Screening for obstructive sleep apnoea in inpatients with schizophrenia: A feasibility study

Recent commentary in this journal has debated the imperative for screening in obstructive sleep apnoea (OSA) in people with schizophrenia. In their systematic review, Myles et al. (2016) reported an estimated prevalence of 1.6–52%. The wide estimated interval is likely attributable to variations in patient population, screening methods and definitions of OSA. We believe the need for screening is underscored by high rates of metabolic syndrome, use of antipsychotic medications, the probable under-reporting of symptoms and misattribution of symptoms to drugs or psychiatric sequelae. Continuous positive airway pressure (CPAP) is a readily available treatment that may yield significant improvements to quality of life in people with schizophrenia.

While Hubbeling et al. (2017) argue that CPAP treatment for OSA has not been shown to affect cardiovascular mortality, it is important to emphasise that patients with severe OSA with significant symptoms (Epworth Sleepiness Scale [ESS] >15) and significant desaturation (>10% total sleep time with SpO₂ <80%) were excluded from the SAVE study (McEvoy et al., 2016). Numerous observational studies have linked untreated OSA to worsening cardiovascular outcomes and so we would argue that identifying severe OSA remains a priority.

Screening for OSA in the schizophrenia population is supported by the Royal Australian and New Zealand College of Psychiatrist’s Clinical Practice Guideline for Schizophrenia (Galletly et al., 2016), but an effective method is yet to be established. Myles et al. suggest the possible utility of screening questionnaires but these tools have been validated in primary care and other populations, not schizophrenia.

Our group screened psychiatry ward inpatients with schizophrenia and schizoaffective disorder diagnosed according to DSM-5 criteria. We used a combination of questionnaire (‘OSA50’) and the ResMed ApneaLink™ two-channel portable diagnostic device, comprising a pulse oximeter and nasal cannula attached to a small palm-sized monitor and 1 m elastic band. This approach was developed and validated for use in primary care with high sensitivity and specificity (Chai-Coetzer et al., 2011). We hypothesised that screening during an inpatient stay with minimally obtrusive equipment would be acceptable to clients. An inpatient setting was chosen for convenience. The study was approved by The Prince Charles Hospital Human Research Ethics Committee.

We approached 33 consecutive inpatients over 50 days from May 2017 on day 3 of their admission. Eligible participants were informed of the nature of OSA, including the typical symptoms and possible high prevalence in schizophrenia. Twelve were excluded based on pre-specified criteria: six due to safety concerns, four had sleep disorders or recent investigation and two could not give consent. Of the 21 eligible participants, nine declined and four were discharged prior to review, leaving eight (24% of approached inpatients) to undergo ApneaLink investigation. Following brief demonstration of the device to participant and nurse, seven completed the overnight study and one participant removed the device due to paranoia. Next day, participant feedback was obtained using Likert-type scale questions.

Of the seven successfully screened, results were (mean, range): age 43 years (20–69), body mass index (BMI) 22.5 (19–35), OSA50 score 1.6 (0–5), ESS 5.4 (0–13), apnoea–hypopnoea index (AHI) 2.7 (0–11) and oxygen desaturation index (3%) 3.6 (0–8). Only one of the participants was not currently on antipsychotic treatment. No patient screened positively for significant OSA (AHI ⩾ 15). Four participants provided feedback: four were satisfied with the explanation of the device, three found the device easy to apply, two reported comfortable sleep, with other responses ‘undecided’. Mental health nurses reported ease of use of the device which was incorporated as part of their usual workload.

The utility of a screening test depends on its acceptability to the participant and the prevalence of the disease in the screened population. While the ApneaLink device was well tolerated by staff and participants, uptake in the inpatient setting was poor. Most subjects who declined participation cited a lack of sleep symptoms or interest in testing. Furthermore, our consenting sample had generally normal BMI and normal ESS scores, suggesting this group had a low pre-test probability for significant OSA.

The ApneaLink device remains unvalidated in this population; however, in research submitted for publication, we have demonstrated its accuracy against Level 1 polysomnogram in a large group of unselected sleep clinic patients, including some on antipsychotic medications.

Despite our results, we believe there remains a role for screening in this group. Future screening may need to target a subgroup of mental health consumers expected to have higher rates of OSA, such as those with a higher BMI or those on high doses of antipsychotics. Consumers may be more inclined to participate when stabilised in a community setting rather than undergo testing while medicated and unwell in a noisy hospital ward. Screening tools must be validated in persons with schizophrenia and further work is needed to find the optimal screening population, setting and methodology.