Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Dec 1.
Published in final edited form as: J Sleep Res. 2017 May 31;26(6):747–755. doi: 10.1111/jsr.12559

Effects of a lifestyle intervention on REM sleep-related OSA severity in obese individuals with type 2 diabetes

Ari Shechter 1, Gary D Foster 2,3, Wei Lang 4, David M Reboussin 4, Marie-Pierre St-Onge 5, Gary Zammit 6, Anne B Newman 7, Richard P Millman 8, Thomas A Wadden 2, John M Jakicic 9, Elsa S Strotmeyer 7, Rena R Wing 10, F Xavier Pi-Sunyer 5, Samuel T Kuna 11,12; the Sleep AHEAD Research Group of the Look AHEAD Research Group
PMCID: PMC5705337  NIHMSID: NIHMS869485  PMID: 28560832

Abstract

The current aim was to determine if an intensive lifestyle intervention (ILI) reduces the severity of obstructive sleep apnea (OSA) in rapid-eye movement (REM) sleep, and to determine if longitudinal changes in glycemic control are related to changes in OSA severity during REM sleep over a 4-yr follow-up. This was a randomized-controlled trial including 264 overweight/obese adults with type 2 diabetes (T2D) and OSA. Participants were randomized to an ILI targeted to weight loss or a diabetes support and education (DSE) control group. Measures included anthropometry, apnea-hypopnea index (AHI) during REM sleep (REM-AHI) and non-REM sleep (NREM-AHI), and glycated hemoglobin (HbA1c) at baseline and year-1, year-2, and year-4 follow-ups. Mean baseline values of REM-AHI were significantly higher than NREM-AHI in both groups. Both REM-AHI and NREM-AHI were significantly more reduced in ILI vs. DSE, but these differences were slightly attenuated after adjustment for weight changes. Repeated-measure mixed-model analyses including data through year 4 demonstrated that changes in HbA1c were significantly related to changes in weight but not to changes in REM-AHI and NREM-AHI. Compared to control, the ILI reduced REM-AHI and NREM-AHI across the 4-year follow-up. Weight, as opposed to REM-AHI and NREM-AHI, were related to changes in HbA1c. The findings imply that weight loss from a lifestyle intervention is more important than reductions in AHI for improving glycemic control in T2D patients with OSA.

Keywords: obstructive sleep apnea, obesity, type 2 diabetes, HbA1c, REM sleep

Introduction

Obesity and obstructive sleep apnea (OSA) are associated with type 2 diabetes (T2D), with a high prevalence of OSA (86%) within obese T2D patients (Foster et al., 2009b). OSA may be causally linked to T2D and impaired glycemic control. A significant association exists between increasing OSA severity and glycated hemoglobin (HbA1c) after controlling for various confounding factors (Aronsohn et al., 2010). OSA may lead to dysregulated glucose metabolism by inducing intermittent hypoxia and excessive sympathetic nervous system activation (Pamidi and Tasali, 2012).

OSA severity during rapid-eye movement (REM) sleep was found to be particularly associated with worsened glycemic control in T2D patients (Grimaldi et al., 2014). This may not be surprising, since REM sleep is associated with increased sympathetic nervous system activation (Somers et al., 1993), longer apnea duration (Findley et al., 1985), and greater hypoxemia during OSA events vs. non-REM (NREM) sleep (Krieger et al., 1997). These may contribute to the higher HbA1c in individuals with more severe OSA during REM sleep (Grimaldi et al., 2014). The degree of OSA-related hypoxia is worsened by excess body weight (Peppard et al., 2009). Specifically, increased body mass index (BMI) is a significant predictor of the severity of oxygen desaturation during OSA events, particularly during REM vs. NREM sleep (Peppard et al., 2009).

Look AHEAD (Action for Health in Diabetes) is a multi-center randomized clinical trial (n=5145) to determine the effects of an intensive lifestyle intervention (ILI) targeted to weight loss on cardiovascular morbidity and mortality in overweight/obese T2D patients (Ryan et al., 2003). In the complete Look AHEAD trial sample, the ILI, consisting of decreased caloric intake and increased physical activity, induced significant and sustained reductions in body weight and HbA1c compared to the diabetes support and education (DSE) control group (Wing et al., 2013). Results from the ancillary Sleep AHEAD study demonstrated that ILI was also associated with reduced OSA severity, quantified as a decrease in the apnea-hypopnea index (AHI), compared to DSE (Kuna et al., 2013). Both body weight loss and intervention group were significant predictors of reduced AHI over the 4-y follow-up (Kuna et al., 2013). It is unknown if OSA severity during REM sleep in particular is improved by the ILI. Furthermore, it is unknown if REM sleep AHI is related to longitudinal changes in glycemic control within this cohort.

The first aim of the study was to determine if ILI vs. DSE results in a reduction of REM sleep-related AHI (REM-AHI). If so, this could present an added health benefit of the multi-component behavioral lifestyle intervention. The second aim was to determine if longitudinal changes in REM-AHI are related to concurrent improvements in glycemic control over a 4-yr follow-up period. Since the lifestyle intervention improves HbA1c, the current analysis presents an opportunity to investigate a potential mechanism linking improvements in sleep and/or body weight to glycemic control. We hypothesized that ILI vs. DSE would result in significantly reduced REM-AHI. Based on the relationship between REM sleep-related OSA and glycemic control (Grimaldi et al., 2014), we hypothesize that a reduction in REM-AHI over the 4-yr follow-up period would be related to improvements in HbA1c over time, after controlling for potential baseline and time-varying confounders.

Methods

Participants

Participants were enrolled in Sleep AHEAD, a 4-center ancillary study of the Look AHEAD trial. Look AHEAD is a 16-center randomized-controlled trial (RCT) to investigate the long-term effects of ILI on cardiovascular morbidity and mortality in overweight/obese T2D patients (Bray et al., 2006, Ryan et al., 2003). Primary inclusion criteria for Look AHEAD were age 45-76 y, physician-verified presence of T2D, HbA1c <11%, BMI ≥25 kg/m2 or ≥27 kg/m2 if taking insulin, and blood pressure <160/100 mm Hg.

