OBJECTIVE

To determine the risk of cardiovascular disease (CVD), microvascular complications, and mortality in patients with type 2 diabetes who subsequently develop obstructive sleep apnea (OSA) compared with patients with type 2 diabetes without a diagnosis of OSA.

RESEARCH DESIGN AND METHODS

This age-, sex-, BMI-, and diabetes duration–matched cohort study used data from a U.K. primary care database from 1 January 2005 to 17 January 2018. Participants aged ≥16 years with type 2 diabetes were included. Exposed participants were those who developed OSA after their diabetes diagnosis; unexposed participants were those without diagnosed OSA. Outcomes were composite CVD (ischemic heart disease [IHD], stroke/transient ischemic attack [TIA], heart failure [HF]), peripheral vascular disease (PVD), atrial fibrillation (AF), peripheral neuropathy (PN), diabetes-related foot disease (DFD), referable retinopathy, chronic kidney disease (CKD), and all-cause mortality. The same outcomes were explored in patients with preexisting OSA before a diagnosis of type 2 diabetes versus diabetes without diagnosed OSA.

RESULTS

A total of 3,667 exposed participants and 10,450 matched control participants were included. Adjusted hazard ratios for the outcomes were as follows: composite CVD 1.54 (95% CI 1.32, 1.79), IHD 1.55 (1.26, 1.90), HF 1.67 (1.35, 2.06), stroke/TIA 1.57 (1.27, 1.94), PVD 1.10 (0.91, 1.32), AF 1.53 (1.28, 1.83), PN 1.32 (1.14, 1.51), DFD 1.42 (1.16, 1.74), referable retinopathy 0.99 (0.82, 1.21), CKD (stage 3–5) 1.18 (1.02, 1.36), albuminuria 1.11 (1.01, 1.22), and all-cause mortality 1.24 (1.10, 1.40). In the prevalent OSA cohort, the results were similar, but some associations were not observed.

CONCLUSIONS

Patients with type 2 diabetes who develop OSA are at increased risk of CVD, AF, PN, DFD, CKD, and all-cause mortality compared with patients without diagnosed OSA. Patients with type 2 diabetes who develop OSA are a high-risk population, and strategies to detect OSA and prevent cardiovascular and microvascular complications should be implemented.

Diabetes-related microvascular complications and cardiovascular disease (CVD) are major causes of morbidity, mortality, and worsening quality of life in patients with type 2 diabetes (13). CVD prevention is one of the main aims of type 2 diabetes treatment (4). Improved health care delivery, including the use of lipid-lowering treatments, antiplatelets (for secondary prevention), and antihypertensives, has resulted in reduced mortality, CVD, and vascular complications in patients with type 2 diabetes; however, the burden of these complications remains large because of the increasing prevalence of type 2 diabetes (4,5). Therefore, identifying risk factors and preventive strategies for the development of vascular disease and mortality in patients with type 2 diabetes is still needed (6).

Obstructive sleep apnea (OSA) is common in patients with type 2 diabetes (24–86%), and patients with type 2 diabetes are at increased risk of developing OSA compared with patients without diabetes (79). Epidemiological studies in patients without diabetes showed that OSA is associated with an increased risk of mortality and CVD (including stroke), which is improved with continuous positive airway pressure (CPAP) in nonrandomized studies (1015). Our group has previously shown a cross-sectional association between OSA and peripheral neuropathy (PN), sight-threatening retinopathy, and chronic kidney disease (CKD) in patients with type 2 diabetes (8,16). We have also shown that OSA is associated with an increased risk of renal function decline, development of CKD, and preproliferative (R2)/proliferative (R3) retinopathy in patients with type 2 diabetes in longitudinal studies (1719). However, these studies were relatively small and from secondary/tertiary care centers. The longitudinal associations between OSA and PN and diabetes-related foot disease (DFD) have not been explored previously. In addition, the relationship between OSA and incident CVD and mortality in patients with type 2 diabetes is largely unknown.

