in people with tetraplegia.

The Theoretical Domains Framework (TDF) is a validated and commonly used set of 12 behavioural domains for use when exploring factors that influence clinical behaviours [14]. The 12 domains of the TDF include: knowledge; skills; social/professional role and identity; beliefs about capabilities; beliefs about consequences; motivation and goals; nature of the behaviour; memory, attention and decision processes; environmental context and resources; social influences; emotion; and behavioural regulation. The TDF enables a comprehensive, theory-based approach to recognising the behaviours that need to be changed, thereby identifying opportunities for improved practice.

The aims of this study are: 1. To describe the OSA screening, diagnosis and treatment practices of specialist doctors managing the rehabilitation of people with tetraplegia. 2. To explore factors that influence the management of OSA in tetraplegia, informed by the Theoretical Domains Framework.

In-depth semi-structured interviews were conducted with specialist doctors managing the rehabilitation of people with tetraplegia in the inpatient and outpatient settings in SCI rehabilitation centres between August 2016 and March 2018. Initially names and contact details of doctors were obtained from websites of hospitals with a specialised SCI Unit. Additionally, a snowball sampling technique, where existing participants recommended future participants from among their professional networks, was utilized. In recognition that practice may vary in different regions because of cultural and healthcare model influences, a purposeful sample was drawn to include a range of regions in the Organisation for Economic Co-operation and Development (OECD; e.g. Australia/New Zealand, North America, Europe, UK). Low and middle-income countries were not included because the availability of resources and infrastructure required to undertake OSA management, and hence culture and subsequent practice, were likely to be very different. Potential participants were approached by telephone or email and invited to participate in the study.

Of the 25 doctors approached, five doctors did not respond to email; none officially declined the invitation to participate. Ethical approval for the study was obtained from the University of Melbourne (School of Health Sciences Human Ethics Advisory Group; Ethics ID 1545475). The University of Melbourne ethics committee provided ethical approval for recruitment of doctors from overseas because of the low-risk nature of the study, and the practical implications of obtaining ethics from multiple countries for the recruitment of small numbers of health professional participants (ie one doctor per site). Further, the participant recruitment strategy detailed above precluded us from pre-emptively knowing which countries we would recruit from and prospectively apply for ethical approval from each country. All participants provided written, informed consent prior to the interview.

Interviews were conducted face to face or via online video technology (e.g. Skype) or telephone at a time suitable to the participants. The researcher conducting the interviews (MG) had experience with qualitative research methods and the clinical area. An interview schedule, based on the TDF, was used to prompt the discussion and guide the analysis (see Additional file 1). The interview questions focussed on describing current practices in the identification and management of OSA, and exploring the domains of the TDF to understand factors that influence practices. See Additional file 1 for interview guide.

All interviews were audio recorded for verbatim transcription. Transcripts were de-identified and imported into NVivo qualitative data analysis software (QSR International Pty Ltd. Version 12, 2018) to aid data management and analysis. The OSA management pathway was divided into screening, diagnosis and treatment components for inpatient and outpatient services, creating six categories. Self-reported practice was initially content analysed into these six categories. Data from the first five interviews were then independently analysed by two researchers (MG and DJB) to identify common clinical practices within each of the categories. For example, for inpatient diagnosis, clinical practice was found to fall within three main clinical practices. They were: conducted by spinal unit (with three sub-clinical practices), referral to sleep specialist and not undertaken. After the first five interviews, the two researchers discussed and resolved any differences in their identified clinical practices. Some changes to wording were required to ensure consistency and clarity. The clinical practice categories are presented in Fig. 1.

This matrix formed the coding structure for the remaining interviews, and as such, six clinical practice categories were assigned for each participant. When clinicians reported more than one clinical practice (e.g. referring to sleep specialist and general practitioner in outpatients), the predominant practice was selected.

Data were next thematically coded according to the 12 constructs of the TDF to assess influences on clinical practice and develop theoretical explanations about the influences on practice. The first five interviews were analysed independently by two researchers (MG and DJB). MG and DJB met after coding two interviews, and again after five, to discuss and resolve any differences and to revise the coding guideline. The remaining interviews were analysed by MG, who discussed any instances of ambiguity with DJB.

