Clinical and psychological predictors of obstructive sleep apnoea in an Australian adult population with severe and complex obesity

Abstract

Background

Obstructive sleep apnoea (OSA) is the most prevalent sleep-related breathing disorder in the general population, with markedly higher rates among individuals with obesity. This elevated prevalence underscores the importance of examining the various risk factors which contribute to increased rates. The current study investigated predictors of receiving a diagnosis of OSA among treatment-seeking individuals with a range of binge-eating frequencies attending a public hospital weight intervention clinic in Western Sydney, Australia.

Methods

Participants (N = 110) were comprised of adults (> 18 years old) seeking weight management treatment at a tertiary level service at Blacktown Hospital, with an average BMI of 51.1 (SD =  10.9) and average age of 46.6 years old (SD = 12.5). Eligible participants completed a series of self-report questionnaires that gathered information on demographics, including age, sex, and education level, as well as body mass index (BMI) and binge-eating frequency.

Results

Our results revealed that older age and higher BMI were significantly associated with increased odds of receiving an OSA diagnosis, while binge-eating frequency, sex, and education level did not significantly predict OSA diagnosis in this sample. Importantly, BMI significantly predicted OSA diagnosis only when binge-eating frequency was excluded from the model, suggesting a potential shared variance.

Conclusion

This study contributes to the existing literature by reinforcing the association between older age, higher BMI, and increased odds of receiving an OSA diagnosis in a medically complex obesity sample. It also emphasises the importance of routinely screening for OSA risk factors, particularly in those presenting with high BMI and older age, which may contribute to early detection, early intervention, and improved outcomes.

Introduction

Obstructive sleep apnoea (OSA) is the most prevalent sleep-related breathing disorder, with estimates ranging from 9 to 38%, and as high as 78% in elderly women and 90% in elderly men [1, 2]. The exact prevalence of OSA is not precisely known and varies across studies, largely due to underreporting and underdiagnosis. A majority of undiagnosed individuals report this to be a by-product of inattention to their symptoms (i.e., snoring, daytime fatigue, etc.) and a recognition failure by healthcare providers [3, 4]. OSA is particularly common among individuals with obesity with up to 69% of individuals with obesity presenting with any degree of OSA, highlighting the importance of understanding the various risk factors which contribute to this elevation [5, 6]. The gold standard for the diagnosis of OSA requires a sleep study and determination of the apnoea-hypopnea index (AHI; number of apnoeas and hypopneas per hour of total sleep) and their respiratory disturbance index (RDI; AHI score in addition to respiratory effort-related arousals per hour of sleep during polysomnography) [7, 8]. The severity of OSA ranges from mild (AHI ≥ 5–14), moderate (AHI 15–29), to severe (AHI ≥ 30 events per hour) [9].

Given the broad range of risk factors, OSA is associated with both economic and societal burdens [10], resulting in working limitations, healthcare costs, and therapeutic costs [11]. With increased research in this area, reliable estimates of burden are beginning to be understood; however, a more focused approach evaluating related risk factors needs to be undertaken, especially in the context of underdiagnosing and misdiagnoses [10].

Research has identified modifiable risk factors which have been linked with a higher likelihood of an OSA diagnosis, including weight and alcohol consumption [12, 13]. While these factors have been deemed modifiable, psychological factors may interfere with change and adherence. For example, binge-eating behaviour and the frequency at which this is engaged in, may interfere with weight change and obesity. The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) defines binge-eating episodes to be the act of consuming unusually large amount of food discretely (usually within any 2-hour period) and feeling a sense of loss of control during the episode [14], which may impede adherence to diet and physical activity, resulting in and maintaining higher weight [15]. Importantly, binge-eating is characterised by disinhibited eating behaviours (e.g., over-eating in response to external cues, emotions, psychological stress outside of hunger), which results in excess energy intake, related metabolic dysfunction, and increased central visceral adiposity [16–18], all of which are well-established risk factors for OSA. Furthermore, disrupted sleep patterns and nighttime eating, which are common in binge-eating episodes, may compromise sleep quality and airway stability, exacerbating presenting OSA symptoms [19, 20]. The treatment of OSA, therefore, may benefit from evaluating the psychological functioning of patients with binge-eating to improve symptom severity and overall functioning.