We hypothesized that in patients with type 2 diabetes, the development of OSA increases the risk of CVD, microvascular complications, and mortality. The primary aims of this study were to determine the risk of incident CVD (ischemic heart disease [IHD], stroke/transient ischemic attack [TIA], or heart failure [HF]), PN, DFD, referable retinopathy, and CKD in patients with type 2 diabetes who go on to develop OSA compared with patients with type 2 diabetes but without diagnosed OSA. Secondary outcomes included the individual components of the composite CVD outcome, peripheral vascular disease (PVD), atrial fibrillation (AF), micro- and macroalbuminuria, and all-cause mortality. Additionally, to explore whether the sequence in which type 2 diabetes and OSA are diagnosed has any impact on observed outcomes, we conducted a second cohort study to determine risk of each of the outcomes in patients with type 2 diabetes and preexisting OSA compared with patients with type 2 diabetes but without diagnosed OSA.

Study Design

This age-, sex-, BMI-, and diabetes duration–matched retrospective cohort study was performed from 1 January 2005 to 17 January 2018.

Data Source

The study data set was extracted from The Health Improvement Network (THIN), a database comprising anonymized electronic primary care records for >15 million patients from 787 general practices across the U.K. It contains coded information on patient demographics, symptoms, diagnoses, medication prescriptions, consultations, and diagnostic tests. The data set is derived from routinely collected patient records and is generalizable to the U.K. population. THIN data have been used in numerous studies in the contexts of type 2 diabetes, OSA, and CVD (7,2022).

Population

General practices were included in the study from the latest of the following dates: 12 months after reporting acceptable mortality rates (a measure of data recording quality) (23), 12 months after starting to use electronic medical records, and study start date (1 January 2005). This was to maximize data and recording quality.

Adults aged ≥16 years registered with an eligible general practice for a minimum of 12 months and with a record of type 2 diabetes were eligible for inclusion. Patients who underwent bariatric surgery or had a record of HIV at any time point were excluded.

Primary Analysis: Incident OSA Exposure

The exposed cohort comprised participants with type 2 diabetes and a subsequent, incident diagnosis of OSA (occurring after the diagnosis of type 2 diabetes and during the study period). Index (study entry) date for exposed patients was the date of OSA diagnosis.

Each person in the exposed group was matched on index date to up to four randomly selected individuals with type 2 diabetes but without diagnosed OSA by age, sex, BMI, and diabetes duration. Age and diabetes duration were matched to within ±1 year; BMI was matched to within ±2 kg/m2. Unexposed participants were assigned the same index date as their corresponding exposed participants to mitigate immortal time bias (24). The study design was an open cohort; therefore, any individual with type 2 diabetes who developed incident OSA during the study period was included in the exposed cohort. As a result, no individuals in the unexposed population developed OSA during follow-up.

Secondary Analysis: Prevalent OSA Exposure

The exposed cohort comprised participants with incident type 2 diabetes (diagnosed during the study period) and prevalent OSA (occurring before the diagnosis of type 2 diabetes). Each exposed participant was matched to up to four randomly selected individuals with incident type 2 diabetes and no existing or subsequent diagnosis of OSA by age, sex, BMI (±2 kg/m2), and index year. Index date was the date of type 2 diabetes diagnosis for both exposed and unexposed participants.

Follow-Up Period

Participants were followed up from the index date until the earliest of the following dates: outcome diagnosis, death, participant left the practice, practice ceased contributing to the database, and study end (17 January 2018).

Outcomes

Primary outcomes were composite CVD (IHD, stroke or TIA, and HF), PN, DFD (defined as foot ulcer, lower-limb amputation, or gangrene of the foot), referable retinopathy (R2, R3, maculopathy, low vision/blindness, or corrective procedures for retinopathy, such as laser and vitreous injections), and CKD stages 3–5. Secondary outcomes included each of the three composite CVD outcomes separately (IHD, stroke/TIA, HF); PVD; AF; albuminuria (albumin-to-creatinine ratio [ACR] >3 mg/mmol), macroalbuminuria (ACR >30 mg/mmol), and severe macroalbuminuria (ACR >300 mg/mmol); and all-cause mortality.

