An update on sleep disordered breathing in spinal cord injury

Abstract

Purpose of review

Sleep disordered breathing (SDB) is a direct consequence of tetraplegic spinal cord injury (SCI), is highly prevalent in both tetraplegia and paraplegia, and is associated with worse daytime functioning and reduced quality of life. Despite this, most people with SCI are undiagnosed and untreated for the disorder. This narrative review summarises research from the last 5 years on the epidemiology, pathophysiology, and consequences of SDB in SCI, as well as the current approaches to screening, diagnosis, and treatment of SDB in this population.

Recent findings

Previous research predominantly focussed on SDB in tetraplegia, however recent studies have established that people with paraplegia also experience substantially higher prevalence than the general population. SDB risk screening questionnaires are not helpful because SDB in SCI is so prevalent, and questionnaires alone cannot exclude true negative cases. Alternative treatments, such as mandibular advancement devices, are feasible and likely effective, and alternative care models may improve rates of diagnosis and access to treatments.

Summary

Recent research into SDB in SCI has identified novel, emergent themes, however researchers must collaborate more to achieve sample sizes that can deliver impact in this relatively rare population.

INTRODUCTION

Spinal cord injury (SCI) is a devastating neurological condition which causes life-altering changes to the person's physical, emotional, and social functioning. Depending on the level and severity, a SCI can result in partial or complete paralysis, loss of sensation, and impaired bodily functions. Quality of life following SCI is often reduced, influenced more by secondary health complications and reduced community engagement than the physical disability itself [1]. In a recent survey of more than 1500 community-dwelling people with SCI, sleep problems were amongst the top five most common secondary health conditions, and the second most likely to go untreated [2]. Over two-thirds experienced poor sleep quality, which was strongly related to lower employment, community participation, and emotional wellbeing [3]. A multitude of factors can disrupt sleep in people with SCI, such as pain, depression and anxiety, bladder routines, poor sleep habits, and a higher prevalence of sleep disorders, including insomnia, periodic limb movements, restless legs syndrome, circadian rhythm sleep-wake disorders, and sleep disordered breathing (SDB) [4].

SDB is the most common and widely studied sleep disorder in SCI and the focus of this study. The term “sleep-disordered breathing” encompasses several conditions whereby breathing during sleep is abnormal, leading to poor sleep quality and daytime sleepiness. These include obstructive sleep apnoea (OSA), central sleep apnoea (CSA), and sleep-related hypoventilation (SRH). Both OSA and CSA are characterised by repeated episodes of cessation (apnoeas) or reduction (hypopneas) in breathing, resulting in oxygen desaturation and sleep fragmentation. In OSA, these episodes are caused by narrowing or closure of the upper airway, with ventilatory drive typically maintained. Conversely in CSA, the upper airway remains open, and the respiratory events are caused by periods of diminished or absent ventilatory drive. SRH refers to periods during sleep where a reduction in ventilation leads to abnormally raised arterial carbon dioxide [5].

Several comprehensive reviews about sleep and breathing in SCI have been published; the most recent published in 2020 [4,6–10]. This review highlights the SDB in SCI literature published over the last 5 years. 

Box 1.

no caption available

EPIDEMIOLOGY OF SLEEP DISORDERED BREATHING

In 1992, Short et al. [11] reported a higher prevalence of sleep apnoea in SCI, a finding replicated by many studies with prevalence estimates ranging from 15 to 97% [12]. Our recent systematic review and meta-analysis synthesised nine studies reporting sleep apnoea prevalence in tetraplegia. The mean prevalence of at least mild (Apnoea Hypopnoea Index; AHI≥5 events per hour), moderate (AHI≥15), and severe (AHI≥30) sleep apnoea were 83% (95% CI = 73–91%), 59% (46–71%), and 36% (26–46%), respectively [12]. This is substantially higher than the general population (9–38%) [13]. Studies reporting prevalence in paraplegia were not included, however two recent publications have estimated the sleep apnoea prevalence (AHI≥5) in paraplegia to be 77% [14^▪] and 85% [15^▪], suggesting they should be afforded the same vigilance as tetraplegia.