Following coding to the TDF domains, common “belief statements” were generated. A belief statement has been defined as a collection of responses with a similar core belief about the barrier or enabler to the behaviour under investigation [15]. Comparisons in belief statements were made between participants with different clinical practices. For example, the influences identified for those routinely screening all their patients with tetraplegia were compared to those who were not. Similarly, the influences of those who were diagnosing and/or prescribing treatment for OSA themselves were compared to those who were referring to sleep specialists for OSA diagnosis and treatment.

After the first five interviews were analysed, sampling continued until saturation of themes was achieved. Saturation was defined as when five consecutive interviews had been analysed with no new belief statements emerging [16].

Interviews were conducted with 20 doctors from 20 spinal units. All doctors were employed as specialist rehabilitation consultants in stand-alone specialist SCI rehabilitation units or general rehabilitation units with substantive SCI caseloads. All had at least five years experience practicing in SCI rehabilitation medicine. Two doctors had initially trained as surgeons while 18 had completed training in physical medicine and rehabilitation. Nine of the doctors were women. Specialist sleep laboratories were on site for six doctors, available in nearby affiliated hospitals for seven, and not available for another seven. Five interviews were conducted face-to-face, 10 with online video technology, and five over telephone. Interviews ranged in length from 22 to 66 (average 41) minutes.

This is the first time the breadth of OSA management practices within a spinal unit has been investigated and documented, and the first time a behavioural model, such as the TDF, has been applied to this area of clinical medicine to explore the influences on clinical practice. We found that 40 to 50% of spinal doctors in our sample were undertaking routine screening for signs and symptoms of OSA in their patients with tetraplegia. The remainder reported being alerted to signs and symptoms from the patient, family or ward staff before any investigation for OSA. This reactive practice may have contributed to an under-diagnosis of OSA in this population. Most doctors in this study felt that routine screening for OSA was their responsibility and was a beneficial practice to either continue or initiate. A comparison of the influences on screening practices between those routinely screening and those who were not, found that time available in outpatients, resources for allied health and nursing support, and reminders to prompt screening were likely to be important.

Providing reminders is a common intervention to prompt clinicians to perform tasks such as screening for a condition or ordering an investigation. A 2012 overview of 35 systematic reviews of reminder interventions aiming to change clinical behaviours found that reminders can lead to modest improvements in clinical practice, and concluded they are an effective intervention across a range of healthcare settings [17]. If reminders for OSA screening were to be implemented in a spinal unit, consideration of the local context to determine the most suitable type of reminder would be important. Further research is needed to explore the feasibility and effectiveness of reminders as an intervention to improve rates of screening for OSA in spinal cord injury settings.

The types of screening varied significantly in the inpatient and outpatient environments, with objective tests (e.g. overnight oximetry) predominant in the inpatient unit, and questions about symptoms (e.g. daytime sleepiness, snoring) prevailing in the outpatient clinics. Two OSA screening questionnaires developed for the non-disabled sleep clinic population have been tested in the SCI and both performed poorly in identifying OSA, however their ability to identify individuals at high risk of OSA, who require objective testing has not been evaluated [3, 18]. Recently a two-stage model for identifying moderate to severe OSA in people with chronic tetraplegia has been validated and published [2]. In this model a screening questionnaire identifies patients who require further testing with overnight oximetry. This simple four-item questionnaire (SOSAT) could be applied in the outpatient setting of a spinal unit to identify high-risk individuals for further investigation.

Our group conducted a recent qualitative study, which sought to understand the experience of using CPAP to treat OSA among people with tetraplegia [19]. We found that people with tetraplegia tended not to recognise their symptoms of OSA until after they had been treated with CPAP and experienced the benefits. As a result, many did not report their symptoms to a health professional. Given this finding, along with the high prevalence of OSA, we would argue that spinal units should be routinely screening for OSA in tetraplegia [19].

