Faculty Bibliography

Introduction: Poor adherence to evaluation and treatment of obstructive sleep apnea (OSA) is a public health challenge. Despite higher prevalence of OSA, blacks are less likely to adhere to physician-recommended OSA care than are whites. Methods: Among black patients with metabolic syndrome, we compared, in an RCT, effectiveness of a telephone-delivered culturally and linguistically tailored OSA health messages over 6 months (Intervention) versus standard patient education (Control) in improving adherence to recommended OSA care. We hypothesized that patients randomized to the intervention arm would exhibit greater adherence to OSA consultation, evaluation, and treatment than those in the control arm. We also evaluated the predictive role of baseline sociodemographics, health risks, comorbidity, and psychosocial factors on adherence status using multivariate-adjusted regression analyses. Results: 380 patients (mean age = 59yrs; 71%, women) were enrolled with 80% retention rate (intervention = 160 and control = 143). Of the sample, 69.4% of patients exposed to the intervention attended initial consultations, compared with 36.7% of patients in the control arm (p < .001); 74.7% versus 66.7% of patients in the intervention and control arms, respectively, completed diagnostic evaluations (p = 0.46), while 86.4% versus 88.9% in the intervention and control arm, respectively, adhered to OSA treatment. Based on adjusted logistic regression, patients in the intervention arm were 3.17 times (95% CI = 1.68-5.99, p < 0.001) more likely to have initial consultations, relative to controls. Treatment self-efficacy was the strongest predictor of OSA adherence (OR = 1.11, 95% CI = 1.03-1.20, p < 0.01). Adjusted models revealed no significant differences between the two arms regarding adherence to OSA evaluation and treatment. Conclusion: The culturally and linguistically tailored OSA health messages were successful in improving initial consultation for OSA diagnosis. However, once patients were in treatment, there was no difference in OSA adherence rates between the two groups

Introduction: Health intervention is successful when messages are culturally and linguistically tailored to a specific population. The current study utilized a comprehensive approach involving multiple stakeholders to develop tailored health messages to promote awareness of sleep apnea among Blacks. Methods: We engaged several stakeholders (community-based organizations, patients, and healthcare providers) to develop and implementan online sleep educational inter vention. First round of focus groups were conducted with patients (N = 35; 71% Female, 100% Black, average age 45.2 years). Next, community leaders from churches, barbershops, and other organizations (N = 8, 75% Female, 87% Black, average age 48.1 years). Finally, interviews were conducted with healthcare providers (N = 6, 16% Female, 83% White, average age 51.2 years). All data collection was focused on barriers to awareness, diagnosis and treatment of sleep apnea. This paper presents results of the qualitative analysis conducted to inform the design of this community-engaged, linguistically and culturally tailored online sleep education program. Results: Analysis illuminated key barriers preventing sleep apnea awareness, including 1) low knowledge about the connection between daytime somnolence and associated sleep difficulties, 2) embarrassment about snoring and sleep apnea, and 3) inadequate healthcare access for effective treatments. The educational tool was designed using evidence-based approaches to diagnosis and treatment of sleep apnea, while acknowledging the primary themes identified in the focus groups. The tool was then refined with feedback from stakeholders (community members, sleep medicine doctors, and health communication experts. The TASHE resource included four key components, 1) tailored, population-appropriate reading level, 2) evidence-based tips and suggestions for sleep health and sleep apnea, 3) partnership with community-based organizations, and 4) cultural context. Conclusion: A conceptual model for tailored interventions in sleep medicine has been developed and implemented based on the principles of community-engaged research to ensure acceptability of tailored health messages and sustainability of the online sleep apnea educational program. The model developed can be used to structure the design and implementation of community-based, tailored sleep education programs that aim to promote sleep health at the population level