Thus far, research has identified several risks factors for OSA across a broad range of studies, including sociodemographic factors [21, 22, 23, 24]. These findings, however, have largely come from non-Australian populations, from the general population, or rural/minority populations within Australia.[25,26,27,28,29] In addition, participants from Australian studies have been recruited through other already established national health care initiatives (e.g., Ten to Men [28]) where the aim of the program was broad in range, or part of random sampling. To date, research has not been undertaken in a treatment-seeking sample of adults with obesity, with complex comorbidities from a public hospital setting in an urban location in Australia.

The current study aimed to examine factors that may increase the odds of receiving a diagnosis of OSA among tertiary treatment-seeking individuals with obesity and its complications in Western Sydney, Australia. Tertiary services are uniquely used as treatment for this population because they provide support through multidisciplinary teams which can treat and see more severe presentations. The use of multidisciplinary teams–rare within a non-hospital service–follow the guidelines of how treatment of higher weight should be managed. As such, we aimed to investigate whether sociodemographic factors (age, sex, educational level), body mass index (BMI), and binge-eating frequency predicted a diagnosis of OSA. In light of previous research, we hypothesised that participants’ sociodemographic information (i.e., sex, age, education) would positively predict OSA diagnosis. Given the existing literature on the relationship between obesity and OSA, we also hypothesised that individuals’ BMI would positively predict OSA diagnosis. Further, based on the understanding of the psychological mechanisms underlying binge-eating behaviour and its potential to exacerbate symptoms of OSA, we hypothesised that binge-eating frequency would positively predict OSA diagnosis, independent of BMI.

Method

Transparency and openness

The data from this study is part of the broader Relationship between Metabolic disease, Inflammation, Microbiome and Obesity (MIMO) study at Blacktown Hospital in the Western Sydney Local Health District. This data is stored in Research Electronic Data Capture (REDCap). The broader MIMO study is currently in progress. Therefore, the data used and analysed in this study are available from the corresponding author on reasonable request. The psychometric measures used in this study are freely available online or through purchase.

Participants

Participants were comprised of adults seeking weight management treatment in Western Sydney, Australia and recruitment took place over the course of 12 months. Participants were enrolled through the Blacktown Hospital Metabolic and Weight Loss Program (BMWLP) and were referred by their primary care physician or specialist if they met one of the two main criteria: BMI ≥ 35 kg/m² with a diagnosis of type 2 diabetes mellitus or BMI ≥ 40 kg/m² with at least two obesity complications (e.g., OSA, hypertension, hyperlipidaemia, severe joint disease, metabolic associated fatty liver disease, cardiovascular disease, intracranial hypertension or polycystic ovarian syndrome). Initially, 145 participants participated in the study; however, due to missing data for OSA diagnosis, 35 participants were excluded from the analysis and the final sample comprised 110 participants with an OSA diagnosis (see Fig. 1). A power analysis was not conducted, though the size of this study is comparable to other studies [29–31].

Fig. 1.

Flow diagram of participants’ inclusion and exclusion criteria

The BMWLP is a tertiary level service that provides multidisciplinary weight interventions for adults with obesity. Participants were excluded if they had a speech or cognitive impairment, neurological disorder, intellectual disability, head injury, or a mental health condition that would interfere with their ability to perform and understand the study. Eligible participants completed a series of self-report questionnaires via REDCap, which is a secure web-based software platform that has been designed for clinical and research purposes [32]. The participant questionnaires took up to 60 min to complete. This study was approved by the Western Sydney Local Health District (5450–2019/ETH01915).

Measures

Demographics

Participants’ sex, age, and education was collected through a self-report questionnaire (see Table 1). Sex was treated as a categorical (nominal) variable and was coded as 0 (male) or 1 (female). Education was treated as a categorical (ordinal) variable, ranging from 0 (< Year 10) to 6 (Master’s Degree). Age was treated as a continuous variable.

Table 1.

Characteristics of the sample

Variable	n	%
Obstructive sleep apnoea
No	37	33.6
Yes	73	66.4
Sex
Males	39	35.5
Females	71	64.5
Highest education
< Year 10	12	11.5
Year 10	21	20.2
Year 11	3	2.9
Year 12	16	15.4
College/TAFE	34	32.7
Bachelor’s Degree	13	12.5
Master’s Degree	5	4.8
	M	SD
Age	46.6	12.5
BMI	51.1	10.9
Binge-eating episodes	9.4	8.4

Note: BMI, body mass index. Binge-eating episodes were assessed across a 28-day period

Body mass index (BMI)

Participant weight and height were measured with specialised wheelchair scales equipped for individuals with high BMI. BMI was calculated by dividing weight by height (kg/m²) and was treated as a continuous variable.