Within our sample, three doctors reported that their spinal unit (15%) was predominantly managing all aspects of non-complicated OSA. Eleven (55%) described referring all patients with suspected OSA to sleep specialists for ongoing management, and six (30%) were performing some components of the diagnosis and/or treatment prescription, usually in the inpatient setting. Those referring to sleep specialists for OSA management tended to lack confidence and skills in interpreting diagnostic tests and prescribing treatments, and felt that OSA management was outside the scope of their specialty and should be managed by a sleep specialist. Seven (35% of the total sample) were also impeded by restrictive regulations from compensatory bodies that limit the diagnosis of OSA to sleep specialists. Spinal units independently managing non-complicated OSA in their patients were well resourced for staff and training, were not impeded by regulations from compensatory funding bodies, and described “clinical champions” who initiated and led the OSA program within their spinal unit. Most of the doctors who were not diagnosing and treating OSA thought that their unit could do so with additional training, equipment, and greater involvement of allied health professionals and/or nurses.

That almost half of the spinal doctors interviewed in this study were undertaking at least some diagnosis and/or treatment of non-complicated OSA suggests that it is entirely possible for spinal doctors to perform these tasks. The perception that OSA is outside of the scope of practice of a spinal doctor may be more likely to reflect local cultural influences, lack of training and resource constraints. The results of this study suggest that with adequate training and resources, spinal units that currently refer to sleep specialists for OSA management may be able to perform these practices within the unit. This is consistent with a non-randomised study in stroke survivors (reported to have a similarly high OSA prevalence to that observed in tetraplegia [20]) which demonstrated that it is feasible and safe to diagnose and treat OSA within a stroke rehabilitation environment [21].

Poor access and high costs of in-laboratory specialist sleep services to diagnose and initiate treatment for OSA have been identified as a problem in the non-disabled population [22, 23]. In response, alternative ambulatory techniques, including automated, home-based diagnosis and treatment initiation, have been compared to specialist sleep laboratory management, with all studies demonstrating non-inferiority of the alternative model [22, 24]. There have now been three non-inferiority randomised controlled trials investigating whether non-sleep specialist health professionals can effectively treat OSA in people without disability using these ambulatory techniques. Two investigated OSA management delivered in primary care settings by general practitioners and practice nurses [25, 26], with the other investigating OSA management provided by nurses in specialist sleep centres [27]. In each of the studies the alternative models were compared to the traditional sleep specialist model, and all concluded that the care provided by non-sleep specialist professionals was not inferior to that provided by the sleep specialists. As yet, there has been no research investigating alternatives to the specialist sleep model for people with tetraplegia.

Ideally, a randomised controlled trial comparing the spinal unit management of non-complicated OSA to specialist sleep laboratory management could determine whether spinal unit management is at least not inferior to the traditional model. The alternative OSA management model could be based on one, or a combination of the three, spinal units found to be independently managing OSA in this study. There are important safety and feasibility considerations, such as the identification and treatment of hypoventilation, to resolve prior to any such clinical trial. However evaluation of safety procedures at the spinal units identified in this study, and consultation with sleep specialists, should enable resolution of these concerns. In addition, our findings suggest that staff training, multi-disciplinary involvement, and resources for equipment are important components of the model. People with tetraplegia have previously identified overnight in-laboratory sleep studies as a major barrier to OSA diagnosis and subsequent treatment [19]. Managing OSA within the spinal unit could eliminate this known barrier and improve diagnosis rates.

People with tetraplegia experience high disability and disadvantage. In this context, while we recognise that knowledge translation interventions should be primarily focused on clinical areas with robust evidence-based recommendations for clinical practice, we are advocating for the translation of best available evidence into practice. We assert that routine screening for a highly prevalent condition, for which there is a relatively cheap, simple and non-invasive treatment available, is both practical and worthwhile. Given the lack of specific, actionable practice recommendations in the existing guidelines, and the wide variation in OSA management practices described in this study, more research into the feasibility and outcomes of spinal unit management of non-complicated OSA is warranted. Interventions that target the factors identified in this study are likely to improve the management of OSA, which may ultimately improve the quality of life of people living with tetraplegia.