Introduction: Although sleep apps are among the most popular commercially available health apps, little is known about how well these apps follow evidenced-based guidelines or are grounded in behavioral theory. We present an analysis of the incorporation of theoretical constructs of behavior change in popular sleep apps created for smartphones using Android and iOS platforms. Methods: Three-hundred and sixty-nine apps were initially identified (n = 272 from the Google play store; n = 97 from iTunes) using the term "sleep" in September 2015. The final sample consisted of 35 apps that met the following inclusion criteria: 1) Stand-alone functionality; 2) Sleep tracker or monitor apps ranked by 100+ users; and 3) Sleep Alarm apps ranked by 1000+ users. Only 5 (14%) of the apps included were paid apps. A coding instrument was developed to assess the presence of 19 theoretical constructs in the following four categories: 1) knowledge; 2) cognitive strategies, 3) behavior strategies, 4) emotionfocused strategies, and 5) therapeutic interventions. Two graduate students downloaded and coded 17 apps to evaluate content and reach consensus with coding procedures (IRR = .993). A "1" was assigned if a construct was present in the app and "0" if it was not. Mean scores were calculated across all apps, and comparisons were made between total scores and app ratings using R. Results: The mean behavior construct scores (BCS) across all apps was 34% (range, 5% to 84%). Behavioral constructs for realistic goal setting (85.7%), time management (77.1%), and self-monitoring (65.7%) were most common. iOS apps (33%) had higher BCS compared to android apps (28%), and a positive association was observed between BCS and user ratings, but neither was found to be statistically significant (p > 0.05). Conclusion: While the overall behavior construct scores were low, an opportunity exists to develop or modify existing apps to support sustainable sleep hygiene practices

Introduction: There is growing evidence suggesting relationships between depression and insomnia symptoms, yet scant evidence indicating directionality of those relationships or potential mediators. Evidence also suggests that social connectedness is a vital, protective factor for depression. Social connectedness may hold promise for helping sleep scientists better understand relationships between disrupted sleep and depression symptomology. Methods: The current study drew on social network analysis, an underexplored approach in sleep medicine, and survey methods to examine social connections, insomnia severity, and depression symptoms. Participants (n = 38) were 44.7% female, with an average age of 56.7 years; 86.8% of the participants self-identified as black. Bivariate correlations and logistic regression were performed to examine relationships between social connectedness, insomnia severity, and depression symptoms. Results: Of the sample, 71.1% reported insomnia and 23.7% reported depression. Participants provided responses to social network items across kin (m = 2.8 people), non-kin (m = 2.4 people), and formal networks (m = 1.6 people). Case by case agreement was strong between kin network size and depression symptomology (chi square < .05), but not between kin network and insomnia (chi square = .658). The logistic regression in the current study showed individuals with depressive symptoms were 6.75 (95% CI 1.45-31.47, p < 0.) times more likely to have severe insomnia versus individuals without clinically significant depression symptoms. There was no significant relationship in the regression between network variables and insomnia or depression symptoms. Conclusion: Our findings are consistent with previous findings and evidence on a strong, positive relationship between depression symptomology and insomnia severity. However, they are not in line with literature suggesting a positive relationship between kin networks and depression symptomology. It is of interest to explore the causal relationship between social connectedness and sleep, and how social networks might serve as a protective (or risk) factor for insomnia, and maybe depression

Introduction: rist actigraphy has been used extensively to measure objective sleep duration in sleep-related research. It has been validated against polysomngraphic and self-reported sleep measures. We sought to compare the accuracy of sleep measurements produced by three algorithms developed for wrist actigraphic scoring. Methods: A random sample of 263 participants were selected from among those participating in the Research on Obesity and Type 2 Diabetes among African Migrants (RODAM) study in Kumasi, Ghana. Each participant completed a sleep diary and wore a wrist actigraph for a period of seven days. Actigraphic data were scored using Actilife software, and th ree sepa rate validated algorithms developed separately by Sadeh, Cole-Kripke, and Jean-Louis were applied. SPSS was used to compare actigraphic sleep durations, derived from each algorithm, with self-reported sleep durations. Results: Valid actigraphic data, defined as having data for at least 5 of the 7-day period, were collected from 255 participants. Total sleep time in minutes varied substantially by sleep algorithm. Average sleep time derived from the Sadeh, Cole-Kripke, and Jean-Louis' algorithms were 346.8 (SD 49.3), 320.1 (SD 53.8), and 453.4 (SD 68.1), respectively. Participants self-reported an average of 468.3 (SD 85.7) minutes of sleep per night. The Sadeh and Cole-Kripke algorithms classified only 2.7% and 6.8% of the sample as sleeping the recommended 7 to 8 hours, respectively, whereas Jean-Louis' classified 28.2% in the 7 to 8 hour range. Conclusion: When employing actigraphy for sleep duration measurement, care should be exercised in choosing the most appropriate algorithm for scoring actigraphic data based on specific study populations in order to increase the accuracy of study results. The Jean-Louis' algorithm seems to fare better than other actigraphic scoring algorithms