There has been debate as to whether sleep apnoea in SCI, especially in tetraplegia, is predominantly OSA, CSA, or both. Our team collated research and clinical sleep studies (polysomnography) from 606 individuals with tetraplegia to investigate the relative contributions of central, obstructive, and hypopnea events. Prevalence of ‘predominant CSA’ (Central Apnoea Index; CAI≥5 and more central than obstructive apnoeas) and ‘any CSA’ (CAI≥5) were 4.3 and 8.4%, respectively [16]. Sankari et al. [17^▪] similarly found that 90% of people with chronic SCI (n = 91) had sleep apnoea; 5% with isolated CSA (AHI≥5 and CAI≥50% of AHI), 69% with isolated OSA (AHI≥5 and CAI<5), and 15% with combined OSA and CSA (AHI≥5 with CAI>5 but <50% of AHI). Bastos et al. [14^▪] compared the SDB characteristics of 40 people with tetraplegia, 48 with paraplegia, and 123 people without SCI undergoing rehabilitation and referred for investigation of sleep complaints. OSA, defined as OAI at least 5 events/hour was identified in 95 and 77% of people with tetraplegia and paraplegia, and CSA in 18 and 13% of people with tetraplegia and paraplegia, respectively. However, their definition of CSA (CAI≥5 event/ hour) did not exclude the presence of OSA. As such, this definition is aligned with Sankari's ‘combined OSA and CSA group’ [17^▪] and our definition of ‘any CSA’ [16]. Together, these studies demonstrate that OSA is the predominant form of sleep apnoea in SCI [14^▪,16,17^▪].

Prevalence of SRH is less well established. We recently identified SRH in 26% (n = 114) of a clinical sample with tetraplegia [16]. Bastos et al. [14^▪] classified SRH in 15.4% (n = 40) and 15.8% (n = 48) of people with tetraplegia and paraplegia, respectively; higher than the non-SCI sample (3.3%, n = 123). Di Maria et al. [15^▪] reported 38.5% (n = 52) prevalence in their retrospective cohort. However, the criteria used to identify SRH in these three studies varied enormously and all estimates were derived from populations that had been referred for investigation of sleep problems. This clinical referral bias may have overestimated the true population prevalence.

PATHOPHYSIOLOGY

It is understood that sleep apnoea is an immediate consequence of acute tetraplegic SCI [18,19], but the exact causes remain unclear [20]. In the last 5 years, a number of mechanistic and physiological experiments and comparative trials have been published [21–24].

Maresh performed a randomised cross-over comparison of placebo with the serotonin agonists, Buspirone (30 mg) and Trazdone (100 mg), for 2 weeks in five people with cervical and three with thoracic SCI and SDB [22]. The trial hypothesis tested if hypocapnic central apnoea in SCI could be ameliorated with serotonin receptor agonists as has previously been demonstrated in heart failure. As discussed by Borrelli [25] in a response letter, Buspirone appeared to increase the carbon dioxide reserve which should stabilise breathing during sleep, but a lack of reported baseline obstructive apnoea data made interpretation of the impact and the clinical utility difficult to understand.

Vivodtzev pooled baseline sleep studies from 13 people with chronic SCI (>T4) enrolled in two separate clinical trials [23]. Participants were categorised posthoc into those with and without paradoxical breathing during sleep. Those with paradoxical breathing had a higher AHI, heart rate, and a lower hypercapnic and exercise ventilatory response. The authors suggested that paradoxical breathing could have been the cause of these abnormalities, but an increase in the resistance of the upper airway during sleep, as occurs with SDB can result in paradoxical breathing. The upper airway collapsibility and/or resistance of participants was not reported, and as such it is difficult to determine the direction of causation.

High nasal resistance is associated with unopposed parasympathetic activation of the upper airway in tetraplegia [26,27] and an increased nasal resistance has been associated with worse adherence to positive airway pressure (PAP). Wijesuriya demonstrated that reducing nasal decongestion with intranasal phenylephrine improved nasal resistance, shifted the balance of respiratory events from apnoeas towards hypopnoeas in most participants, but did not significantly alter important sleep indices overall [24]. These findings illustrated a possible treatment target, but much work remains to identify longer acting nasal decongestant agents able to be used chronically to reduce sleep apnoea severity.