Inclusion criteria for Sleep AHEAD included all primary inclusion criteria for the Look AHEAD trial, and some further sleep-related items. Only those Sleep AHEAD participants with OSA at baseline (AHI ≥5 events/h) were included in the current analysis (264 of 306 participants). Central apnea was exclusionary. In the initial design of the Sleep AHEAD protocol, it was decided that the primary analysis would focus on individuals with OSA at baseline, defined as AHI ≥5 events/h, as the main group of interest. Individuals with AHI <5 events/h at baseline had a less complete and thorough follow up, and are not included in these analyses. Use of continuous positive airway pressure (CPAP) during the follow-up was not exclusionary for enrollment in the Sleep AHEAD trial. As described in prior Sleep AHEAD publications, participants refrained from CPAP for 3 nights preceding the study, as this duration is sufficient to reverse CPAP-related effects on AHI (Kohler et al., 2011). However, CPAP use may have affected HbA1C levels (a primary outcome), and refraining from CPAP for 3 days is unlikely to reverse these effects. Accordingly, a separate analysis was completed after excluding participants who used CPAP, as was done by others (Young et al., 2008). A total of 42 participants used CPAP at some point during follow up.

All participants provided written informed consent, and experimental procedures, including administering the interventions and obtaining measures, were approved by the Institutional Review Board of each of the four centers participating as Sleep AHEAD sites (University of Pennsylvania; University of Pittsburgh; St. Luke's-Roosevelt Hospital/Columbia University; and The Miriam Hospital/Brown University).

Intervention

Look AHEAD design and intervention have been described (Ryan et al., 2003). Briefly, participants were randomly allocated to either ILI or DSE. ILI was a group behavioral weight loss program developed for overweight/obese T2D patients, consisting of reduced daily caloric intake and increased physical activity (Ryan et al., 2003, Kuna et al., 2013). Daily caloric intake was 1200-1500 kcal for individuals <113.6 kg, and 1500-1800 kcal for those with body weight >113.6 kg. The physical activity component consisted of 175 min/wk of moderate intensity activity. In DSE, participants were provided 3 group support and education sessions per year, focusing on diet and physical activity education, and social support. There were no specific behavior change strategies. Technical staff involved in the sleep studies and scoring of polysomnographic (PSG) recordings remained blinded to the randomization.

Measures

Unattended overnight PSG using a portable system were obtained at baseline, and at year-1, year-2, and year-4 follow-ups as previously described (Shechter et al., 2014, Kuna et al., 2013, Foster et al., 2009a). Recorded signals included electroencephalogram (C3-A2 and C4-A1), bilateral electrooculogram, chin electromyogram, electrocardiogram, rib cage and abdominal excursion, airflow via oronasal thermistry and nasal pressure cannula, oxygen saturation via pulse oximetry. Recordings were scored at a single reading laboratory using recommended criteria (Kushida et al., 2005). Total sleep time (TST) was the sum of the duration of stage 1, 2, slow wave sleep (SWS), and REM sleep. Time in NREM sleep was the sum of the duration of stage 1, 2, and SWS. Apnea was defined as the cessation of airflow ≥10 sec with concomitant respiratory-related chest wall movement for obstructive apnea, or in the absence of respiratory-related chest wall movement for central apnea. Hypopnea was defined as ≥30% reduction in airflow or thoraco-abdominal movement for ≥10 sec with ≥4% oxygen desaturation. Total AHI was the sum of the number of apneas and hypopneas per hour of sleep. REM-AHI was the sum of the number of apneas and hypopneas occurring during REM sleep, expressed per hour of REM sleep. NREM-AHI was the sum of the number of apneas and hypopneas occurring during NREM sleep, expressed per hour of NREM sleep. In the first analysis, no minimum duration of REM sleep was used in the calculation of REM-AHI. However, in a secondary analysis, a minimum REM sleep duration threshold of 30 minutes was used as a requirement to calculate REM-AHI (Mokhlesi and Punjabi, 2012). Intrascorer reliability across the 4-year follow-up was assessed and confirmed (Shechter et al., 2014).

Height and weight were assessed at clinic visits within 1 week of PSG, and used to calculate BMI (kg/m2). Fasting blood samples were obtained and assayed for HbA1c at annual clinic visits (Ryan et al., 2003).

Statistical analyses

Analyses were conducted to determine if missing follow-up data could potentially bias the results. These were described in detail in a prior publication (Kuna et al., 2013). Using generalized estimating equation models, the probability of a missing PSG was not related to baseline AHI, intervention arm, or follow-up weight measurements (Kuna et al., 2013). Furthermore, in pattern-mixture models, missing data patterns, either as a main effect or as moderated by intervention arm or year, were not related to observed changes in AHI (Kuna et al., 2013).

Adjusted change-from-baseline was calculated for NREM-AHI and REM-AHI at year 1, year 2, and year 4. Changes-from-baseline for these parameters were adjusted for age, sex, race, baseline BMI, and site. Repeated-measure linear-mixed model analysis was used to determine the between-group differences in adjusted NREM-AHI and REM-AHI change-from-baseline to year 1, year 2, and year 4. This analysis was also done after adjustment for body weight change over time to see how weight change affects between-group differences in NREM-AHI and REM-AHI. Repeated-measure mixed-model analyses was used to assess the relationship between changes in OSA (NREM-AHI, REM-AHI) over time and changes in HbA1c at year 1, year 2, and year 4 when data from all participants were pooled together. For the mixed-model analyses, individuals without baseline were excluded. Dependent variables were changes-from-baseline in HbA1c at year 1, year 2, and year 4. Predictor variables for the models were group (ILI vs. DSE), year, baseline REM-AHI or NREM-AHI, and change in REM-AHI or NREM-AHI. An additional analysis was done by adding baseline body weight and change in body weight to the model as predictor variables to determine whether REM-AHI and NREM-AHI contribute to predicted changes in HbA1c after adjusting for the effects of weight loss. A treatment group x year interaction was also included in the models. The reference level for the predictor “group” was DSE and for the predictor “year” was year 4. Models were adjusted for relevant covariates including age, sex, race, baseline BMI, site, and baseline AHI and HbA1c. Analyses were repeated after excluding participants who used CPAP at any time point during the trial and those who did not reach the minimum REM sleep duration threshold of 30 min.