Binge-eating

Binge-eating was measured using item 14 of the Eating Disorder Examination-Questionnaire (EDE-Q), and was rated on a continuous scale. The EDE-Q is a 28-item self-report measure often used in research, derived from the gold standard Eating Disorder Examination (EDE) interview [33–35]. The EDE-Q evaluates eating disorder psychopathology and is comprised of a global score and four subscale scores: Weight Concern, Eating Concern, Shape Concern, and Restraint [36, 37]. Item 13 of the EDE-Q examines how many occasions one has eaten an unusually large amount of food in the past 28 days, while item 14 assesses how many times they experienced a sense of loss of control of over eating on those occasions when they had an unusually large amount of food in the past 28 days. Higher scores indicate a higher frequency of binge-eating. As this study did not focus on all aspects of eating disorders (e.g. dietary restraint, shape concerns, etc.), only item 14 was examined in association with OSA. [12] The psychometric properties of the EDE-Q have been examined, and both its reliability and validity are supported [38, 39].

Obstructive sleep apnoea (OSA)

OSA was assessed by the bariatric endocrinologist and general physician (R.B) as part of the medical evaluation of participants enrolled in the BMWLP. The diagnosis was based on a prior diagnosis of OSA established on a formal sleep study, or by the medical records provided by the referring primary care physician, or from prior hospital and medical discharge reports or letters. In the absence of the above, if participants had an elevated Epworth Sleepiness [40] score > 10 [41] in addition to typical symptoms of OSA (i.e., daytime somnolence, waking unrefreshed, headaches, mood changes, or irritability, etc.), they were evaluated through a clinical assessment by the endocrinologist, along with an assessment of neck circumference (> 42 cm in men, > 40 cm in women) [42, 43], and were presumed to have OSA in congruence with historical data. Approximately 46% of patients had a formal diagnosis of OSA, whereas the remainder were based on the results of the ESS, daytime symptoms, and likelihood of OSA given the presence of severe obesity. The validity of diagnosis is similar to previous tertiary cohorts [44] and is justified given the prevalence of OSA in obesity populations is very high. OSA diagnosis was treated dichotomously, with each participant coded as 0 (no diagnosis) or 1 (diagnosis).

Statistical analysis

All analyses were carried out using SPSS Version 28.0.1.1. Initially, the data were inspected for outliers and missing values, and assumption testing was performed [45]. Logistic regression was chosen due to its robustness to assumptions (i.e., normality and homoscedasticity of the distribution of independent variables) relative to other regression models. The Nagelkerke R2 indicator was used to analyse the contribution of all five predictor variables to the variability of the dependent variable [46]. It is recognised that Cox and Snell R2 indicator underestimates the real value, as noted in previous literature [47–49]. Likelihood ratio (LR) chi-square test is one method for evaluating the overall model fit. A significant likelihood ratio chi‐square test indicates the model containing the predictors is an overall significant improvement in comparison to the intercept-only model [50]. The Hosmer and Lemeshow test is another method for testing the overall model fit [51]. This examines whether the observed event rates match the expected event rates in subgroups of the model population. A non-significant test result indicates a good-fitting model.

The present study analysed whether or not participants had a current diagnosis of OSA as the dependent variable (0 = no; 1 = yes). The sex, age, education level, BMI, and binge-eating frequency of each participant were used as the predictor variables. To assess the odds of receiving an OSA diagnosis, binary logistic regression was performed using sociodemographic factors (Hypothesis 1; H1), BMI (Hypothesis 2; H2), and binge-eating frequency (Hypothesis 3; H3) as explanatory variables, with odds ratios calculated for each. The final analysis used data from 110 participants based on available OSA data, and the variables were input using three analyses (Analysis 1, Analysis 2, and Analysis 3), parsing and reporting on BMI and binge-eating frequency separately because our interest was on their independent, unique effects.

Results

Participant characteristics

Participant age ranged from 19 to 75 years (M = 46.6 years, SD = 12.5), with more female participants (64.5%). BMI ranged from 30.6 to 80.3 kg/m² (M = 51.1, SD = 10.9) and 61% of participants reported engaging in ≥ 1 binge-eating episode in the previous 28 days. Participant characteristics are shown in Table 1.