Pharyngeal mechanics and function were measured using 3-Tesla MRI during quiet, awake breathing in eight people with cervical SCI and OSA, 13 without SCI but with OSA, and 12 healthy controls without either SCI or OSA [21]. Despite considerable variability between participants, there was no consistent differences in respiratory-related upper airway movement in participants with SCI and OSA compared to able-bodied OSA participants, although most of those with SCI and OSA dilated their upper airways more. As with the paradoxical breathing study [23], this may reflect a compensatory response to the upper airway dysfunction, rather than a cause of OSA in SCI.

CONSEQUENCES OF SLEEP DISORDERED BREATHING IN SPINAL CORD INJURY

SDB has previously been associated with several negative outcomes in SCI, including reduced QoL, impaired cognition, and daytime sleepiness [4,28]. Most studies reported cross-sectional associations, however two clinical trials in SCI have shown that PAP treatment improves daytime sleepiness, sleep quality, and depression, suggesting a causal link between SDB and these outcomes [29^▪,30]. Strong cross-sectional associations between worse sleep quality and lower community participation and employment in SCI have been reported [3], but, no studies have investigated associations between SDB as the driver of poor sleep quality and community participation or vocational outcomes. Conti et al. [31] found significant associations between better sleep quality on the PSQI and more frequent and satisfactory participation in rehabilitation, supporting calls from review authors [32] for rehabilitation trials to measure and adjust for SDB.

The Handisom database from the Hôpital Raymond Poincaré in France was interrogated to examine associations between sleep apnoea in SCI and both nocturia [33] and Major Adverse Cardiovascular Events (MACE) [34]. Nocturia was associated with neurogenic detrusor overactivity, but not SDB, although both conditions were highly prevalent. Nocturia rates in SCI were the same as in the noninjured population with sleep apnoea. The authors attempted to assess the effect of CPAP treatment on nocturia, but the sample was too small for meaningful interpretation [33]. Despite a sample of 269 patients, only 32 people with SCI experienced a MACE in the second paper, limiting the opportunity to observe associations with SDB once confounders were accounted for. The Kaplan--Meier curve of MACE event-free time suggested a nonsignificant stepwise increase from the untreated sleep apnoea group, to those with treated OSA, to those without OSA [34]. These studies highlight the value of prospective, clinical data collection for hypothesis generation, research, and clinical care improvement.

Lastly, we note two studies that speculate on new consequences of SDB in SCI. Alsaqobi reported a case of autonomic dysreflexia where all other likely causes of dysreflexia were excluded. Severe sleep apnoea was eventually identified, and once treated with CPAP, the dysreflexia resolved [35]. Weintraub and Chen [36] hypothesised that the intra-cellular oxidative stress that occurs when apnoeas are terminated, along with other cardiovascular consequences of sleep apnoea, may be important in the development and delayed healing of pressure injuries in SCI.

The pathophysiology and consequences of SDB in SCI sections in this article all provide mechanistic insights, but overall, incomplete reporting, small sample sizes, and other methodological issues limit the impact of their findings. Adequately powered prospective trials of therapy are urgently needed.

SCREENING AND DIAGNOSIS

In the last 5 years, several studies have investigated simplified methods for detecting SDB in SCI. Di Maria et al. [15^▪] compared different screening strategies for identifying sleep apnoea in 190 patients with SCI from the previously discussed Handisome database [15^▪]. The performance of two questionnaires, the modified Screening for OSA in Tetraplegia (mSOSAT) [37] and the Epworth Sleepiness Scale (ESS) [38], for detecting moderate-severe sleep apnoea (AHI≥15 on PSG) were assessed. Using published thresholds of mSOSAT at least 5 and ESS at least 10, both questionnaires performed poorly overall, correctly classifying 70 and 40%, respectively. However, an overnight oximetry derived 3%ODI at least 13 events/h [37] demonstrated good sensitivity and specificity and correctly classified more than 80% of the sample. Another study similarly investigated the ability of two questionnaires (mSOSAT [37] and Stop-BANG [39]) and the 3%ODI at least 13 [37] to detect “treatment recommended sleep apnoea”, defined as AHI at least 15, followed by a clinical examination to consider comorbidities and symptoms and to decide if treatment is indicated (n = 24). The questionnaires also performed poorly, however the ODI accurately identified 100% of the sample [40]. Finally, Furlan et al. [41] assessed the accuracy of four commonly used sleep apnoea risk screening questionnaires in 28 people with SCI. All four questionnaires showed insufficient discriminatory ability to identify moderate-severe sleep apnoea (AHI ≥15). These three studies clearly demonstrate the inability of risk screening questionnaires to classify sleep apnoea in SCI [15^▪,40,41]. Given the high pretest probability of sleep apnoea in paraplegia and tetraplegia, screening with questionnaires is arguably pointless. However, overnight oximetry appears to be a cheap, accessible, and accurate method of identifying moderate-severe sleep apnoea in SCI when PSG is not possible [15^▪,37,40].