Introduction: Sleep is an important domain of health. Most data come from US or European samples. This study explored sleep in a university student population in Jamaica. Methods: As part of an ongoing study, N = 361 students were administered the Sleep Practices and Attitudes Questionnaire. To broadly evaluate differences between men and women t-tests for continuous variables and chi-square tests for categorical variables were evaluated. Results: Women reported a greater sleep need on average (7.97h vs 7.57h; p = 0.047). They were also more likely to feel tired (78% vs 67%; p0.019) and share their bed with other family members (19% vs 3%; p < 0.0001). If they felt sleepy, women were more likely to nap (p = 0.002) or exercise (p = 0.028). Women reported worse stimulus control, with greater likelihood of reading (p = 0.0002), worrying (p = 0.0001), arguing (p = 0.0065), and working (p = 0.0086) in bed. Men were more likely to get sleep information from the community (p = 0.022) and trust this information (p = 0.023), as well as information from neighbors (p = 0.007). Men were more likely to consider sleep an important safety issue for pilots (p = 0.0007), drivers (p = 0.013), and less important for police (p = 0.014). Women were more likely to believe that it was important to keep a consistent bedtime (p = 0.030), that st ress affects their sleep (p = 0.006), and that sleep is impor tant for health (p = 0.008). Women were more likely to believe that sleep loss impairs sex drive (p = 0.0008) and makes you tired (p = 0.012). Men were more likely to consider sleep important if it could be shown to be related to driving (p = 0.003), and heart disease (p = 0.0007). Women were more likely to consider sleep important if it could be shown to be related to weight (p = 0.047), hypertension (p = 0.046), missed work (p = 0.019), cognitive performance (p = 0.030), concentration (p = 0.006), and diabetes (p = 0.025). Conclusion: Overall, women reported a greater sleep need and more positive beliefs about sleep, but worse sleep hygiene

Introduction: Older age is generally characterized by increased risk for chronic conditions, such as obesity, dyslipidemia, diabetes, and hypertension, and significant changes in sleep patterns. It is unclear whether sleep duration (short or long sleep), contributes to chronic conditions differentially contrasting retirees (> 65 years) and non-retirees (18-65 years). Methods: The study utilized data from the 2004-2013 National Health Interview Survey. NHIS applies a stratified multistage sample survey of the resident civilian non-institutionalized US population. Respondents provided sociodemographic and physician-diagnosed chronic conditions. We defined an unhealthy cohort as a subset of the retired population who reported at least one of four chronic conditions: obesity, dyslipidemia, diabetes, and hypertension. The healthy cohort included individuals who reported none of these conditions. Data was analyzed using SPSS 20. Results: Of the sample, 56.4% of the retirees were female and 81.7% were white. Among non-retirees, 52.0% were female and 76.5% were white. Non-retirees and retirees had an average sleep duration of 7.08 and 7.49, respectively (p < .01). Adjusted logistic regression analysis indicated that overall retirees were less likely to report short sleep (< 7hrs) [OR = .92, 95%CI = .89-.95, p8hrs) [OR = 1.89, 95%CI = 1.801.98, p < .01] compared to non-retirees. Healthy retirees had a 41% greater odds of reporting long sleep, but were no more or less likely to report short sleep, compared to non-retirees. Unhealthy retirees had a two-fold greater odds of reporting long sleep, but 5% lower odds of reporting short sleep, relative to non-retirees. Conclusion: Retirees had a higher mean sleep duration and were characterized by significantly greater odds of long sleep compared to non-retirees regardless of health status. Although retirees overall were more likely to report long sleep, those with 1 or more chronic health conditions had greater odds of reporting long sleep duration compared to healthy retirees

Introduction: Although the association between sleep disturbance and cancer is well documented, there is little evidence regarding how sleep duration among cancer survivors may be associated with other chronic diseases. Growing evidence suggests that cancer and diabetes may share common risk factors such as age, gender, race, being over weight, physical inactivity, smoking and alcohol. However, it is yet unclear how unhealthy sleep duration (a known cardiometabolic risk factor) may affect the relationship between cancer and diabetes. The aim of this study was to investigate whether sleep duration moderated the relationship between physician-diagnosed cancer and diabetes. Methods: Data was extracted from the NHIS dataset (2004-2013), providing demographics, chronic diseases and sleep duration. For the present analysis, we used a subset of individuals providing complete data for the following variables: physician-diagnosed cancer and diabetes and self-reported habitual hours of sleep. Data were analyzed to assess the moderating effect of sleep duration on cancer and diabetes risk. Results: Of the total sample of 283,086 participants, 15.8% were black and 77.2% were white; 55.7% were female and the mean age was 47.7 (18.0) years. In the first adjusted regression model, short sleep duration [< 7 hours] (Beta = 0.15, p < .001) and cancer (Beta = 0.91, p8 hours] (Beta = 0.28, p < .001) and cancer (Beta = 0.14, p < .001) were independently associated with diabetes. However, moderation analysis indicated that only long sleep significantly moderated relationships between cancer and diabetes (Beta = -0.218, S.E. = 0.055, p < .0001, 95% CI = -0.326-0.110). Short sleep did not significantly moderate those relationships. Conclusion: Our findings demonstrate significant associations of short and long sleep with cancer and diabetes. We should note that among people with long sleep, having a cancer diagnosis did not increase diabetes r isk. However, among people with a cancer diag nosis, short sleep seemed to have increased diabetes risk