Preliminary analyses

Basic assumptions for a binary logistic regression were met, including independence of errors, linearity in the logit for continuous variables, and lack of strongly influential outliers [45]. Little’s MCAR test was conducted on all variables as a missing values analysis, with a significant level > 0.05 indicating a failure to reject the null hypothesis and suggesting the data are missing completely at random [52] (p = .917) and parameter estimates are unbiased. Multicollinearity was checked for all variables and the VIF statistics were below 3, suggesting that the assumption had not been violated [53].

Binary logistic regression

The values of the regression coefficients and their statistical significance were obtained using binary logistic regression [54]. Based on the LR chi-square test, we inferred that the full model represented a significant improvement in fit relative to the null model, LR χ²(5) = 13.445, p = .020. Within this analysis, the Hosmer and Lemeshow test is non-significant, χ²(8) = 7.632, p = .470, suggesting a good fitting model. The results of the binary logistic regression analysis provided information on the impact of the independent variables on OSA diagnosis through examining the ‘Yes’ responses (see odds ratio [OR]).

In analysis 1, the full model (see Table 2) explained 21.4% (Nagelkerke R2) of the variance in OSA and correctly classified 71.6% of cases. The regression slope for age was positive and statistically significant (β = .054, p = .013), indicating that age was significantly associated with an increased likelihood of OSA diagnosis. Specifically, for each additional year of age, the odds of receiving an OSA diagnosis increased by 5.5% (OR = 1.055). In the full model, no significant associations were found for education, sex, binge-eating, and BMI, although BMI approached statistical significance (p = .051).

Table 2.

Logistic regression analysis 1: Full model

Predictors	Units	β	SE	Wald test	p	OR
Age	Years	.054	.022	6.136	.013	1.055
Sex	F/M	− .823	.644	1.633	.201	0.439
Education	Level	.075	.160	0.222	.638	1.078
Binge-eating	Episodes	− .040	.035	1.261	.261	0.961
BMI	W/H2	.057	.029	3.824	.051	1.059

Note: N = 110. BMI, body mass index

In analysis 2 (see Table 3), the predictor variables were re-examined without binge-eating in the model. Results indicated that age (β = .041, p = .037) and BMI (β = .100, p < .001) were significantly associated with an increased likelihood of receiving an OSA diagnosis. Specifically, each additional year of age was associated with a 4.2% increase in the odds of receiving an OSA diagnosis (OR = 1.042). In addition, each unit increase in BMI was associated with a 10.5% increase in the odds of receiving an OSA diagnosis (OR = 1.105). In analysis 2, no significant associations were found for education level and sex.

Table 3.

Logistic regression analysis 2: Without binge-eating

Predictors	Units	β	SE	Wald test	p	OR
Age	Years	.041	.020	4.349	.037	1.042
Sex	F/M	− .999	.514	3.774	.052	0.368
Education	Level	− .062	.137	0.201	.654	0.940
BMI	W/H2	.100	.028	12.937	< .001	1.105

Note: N = 110. BMI, body mass index

In analysis 3 (see Table 4), the predictor variables were re-examined without BMI in the model. Results indicated that only age (β = .048, p = .021) was significantly associated with an increased likelihood of receiving an OSA diagnosis. Specifically, each additional year of age was associated with a 4.9% increase in the odds of receiving an OSA diagnosis (OR = 1.049). In analysis 3, no significant associations were found for education level, sex, and binge-eating.

Table 4.

Logistic regression analysis 3: without body mass index

Predictors	Units	β	SE	Wald test	p	OR
Age	Years	.048	.021	5.363	.021	1.049
Sex	F/M	− .683	.629	1.179	.278	0.505
Education	Level	.031	.156	0.041	.840	1.032
Binge-eating	Episodes	− .049	.034	2.052	.152	0.952

Note: N = 110

Furthermore, post hoc analyses were conducted to explore the relationship between education level and OSA diagnosis, considering BMI. Results indicated a weak and non-significant negative correlation between education level and OSA diagnosis (r = –.149, p = .131). Additionally, results showed a significant but weak negative correlation between education level and BMI (r = –.204, p = .038). Overall, our findings suggest that education level is unlikely to be a meaningful predictor of OSA in this sample, and that BMI is unlikely to account for any potential association between education and OSA.

