BACKGROUND: Obstructive sleep apnea (OSA) is associated with cardiovascular disease, decreased quality of life and mortality. Given availability of treatment, the American Academy of Sleep Medicine recommends routine evaluation for symptoms in high risk patients (e.g. BMI>35, type 2 diabetes and hypertension). Data regarding OSA evaluation in primary care is limited. We examined prevalence of sleep study orders across individuals with varying levels of risk and markers of sleep disturbance.

METHODS: We analyzed EHR and survey data collected by the PaTH Cohort Study of Healthy Lifestyles, Body Weight and Health Care. Survey data collected from primary care patients in 5 health systems were linked with de-identified EHR data collected from the PCORnet Common Data Model. Patient vital signs, comorbid conditions and sleep study orders were extracted from the EHR and analyzed in relation to survey responses. We categorized 5-point Likert responses for PROMIS-29 data related to sleep: sleep quality [worse >3]; [more: </=3; less: >3]; and fatigue [more: >48.6; less: </=48.6]. T-scores for PROMIS-29 subscales were calculated for patients overall and by weight class. Linear regression models were used to perform ANOVA and test for trends in t-scores across classes.

RESULTS: EHR data were analyzed from 3,191,225 patients; a subset of 1,070 patients completed the PROMIS-29 survey. Among the EHR cohort, 30% had a normal BMI, 31% had overweight, 29.5% had Class I/II obesity and 7.6% had Class III obesity. Mean age was 52.4 years, 58% were female, 79% were white and 11.7% were African American. The prevalence of weight-related comorbidities increased with increasing weight category; 3.1% of all patients, 6.1% of patients with Class II obesity and 8.9% of those with Class III obesity had a sleep study ordered between 2011 and 2018. Among individuals with BMI<40 (n=892), the proportion of sleep studies ordered in 2017-18 was significantly higher in those with more fatigue (16.7%) vs. those with less fatigue

(8.6%;p<0.001). However, that proportion was similar for individuals with better (11.7%) vs. worse (12.6%) self-reported sleep quality (p=0.76). Furthermore, those with greater refreshment from sleep were significantly more likely to have a sleep study ordered (16.2%) vs. those with less refreshment (6.1%,p<0.001). Among individuals with Class III obesity, the percentages with a sleep study order were similar regardless of sleep quality (21.4 vs. 19.2%;p=0.75), refreshment from sleep (21.3 vs. 19.6%;p=0.80) and fatigue (19.4 vs. 22.4%;p=0.62).

CONCLUSIONS: OSA may be underdiagnosed in high risk patients given the low rate of sleep studies ordered compared with the high prevalence of OSA in this population. A lack of correlation between sleep study orders and markers of sleep disturbance suggests that despite guideline recommendations, these tests are not being ordered based on evaluation of sleep-related symptoms.