Does unconscious socioeconomic bias influence tele-evaluation of obstructive sleep apnea? An exploratory analysis.

STUDY OBJECTIVES: We examined the association between a patient's income and a provider's ability to identify risk for obstructive sleep apnea (OSA) when assessed in-person versus via telemedicine. METHODS: We utilized data from a randomized interrater reliability study of 58 patients who were referred to a university sleep center. Participants volunteered their annual income bracket as part of data collection, although raters were blinded to these data. We assessed the inter-method reliability between the clinical impressions of a telemedicine rater and those of an in-person rater for pretest probability of OSA, stratified by income levels. RESULTS: Inter-method reliabilities, assessed using weighted kappa, were 0.83, 0.24, and 0.66 for subjects with low (<$50,000/year), moderate ($50,000-$100,000/year) and high (>$100,000) incomes, respectively. The kappa statistics were significantly different (p = 0.005) between the low and moderate income groups. CONCLUSIONS: There was a significant difference in the reliability values of telemedicine versus in-person assessments between the low and middle income brackets. This is despite the raters being unaware of the patients' income levels. This association might suggest possible unconscious bias in evaluating for OSA. With telemedicine in early development, it is important to create processes that will minimize bias that might result from patients' economic disparities.

Functional MRI Correlates of Sleep Quality in HIV.

OBJECTIVE: To examine resting-state functional MRI (rs-fMRI) networks related to sleep in the context of HIV infection. METHODS: rs-fMRI data were collected in 40 HIV-infected (HIV+) individuals at baseline (treatment-naive), 12 week (post-treatment) and one year timepoints. A group of 50 age-matched HIV-negative (HIV-) individuals were also imaged at baseline and one year timepoints. The Pittsburgh Sleep Quality Index (PSQI) questionnaire was administered at all timepoints. Using group independent component analysis (ICA), maps of functional networks were generated from fMRI data; from these, sleep-related networks were selected. A generalized linear model (GLM) was used to analyze if these networks were significantly associated with the PSQI score, and if this relationship was influenced by HIV status/treatment or timepoint. RESULTS: HIV+ individuals had significantly lower PSQI score after treatment (p=0.022). Networks extracted from group ICA analysis included the anterior and posterior default mode network (DMN), central executive network (CEN), bilateral frontoparietal networks (FPNs), and the anterior cingulate cortex salience network (ACC SN). We found the posterior DMN, right FPN, and ACC SN GLMs showed significantly higher goodness-of-fit after incorporating PSQI data (p = 0.0204, 0.044, 0.044, respectively). Furthermore, the correlation between ACC SN and posterior DMN connectivity was significantly decreased in the HIV+ cohort. CONCLUSION: Functional networks such as the DMN, FPN, CEN, and ACC SN are altered in poor sleep, as measured by the PSQI score. Furthermore, the relationship between these networks and PSQI is different in the HIV+ and HIV- populations.

Interrater reliability between in-person and telemedicine evaluations in obstructive sleep apnea.

STUDY OBJECTIVES: We examined how telemedicine evaluation compares to face-to-face evaluation in identifying risk for sleep-disordered breathing. METHODS: This was a randomized interrater reliability study of 90 participants referred to a university sleep center. Participants were evaluated by a clinician investigator seeing the patient in-person, then randomized to a second clinician investigator who performed a patient evaluation online via audio-video conferencing. The primary comparator was pretest probability for obstructive sleep apnea. RESULTS: The primary outcome comparing pretest probability for obstructive sleep apnea showed a weighted kappa value of 0.414 (standard error 0.090, P = .002), suggesting moderate agreement between the 2 raters. Kappa values of our secondary outcomes varied widely, but the kappa values were lower for physical exam findings compared to historical elements. CONCLUSIONS: Evaluation for pretest probability for obstructive sleep apnea via telemedicine has a moderate interrater correlation with in-person assessment. A low degree of interrater reliability for physical exam elements suggests telemedicine assessment for obstructive sleep apnea could be hampered by a suboptimal physical exam. Employing standardized scales for obstructive sleep apnea when performing telemedicine evaluations may help with risk-stratification and ultimately lead to more tailored clinical management.