Discussion

This study was the first to examine the relationship between binge-eating frequency and obstructive sleep apnoea (OSA) in a complex cohort of adults with obesity attending a public hospital weight management program in Western Sydney, Australia. In addition, we investigated associations between OSA and sociodemographic factors (age, sex, and education), as well as BMI, given its established link in previous research. Overall, we found that age significantly predicted OSA diagnosis, and BMI was a significant predictor when binge-eating frequency was excluded from the model. However, binge-eating frequency, sex, and education level did not significantly predict OSA diagnosis in this sample.

First, contrary to our hypothesis, binge-eating frequency did not significantly predict OSA diagnosis. That is, increases in binge-eating frequency were not associated with higher odds of receiving an OSA diagnosis. Previous findings on this relationship are mixed, with most research focusing on eating patterns in the context of obesity and OSA [55,56]. Given that BMI is a well-established predictor of OSA and there is shared variance with binge-eating, the effect of binge-eating might be captured by BMI in this sample. As many individuals with OSA may present with both binge-eating difficulties and high BMI, improving clinical education and tailoring patient care may enhance diagnostic outcomes and treatment adherence in those with OSA [57, 58]. Fragmented care remains a challenge in OSA management, with delays in diagnosis and inconsistent treatment contributing to patient dissatisfaction and unequal access to services across Australia [59]. Importantly, however, binge-eating is not exclusive to obesity; those at a healthy weight may also present clinically with binge-eating behaviours [60]. Given the increasing involvement of physicians (e.g., general practitioners) in diagnosing OSA due to its complexity and high service demand, it is essential to educate physicians not only on the physical signs and symptoms, but also on the potential mental health implications that may precede and accompany an OSA diagnosis, particularly for those experiencing complications related to binge-eating [61]. Future studies are required to evaluate these relationships further, especially since the inclusion of binge-eating criteria in the latest iteration of the DSM-5.

Second, consistent with our hypothesis and previous research [62], BMI significantly predicted OSA diagnosis. Specifically, higher BMI was associated with increased odds of receiving an OSA diagnosis. This finding is consistent with previous research [63], reinforcing the co-existence of obesity and OSA as two related major health concerns, and parallels findings that obesity serves as a consistent risk factor for a diagnosis of OSA [64]. An important consideration in our findings is that BMI significantly predicted OSA diagnosis only when binge-eating frequency was excluded from the model. This pattern suggests that binge-eating frequency may share variance with BMI, potentially acting as a confounding or mediating variable. When both variables are included, the individual contribution of BMI may be supressed, reducing the likelihood of BMI emerging as a significant predictor. In contrast, when binge-eating frequency is removed, BMI becomes significant, possibly reflecting the indirect effect of binge-eating on OSA through its contribution to increased body weight. These findings underscore the importance of considering behavioural contributors to weight when examining risk factors for OSA. These results also highlight the importance of routinely assessing BMI across healthcare settings (e.g., hospital, general practitioners) through commonly used OSA screening tools (e.g. OSA50, Berlin Questionnaire, STOP-Bange questionnaire), which may contribute to improved detection of individuals at risk of developing OSA.

Third, partially congruent with our hypothesis and consistent with previous research [2], age significantly predicted OSA diagnosis, whereas sex and education did not. Specifically, older age was associated with increased odds of receiving an OSA diagnosis. In line with previous results, older age can serve as a factor in identifying at-risk participants for OSA within this population [65, 66]. This parallels the results found in a similar study that demonstrated age as a key factor in predicting moderate OSA [67]. However, contrary to our hypothesis, no significant associations were found between sex and education, and OSA. These findings may be understood within the context of our complex cohort. Given that most participants (86%) were in the highest obesity risk category (BMI ≥ 40), our sample may not be reflective of typical risk distribution. While a majority of Australian adults fall into the overweight or obesity category [68, 69], only a small percentage of adults are classified with a BMI above 40, and even less represented in related research. [13] Additionally, 50% of our sample completed high school or lower, which is considerably less than those from the South Australian Health Omnibus Survey, which reported that 37.8% of participants with OSA had completed high school or lower [70]. Although the Omnibus population survey recruited South Australian participants aged ≥ 15 years, the discrepancy in education level may account for differences in our findings. Moreover, the uneven representation of men and women in our study, might preclude sex from emerging as a meaningful predictor in our model. Future research should investigate these sociodemographic effects further to determine whether the null findings were due to sampling variance and/or driven by the very high BMI in our sample.

Clinical implications

Past research has identified the significance of understanding risk factors for OSA in attempts to minimise both economical and societal burdens [10]. In the current study, we aimed to extend this knowledge by investigating risks factors associated with OSA in an Australian sample seeking weight intervention in a public hospital. Our findings that older age and higher BMI are related to increased odds of receiving OSA diagnosis, highlight the cumulative risk for individuals already vulnerable to chronic medical conditions. In addition, they highlight the importance of routine screening for OSA, particularly for those that present with both high BMI and older age. Increased screening may contribute to early detection and early intervention, and positively benefit treatment outcomes. Furthermore, these results demonstrate the importance of promoting awareness and education among patients susceptible to OSA, which may serve to develop additional insight for patients on their physical health risk. As future studies continue to investigate other factors associated with OSA, further insights may be developed which contribute to a more comprehensive understanding of the cumulative risk factors for patients.

Limitations, strengths, and future directions

This study should be understood within the context of several limitations. Firstly, the use of self-report measures in this study highlights the risk of self-reporting errors; specifically, the accuracy of participants reporting their binge-eating episodes. Binge-eating frequencies may have been misrepresented in this study, as previous research has shown that individuals with obesity may underreport clinical symptoms due to fear of being ineligible for treatment [71]. In particular, the use of the EDE-Q as a self-report tool may not detect symptoms or capture symptom severity equally well compared to clinical interviews with a clinician (i.e., EDE), which may provide less discrepant responses than self-report questionnaires. Future research should consider this self-report bias and develop multi-method assessment processes to increase the reliability of this data. Secondly, our sample size was comprised exclusively of treatment-seeking individuals with obesity attending a tertiary weight intervention service in Western Sydney, Australia, and therefore, was not representative of the overall population. Future research and clinical interventions should aim to minimise this inequity, particularly in public intervention programs. Similarly, because of the relatively small sample size, it is possible that our study was underpowered. Consequently, this may have resulted in an increase in the type II error rate and could account for the null effect between binge-eating and OSA. Nonetheless, future research with larger sample sizes is recommended to further assess these findings in order to determine if our effects can be replicated.

The main strength of the study was the recruitment of a clinical sample within a public hospital setting. Using clinical data within a hospital population is best practice as it avoids study repetition and overworking in vulnerable populations. This also strengthens the external validity of the study, particularly to other tertiary treatment-seeking samples [72]. Secondly, because OSA was assessed medically, and not self-reported, this minimises the risk of self-reporting errors. The rate of self-reporting errors in participants with obesity and related conditions has been shown to trend toward underreporting [73], so mitigating misreporting was critical in understanding OSA diagnoses. Thirdly, the most of the variables of interest (i.e., age, sex, education, and BMI) were all concrete, which also attenuated the risk of self-reporting errors and is an important factor in reducing the self-report bias that has emerged in prior studies [74].

Conclusion

Understanding the risk factors for a diagnosis of obstructive sleep apnoea (OSA) is essential for minimising its societal and economic burden, and improving patient outcomes. This study contributes to the existing literature by examining the risk factors of OSA diagnosis in a sample seeking weight intervention within a tertiary-level public hospital program in Western Sydney, Australian. Overall, our results found that older age and higher BMI were positively associated with increased odds of OSA diagnosis, while binge-eating frequency, sex, and education level were not significantly associated with OSA diagnosis. These results highlight the importance of continued investigation into risk factors within complex clinical populations, given their relevance to public health. Routine screening of these risk factors, particularly in treatment-seeking populations with obesity, may support early detection and intervention for patients, and contribute to improved treatment outcomes.

Author contributions

E.S., R.B., and D.S. conceived the study. N.A., D.S., and E.S. analysed and interpreted the data. N.A. drafted significant parts of the original draft. All authors critically revised and approved the final manuscript.

Funding

This research was supported by an Australian Government Research Training Program Scholarship, the NSW Health Education and Training Institute Mental Health Research Award, and the Ainsworth donation.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The study was approved by the Western Sydney Local Health District Human Research Ethics Committee (5450–2019/ETH01915). All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable.

Competing interests

E.S. receives royalties from Taylor and Francis. The remaining authors declare that they have no conflicts of interest. This research was supported by an Australian Government Research Training Program Scholarship, the NSW Health Education and Training Institute Mental Health Research Award, and the Ainsworth donation.