Participants with a record of the outcome of interest at index date were excluded from the corresponding analysis; for example, for the composite CVD outcome, patients with a record of IHD, stroke/TIA, or HF at baseline were excluded. In the analysis for PN as the outcome, participants with folate or B12 deficiency were excluded because these deficiencies are associated with development of the outcome.

Definitions of Variables

Type 2 diabetes was defined as a record of any diabetes clinical (Read) code and no record of type 1 diabetes. Individuals with prevalent or incident (first recorded during the study period) diabetes were included.

OSA, IHD, stroke/TIA, HF, PVD, AF, PN, DFD, referable retinopathy, hypertension, and conditions contributing to the Charlson comorbidity index (CCI) (25) were defined by a record of a relevant diagnostic clinical (Read) code indicating the presence of the condition. CKD stage 3–5 was defined by the presence of a relevant clinical code or by an estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2 in two records separated by at least 90 days (the latter aligns with the definition used in clinical guidelines) (26). All cardiovascular outcomes are included in the Quality and Outcomes Framework (QOF), a payment system that incentivizes general practices in the U.K. to maintain disease registers (27); these outcomes are therefore well recorded in primary care.

Covariates that might impact the outcomes were selected on the basis of biological plausibility and previous literature. Covariates included age; sex; BMI; Townsend deprivation quintile; smoking status; current prescription of lipid-lowering drugs, antihypertensives, antiplatelets, and insulin; ethnicity; diabetes duration; HbA1c; eGFR; ACR; hypertension; and CCI (mortality outcome only).

Covariates were measured at baseline. Physiological measures were taken as the latest value recorded before the index date. Current medication prescriptions were defined as those issued within 60 days before the index date. Insulin prescription was used as an indication of disease severity.

BMI was categorized as <25 kg/m2 (normal or underweight), 25–30 kg/m2 (overweight), and ≥30 kg/m2 (obesity). Smoking was categorized as current smoker, ex-smoker, and nonsmoker. HbA1c was categorized as ≤47.5, 47.5–58.5, 58.5–69.4, and >69.4 mmol/mol. eGFR was calculated from serum creatinine values and ethnicity data (where available) using the Chronic Kidney Disease Epidemiology Collaboration equation (28) and categorized as ≥60, 30–59, and <30 mL/min/1.73 m2. ACR was categorized as <3, 3–30, and >30 mg/mmol (26).

Missing Data

Missing categories were used where values for BMI, smoking status, Townsend quintile, ethnicity, HbA1c, eGFR, and ACR were not recorded. Implausible measurements of BMI, HbA1c, eGFR, and ACR were considered data entry errors and were treated as missing. In a sensitivity analysis, missing values of BMI, HbA1c, eGFR, and ACR were replaced using multiple imputation, using chained equations with predictive mean matching. The absence of a diagnostic code or medication code was taken to indicate absence of the condition or prescription, respectively.

Analysis

Crude incidence rates were calculated for each outcome. Crude and adjusted hazard ratios (HRs) and their corresponding 95% CIs were calculated using Cox proportional hazards models. The proportional hazards assumption was checked using log-log plots and the Schoenfeld residuals test. Adjusted models for all cardiovascular outcomes (composite CVD, IHD, stroke/TIA, HF, PVD, AF) and microvascular diabetes complications (PN, DFD, referable retinopathy) included the following covariates measured at baseline: age category; sex; BMI category; smoking category; Townsend deprivation quintile; hypertension; prescription of lipid-lowering drugs, antihypertensives, antiplatelets, and insulin; ethnicity; diabetes duration; HbA1c category; eGFR category; and ACR category. For renal outcomes (CKD, albuminuria), covariates were as above with the exclusion of eGFR and ACR and with the addition of baseline CVD (IHD, HF, stroke/TIA). The adjusted model for all-cause mortality outcome included age, sex, BMI, smoking status, Townsend quintile, lipid-lowering drug prescription, antiplatelet prescription, antihypertensive prescription, insulin prescription, ethnicity, diabetes duration, HbA1c category, and CCI.

In primary analysis, all identified exposed patients and their corresponding control participants were included. It was not possible to identify matched control participants for a small proportion of patients with OSA; therefore, sensitivity analyses were performed, restricting to those exposed patients for whom one or more control participants were identified together with the corresponding control participants.

All analyses were performed using Stata/IC 14 statistical software. Two-sided P values were obtained, and P < 0.05 was considered statistically significant.

Ethics

The THIN data collection scheme and research carried out using THIN data were approved by the NHS South-East Multicentre Research Ethics Committee in 2003. Under the terms of the approval, studies must undergo independent scientific review. Approval for this study was obtained from the Scientific Review Committee (for the use of THIN data) in July 2018 (SRC reference 18THIN062).

Primary Analysis: Incident OSA Exposure

In the primary analysis, we compared participants with type 2 diabetes and a subsequent diagnosis of OSA with participants with type 2 diabetes and no OSA. A total of 3,667 adults with type 2 diabetes and a subsequent diagnosis of OSA and 10,450 control participants with type 2 diabetes but no diagnosis of OSA were included in the analysis (Fig. 1). Baseline characteristics are summarized in Table 1. The overall study population was middle-aged and mostly male, with a high prevalence of IHD (approximately one-fifth) and hypertension (almost two-thirds) at baseline. The majority were prescribed antihypertensives and lipid-lowering medications, and the average HbA1c was 7.5% (59 mmol/mol). Approximately 15% of the study population were current smokers. Age, sex, BMI, and diabetes duration (matching parameters) were broadly similar between exposed and unexposed participants (Table 1). Compared with control participants without diagnosed OSA, patients with type 2 diabetes who went on to develop OSA had marginally higher deprivation scores; were slightly more likely to be of South Asian ethnicity (3.0% vs. 2.2%); had a higher prevalence of hypertension and IHD; had a slightly higher mean HbA1c; and were more likely to be prescribed medications, particularly insulin (19.3% vs. 8.9%), at baseline (Table 1).

Cardiovascular Outcomes

A total of 250 (10.1%) participants with type 2 diabetes and incident OSA and 564 (7.0%) control participants with type 2 diabetes and without diagnosed OSA developed composite CVD (IHD, stroke/TIA, and HF) during follow-up (Supplementary Table 1). Median (interquartile range [IQR]) follow-up was 3.1 (1.6–5.5) and 3.5 (1.7–5.9) years in exposed and unexposed patients. Crude incidence rates were 26.6 and 17.4 per 1,000 person-years in exposed and unexposed participants, respectively. Crude and adjusted HRs were 1.54 (95% CI 1.32, 1.78) and 1.54 (1.32, 1.79), respectively (Fig. 2 and Supplementary Table 1). These results suggest that incident OSA in patients with type 2 diabetes was associated with increased risk of incident composite CVD compared with patients with type 2 diabetes but without diagnosed OSA.

Adjusted HRs for each component CVD were as follows: IHD 1.55 (95% CI 1.26, 1.90), stroke/TIA 1.57 (1.27, 1.94), and HF 1.67 (1.35, 2.06), showing similar associations to those between OSA and composite CVD. For AF, the adjusted HR was 1.53 (1.28, 1.83). There was no statistically significant difference in hazard of PVD (adjusted HR 1.10 [0.91, 1.32]).

Diabetes-Related Microvascular Outcomes

For each of the microvascular outcomes, the median follow-up period was ∼3 years for exposed participants with type 2 diabetes and comorbid OSA and ∼3.5 years for unexposed participants with type 2 diabetes but without diagnosed OSA. A total of 288 (10.5%) exposed participants and 697 (8.2%) unexposed participants developed PN during follow-up (Supplementary Table 2). The hazard of PN was significantly higher in participants with OSA than in those without diagnosed OSA (crude HR 1.37 [95% CI 1.19, 1.57], adjusted HR 1.32 [1.14, 1.51]) (Fig. 3 and Supplementary Table 2).

A total of 161 (4.7%) and 289 (2.9%) exposed and unexposed participants, respectively, developed DFD. The hazard of DFD was significantly higher in participants with OSA than in those without diagnosed OSA (crude HR 1.76 [95% CI 1.45, 2.13], adjusted HR 1.42 [1.16, 1.74]).

A total of 156 (4.7%) and 395 (4.0%) exposed and unexposed participants, respectively, developed diabetes-related referable retinopathy. Crude HR was 1.22 (95% CI 1.01, 1.47); after adjustment for potential confounders, the result was not statistically significant (adjusted HR 0.99 [0.82, 1.21]).

Renal Outcomes

A total of 276 (9.7%) and 740 (8.6%) exposed and unexposed participants, respectively, developed CKD stage 3–5. There was a significant increase in hazard of CKD in the exposed group compared with the unexposed group (crude HR 1.20 [95% CI 1.05, 1.38], adjusted HR 1.18 [1.02, 1.36]) (Fig. 3 and Supplementary Table 2). Hazard of albuminuria was higher in participants with type 2 diabetes and incident OSA than in those without diagnosed OSA, although this did not reach statistical significance for severe macroalbuminuria (adjusted HRs for micro-/macroalbuminuria [ACR >3 mg/mmol], macroalbuminuria [ACR >30 mg/mmol], and severe macroalbuminuria [ACR >300 mg/mmol] 1.11 [1.01, 1.22], 1.33 [1.13, 1.55], and 1.33 [0.92, 1.93], respectively).

All-Cause Mortality

A total of 434 (11.8%) exposed and 887 (8.5%) unexposed participants died during follow-up. Median (IQR) follow-up was 3.2 (1.7–5.8) and 3.6 (1.8–6.1) years in the exposed and comparator groups, respectively. Crude mortality rates were 29.9 and 20.3 per 1,000 person-years in exposed and unexposed patients, respectively. Adjusted HR was 1.24 (95% CI 1.10, 1.40) (Fig. 2 and Supplementary Table 1), showing that OSA was associated with increased all-cause mortality in patients with type 2 diabetes.

Sensitivity Analysis

A sensitivity analysis was performed excluding exposed patients with no matched control participants; this made no difference to the observed findings (Supplementary Tables 13). A further supplementary analysis was performed replacing missing values by multiple imputation; this did not affect the results (Supplementary Tables 1 and 2).

Prevalent OSA Exposure

In secondary analysis, we explored outcomes in participants with incident type 2 diabetes and preexisting OSA (prevalent at the time of the type 2 diabetes diagnosis) compared with participants with type 2 diabetes and no OSA. A total of 4,564 participants with type 2 diabetes and prevalent OSA and 15,589 control participants with diabetes and without diagnosed OSA were included in the analysis. Baseline characteristics are presented in Table 1.

Cardiovascular Outcomes

Adjusted HRs for composite CVD, IHD, stroke/TIA, and HF were 1.23 (95% CI 1.05, 1.43), 1.14 (0.95, 1.38), 1.20 (0.94, 1.52), and 1.26 (0.98, 1.63), respectively (Fig. 2 and Supplementary Table 4). Adjusted HRs for AF and PVD were 1.11 (0.91, 1.34) and 1.27 (1.17, 1.47), respectively.

Microvascular Outcomes

Adjusted HRs for PN, DFD, and referable retinopathy were 1.22 (95% CI 1.11, 1.34), 1.36 (1.08, 1.71), and 0.93 (0.75, 1.16), respectively (Fig. 3 and Supplementary Table 5). Adjusted HRs for CKD (stage 3–5), micro-/macroalbuminuria, macroalbuminuria, and severe macroalbuminuria were 1.01 (0.89, 1.15), 1.16 (1.09, 1.24), 1.21 (1.04, 1.41), and 1.37 (0.78, 2.41), respectively (Fig. 3 and Supplementary Table 5).

All-Cause Mortality

The adjusted HR for all-cause mortality was 1.00 (95% CI 0.88–1.14) (Fig. 2 and Supplementary Table 4).

