The Top End Sleepiness Scale (TESS): A New Tool to Assess Subjective Daytime Sleepiness Among Indigenous Australian Adults

Plain Language Summary

It is estimated that there are approximately 370 million people of Indigenous descent worldwide living in over 90 countries who make up five percent of the global population. In the Australian context, about 3.3% of the population identify as Indigenous Australians. In the Northern Territory of Australia about 30% of the population identify as Indigenous Australians, and moreover, 81% of them reside in isolated rural and remote communities. Obstructive sleep apnea, a global epidemic, is also being increasingly recognised among Indigenous people. Currently, the Epworth Sleepiness Scale (ESS) is the most common tool utilised in the clinical assessment of subjective daytime sleepiness. However, some parameters in the self-administered ESS questionnaire may not be appropriate or relevant for Indigenous people, especially among those living in isolated regional and remote communities. Hence, there is a desperate need to develop a culturally relevant sleepiness assessment tool specific to Indigenous people. Therefore, we developed and prospectively evaluated a culturally and clinically relevant sleepiness assessment tool, the “Top End Sleepiness Scale” (TESS) and its correlation to apnea-hypopnea index specific to Indigenous Australian patients. The results demonstrated that the TESS tool could be adopted and potentially complement the other currently used sleep assessment tools, such as ESS, Berlin Questionnaire and OSA 50 for Indigenous population.

Introduction

The prevalence of Obstructive Sleep Apnoea (OSA) in the adult population is reported to be between 6 to 17% worldwide.^1,2 OSA can have detrimental effects on individuals’ health, being associated with increased all-cause mortality, cardiovascular disease, neurocognitive dysfunction, hypertension, stroke and metabolic syndrome.^3–8 The burden of chronic health conditions is noted to be higher not only in the Indigenous Australian population,^9–12 but also amongst other Indigenous people living in the English speaking Organisation for Economic Co-operation and Development (OECD) countries, such as New Zealand Māori’s, and First Nations people of Canada and United States of America .¹³ Moreover, occurrence of OSA is being increasingly recognised across diverse ethnic and socioeconomic spectrums,^14–16 including in the Indigenous Australians and other Indigenous populations globally.^17,18 Studies have demonstrated that Indigenous people are more likely to return positive results for the presence of OSA in comparison to their non-Indigenous counterparts.^17,19 Earlier studies from our centre, the Top End Health Service (TEHS), in the Northern Territory (NT) of Australia have documented higher prevalence of OSA among Indigenous Australian people living in the regional and remote communities of NT of Australia.^20–24

In the clinical assessment of patients presenting with sleep disorders, in particular, OSA, an integral component in day to day clinical practice, specifically, in clinical decision-making at the primary health care level is to assess individuals daytime sleepiness and the Epworth Sleepiness Scale (ESS); Supplementary Material) is a well-established tool that is widely utilised for this purpose.²⁵ The ESS has been translated into many languages other than English and has been widely adapted for use in various ethnic populations, including non-driving populations in Peru, India and China.^26–30 The ESS has also been adapted for use with low literacy populations with pictorial representations of the questionnaire.³¹

Although there is evidence in the literature to suggest that the ESS can be adapted to assess daytime sleepiness in various ethnic and socioeconomic populations, there is to our knowledge no specific tool or adaptation of the ESS for the Indigenous Australian population currently. Moreover, there is limited knowledge of the validity of the ESS among Indigenous populations in general. Some parameters in the self-administered ESS questionnaire may not be appropriate or relevant for Indigenous people, especially among those living in isolated remote and regional locations/communities. Furthermore, the deficiency of the ESS tool in providing a valid assessment of daytime sleepiness in Indigenous Australian populations has been acknowledged in previously published literature.¹⁹

Currently, the ESS is the only tool available to assess subjective day time sleepiness among Indigenous Australians and other Indigenous populations around the world. Development of a culturally appropriate sleepiness assessment tool specific to Indigenous people will facilitate appropriate clinical assessment and management of OSA among such at risk populations to recommend polysomnography (PSG) testing. Therefore, we developed and prospectively evaluated the utility of a culturally and clinically relevant sleepiness assessment tool, the “Top End Sleepiness Scale” (TESS) specific to Indigenous Australians undergoing a diagnostic PSG in the TEHS region of the NT of Australia.

Methods

Background and Setting

This prospective study was conducted at the Respiratory and Sleep service based at the Royal Darwin Hospital (RDH) and Darwin Respiratory and Sleep Health (DRSH), based at Darwin Private Hospital (DPH), a tertiary care, Flinders university affiliated teaching hospital for the TEHS region of the NT of Australia. The population profile of the Top End, NT of Australia consists of about 249,220 people of whom about 30% are of Indigenous Australian descent (the highest proportion among all Australian States and Territories), living in a vast geographical area stretched over 245,000 km² (94,595 square miles) giving a population density of 0.16 people per kilometre. The majority (81%) of Indigenous Australians and a minor proportion of non-Indigenous Australians live in remote and regional communities.^32,33

Institution Where Work Was Performed

Department of Respiratory and Sleep Medicine, Royal Darwin Hospital, Darwin, Northern Territory, Australia and Darwin Respiratory and Sleep Health, Darwin Private Hospital, Darwin, Northern Territory, Australia. 0810.

Development of the “Top End Sleepiness Scale” (TESS) Questionnaire

A focus group from the Top End NT region was formed in the development of the TESS questionnaire. The focus group consisted of eleven individuals, including Indigenous representatives (Traditional Indigenous custodians), Anthropologist, Sleep Physician, remote clinical nurse consulatants and Flinders university academics with in-depth knowledge in Indigenous Australians health issues. Indigenous Australian-specific public health resources and clinical literature materials were provided to the focus group for review prior to the development of the TESS questionnaire.

A nominal group technique³⁴ was undertaken to identify possible specific unsuitable questionnaire items within the ESS and to identify alternative culturally relevant daily activities that would likely induce the onset of sleep to a similar degree. The nominal group process concluded that the current ESS tool is unsuitable for the Indigenous Australian people living in the regional and remote communities and suggested alternative questions, simplification of language and addition of pictorial representations for each questionnaire item. The group also suggested for the pictorial representation to be generic and not to represent any specific Indigenous Australian populations appearance, as such various Indigenous Australian people by ethnic descent can vary in their appearance and corpulence or stature. Thus, these (pictorials) were represented as cartoons so that this could be adopted across wider Indigenous Australian population. Several meetings were held over a nine-month period, with each meeting lasting over one to three hours before arriving at the final approved version by the group.

The “Top End Sleepiness Scale” (TESS) Questionnaire

After several reviews and alterations of the art work (drawn by author TS), the focus group approved the (TESS) sleepiness assessment tool be tested (piloted) among the NT Indigenous Australian people. The TESS questionnaire consisted of items representing daily realistic activities/scenarios that would induce sleepiness to similar degree (ESS) for Indigenous people living in the regional and remote regions of the NT of Australia: 1. Siting and fishing, waiting for the fish to bite; 2. Sitting with one other person talking to you; 3. Riding in a car or bus for some time; 4. Sitting in a group listening to other people talk or tell stories; 5. Waiting for your turn at the medical clinic; 6. Watching television (TV). Patients were asked to score themselves on a subjective four point scale rating the chance of them falling asleep in the situation; never (score of 0), sometimes (score of 1), most times (score of 2) or always (score of 3). The final proposed tool that was approved by the group for trial in this study is illustrated in Figure 1.

Figure 1 (Continue).

Figure 1 Top End Sleepiness Scale.

Study Participants

Participants were recruited for this study between 2017 and 2020. Included were only adult Indigenous Australian patients aged 18 years and above, undergoing a diagnostic PSG from the TEHS region, willing and able to give informed consent for the study. Patients were recommended to undergo a diagnostic PSG as per the discretion of treating Respiratory and Sleep specialist following an initial clinical consultation. The patients were referred to the specialist sleep service by primary health physicians, general practitioners and specialist practitioners for assessment of clinically suspected OSA.

Questionnaire Administration

Prior to undergoing a diagnostic PSG, all study participants were asked to complete TESS and ESS questionnaires. The questionnaires were administered by the sleep technologists who were briefed/trained by the research team on the research project. Indigenous language interpreters were not used to assist in administration of the questionnaires where participants did not speak English as a first language, in order to reflect the intended use of the tool in a primary health care level. Clinical and demographic characteristics were recorded including: age, sex, height, weight, body mass index (BMI) and medical co-morbid conditions; such as the presence of diabetes, hypertension and heart disease. Further, alcohol and smoking history were also recorded.

Polysomnography (PSG) Data

All PSGs were performed at the Darwin-based sleep service (DRSH) facility accredited by the National Association of Testing Authorities, Australia (NATA) and the Australian Sleep Association (ASA). Sleep studies were performed and analysed in accordance with the American Academy of Sleep Medicine recommendations and as described in a recent publication from this centre.^20–22 Polysomnography data was extracted to include information on the severity of sleep disordered breathing using Apnoea-Hypopnoea Index (AHI) criteria: AHI <5 (normal range), AHI ≥5 and <15 (mild sleep apnoea), AHI ≥15 and <30 (moderate sleep apnoea), AHI ≥ 30 (Severe sleep apnoea).

Statistical Methods

Continuous parameters were tested for normality with the Shapiro Wilks distribution test. Age appeared normally distributed while body mass index (weight/height²), AHI, ESS and TESS score were found to deviate from normality to a significant extent and thus were treated as non-parametric. Non-parametric parameters were presented as medians (interquartile ranges (IQRs)), normally distributed parameters as means (95% confidence intervals (CIs)), and categorical parameters as numbers (%). Total TESS scores were also presented with means (95% CI’s). Median TESS scores were compared between sex, age categories (age in years 18<35, 35<45, 45<55 and ≥55), BMI categories (BMI <25, 25<30, 30<35 and ≥35) and presence of chronic disease utilising equality of medians test.

Split half reliability of the TESS was tested by correlating the sum of scores for odd questions with those of even questions using Spearman correlation coefficient, and an equality of medians test. Cronbach’s α was used to assess internal consistency and given for the scale as a whole and on deletion of each individual question. An α greater than 0.7 was considered sufficient for consistency. Consistency was further assessed with Spearman correlation coefficient between individual questions and rated as poor (0–0.3), fair (0.3–0.5), moderate (0.5–0.7) or strong (>0.7). Sensitivity, specificity, positive predictive power and negative predictive power for abnormal scores of TESS on OSA outcome and OSA severity were calculated in the normal fashion and stratified by sex. All data were analysed in STATA IC 15 (StataCorp, Texas) and alpha was set to 0.05 throughout.

Ethical Consideration

This study was approved by the Human Research Ethics Committee of the NT Department of Health/TEHS and Menzies School of Health Research. (Reference No: HREC 2017–2831). All included study participants in this study provided informed consent to participate in this study and the study was conducted in accordance with the Declaration of Helsinki. Permission to use the Epworth Sleepiness Scale for this study has been granted by respective authorities.

Results

A total of 82 Indigenous Australian patients were included in this study. Three patients declined participation. Half (50%) were male, with a mean age of 45 years (95% CI 42.9, 48.05) and the majority were obese (78%, n=63), with a median BMI 33.6 (IQR 30.86, 38.95) (Table 1). The majority of the study participants were recorded to have some form of chronic medical co-morbidities (72%, n=59). According to the AHI criteria (AHI > 5/hour), OSA was present in the majority of the study participants (90%, n=72) with most of them noted to have severe OSA (57%, n=41).

Table 1 Demographic and Clinical Parameters of the Study Participants

The TESS total score was non-parametrically distributed (Shapiro–Wilk W=0.938, p<0.001) and skewed towards zero (Figure 2). Median TESS score was 5 (IQR 3, 7), and when stratified by demographic parameters did not significantly differ (age category p=0.224, sex p=0.657, BMI category p=0.188, chronic disease p=0.912) (Table 2).

Figure 2 Frequency plot of Top End Sleepiness Scale scores. Abbreviation: TESS, Top End Sleepiness Scale.

Table 2 Median and Mean Top End Sleepiness Scale Scores by Demographic Breakdown

Individual TESS question data was available for 65 (78%) patients. All patients answered each of the six questions, bar one patient who answered five (Table 3). Median scores for each individual question did not significantly differ, nor did split half scores (2 (1, 4) vs 3 (1, 4) p=0.790) which recorded a Spearman correlation coefficient of 0.71. Cronbach’s α was 0.81 and ranged from 0.75 to 0.80 on stepwise deletion of questions.

Table 3 Breakdown of Scores for Individual Top End Sleepiness Scale Questions with Cronbach’s α and Split Half Reliability Value

Overall, correlation between scores on individual questions was fair (average Spearman’s coefficient 0.39) (Table 4). The lowest values were recorded for question one (typically poor (0.2–0.3) to fair (0.3–0.4)), and the highest for question three (typically fair (0.4–0.5)). The correlation to the total TESS score was fair for question one (0.5–0.6), and moderate (0.6–0.7) to strong (0.7–0.8) for each other question.

Table 4 Spearman Correlation Coefficient Between Questions Within Top End Sleepiness Scale and Top End Sleepiness Scale Total Score

Univariate linear regression between ESS & total AHI did not show a significant correlation, while TESS & total AHI did show a significant correlation, though the fit was poor (p=0.072, R2=0.032 and p=0.049, R2=0.037 for ESS and TESS, respectively) (Figure 3). Comparison of area under the curve statistic showed the TESS to consistently perform better in the prediction for the presence of moderate to severe OSA (AHI ≥15) compared to the ESS. For the TESS instrument assessed overall, the AUC (AUC=0.630, 95% CI 0.489, 0.771) was greater than that for the ESS (AUC=0.553 95% CI 0.409, 0.697), though the difference was not significant (p=0.149) (Figure 4A). The greatest AUCs were noted at TESS cut-off points of “3” and “8”. When utilising cut-offs of ≥3 or ≥8 for the TESS and ≥8 for the ESS, the performance of TESS remained superior (≥3 AUC=0.607, 95% CI 0.497, 0.717 and ≥8 AUC=0.619, 95% CI 0.535, 0.703 vs ESS ≥8 AUC=0.501, 95% CI 0.372, 0.630) and trended towards significance (p=0.177 and p=0.072 for TESS ≥8 and TESS ≥3, respectively) (Figure 4B).

Figure 3 Scatterplots showing the relationship between TESS (left) and ESS (right) scores and total AHI. Abbreviations: TESS, Top End Sleepiness Scale; ESS, Epworth Sleepiness Scale; AHI, apnea-hypopnea index.

Figure 4 Area under the curve (AUC) receiver operator characteristic (ROC) for TESS and ESS tools sensitivity and 1-specificity overall (A) and for cut-offs TESS ≥3, TESS ≥8 and ESS ≥8 (B). Abbreviations: TESS, Top End Sleepiness Scale; ESS, Epworth Sleepiness scale.

Sensitivity and specificity of TESS were tested using cut-offs of TESS ≥3 and TESS ≥8 (Table 5). The lower cut-off score showed significantly greater sensitivity than the higher cut-off score (84% and 32% respectively), and this was reversed for specificity (38% vs 92%). Positive predictive values were comparable for both cut-offs (76% and 90%, respectively) and when predicting severity of OSA, though the negative predictive value was higher for the lower cut-off score (50% vs 37%). Sensitivity, specificity and positive predictive power were greater among males for both cut-off scores, however, the difference in sensitivity was minor.

Table 5 Sensitivity, Specificity, Positive and Negative Predictive Values of Top End Sleepiness Scale in the Prediction of Obstructive Sleep Apnea (AHI ≥15), Severity, and by Sex Stratification

Sensitivity and specificity of the TESS tool were comparable to other common questionnaires currently used in clinical practice in the prediction of presence of moderate to severe OSA (Table 6). Positive predictive value changed when a cut-off value of TESS ≥8 was used instead of TESS ≥3 (90% and 76%, respectively), sensitivity fell and specificity increased.

Table 6 Sensitivity, Specificity, Positive and Negative Predictive Values of Top End Sleepiness Scale in the Prediction of Moderate to Severe Obstructive Sleep Apnea for Scores ≥3 and ≥8 Compared to Other Alternative Obstructive Sleep Apnea Specific Screening Questionnaires

Discussion

To the best of our knowledge, this is the first study to evaluate a culturally and clinically relevant sleepiness assessment tool for use in the clinical assessment of OSA among Indigenous Australian people. Our study demonstrates that the proposed TESS scoring scale to assess daytime sleepiness correlates reasonably in predicting the presence of OSA in comparison to the other currently available sleep assessment tools among Indigenous Australian people.

In the Australian perspective, literature evidence suggests that OSA is highly prevalent in the non-Indigenous population.^35,36 However, more recently published reports show that sleep health issues could be equally prevalent among Indigenous Australian people.^18–24,37 Moreover, a higher proportion of chronic health conditions^9,11,13 and obesity^38,39 are frequently observed among the Indigenous Australian population, and obesity is considered to be one of the main risk factors for OSA.⁴⁰ Hence, it may be reasonable to presume that in the absence of a culturally appropriate sleepiness screening tool, OSA among Indigenous people is likely to be under or undiagnosed, and thus not be treated, which may in turn potentially contribute to heightened adverse health consequences. Therefore, we believe our proposed TESS scale may be of value in screening high-risk Indigenous patients in recommending appropriate investigations, such as to undergo a diagnostic PSG to confirm the presence of OSA.

It is estimated that there are approximately 370 million people of Indigenous origin worldwide living in over 90 countries and making up 5% of the global population.^41,42 Indigenous people are noted to have poorer health status and its outcomes, and also face other unfavourable social determinants of health and geographic isolation.^{9–13,43–46} The absence of culturally appropriate screening tools for the diagnosis and management of chronic health conditions, including OSA, may lead to higher morbidity and mortality and further increase the health care costs and expenditure in these populations.⁴⁷

In the recent past, sleep health issue have been increasingly acknowledged to be a major contributor to the chronic health burden among Indigenous populations, especially in the English speaking OECD countries.^{18–24,37,48–54} Currently, in the Australian context, the ESS,²⁵ Berlin Questionnaire,⁵⁵ OSA 50, and STOP-Bang^56–58 are the main screening tools recommend in the assessment or screening of patients prior to undergoing or qualifying for a diagnostic sleep study to confirm the presence of OSA, in particular, directly through primary health care setting. Moreover, several studies have reported that ESS may lack accuracy in predicting OSA severity,^59–61 including patients with hypertension.⁶² However, some of the aforementioned assessment tools may not be applicable, appropriate or relevant to Indigenous Australian people. Moreover, a recent study assessing⁵⁸ the three screening questionnaires plus ESS at primary health care level in clinical decision-making for patients directly to undergo a diagnostic sleep study concluded that the STOP-Bang, Berlin Questionnaire, and OSA-50 questionnaires are useful only to rule in, but not to rule out presence of OSA. In a similar vein as demonstrated in this study, our proposed TESS tool is not inferior to the currently available screening questionnaires to rule in for the presence of OSA among an Indigenous cohort.

The sensitivity and specificity of the TESS demonstrated that a cut-off of ≥3 had greater sensitivity, while ≥8 had lower sensitivity, but much higher specificity. Although typically for a screening tool, it may be more meaningful to use higher sensitivity compared to higher specificity. In the authors opinion, in this context a lower sensitivity/higher specificity tool may be of greater benefit. Using a cut-off score of ≥3 with higher sensitivity may result in more tests being ordered for Indigenous people in this resource poor setting, many of which may yield negative results. Therefore, it may be reasonable to use a TESS cut-off score of ≥8 with a higher specificity in recommending or considering patients to undergo a diagnostic sleep study.

Using culturally and clinically relevant tools, adapted specifically to Indigenous Australian people such as the TESS tool proposed in this study may complement the holistic management of sleep disorders in this population. Furthermore, the TESS concept could be adapted and modified according to other ethnic Indigenous populations globally. Although our proposed TESS tool appears to be useful among the smaller sample Indigenous Australian population we tested in this study, however, the true strengths of the TESS tool will only be apparent when the TESS tool be used widely among wider Indigenous population in the primary health care level. Hence, further studies are however warranted to test the validity and acceptability of this proposed TESS tool.

Limitations of the Study

We acknowledge the number of study participants was small in our study, however, it is not surprising given the logistical, cultural and financial barriers to accessing sleep studies in this target study Indigenous Australian population. Many patients are flown to capital Darwin from remote communities which, in some circumstances are only accessible seasonally by light aircraft. The nature of the study design resulted in a high prevalence of OSA in the study group due to high pre-test clinical suspicion of OSA. The ESS was validated using multiple sleep latency test (MSLT) as an objective measure of subjective daytime sleepiness. It would have been useful in our study to validate the proposed TESS with MSLT data. However, carrying out MSLT testing purely for the purpose of such research will result in unprecedented challenges due to ethical, cultural, logistic and financial barriers, so AHI was used as the primary outcome in our study. Furthermore, during initial designing of our study we did not record Berlin questionnaire, OSA 50, of STOP-Bang in our study participants. In hindsight, we believe this information would have been very valuable and would have further strengthened our study outcome. Nevertheless, there is room for further research in validating/comparing TESS with other sleep assessment tools.

Conclusion

The TESS subjective daytime sleepiness assessment tool correlates with OSA in Top End NT Indigenous Australian patients when judged against AHI. Furthermore, the TESS tool showed sensitivity and specificity similar to that of the Berlin Questionnaire, STOP-Bang, OSA 50 and ESS. As such, utilising the TESS tool with a cut-off score ≥8 may complement other currently available screening tools, which may offer greater utility in informed referral for further assessment and management of sleep disorders among Indigenous Australian people.