Sleep laboratory test referrals in Canada: Sleep Apnea Rapid Response Survey

1Centre for Centre for Chronic Disease Prevention, Public Health Agency of Canada, Ottawa, Ontario; 2Department of Medicine, University of Saskatchewan, Saskatoon, Saskatchewan; 3Sleep Disorders Laboratory, Kingston General Hospital and Department of Medicine, Queen’s University, Kingston, Ontario; 4Division of Respirology, Department of Medicine, University of Saskatchewan, Saskatoon, Saskatchewan; 5Department of Medicine, University of Alberta, Edmonton, Alberta; 6Department of Paediatrics, University of Ottawa, Ottawa; 7Child Health Evaluative Sciences, The Hospital for Sick Children,Toronto, Ontario; 8Department of Medicine, University of British Columbia, Vancouver, British Columbia Correspondence: Ms Jessica Evans, Centre for Chronic Disease Prevention, Public Health Agency of Canada, 785 Carling Avenue, Ottawa, Ontario K1A 0K9. E-mail jessica.evans@mail.mcgill.ca Obstructive sleep apnea (OSA) is a common chronic condition; its prevalence, estimated to be 3% to 7% among adult males and 2% to 5% among adult females in Western countries, along with the prevalence of its risk factors, such as obesity and aging, are increasing (1). A recent Public Health Agency of Canada (PHAC) survey indicated that 22% (5.4 million) of adult Canadians report either being diagnosed with sleep apnea (3%) or are at high risk for OSA (19%) (2). The exact proportion of high risk for OSA cases that are true cases of OSA remains unknown; however, if one conservatively assumes that only 15% of the high-risk cases have OSA (3% of the overall population), in addition to the 3% who report being diagnosed with sleep apnea, the prevalence of OSA approaches that of reported diagnosed diabetes (6%) (3). With the growing awareness of OSA and its associated outcomes, such as increased risk of systemic hypertension (4), atherosclerosis and cardiovascular events (5), and motor vehicle collisions (6) and, with evidence demonstrating the effectiveness of therapy (7-8), there has been a progressive increase in sleep laboratory tests for OSA. Sleep laboratory tests in Australia increased from 123 per 100,000 in 1995 to 308 per 100,000 in 2004 (9). In Canada, increases in sleep laboratory test rates have been correlated with increases in obesity prevalence rates over a nine-year period among patients referred to a sleep laboratory in Winnipeg, Manitoba (10). During 2001, it was estimated that at least 1.17 million people in the United States were examined for sleep disorders in 1292 sleep laboratories (11). Only one study has characterized access to in-laboratory polysomnography (PSG) in Canada and reported an overall sleep laboratory test rate of 370.4 per 100,000 Canadians, and revealed wide discrepancies in per-capita testing and wait times among provinces (12). There are no recent data regarding access to and predictors of referral for sleep laboratory diagnostic tests for OSA in Canada. The objectives of the present study were to determine the number of Canadian adults with sleep apnea or at high risk for OSA who were referred for sleep laboratory testing; determine the number of sleep oRiginAl ARtiCle


Sleep laboratory referrals and capacity in Canada
Can Respir J Vol 21 No 1 January/February 2014 e5 laboratory beds per capita in Canada; and to examine the predictors of sleep laboratory testing in the adult Canadian population.

MEtHoDS the Sleep Apnea Rapid Response, 2009 Canadian Community Health Survey
The Sleep Apnea Rapid Response (SARR) questionnaire was developed and funded by PHAC, and was conducted by Statistics Canada as a component of the 2009 Canadian Community Health Survey (CCHS).The CCHS is a cross-sectional survey that annually samples 65,000 respondents ≥12 years of age and provides estimates at the health region level.Residents of Indian reserves, Crown lands, institutions and the Territories, as well as full-time members of the Canadian Forces are not included.The CCHS uses three sampling frames to select the sample of households and generate a sample representative of the Canadian population.The area frame is a complex, two-stage stratified design in which each stratum is formed of clusters.The clusters are selected using a sampling method with a probability proportional to size; the final sample is generated by a systematic sampling of dwellings stratified according to health regions within the clusters.In addition, random household telephone numbers were selected using a list frame of telephone numbers and a random digit dialling sampling frame that complemented the area frame and that were also stratified according to health region.

Analysis
The analyses were performed using SAS Enterprise Guide version 4.1 (SAS, USA).Estimates were weighted to the Canadian population by applying sampling weights that accounted for the unequal probability of being selected into the survey.SDs and 95% CIs were estimated using the bootstrap technique (13).The bootstrap method allows for variance estimation while accounting for the complex multistage survey design.The process involves drawing repeated random samples, with replacement, from the observations to obtain a set of estimates.
Estimates associated with a coefficient of variation >33.3 or a cell size <10 were not reported due to high sampling variability.Statistical significance was defined at P<0.05.Potential predictors of reported sleep laboratory test referral were identified a priori.Bivariate log-linked binomial modelling was used to assess the relationship between the variables of interest and reported sleep laboratory test referral; prevalence rate ratios (PRR) were subsequently estimated.All variables found to be significant in the bivariate regression model, as well as all confounders were included in the final adjusted multivariable log-linked binomial regression models.Potential confounders were identified as such if they resulted in a 10% change in the relative risk for the association between the exposure and outcome (14).Potential effect modification was explored by including multiplicative interaction terms into the model; however, none of the interactions assessed were significant.Variance inflation factors were also assessed, and multicollinearity was not observed in any of the models (variance inflation factor <10) (15).

Defining select covariates
Individuals with self-reported diagnosed sleep apnea were identified as those who responded 'yes' to the following question: 'Have you been told by a health professional that you have sleep apnea?' Individuals with self-reported sleep laboratory testing referral were identified as those who responded 'yes' to the following question: 'Have you ever been referred to a sleep lab for overnight testing?' Individuals at high risk for OSA, excluding those who reported diagnosed sleep apnea, were identified using the STOP tool, which screens patients for OSA based on Snoring, Tiredness, Observed apneas and high blood Pressure (16).Given specific length and time guidelines for the CCHS Rapid Response modules, which dictated restrictions on the number and length of questions, the use of the STOP tool for estimating risk of OSA was a logistical decision.

Sleep laboratory testing capacity data
Data regarding numbers of sleep laboratories and numbers of sleep laboratory beds for each province was provided by a Canadian Sleep Society (CSS) database, which included data from provincial regulatory bodies and home care companies.Confirmation of the data was performed by telephone survey conducted in late 2010 and early 2011, and was additionally verified by members of the CSS and the Canadian Thoracic Society.Facilities that only performed home-based sleep testing with portable monitors (ie, level 3 testing) were excluded from the list.

Population characteristics
Population distributions according to various characteristics are presented in Table 1 for the total population and for individuals reporting having been referred to a sleep laboratory.The proportion of individuals who reported a sleep laboratory test referral was similar among the total population (4.8%) and those at high risk for OSA (5.1%) (Table 2).Nearly one in four individuals with diagnosed sleep apnea had not been referred to a sleep laboratory for overnight testing.

Predictors of referral for overnight sleep laboratory testing
The prevalence of reported referral to a sleep laboratory according to various population characteristics and the associated rate ratios are presented in Table 3 and Figure 1.
In the adjusted model, significant predictors of reported sleep laboratory test referral were body mass index (BMI), number of chronic conditions, regular medical doctor, age, sex and region (Figure 1).Individuals who were male, 45 to 64 years of age, were overweight or obese, had one or more of the five chronic conditions of interest, had a regular medical doctor and resided in Ontario were more likely to report sleep laboratory test referral.
The PRRs of sleep laboratory test referral for symptoms of sleep apnea, including those of the STOP tool, are reported in Table 4. Reported testing referral was greater among individuals reporting symptoms of sleep apnea -loud snoring, tiredness and observed apnea -compared with those not reporting these symptoms.Observed apnea was the strongest symptom predictor for sleep laboratory test referral.The proportion of individuals at high risk for OSA who were referred for sleep laboratory testing was nearly two times greater than in those who were not at high risk for OSA.

Chronic conditions and sleep laboratory test referral
The prevalence of high risk for OSA as well as the proportion reporting sleep laboratory test referral according to chronic condition is presented in Table 5.Compared with the prevalence of high risk of OSA (19.0%[95% CI 17.6% to 20.4%]) and the proportion reporting sleep laboratory test referral (4.8% [95% CI 4.1% to 5.5%]) in the general population, the prevalence of high risk for OSA and proportion reporting sleep laboratory test referrals was significantly higher for individuals with chronic conditions, except for migraine headaches (Table 5).
The prevalence of high risk for OSA was greatest among individuals with hypertension, chronic obstructive pulmonary disease (COPD), heart disease and diabetes, while the largest proportion reporting a sleep laboratory test was among respondents with COPD, diabetes, heart disease, and mood or anxiety disorder.Reporting sleep laboratory testing among individuals with hypertension was relatively low given the large proportion at high risk for OSA in this group.The proportion reporting a test referral did not differ significantly according to chronic condition.

geography and sleep laboratory testing referral
Geographical region region was a strong predictor of reported sleep laboratory test referrals (Table 3).The prevalence according to geographical region of reported diagnosed sleep apnea, high risk for OSA and the proportion reporting sleep laboratory test referrals are presented in Table 6.After adjustment for confounders, the proportion reporting sleep laboratory referral was significantly less in the Prairie region, Quebec and the Atlantic regions compared with Ontario.The nonsignificance of the adjusted PRR for British Columbia was potentially due to limited sample power, as indicted by the degree of variability associated with the estimate as well as the strength of the association represented by the unadjusted and adjusted point PRR estimate.The CSS database of level 1 sleep laboratories was generated using provincial regulatory databases and was supplemented by national home care company lists in which level 1 laboratory activity is not regulated.Database accuracy was verified in each province by Canadian Thoracic Society and CSS members from those regions, and all laboratories on the list were called to confirm their operation and the number of beds.It is estimated that >90% of in-laboratory (level-1) testing activity has been captured using this methodology.Conversely, unattended portable sleep monitoring (level-3 testing; excluded from the analysis) remains unregulated; it is currently difficult to ascertain the availability of portable testing in Canada.Regional discrepancies in the number of beds per 100,000 population were observed, with estimates being four to 10 times greater in Ontario compared with other provinces where sleep laboratory beds exist.The regional differences in beds per 100,000 population estimates and, subsequently, rates of sleep laboratory test referral and diagnosed sleep apnea, may reflect different practice models and public coverage among the provinces.The potential underdiagnosis of OSA is a Canada-wide issue, as is suggested by the large differences observed between the prevalence of individuals at high risk for OSA and diagnosis of sleep apnea across all provinces.The ratio of individuals at Recent Canadian guidelines recommend that portable sleep monitoring studies can be used to confirm the diagnosis of OSA and institute appropriate treatment in patients with a moderate to high pretest probability of OSA (19).As such, portable monitoring is likely a viable and cost-effective option for increasing access to sleep testing in Canada.However, portable sleep monitoring has limited application in patients with medical or psychiatric comorbid diseases as well as for the diagnosis of other forms of sleep apnea.In the majority of these patients, level-1 overnight sleep testing is required.Moreover, patients with obesity-hypoventilation currently require overnight PSG in a sleep laboratory for treatment to be initiated.Thus, even if portable monitoring becomes more widely available, the need for overnight sleep laboratory facilities is still apparent, particularly when one considers the trend of increasing obesity prevalence (and associated comorbidities).OSA management pathways should focus on early identification at the primary care level, with proper risk stratification and triaging to either in-laboratory PSG or home portable monitoring with management based on a validated ambulatory pathway protocol.
The STOP tool was validated in an older, preoperative population with higher average BMI measures as compared with the general population (16).Thus, our use of the tool for estimating the prevalence of high risk for OSA may have limited generalizability to the Canadian population.Furthermore, by using male-oriented OSA symptom criteria (eg, snoring), the tool may have been biased toward

Evans et al
Can Respir J Vol 21 No 1 January/February 2014 e10 male selection.Women may present with different OSA symptoms compared with men, and are more likely to report insomnia, fatigue, to have hypothyroidism, to be treated for depression and be more overweight than men (20,21).
The SARR questionnaire asked individuals if they had 'diagnosed sleep apnea' and, as such, we have no information regarding the type of sleep apnea.In our interpretation of the data, we assumed that the majority of these individuals had OSA.A few studies have demonstrated that the prevalence of central sleep apnea among individuals referred for sleep laboratory tests is <5% (22,23).
In the majority of provinces, sleep laboratory tests are covered by provincial heath plans; however, in some (eg, Alberta, British Columbia, Nova Scotia and Quebec), privately funded sleep laboratories exist.We included privately funded sleep laboratories in our sleep laboratory survey recognizing that, for many Canadians, the costs of such testing constitutes a significant barrier to access.
Of the estimated 5.4 million Canadian adults who have either been diagnosed with sleep apnea or who are estimated to be at high risk for OSA, 4.5 million did not report having been referred for sleep laboratory testing.Canada currently has 131 diagnostic sleep laboratories, which is inadequate given the number of individuals at high risk for OSA, and the proportion of these individuals who would potentially benefit from testing and appropriate treatment.
AutHoR ContRiButionS: J Evans performed the data analysis and contributed to interpretation and writing of the manuscript; C Rusu performed data quality review and contributed to data analysis; R Skomro and HS Driver contributed to data interpretation and writing of the manuscript; B Graham, I Mayers, J Reisman, T To and J Fleetham contributed to data interpretation and reviewing of the manuscript; L McRae contributed to survey design and implementation and review of the manuscript.

TablE 2 Proportion reporting sleep laboratory testing referral among adults ≥18 years of age Proportion reporting sleep laboratory testing referral*, % (95% CI)
*Estimates weighted to the Canadian population; † High risk for obstructive sleep apnea (OSA) excludes individuals who reported diagnosed sleep apnea and is defined by the STOP tool (see text) Figure 1) Adjusted prevalence rate ratios (PRR) for reported sleep laboratory testing referrals among adults ≥18 years of age and older.Estimates are weighted to the Canadian population; * Includes hypertension, heart disease, diabetes, stroke and mood disorder; † Excludes the Territories; ‡ Adjusted for sex, age, ethnicity, body mass index, smoking, marital status, educational attainment, household income, chronic conditions, having a regular medical doctor and region.Bolded values indicate P<0.05.vVersus Figure 2) Number of sleep testing beds per 100,000 population in Canada, 2010.Estimates do not include level 3 or pediatric testing facilities.AB Alberta; BC British Columbia; MB Manitoba; NB New Brunswick; NL Newfoundland and Labrador; NS Nova Scotia; NT Northwest Territories; NU Nunavut; ON Ontario; PE Prince Edward Island; QC Quebec; SK Saskatchewan; YT Yukon Territory Sleep laboratory referrals and capacity in Canada Can Respir J Vol 21 No 1 January/February 2014e7 mate for Canada.Overall, the number of sleep laboratory beds per 100,000 population in Canada has increased to 2.0 in 2010 from the 1.4 estimated in 2004.The largest relative increase in the number of beds was observed in the Prairie region, followed by Quebec, Ontario and British Columbia.In the Atlantic region, the number of beds per 100,000 population decreased slightly, while estimates for the Territories remained at 0. Within the regions, a high degree of provincial variability exists in the number of sleep laboratory beds per 100,000 population.ings that male sex, increased BMI and multiple comorbid chronic conditions are associated with OSA.Seventy one per cent of the individuals referred to a sleep laboratory were diagnosed with sleep apnea or were at high risk for OSA, suggesting that of the potential sleep disorders being investigated, suspicion of sleep apnea is likely the most common reason for referral to sleep laboratory.While several sleep apnea symptoms and risk factors were associated with sleep laboratory test referral, systemic hypertension, a key clinical

(3.5-8.9) 2.5 (1.3-4.9)
*Estimates weighted to the Canadian population; † Adjusted for sex, age, ethnicity, body mass index, smoking, marital status, educational attainment, household income, chronic conditions, having a regular medical doctor and region; ‡ Interpret with caution, estimate associated with high sampling variability; § Includes hypertension, heart disease, diabetes, stroke and mood disorder; ¶ Excludes the Territories feature of sleep apnea, was not a significant predictor of referral.Although 60.3% of the respondents with systemic hypertension were at high risk for OSA, only 8.3% of this population reported being referred for sleep laboratory testing, which was lower than the proportion referred for testing among several other chronic disease categories (eg, COPD, diabetes, heart disease, mood or anxiety disorder).

TablE 4 Prevalence rate ratios of sleep laboratory testing referral according to symptoms of obstructive sleep apnea (OSa) among adults ≥18 years of age
(16)ed values indicate P<0.05, 1.0 represents the reference category.*Estimatesweightedto the Canadian population; † Adjusted for sex, age, ethnicity, body mass index, smoking, marital status, educational attainment, household income, chronic conditions, regular medical doctor and region (note: ratio for hypertension not adjusted for chronic conditions due to colinearity issues); ‡ Components of the STOP tool for estimating high risk for OSA(16); § High risk for OSA excludes individuals who reported diagnosed sleep apnea and is defined by the STOP tool (see text)

TablE 6 Prevalence and rate ratios of reported diagnosed obstructive sleep apnea (OSa) and sleep laboratory testing referral according to region among adults ≥18 years of age Prevalence* (95% CI) Prevalence rate ratio for sleep laboratory testing referral* (95% CI) Reported sleep laboratory testing referral
Bolded values indicate P<0.05, 1.0 represents the reference category.*Estimates weighted to the Canadian population; † High risk for OSA excludes individuals who reported diagnosed sleep apnea and is defined by the STOP tool; ‡ Adjusted for sex, age, ethnicity, body mass index, smoking, marital status, educational attainment, household income, chronic conditions and regular medical doctor; § Interpret with caution, estimate associated with high sampling variability; ¶ Includes Alberta, Saskatchewan and Manitoba; **Includes New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador

TablE 7 Number of overnight sleep laboratories and number of sleep testing beds according to province and region, Canada, 2004 and 2010 2004 (Flemons et al [17]) 2010*
(18)imates do not include level 3 or pediatric testing facilities or data for the Territories; † Statistics Canada, 2010(18)