Only Di Maria et al. [15^▪] also investigated the accuracy of simplified methods to detect SRH, conservatively defined as transcutaneous CO₂ at least 49 mmHg for 10% of recording time, in 52 people with SCI. Morning arterial paCO₂ at least 45 mmHg, morning bicarbonate at least 27 mmol/l, and SpO₂ less than 90% for more than 5% of recording time all performed poorly. The authors argue that transcutaneous CO₂ monitoring should be routinely used in SCI to assess SRH [15^▪]. However, the evidence for this recommendation, the specificity of these cutoffs, and the subsequent modification of the risk of developing pathological as opposed to detectable SRH are unclear. As such, current best practice is arguably conservative and risks adding “unnecessary” pressure support to PAP therapy.

Feasibility of a novel smartphone system for identifying SDB in SCI was investigated in 19 people with SCI and 19 non-SCI controls [42]. The smartphone “mHealth” system had previously demonstrated excellent agreement with polysomnography and level 3 sleep studies in the general population. It utilises three signals – audio and accelerometery from the built-in smartphone device and a wireless finger oximeter – to automatically generate SDB indicators, such as AHI, ODI, time SpO₂ less than 90%, sleep position, and oral vs. nasal breathing [42]. This proof-of-concept study highlights the potential for accessible, cheap, and simple systems to detect SDB in SCI. More research is needed to validate device accuracy in SCI given the high pretest probability.

TREATMENT

The COSAQ (CPAP for OSA in Quadriplegia) trial is the only randomised controlled trial (RCT) to investigate the efficacy of PAP for treating sleep apnoea in SCI [30]. This multicentre, multinational trial of auto-titrating CPAP in acute (<1 year postinjury) tetraplegia found that CPAP did not improve the primary neurocognitive outcome (attention and information processing) but significantly reduced daytime sleepiness. Overall adherence with CPAP (>4 h/night over 3 months) was 33% (n = 26/79), comparable to other studies in SCI [43,44].

Most recently, Di Maria et al. [45] analysed predictors of CPAP use in a retrospective cohort of 60 people with SCI and moderate-severe sleep apnoea. Only average CPAP use in the first week following initiation was predictive of adherence at 1, 6 and 12 months. Our secondary analysis of the COSAQ trial similarly found strong associations with CPAP use in the first week and higher use at 3 months [44]. Two qualitative studies have sought to understand the factors influencing CPAP adherence from the perspectives of people with SCI [46,47]. Complementary to Di Maria's findings [45], the early experience of using CPAP was described as critical for ongoing use. Immediate daytime benefits of treatment, such as reduced sleepiness and improved social relationships, were important motivators for ongoing use [46,47].

To address sub-optimal PAP usage in SCI, a recent RCT (n = 63) assessed the efficacy of a behavioural intervention, combining education, goal setting, and motivational coaching, on PAP adherence in chronic SCI [29^▪]. PAP adherence (>4 h/night over 3 months) in the group receiving the 3-month intervention was compared to an educational control group. Overall PAP adherence was low (23%) and did not differ between the two groups, with the authors concluding there were no additional benefits from the behavioural intervention [29^▪]. A target of more than 4 h/night of PAP use is clearly challenging for people with SCI to achieve. Secondary analysis of the Badr trial found greater hours of PAP use was associated with significant improvements in sleep quality, sleepiness, and depression, despite the low usage overall [29^▪]. This provides additional evidence that when used, PAP effectively improves outcomes in SCI, and suggests that less than 4 h/night may still confer important benefits.