Results

In the Sleep AHEAD trial, 264 participants with OSA were included. Baseline characteristics, including demographics, anthropometry, OSA severity, sleep duration and architecture (TST, stage 1, 2, SWS, REM sleep), and HbA1c have been previously reported (Kuna et al., 2013, Shechter et al., 2014). No baseline between-group differences were observed. As reported (Kuna et al., 2013), of the 264 participants, n=125 were enrolled in ILI and 139 were enrolled in DSE at baseline. At year 1, n=103 (82%) of those in ILI and n=116 (83%) of those in DSE returned for follow up. At year 2, n=99 (79%) returned for ILI, and n=111 (80%) returned for DSE. Patients returning for follow-up at year 4 included n=82 (66%) for ILI and n=83 (60%) for DSE.

Table 1 contains participant characteristics at baseline, for all Sleep AHEAD participants (top panel; n=264) and after excluding for CPAP use and minimum REM sleep duration (n=192, bottom panel). No between-group differences at baseline were observed for AHI-total, NREM-AHI, or REM-AHI (p-values ≥0.75). In all participants at baseline, regardless of treatment arm, AHI was significantly higher in REM sleep vs. NREM sleep (p<0.001).

Table 1.

Participant characteristics at baseline.

All Sleep AHEAD participants (n=264)
Variable All ILI DSE
Sample size 264 125 139
Age, y 61.2 (6.5) 61.2 (6.6) 61.3 (6.4)
BMI, kg/m2 36.7 (5.7) 36.8 (5.8) 36.5 (5.7)
Weight, kg 102.4 (18.3) 102.9 (19.6) 102.0 (17.1)
AHI-Total, events/h 23.2 (16.5) 22.9 (18.0) 23.5 (15.0)
REM-AHI, events/h 40.3 (24.3) 39.8 (26.0) 40.8 (22.7)
NREM-AHI, events/h 19.1 (17.3) 18.8 (18.8) 19.4 (15.9)
Participants after excluding for minimum REM sleep duration and CPAP use at follow-up (n=192)
Variable All ILI DSE
Sample size 192 98 94
Age, y 61.1 (6.5) 61.3 (6.5) 60.9 6.5)
BMI, kg/m2 36.8 (5.8) 36.5 (5.7) 37.1 (6.0)
Weight, kg 102.0 (18.2) 101.6 (18.7) 102.5 (17.7)
AHI-Total, events/h 21.3 (15.3) 21.0 (16.3) 21.6 (14.4)
REM-AHI, events/h 38.5 (24.3) 38.1 (26.1) 38.9 (22.4)
NREM-AHI, events/h 16.7 (15.9) 16.5 (17.0) 16.9 (14.9)
Open in a new tab

Baseline data are shown for all participants (top panel), and for participants included after removing those not meeting the minimum REM sleep duration threshold and those who used CPAP during follow up (lower panel). Values are expressed as mean (SD). AHI: Apnea-hypopnea index; BMI: Body mass index; DSE: Diabetes support and education; ILI: Intensive lifestyle intervention; NREM: non rapid eye movement sleep; REM: rapid eye movement sleep

Table 2 presents the estimated mean changes in REM-AHI and NREM-AHI from baseline to year 1, year 2, and year 4, without adjustment for changes in weight over time. When considering all participants, significant between-group treatment differences in REM-AHI were observed at each year, with REM-AHI more reduced in ILI vs. DSE at each follow-up. Between-group comparisons revealed significantly reduced NREM-AHI in ILI vs. DSE at year-1, year-2, and year-4 follow-ups. When participants using CPAP or not reaching the REM sleep duration threshold were excluded, results were the same as above, except year 1 between-group differences in NREM-AHI were no longer significant (Table 2).

Table 2.

Estimated adjusted mean (standard error) changes in OSA measures from baseline to year 1, year 2, and year 4, with and without adjustment for weight change over time. Data are presented for all participants (top panel) and for participants remaining after excluding for REM sleep duration threshold and CPAP use (bottom panel).

All Sleep AHEAD participants (n=264)
Not adjusted for weight change Adjusted for weight change
Measure Year ILI DSE P-value ILI DSE P-Value
REM-AHI 1 -10.6 (3.7) 5.4 (3.7) <0.01 -8.0 (3.7) 4.0 (3.6) <0.01
2 -7.1 (3.6) 1.3 (3.6) <0.01 -5.8 (3.6) 0.2 (3.6) 0.05
4 -12.0 (3.8) -3.9 (3.8) 0.02 -11.0 (3.7) -5.6 (3.8) 0.10
NREM-AHI 1 -5.1 (2.6) 3.0 (2.6) <0.01 -2.5 (2.6) 0.8 (2.5) 0.12
2 -3.7 (2.6) 4.0 (2.6) <0.01 -2.5 (2.6) 2.1 (2.6) 0.03
4 -3.3 (2.7) 3.0 (2.8) <0.01 -3.2 (2.6) 1.0 (2.7) 0.07
Participants after excluding for minimum REM sleep duration and CPAP use at follow-up (n=192)
Not adjusted for weight change Adjusted for weight change
Measure Year ILI DSE P-value ILI DSE P-Value
REM-AHI 1 -10.0 (3.7) 1.8 (3.9) <0.01 -8.7 (3.8) 0.7 (3.9) <0.01
2 -7.4 (3.8) 3.2 (4.0) <0.01 -7.1 (3.8) 2.2(4.1) <0.01
4 -12.8 (4.1) -2.0 (4.2) <0.01 -12.3 (3.9) -3.1 (4.2) 0.02
NREM-AHI 1 -4.6 (2.4) -1.8 (2.5) 0.1601 -3.8 (2.5) -3.3 (2.6) 0.84
2 -3.0 (2.6) 3.6 (2.7) <0.01 -2.7 (2.6) 2.0 (2.7) 0.04
4 -3.9 (2.7) 3.1 (2.9) <0.01 -3.6 (2.7) 1.5 (2.8) 0.05
Open in a new tab

Values are adjusted mean change from baseline (SE) at year 1, year 2, and year 4. Analyses are adjusted for age, sex, race, baseline body mass index, site, and were done with and without adjustment for body weight change over time. AHI: apnea-hypopnea index; DSE: diabetes support and education; ILI: intensive lifestyle intervention; NREM: non-rapid eye movement; REM: rapid eye movement. Statistically significant p-values (<0.05) are denoted in bold.

Table 2 also presents the estimated mean changes in REM-AHI and NREM-AHI from baseline to year 1, year 2, and year 4, after adjustment for weight change over time. Values of REM-AHI remained more reduced in ILI vs. DSE at each follow-up. Between-group differences were of a smaller magnitude than non-weight loss adjusted differences, and only showed statistically significant differences at year 1 and 2, but not year 4. Values of NREM-AHI also remained more reduced in ILI vs. DSE at each follow-up, but were of smaller magnitude than non-weight loss adjusted differences, and were only statistically significantly different at year 2. When participants using CPAP or not reaching the REM sleep duration threshold were excluded, results were the same as above, except year 4 between-group differences in REM-AHI were statistically significant (Table 2).

The relationships between changes in REM-AHI and HbA1c across years 1, 2 and 4, with and without inclusion of weight change over time, are presented in Table 3. Changes in REM-AHI were not associated with changes in HbA1c, regardless of whether weight loss was included in the model. When weight changes were not included in models, significant group x year interactions were observed, indicating that the pattern of change in HbA1c from baseline was significantly different (more reduced) in ILI vs. DSE throughout the follow-up. When weight changes were included in the models, weight loss was a significant predictor of reduced HbA1c. Results were similar after excluding for CPAP use and REM sleep duration threshold, except the ILI x Year 2 interaction was no longer significant. Change in body weight remained the most important predictor of change in HbA1c.

Table 3.

Repeated measure mixed-model analysis showing the relationship between change in REM-AHI and change in HbA1c over year 1, year 2, and year 4.

All Sleep AHEAD participants (n=264)
REM-AHI and HbA1c
Not including weight change Including weight change
Predictor Coefficient SE P-value Coefficient SE P-value
Group (ILI vs. DSE) -0.214 0.177 0.229 -0.072 0.175 0.680
Year 1 -0.377 0.142 0.009 -0.177 0.139 0.206
Year 2 -0.142 0.120 0.239 -0.057 0.120 0.632
Baseline REM-AHI 0.002 0.003 0.352 0.001 0.002 0.653
Change in REM-AHI 0.001 0.002 0.479 0.000 0.002 0.995
ILI x Year 1 -0.462 0.198 0.021 -0.246 0.192 0.202
ILI x Year 2 -0.352 0.166 0.035 -0.267 0.166 0.109
Baseline body weight ------ ------ ------ 0.000 0.004 0.944
Change in body weight ------ ------ ------ 0.037 0.006 <0.001
Participants after excluding for minimum REM sleep duration and CPAP use at follow-up (n=192)
REM-AHI and HbA1c
Not including weight change Including weight change
Predictor Coefficient SE P-value Coefficient SE P-value
Group (ILI vs. DSE) -0.283 0.236 0.232 -0.098 0.229 0.668
Year 1 -0.412 0.185 0.027 -0.202 0.177 0.256
Year 2 -0.102 0.163 0.531 -0.020 0.161 0.902
Baseline REM-AHI 0.0004 0.003 0.914 -0.000 0.003 0.989
Change in REM-AHI 0.002 0.003 0.594 0.0007 0.003 0.798
ILI x Year 1 -0.522 0.261 0.047 -0.281 0.248 0.260
ILI x Year 2 -0.324 0.233 0.166 -0.254 0.229 0.270
Baseline body weight ------ ------ ------ -0.001 0.004 0.754
Change in body weight ------ ------ ------ 0.042 0.007 <0.001
Open in a new tab

AHI: apnea-hypopnea index; DSE: diabetes support and education; ILI: intensive lifestyle intervention; NREM: non-rapid eye movement; REM: rapid eye movement. Analyses are adjusted for age, sex, race, baseline body mass index, site, and baseline HbA1c. Reference for Group is DSE, and reference for Year is year 4. Statistically significant P-values (< 0.05) are shown in bold type.

The relationships between changes in NREM-AHI and HbA1c across years 1, 2 and 4, with and without inclusion of weight change over time, are presented in Table 4. Changes in NREM-AHI were not associated with changes in HbA1c, regardless of whether weight loss was included in the model. When weight changes were not included in models, significant group x year interactions were observed, indicating that the pattern of change in HbA1c from baseline was significantly different (more reduced) in ILI vs. DSE during the follow-up. When included, change in body weight was again a significant predictor of change in HbA1c, for all participants and after excluding for CPAP use and REM sleep threshold.

Table 4.

Repeated measure mixed-model analysis showing the relationship between change in NREM-AHI and change in HbA1c over year 1, year 2, and year 4.

All Sleep AHEAD participants (n=264)
NREM-AHI and HbA1c
Not including weight change Including weight change
Predictor Coefficient SE P-value Coefficient SE P-value
Group (ILI vs. DSE) -0.160 0.164 0.331 -0.010 0.161 0.952
Year 1 -0.435 0.127 0.001 -0.233 0.124 0.062
Year 2 -0.183 0.112 0.103 -0.101 0.110 0.361
Baseline NREM-AHI -0.0004 0.003 0.881 -0.001 0.003 0.800
Change in NREM-AHI 0.0001 0.003 0.995 -0.004 0.003 0.101
ILI x Year 1 -0.527 0.176 0.003 -0.296 0.171 0.086
ILI x Year 2 -0.431 0.157 0.007 -0.356 0.155 0.022
Baseline body weight ------ ------ ------ 0.001 0.003 0.803
Change in body weight ------ ------ ------ 0.040 0.006 <0.001
Participants after excluding for minimum REM sleep duration and CPAP use at follow-up (n=192)
NREM-AHI and HbA1c
Not including weight change Including weight change
Predictor Coefficient SE P-value Coefficient SE P-value
Group (ILI vs. DSE) -0.313 0.235 0.186 -0.130 0.226 0.565
Year 1 -0.413 0.184 0.026 -0.206 0.175 0.239
Year 2 -0.091 0.161 0.574 -0.017 0.159 0.914
Baseline NREM-AHI -0.004 0.005 0.475 -0.002 0.005 0.750
Change in NREM-AHI -0.002 0.005 0.611 -0.007 0.005 0.128
ILI x Year 1 -0.520 0.260 0.048 -0.251 0.247 0.311
ILI x Year 2 -0.321 0.232 0.168 -0.256 0.227 0.262
Baseline body weight ------ ------ ------ -0.002 0.004 0.714
Change in body weight ------ ------ ------ 0.044 0.007 <0.0001
Open in a new tab

AHI: apnea-hypopnea index; DSE: diabetes support and education; ILI: intensive lifestyle intervention; NREM: non-rapid eye movement; REM: rapid eye movement. Analyses are adjusted for age, sex, race, baseline body mass index, site, and baseline HbA1c. Reference for Group is DSE, and reference for Year is year 4.

Discussion

We report an improvement in REM-AHI after weight loss in patients with obesity, T2D and OSA. After controlling for confounding variables, REM-AHI and NREM-AHI were reduced in the ILI vs. DSE. Improvements in AHI during REM- or NREM sleep were not related to changes in HbA1c. Rather, participation in the ILI, and more importantly, weight loss, similar to Look AHEAD participants overall (Wing et al., 2013), was found to improve HbA1c over the 4-year follow-up. Weight loss, and other beneficial effects of the multi-component ILI, seem to improve HbA1c, as opposed to being caused directly by improvements in REM or NREM-AHI.

RCTs utilizing ILIs via low caloric intake (Johansson et al., 2009, Tuomilehto et al., 2009), or a combination of diet and physical activity (Foster et al., 2009b) explored how targeted weight loss affects AHI. These RCTs, similar to baseline-post-treatment studies, demonstrate the effectiveness of lifestyle interventions in producing significant improvements in OSA severity (Araghi et al., 2013). However, none of the controlled or prospective intervention studies have described the effects of lifestyle intervention and weight loss on REM-AHI, in addition to total AHI.

REM-AHI appears to be a relevant, but often overlooked, clinical outcome which increases the risk for adverse cardiometabolic parameters (Kendzerska and Ayas, 2015). REM sleep is associated with increases in heart rate, blood pressure, blood pressure variability, and sympathetic-nerve activity (Somers et al., 1993). Apnea events are of longer duration and are associated with a greater degree of hypoxemia in REM vs. NREM sleep (Findley et al., 1985, Muraki et al., 2008, Series et al., 1990). Significant longitudinal relationships were found for REM-AHI with hypertension development (Mokhlesi et al., 2014) and the risk of developing nocturnal non-dipping blood pressure (Mokhlesi et al., 2015). We report that AHI was worse in REM vs. NREM sleep, and accordingly, patients were at a high risk for the aforementioned consequences of REM-related OSA. Thus, the reduction of REM-AHI we observed after ILI presents a new beneficial result of the intervention.

The prevalence of OSA in T2D is about 71% (Pamidi and Tasali, 2012). Studies investigating the relationship between OSA and glycemic control in T2D have been inconsistent, with some showing a significant association between total AHI and HbA1c (Aronsohn et al., 2010, Grimaldi et al., 2014), while others did not (Einhorn et al., 2007, Lam et al., 2010). In a large group of participants including those with normal glucose tolerance, impaired glucose tolerance, and T2D, HbA1c was significantly higher in those with AHI ≥30 vs. those with AHI <5 (Tamura et al., 2012). However, when only considering those with T2D, no relationships were seen between AHI and HbA1c (Tamura et al., 2012). In the large European Sleep Apnea Cohort study (n=6,616), not only was T2D prevalence increased with greater severity of OSA, but within individuals with T2D, increasing OSA severity predicted worsened glycemic control (Kent et al., 2014). Together, this implies that the relationship between OSA and glucose homeostasis is complex. In a recent study REM-AHI but not NREM-AHI was related to HbA1c, with a statistically and clinically significant increase in HbA1c between the lowest and highest quartiles of REM-AHI (Grimaldi et al., 2014). The cross-sectional nature of their investigation lacks the ability to infer causality within the relationship (Grimaldi et al., 2014). Our current longitudinal analyses further explored, within a controlled intervention, how OSA during REM sleep affects HbA1c. To our knowledge, this is the first investigation of the relationship between longitudinal changes in REM-AHI and HbA1c. Contrary to our hypothesis, longitudinal improvements in REM-AHI were not related to changes in HbA1c. Rather, we observed that participation in ILI and weight loss were associated with improvements in HbA1c, independent of changes in OSA severity.

These observations should be considered in light of trials investigating the effects of CPAP on HbA1c. Although some have been positive (Shpirer et al., 2012, Steiropoulos et al., 2009, Hassaballa et al., 2005), others (West et al., 2007, Craig et al., 2012) have shown no effect of CPAP use on HbA1c. The inconsistent results were highlighted by two recent RCTs which appeared simultaneously in the same journal and demonstrated either an improvement in HbA1c after 6 mo of CPAP treatment (Martínez-Cerón et al., 2016), or no effect of CPAP on glycemic control (Shaw et al., 2016). The results of a recent meta-analysis of RCTs (though not including the aforementioned pair of studies) and prospective observational studies on the effects of CPAP on HbA1c in patients with OSA and T2D concluded that CPAP does not alter HbA1c (Feng et al., 2015). Moreover, CPAP does not appear to reduce body weight, and in fact may lead to weight gain (Drager et al., 2015). Together with our current findings, this suggests that a reduction in AHI, as is achieved with CPAP, is not sufficient to improve glycemic control. Conversely, Grimaldi et al. suggested that insufficient CPAP use, which would fail to alleviate REM-related OSA events in the latter portion of the sleep episode, may explain the lack of an effect of CPAP on HbA1c (Grimaldi et al., 2014). They propose that CPAP use of over 6 h/night would be needed to achieve a significant improvement in glycemic control. Current findings alternately indicate that a significant reduction in nocturnal REM-AHI can be achieved by the ILI and/or weight loss, and that such lifestyle interventions targeting weight loss can reduce HbA1c. Our data suggest that a lifestyle intervention can be an effective and useful adjunctive to CPAP in the clinical care of T2D patients with OSA.

Despite our findings, it is important to consider the relatively modest effects of the lifestyle intervention on REM-AHI within the proper context. While we did observe statistically significant reductions in REM-AHI, the mean treatment-related reduction of ∼12 events/h (from a baseline value of ∼40 events/h) is expected to leave a considerable amount of REM-AHI. It has been suggested that insufficient CPAP use may account for the lack of observed effects of CPAP on various cardiometabolic outcomes, including HbA1c, reported in some trials (Grimaldi et al., 2014, Pamidi et al., 2015). It is possible that a complete alleviation of REM sleep-related OSA, which we were not able to demonstrate here, would have a more appreciable effect on improving glycemic control in individuals with OSA and T2D. More work should therefore be done to determine this.

Current strengths are that we utilized data from a multi-center RCT with a 4-year follow-up. The effects of intervention REM and NREM-AHI were not described in the prior interventions which demonstrated an association between weight loss and improved AHI. Despite the large baseline sample size, 63% of participants returned at year 4. This high dropout rate (∼37%), likely due to the added burden asked of Look AHEAD participants to undergo additional PSG recordings, should be emphasized as a limitation in the study. However, the probability of missing PSG was not related to baseline AHI, treatment group, or follow-up body weight, and patterns of missing data were not related to observed change in AHI. Moreover, even with dropouts, the sample size remained larger than all prior lifestyle intervention trials investigating the relationship between weight loss and OSA severity (Johansson et al., 2009, Tuomilehto et al., 2009). Sleep data were collected from single overnight recordings at each time point, and the lack of an acclimatization night may affect recorded sleep quality. However, the at-home nature of the recording may mitigate some of the possible “first-night effect”. As our study included only T2D participants, it is unclear if results can be extended to individuals with prediabetes aiming to improve HbA1c. An additional limitation is that the analysis included only individuals who demonstrated some degree of OSA at baseline, based on a total AHI ≥5 events/h. It is possible that in focusing on these individuals, we may have missed some participants who have clinically relevant REM sleep-related OSA, despite not achieving minimum requirements for a standard OSA diagnosis (Mokhlesi and Punjabi, 2012).

We observed that the ILI for weight loss was successful in reducing AHI during REM and NREM sleep, vs. control in a large group of overweight/obese individuals with T2D and OSA. Weight loss, and not reduced REM-AHI, was associated with improved HbA1c. It should be emphasized that the initial Sleep AHEAD study was designed primarily to assess the effects of weight loss via an intensive lifestyle intervention on OSA severity in overweight and obese individuals with T2D (Kuna et al., 2013, Foster et al., 2009a). The current posthoc analysis reported here was concerned with two main questions not addressed in the initial reports: 1) The effects of weight loss induced by the intensive lifestyle intervention on REM sleep-related AHI and 2) The potential relationship between longitudinal changes in REM sleep-related AHI and improvements in glycemic control. The current study was therefore not designed specifically to address the latter question of how improvements in REM-AHI affect metabolic health. To more thoroughly and clearly investigate this, future well-controlled and sufficiently powered RCTs should be designed and conducted to determine how interventions to improve REM-AHI affect cardiometabolic outcomes.

Acknowledgments

Members of the Sleep AHEAD research group at each site consisted of the following individuals: St. Luke's-Roosevelt Hospital/Clinilabs: Jon Freeman, PPSGT, Ph.D., Jennifer Patricio; University of Pennsylvania: Brian McGuckin, Stephanie Krauthamer-Ewing, Allan Pack, M.B., Ch.B., Ph.D., Richard Schwab, M.D., Mary Jones-Parker, RPSGT, Matthew Anastasi, RPSGT, Beth Staley, RPSGT, Liz Roben, Andrea Sifferman; Brown University: Marie Kearns, Caitlin Egan; Temple University: Nida Cassim, Valerie Darcey, Sakhena Hin, Stephanie Vander Veur. A detailed list of the Look AHEAD Research Group is provided in Bray et al. Members of the Observational Safety and Management Board were: Kingman P. Strohl, M.D. (chair), Donald L. Bliwise, Ph.D., and Helaine E. Resnick, Ph.D.

Funding: The study was supported by the National Institutes of Health NHLBI grant HL070301 and NIDDK grants DK60426, DK56992, DK057135, and DK007559.

Footnotes

Clinical trial registration number: NCT00194259

Disclosures: This was not an industry supported study. Dr. Kuna has received research support from Philips Respironics. Dr. Zammit is a consultant for Actelion, Alexza, Arena, Aventis, Biovail, Boehringer-Ingelheim, Cephalon, Elan, Eli Lilly, Evotec, Forest, Glaxo Smith Kline, Jazz, King Pharmaceuticals, Ligand, McNeil, Merck, Neurocrine Biosciences, Organon, Phizer, Renovis, Sanofi-Aventis, Select Comfort, Sepracor, Shire, Somnus, Takeda Pharmaceuticals, Vela, Wyeth-Ayerst Research; provides expert testimony for Acorda; has grants or grants pending from Abbott, Actelion, Ancile, Apnex, Arena, Aventis, Cephalon Inc, CHDI, Elan, Epis, Evotec, Forest, Galderma, Glaxo Smith Kline, H. Lundbeck A/S, King, Merck and Co., National Institutes of Health (NIH), Neurim, Neurocrine Biosciences, Neurogen, Organon, Orphan Medical, Otsuka, Pfizer, Predix, Respironics, Sanofi-Aventis, Sanofi-Synthelabo, Schering-Plough, Sepracor, Shire, Samaxon, Takeda Pharmaceuticals North America, Targacept, Thymon, Transcept, UCB Pharma, Predix, Vanda, Wyeth-Ayerst Research. And has received payment for lectures from Neurocrine Biosciences, King Pharmaceuticals, McNeil, Sanofi-Aventis, Sanofi-Synthelabo, Sepracor, Takeda Pharmaceuticals, Vela Pharmaceuticals, Wyeth-Ayerst Research. Dr. Wadden serves on advisory boards for Orexigen Therapeutics and Novo Nordisk and has received grant support from Novo Nordisk and NutriSystem. Dr. Jakicic is on the Weight Watchers International Scientific Advisory Board, is a co-investigator for research grants awarded to the University of Pittsburgh by Weight Watchers International and Human Scale. Dr. Foster is an employee and shareholder of Weight Watchers International. The other authors have indicated no financial conflicts of interest.

Author contributions: Study concept and design: GDF, RPM, GZ, ABN, TAW, RRW, FXP, STK. Acquisition of data: GDF, DMR, RPM, GZ, ABN, TAW, RRW, FXP, STK. Analysis and interpretation of data: AS, WL. Drafting of manuscript: AS. Critical revision of the manuscript for intellectual content: AS, GDF, WL, DMR, MPSO, GZ, ABN, RPM, TAW, JMJ, ES, RRW, FXP, STK. Final approval of the manuscript: AS, GDF, WL, DMR, MPSO, GZ, ABN, RPM, TAW, JMJ, ES, RRW, FXP, STK.

References

  1. Araghi MH, Chen YF, Jagielski A, et al. Effectiveness of lifestyle interventions on obstructive sleep apnea (OSA): systematic review and meta-analysis. Sleep. 2013;36:1553–62. doi: 10.5665/sleep.3056. 62A-62E. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Aronsohn RS, Whitmore H, Van Cauter E, Tasali E. Impact of untreated obstructive sleep apnea on glucose control in type 2 diabetes. Am J Respir Crit Care Med. 2010;181:507–13. doi: 10.1164/rccm.200909-1423OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bray G, Gregg E, Haffner S, et al. Baseline characteristics of the randomised cohort from the Look AHEAD (Action for Health in Diabetes) study. Diab Vasc Dis Res. 2006;3:202–15. doi: 10.3132/dvdr.2006.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Craig SE, Kohler M, Nicoll D, et al. Continuous positive airway pressure improves sleepiness but not calculated vascular risk in patients with minimally symptomatic obstructive sleep apnoea: the MOSAIC randomised controlled trial. Thorax. 2012;67:1090–6. doi: 10.1136/thoraxjnl-2012-202178. [DOI] [PubMed] [Google Scholar]
  5. Drager LF, Brunoni AR, Jenner R, Lorenzi-Filho G, Bensenor IM, Lotufo PA. Effects of CPAP on body weight in patients with obstructive sleep apnoea: a meta-analysis of randomised trials. Thorax. 2015;70:258–64. doi: 10.1136/thoraxjnl-2014-205361. [DOI] [PubMed] [Google Scholar]
  6. Einhorn D, Stewart DA, Erman MK, Gordon N, Philis-Tsimikas A, Casal E. Prevalence of sleep apnea in a population of adults with type 2 diabetes mellitus. Endocr Pract. 2007;13:355–62. doi: 10.4158/EP.13.4.355. [DOI] [PubMed] [Google Scholar]
  7. Feng Y, Zhang Z, Dong ZZ. Effects of continuous positive airway pressure therapy on glycaemic control, insulin sensitivity and body mass index in patients with obstructive sleep apnoea and type 2 diabetes: a systematic review and meta-analysis. NPJ Prim Care Respir Med. 2015;25:15005. doi: 10.1038/npjpcrm.2015.5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Findley LJ, Wilhoit SC, Suratt PM. Apnea duration and hypoxemia during REM sleep in patients with obstructive sleep apnea. Chest. 1985;87:432–6. doi: 10.1378/chest.87.4.432. [DOI] [PubMed] [Google Scholar]
  9. Foster GD, Borradaile KE, Sanders MH, et al. A randomized study on the effect of weight loss on obstructive sleep apnea among obese patients with type 2 diabetes: the Sleep AHEAD study. Arch Intern Med. 2009a;169:1619–26. doi: 10.1001/archinternmed.2009.266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Foster GD, Sanders MH, Millman R, et al. Obstructive sleep apnea among obese patients with type 2 diabetes. Diabetes Care. 2009b;32:1017–9. doi: 10.2337/dc08-1776. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Grimaldi D, Beccuti G, Touma C, Van Cauter E, Mokhlesi B. Association of obstructive sleep apnea in rapid eye movement sleep with reduced glycemic control in type 2 diabetes: therapeutic implications. Diabetes Care. 2014;37:355–63. doi: 10.2337/dc13-0933. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hassaballa HA, Tulaimat A, Herdegen JJ, Mokhlesi B. The effect of continuous positive airway pressure on glucose control in diabetic patients with severe obstructive sleep apnea. Sleep Breath. 2005;9:176–80. doi: 10.1007/s11325-005-0033-y. [DOI] [PubMed] [Google Scholar]
  13. Johansson K, Neovius M, Lagerros YT, et al. Effect of a very low energy diet on moderate and severe obstructive sleep apnoea in obese men: a randomised controlled trial. BMJ. 2009;339:b4609. doi: 10.1136/bmj.b4609. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kendzerska T, Ayas N. Turning Dreams into Nightmares: Different Metabolic Effects of Obstructive Events in Rapid Eye Movement versus Non-Rapid Eye Movement Sleep. Am J Respir Crit Care Med. 2015;192:1034–5. doi: 10.1164/rccm.201507-1504ED. [DOI] [PubMed] [Google Scholar]
  15. Kent BD, Grote L, Ryan S, et al. Diabetes mellitus prevalence and control in sleep-disordered breathing: the European Sleep Apnea Cohort (ESADA) study. CHEST Journal. 2014;146:982–90. doi: 10.1378/chest.13-2403. [DOI] [PubMed] [Google Scholar]
  16. Kohler M, Stoewhas AC, Ayers L, et al. Effects of continuous positive airway pressure therapy withdrawal in patients with obstructive sleep apnea: a randomized controlled trial. Am J Respir Crit Care Med. 2011;184:1192–9. doi: 10.1164/rccm.201106-0964OC. [DOI] [PubMed] [Google Scholar]
  17. Krieger J, Sforza E, Boudewijns A, Zamagni M, Petiau C. Respiratory effort during obstructive sleep apnea: role of age and sleep state. Chest. 1997;112:875–84. doi: 10.1378/chest.112.4.875. [DOI] [PubMed] [Google Scholar]
  18. Kuna ST, Reboussin DM, Borradaile KE, et al. Long-term effect of weight loss on obstructive sleep apnea severity in obese patients with type 2 diabetes. Sleep. 2013;36:641–49A. doi: 10.5665/sleep.2618. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kushida CA, Littner MR, Morgenthaler T, et al. Practice parameters for the indications for polysomnography and related procedures: an update for 2005. Sleep. 2005;28:499–521. doi: 10.1093/sleep/28.4.499. [DOI] [PubMed] [Google Scholar]
  20. Lam DC, Lui MM, Lam JC, Ong LH, Lam KS, Ip MS. Prevalence and recognition of obstructive sleep apnea in Chinese patients with type 2 diabetes mellitus. Chest. 2010;138:1101–7. doi: 10.1378/chest.10-0596. [DOI] [PubMed] [Google Scholar]
  21. Martínez-Cerón E, Barquiel B, Bezos AM, et al. Effect of continuous positive airway pressure on glycemic control in patients with obstructive sleep apnea and type 2 diabetes. A Randomized Clinical Trial American journal of respiratory and critical care medicine. 2016;194:476–85. doi: 10.1164/rccm.201510-1942OC. [DOI] [PubMed] [Google Scholar]
  22. Mokhlesi B, Finn LA, Hagen EW, et al. Obstructive sleep apnea during REM sleep and hypertension results of the Wisconsin Sleep Cohort. Am J Respir Crit Care Med. 2014;190:1158–67. doi: 10.1164/rccm.201406-1136OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Mokhlesi B, Hagen EW, Finn LA, Hla KM, Carter JR, Peppard PE. Obstructive sleep apnoea during REM sleep and incident non-dipping of nocturnal blood pressure: a longitudinal analysis of the Wisconsin Sleep Cohort. Thorax. 2015;70:1062–9. doi: 10.1136/thoraxjnl-2015-207231. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Mokhlesi B, Punjabi NM. REM-related” obstructive sleep apnea: an epiphenomenon or a clinically important entity. Sleep. 2012;35:5–7. doi: 10.5665/sleep.1570. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Muraki M, Kitaguchi S, Ichihashi H, et al. Apnoea-hypopnoea index during rapid eye movement and non-rapid eye movement sleep in obstructive sleep apnoea. J Int Med Res. 2008;36:906–13. doi: 10.1177/147323000803600506. [DOI] [PubMed] [Google Scholar]
  26. Pamidi S, Tasali E. Obstructive sleep apnea and type 2 diabetes: is there a link? Front Neurol. 2012;3:126. doi: 10.3389/fneur.2012.00126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Pamidi S, Wroblewski K, Stepien M, et al. Eight hours of nightly continuous positive airway pressure treatment of obstructive sleep apnea improves glucose metabolism in patients with prediabetes A randomized controlled trial. American journal of respiratory and critical care medicine. 2015;192:96–105. doi: 10.1164/rccm.201408-1564OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Peppard PE, Ward NR, Morrell MJ. The impact of obesity on oxygen desaturation during sleep-disordered breathing. Am J Respir Crit Care Med. 2009;180:788–93. doi: 10.1164/rccm.200905-0773OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ryan DH, Espeland MA, Foster GD, et al. Look AHEAD (Action for Health in Diabetes): design and methods for a clinical trial of weight loss for the prevention of cardiovascular disease in type 2 diabetes. Control Clin Trials. 2003;24:610–28. doi: 10.1016/s0197-2456(03)00064-3. [DOI] [PubMed] [Google Scholar]
  30. Series F, Cormier Y, La Forge J. Influence of apnea type and sleep stage on nocturnal postapneic desaturation. Am Rev Respir Dis. 1990;141:1522–6. doi: 10.1164/ajrccm/141.6.1522. [DOI] [PubMed] [Google Scholar]
  31. Shaw JE, Punjabi NM, Naughton MT, et al. The effect of treatment of obstructive sleep apnea on glycemic control in type 2 diabetes. American journal of respiratory and critical care medicine. 2016;194:486–92. doi: 10.1164/rccm.201511-2260OC. [DOI] [PubMed] [Google Scholar]
  32. Shechter A, St-Onge MP, Kuna ST, et al. Sleep architecture following a weight loss intervention in overweight and obese patients with obstructive sleep apnea and type 2 diabetes: relationship to apnea-hypopnea index. J Clin Sleep Med. 2014;10:1205–11. doi: 10.5664/jcsm.4202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Shpirer I, Rapoport MJ, Stav D, Elizur A. Normal and elevated HbA1C levels correlate with severity of hypoxemia in patients with obstructive sleep apnea and decrease following CPAP treatment. Sleep Breath. 2012;16:461–6. doi: 10.1007/s11325-011-0525-x. [DOI] [PubMed] [Google Scholar]
  34. Somers VK, Dyken ME, Mark AL, Abboud FM. Sympathetic-nerve activity during sleep in normal subjects. N Engl J Med. 1993;328:303–7. doi: 10.1056/NEJM199302043280502. [DOI] [PubMed] [Google Scholar]
  35. Steiropoulos P, Papanas N, Nena E, et al. Markers of glycemic control and insulin resistance in non-diabetic patients with Obstructive Sleep Apnea Hypopnea Syndrome: does adherence to CPAP treatment improve glycemic control? Sleep Med. 2009;10:887–91. doi: 10.1016/j.sleep.2008.10.004. [DOI] [PubMed] [Google Scholar]
  36. Tamura A, Kawano Y, Watanabe T, Kadota J. Obstructive sleep apnea increases hemoglobin A1c levels regardless of glucose tolerance status. Sleep Med. 2012;13:1050–5. doi: 10.1016/j.sleep.2012.04.007. [DOI] [PubMed] [Google Scholar]
  37. Tuomilehto HP, Seppa JM, Partinen MM, et al. Lifestyle intervention with weight reduction: first-line treatment in mild obstructive sleep apnea. Am J Respir Crit Care Med. 2009;179:320–7. doi: 10.1164/rccm.200805-669OC. [DOI] [PubMed] [Google Scholar]
  38. West SD, Nicoll DJ, Wallace TM, Matthews DR, Stradling JR. Effect of CPAP on insulin resistance and HbA1c in men with obstructive sleep apnoea and type 2 diabetes. Thorax. 2007;62:969–74. doi: 10.1136/thx.2006.074351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Wing RR, Bolin P, Brancati FL, et al. Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. N Engl J Med. 2013;369:145–54. doi: 10.1056/NEJMoa1212914. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Young T, Finn L, Peppard PE, et al. Sleep disordered breathing and mortality: eighteen-year follow-up of the Wisconsin sleep cohort. Sleep. 2008;31:1071–78. [PMC free article] [PubMed] [Google Scholar]

RESOURCES