Introduction: There is growing use of electronic cigarettes (e-cigarettes) as an alternative to cigarette smoking. However, little is known about the cardiometabolic consequences of e-cigarette use. The current study therefore evaluated the associations between e-cigarette use, and diabetes as well as the interactions between e-cigarette use, and difficulty falling asleep. Methods: We used data from the 2014 National Health Interview Survey (NHIS) supplement. The NHIS is an in-person household survey that provides estimates on health indicators, health care utilization and access, and health-related behaviors of the civilian, non-institutionalized US population. The main independent variable for the study was use of electronic cigarettes. Description of demographics was performed for the studied population using survey frequency procedures. The moderating role of sleep on electronic cigarette and diabetes was also assessed using interaction effects. Results: From the 36,521 survey participants included in the study, mean age = 49.3 years +/- 0.10 (+/- SEM), mean BMI kg/m(2) = 27.9 +/- 0.03. Of the entire sample, 52% were female; White 66.9%; 11.9% Black; 5.6% Asian; and Hispanics 15.6%. The rate of electronic cigarette use was 12.8%. Two percent of these individuals used electronic cigarettes alone, 10.8% were combined e-cigarette users and cigarette smokers, 28% st r ict non-electronic cigarette smokers, and 59.4% non-e-cigarette use/non-smoker. After multivariate logistic regression analysis, strict electronic cigarette users had lower odds of having diabetes than nonelectronic-cigarette users/non-smokers, OR [95% CI] = 0.33 [0.160.69]; p = 0.01. There was a significant interaction effect for str ict users of electronic cigarettes and difficulty falling asleep on diabetes (beta estimate = 0.7428, p < 0.01). There were no significant interactions for strict non-electronic-cigarette smokers and difficulty falling asleep on diabetes (beta estimate = 0.2558, p < 0.10). Conclusion: The use of e-cigs lowered the risk of diabetes, compared to non-electronic-cigarette users/non-smokers. However, among those strict e-cig users who have difficulty falling asleep, there was increased association with diabetes. Further studies are needed to confirm these findings and to examine the underling mechanisms of these associations. Given the overall serious and well established health hazards, we do not endorse any form of smoking

Introduction: Previous research has shown that Blacks/African-Americans (vs Non-Hispanic Whites) are more likely to be obese, suffer from stroke, and experience short sleep(SS8hrs/day) durations. Also, sleep duration itself is related to obesity and stroke risk, and the relationship between sleep duration and obesity is stronger in Blacks/African-Americans. This study explored the mediating role of obesity on relationships of SS and LS with stroke, while also contrasting traditional and newer multivariate machine modeling approaches. Methods: Data from the National Health Interview Survey from 20042013 (N = 288,888) was used. Structural equation modeling (SEM) and Bayesian Belief Network (BBN) analysis assessed the mediating effects of BMI on the relationship between SS, LS, and stroke, and whether race/ethnicity differences in obesity moderated relationships. Covariates included age, gender, marital status, and income. Results: Based on SEM results, BMI positively mediated relationships between SS and stroke (Path Coefficient Estimates < 0.027;p < 0.001), and between LS and stroke (Path Coefficient Estimates = 0.024; p < .001), adjusting for covariates. In SEM, race/ethnicity did not significantly moderate relationships between SS or LS and obesity. In contrast, BBN analysis showed that these relationships differed between blacks and whites. Blacks who were SS and obese had a 5.14% stroke probability, while white counterparts had a 3.73% stroke probability, with a significant difference of 37.8% (p < 0.001). Blacks who were LS and obese had an 11.71% stroke probability compared to whites with an 8.66% stroke probability and a significant difference of 35.21% (p < 0.001). Conclusion: No racial/ethnic influences on the mediating effect of BMI on the sleep-CVD relationship were detected using SEM. However, BBN analysis (but not SEM) showed racial/ethnic influences on the mediating effect of BMI on the sleep-stroke relationship, suggesting that obese blacks who reported short or long sleep were at greater risk for stroke. Findings also highlight the power of BBN analysis to elucidate disparities in complex chronic diseases