Given these challenges, research investigating alternative treatments in SCI is urgently needed. Researchers from the Czech Republic have recently published the first study investigating mandibular advancement devices (MADs) for the treatment of sleep apnoea in chronic tetraplegia. In this uncontrolled study, 40 people with SCI and moderate-severe sleep apnoea were fitted with a MAD and 35 completed the study. After 6 months, the average AHI reduced from 36 to 19 events/hour, and 49% of the cohort achieved an AHI reduction of more than 50%. Daytime sleepiness, measured by the ESS, also improved significantly at 6 months, by an average of 2.5 points. MADs were well tolerated, and self-reported adherence (>4 h/night) was 70% at 6 months. This study provides, for the first time, promising evidence that MADs are feasible in SCI and may be an effective treatment option [48^▪].

CARE MODELS

While most people with SCI have sleep apnoea, only 15–20% are treated for it [43]. The predominant care pathway involves referral of people with signs and symptoms of sleep apnoea to specialised sleep services for assessment and ongoing management [49]. However, we have shown that both people with SCI and their referring clinicians find these services difficult to access [47,49]. Furlan et al. [41] recently surveyed all sleep laboratories in Ontario to evaluate their capacity to provide care for people with SCI. Of the 71 sleep laboratories that responded, 22.5% were unable to accommodate people with SCI in wheelchairs, and the remaining 77.5% required the person to be accompanied by a caregiver to provide all overnight care [41]. To address these access problems, our group studied the unique care models of three rehabilitation centres, in Switzerland, the Netherlands, and Canada, who overcame the barriers to accessing specialist sleep services. These three services all developed “in-house” models involving simplified ambulatory assessments and treatment with PAP, with referral to specialist sleep services reserved for complicated cases (i.e. hypoventilation, co-morbid lung disease). We distilled the common elements of care models [50], and have adapted and implemented this a similar model in a SCI rehabilitation centre in Sydney, Australia. Our 12-month evaluation found that the “rehabilitation-led” care model was feasible to implement, provided additional staffing, ongoing education, and external respiratory/sleep specialist support were available [51^▪,52]. We hypothesise that rehabilitation-led care for OSA (the predominant form of SDB in SCI) is as effective as specialist sleep centres at improving outcomes such as daytime sleepiness, but more accessible. These hypotheses need to be tested in prospective, randomised trials comparing the two care models. Figure 1 summarises the common sleep apnoea care pathways available for people with SCI.

FIGURE 1.

Common sleep apnoea care pathways for people with spinal cord injury. ^aSleep study more likely to be level 1 polysomnography for respiratory/ sleep specialist led care. Treatment success more likely to be judged on ^bresolution of respiratory events for respiratory/sleep specialist led care and ^cresolution of symptoms for SCI rehabilitation led care. ABG, arterial blood gas; bi-PAP, bi-level positive airway pressure; CO₂, carbon dioxide; CPAP, continuous positive airway pressure; MAD, mandibular advancement device; RFTs, respiratory function tests; SCI, spinal cord injury.

CONCLUSION

This review has identified novel, emergent themes in recent research into SDB in SCI. It is apparent that risk screening questionnaires are not helpful because sleep apnoea in SCI is so prevalent, but importantly, home oximetry and clinical assessment will likely rule out sleep apnoea in the minority of true negative cases. More research is required to understand clinically relevant SRH and how to identify this in the community. Emerging evidence suggests that sleep apnoea prevalence is high in paraplegia as well as in tetraplegia, and OSA is substantially more prevalent in SCI than CSA. While PAP therapy remains challenging, benefits appear to be achieved with fewer than 4 h/night of usage, and alternative therapies are emerging. Arguably the largest challenge in the field is for researchers to better collaborate on trials to ensure sufficient statistical power, and to focus on testing equitable, accessible care models.

Acknowledgements

None.

Financial support and sponsorship

In the past 3 years, M.G. and D.B. have received competitive grant funding from the Australian and New Zealand Spinal Cord Society (ANZSCoS) for research referenced in this review.

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest