Comparison of Sleep Apnea Assessment Strategies to Maximize TBI Rehabilitation Participation and Outcome

Overview

Background: Sleep disorders, including sleep apnea, are common after traumatic brain injury and affect recovery and negatively influence participation in rehabilitation. Sleep apnea is a breathing problem while persons sleep and causes further brain damage and problems with thinking, daily functioning, and overall health. Earlier diagnosis and treatment is important for traumatic brain injury (TBI) survivors to maximize the recovery process. There is little information that guides TBI doctors on how to identify sleep apnea during inpatient TBI rehabilitation, a phase in which people experience the potential for a rapid pace of improvement. The Agency for Healthcare Research has highlighted gaps in best methods for identifying sleep apnea and separately in helping consumers with TBI rehabilitation choices. Partnering with survivors, caregivers, and administrators, investigators developed this study to compare sleep apnea screening and diagnostic tools in TBI rehabilitation settings. This information will provide clinicians, providers, and patients with the best information for early identification of sleep apnea to remove negative influence on the pace of recovery in early phases after TBI. The Goal: Investigators will compare existing screening (Aim 1) and diagnostic tools (Aim 2) in TBI patients undergoing inpatient rehabilitation. For the second aim, investigators will determine if a more accessible diagnostic test is sufficient to diagnose sleep apnea compared to the traditional method used which is less accessible to consumers. If the more accessible test is good enough, this will increase recognition of this problem and increase patient access to earlier sleep apnea treatment. Stakeholders and Products. TBI survivors, caregivers, researchers, and policymakers working together on this study helped develop the study questions. Idea exchanges included ways to reach clinicians and TBI survivors/caregivers via existing educational programming and online tools for consumers such as fact sheets and patient/caregiver-focused videos. Other traditional methods will include targeting professional magazines, conferences, and research journals that reach professionals working with TBI survivors and their families at the time of admission to rehabilitation and during the recovery process. This study will occur at rehabilitation hospitals around the country who enroll TBI survivors into a lifetime study called the TBI Model System funded by the Department of Health and Human Services and Veterans Affairs (VA).

Background and Significance: Meta-analyses of TBI studies found prevalence of obstructive sleep apnea (OSA) was 12 times higher than community-based studies. Yet, sleep apnea in those with TBI is largely undiagnosed. Given that sleep is critical for neural repair and disordered sleep may play a role in slowing functional recovery and prolonging rehabilitation, early detection of sleep apnea is critical (Stakeholder Input). Minimal guidance exists informing TBI and sleep medicine physicians regarding assessment of sleep disorders in the acute rehabilitation setting. The Agency for Healthcare Research and Quality's (AHRQ) Comparative Effectiveness Review highlighted insufficient comparative effectiveness evidence for diagnostic and screening tools. This study proposes to leverage the existing TBI Model System research consortium (TBIMS), funded by the Departments of Health and Human Services and Veterans Affairs, to improve early detection of sleep apnea in patients with TBI undergoing inpatient rehabilitation at six study sites. Study aims are focused around components of phased testing for sleep apnea. Findings will inform clinician's decisions for screening and diagnostic tests to facilitate earlier recognition of sleep apnea (outcome) and subsequently prescribe appropriate treatments. Study Aims: Aim I: (Screening) For individuals with TBI, determine comparative effectiveness of 2 American Academy of Sleep Medicine (AASM) endorsed screening tools (STOPBang Questionnaire \[STOPBang\] vs. Actigraphy \[ACG\]) to identify those at high risk of sleep apnea as diagnosed through Level 1 polysomnography (PSG). Long-term Objectives: To inform clinician's choices for screening measures to determine presence of sleep apnea. Exploratory Aim: Evaluate alternative screening tools (Berlin Questionnaire, Multivariate Apnea Prediction Index \[MAPI\]) to Level 1 PSG in Aim 1. Aim II: (Diagnosis) This study will determine the diagnostic accuracy (non-inferiority) of Level 3 PSG in determining presence of sleep apnea in patients with acute TBI patients in the rehabilitation setting. Level 1 PSG will be used as the criterion standard. Long-term Objectives: To inform clinician's choices of diagnostic tests to determine presence of sleep apnea and prescribe appropriate treatments. If Level 3 portable PSG is equivalent to the less accessible Level 1 PSG, this will increase consumer accessibility. Study Design: This is a 3-year, multi-center prospective observational cohort study. Intervention and Comparators: In Aim I, the screening tools STOPBang and ACG will be compared against one another to determine their sensitivity and specificity in identifying patients with TBI at risk for sleep apnea. In Aim II, the non-inferiority of Level 3 PSG in determining presence and severity of sleep apnea in patients with moderate to severe TBI will be compared to Level 1 PSG. Study Population: Consecutive participants to neurorehabilitation at an existing National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR) or VA funded TBIMS Center will make up the study population. Patients will be age 16 and older (18 or older at VA sites), with moderate to severe TBI, admitted for rehabilitation in a U.S. TBIMS inpatient rehabilitation facility, and sleeping \> 2 hours/night suggesting habitual sleep duration. Based on historical data from the six study sites participating in this study, investigators in this study conservatively estimated that 70% of patients admitted for inpatient TBI rehabilitation would consent into this study (N=20/month), resulting in a 13-month enrollment estimate of 259 patients. Power analyses of each study aim suggest that a sample size of 237 is the largest sample size needed for sufficient power. The sites participating in this study aim to collectively enroll 259 patients to account for possible further attrition. Based on TBIMS data acquired over the past 25-years, 44% of the sample is female, 64% are categorized as Non-White, median age is 38 years, with predominantly moderate to severe TBI. Primary outcomes consist of: 1\) Positive predictive power of screening tools: STOPBang and ACG, 2) Equivalency in detection of sleep apnea via Level 1 and Level 3 PSG, with diagnosis of sleep apnea based on overall apnea-hypopnea index (AHI) \> 5. Analytic Methods: Aim 1: Investigators expect that STOPBang values \> 5 and abnormal ACG total sleep time (\< 5 hours with desaturation) will produce sufficiently high SEs with reasonable tradeoff in SP (noting that SE and SP are inversely related). Similarly, for Level 1 PSG, a diagnosis of sleep apnea will be made if AHI ≥ 5. Cross-tabulations of STOPBang and ACG screening for high risk of sleep apnea (positive/negative) versus Level 1 PSG diagnosis (positive/negative) will be constructed and estimates of diagnostic accuracy (i.e., sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy) will be estimated along with 95% confidence intervals. To address hypothesis 1.1, two-sided McNemar's tests will be used to compare the paired SEs and paired SPs between STOPBang and ACG assuming a significance level of α = 0.05. Aim 2: The sensitivity (SE) and specificity (SP) of Level 3 PSG (compared to Level 1 PSG, the reference gold standard) will each be estimated and compared to fixed hypothesized rates of SE and SP of 90% and 60%, respectively, considered sufficiently large enough. Level 1 PSG (reference) diagnosis of sleep apnea will be made based on AHI ≥ 5. Level 3 PSG (test) diagnosis will be made based AHI ≥ 5. A cross-tabulation of Level 3 PSG (positive/negative) versus Level 1 PSG (positive/negative) will be constructed and standard measures of diagnostic accuracy will be estimated along with 95% confidence intervals (i.e., sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy).To test for non-inferiority of the SE to 90% and the SP to 60%, a non-inferiority threshold of 0.1 is assumed. If the lower bound on the 95% confidence interval for SE is strictly above 80% and the lower bound on the confidence interval for SP is strictly above 50% then non-inferiority of Level 3 PSG (compared to Level 3 PSG) can be established.