In this cohort study, participants who developed OSA after their diagnosis of type 2 diabetes had a >50% increase in risk of composite CVD, IHD, HF, and stroke/TIA; a 53% greater risk of developing AF; a 32% increase in risk of PN; a 42% increase in risk of DFD; an 18% increase in risk of CKD stages 3–5; an 11% increased risk of albuminuria; and a 24% increase in risk of all-cause mortality compared with participants with type 2 diabetes and no OSA. These results were observed after matching for age, sex, BMI, and diabetes duration and adjusting for a range of potential confounders. There were no significant associations between OSA and the risk of PVD or referable retinopathy in this study.

This study adds novel insights and findings to the limited published literature describing the impact of OSA on cardiovascular outcomes in patients with type 2 diabetes. A previous study showed an association between OSA and stroke; however, this was cross-sectional, relied on self-reported outcomes, and was of small sample size (29). Two previously published cohort studies showed an association between OSA and CVD/mortality (30,31), but these studies were in highly selected populations (patients with type 2 diabetes referred to cardiology units for the investigation of coronary artery disease in one study and patients referred for percutaneous coronary intervention in the other). In addition, these studies were of smaller sample size, the analysis adjusted for a limited number of variables, and one of the studies included people at high risk of OSA rather than diagnosed OSA. Hence, our study provides significant findings and adds methodological rigor compared with the limited published literature in that it was a population-based study of large sample size, allowing for adjustment for a large number of potential confounders, and examined the association between OSA and multiple individual CVD outcomes, as well as all-cause mortality and microvascular complications, in patients with type 2 diabetes.

The association of OSA with incident AF in our study was mirrored by a similar association with incident stroke. Hence, our data suggest that identifying patients with type 2 diabetes who have OSA might provide an opportunity to examine for AF and implement appropriate stroke prevention strategies. There is interest in the impact of OSA in patients with AF on the risk of stroke and AF recurrence following ablation (32,33).

Our study also shows that having OSA identifies a high-risk population with type 2 diabetes in which CVD prevention measures should be maximized. To our knowledge, our study is the first to report the association between OSA and mortality in patients with type 2 diabetes, which is similar to that observed in general population studies without diabetes (1215).

Our study is the first to show the relationship between OSA and PN and DFD in patients with type 2 diabetes in a longitudinal, population-based study. We have previously shown an association between OSA and PN in a cross-sectional study (16). We have also previously shown OSA to be associated with CKD in a cross-sectional analysis, and with eGFR decline in a longitudinal study, in secondary/tertiary care centers in the U.K. (19,34). The current study allowed us to expand our findings to a more representative population and to examine the impact of OSA on albuminuria. Unlike our previous longitudinal study that showed an association between OSA and increased risk of R2 and R3 retinopathy (17), this study showed no significant association between OSA and referable retinopathy after adjustment. This difference may be explained by the different study populations—secondary/tertiary versus primary care—and the use of a broader outcome measure (referable retinopathy) compared with our previous study, which was based only on the development of R2 or R3.

There are multiple mechanisms that link OSA to CVD and microvascular complications, including insulin resistance, hypertension, increased oxidative and nitrosative stress, sympathetic activation, increased inflammation, and endothelial dysfunction, which can improve with CPAP (34,35). Our findings pose the question of whether CPAP should form part of CVD and mortality prevention in patients with type 2 diabetes and comorbid OSA. Our study does not address this question, but despite observational studies showing an association between use of CPAP and vascular benefits in patients with OSA, data from randomized controlled trials (RCTs) remain lacking (36,37), with the exception of one RCT that showed that CPAP reduced cardiovascular events and mortality after a first ischemic stroke (11). However, CPAP trials are always challenging because of the lack of adherence to treatment. Nonetheless, the impact of CPAP on CVD in patients with type 2 diabetes needs to be assessed in well-designed RCTs.

In the secondary analysis exploring outcomes in participants with type 2 diabetes and prevalent OSA compared with type 2 diabetes only, participants with OSA continued to be at increased risk of composite CVD, PN, DFD, and albuminuria. The association between OSA and increased risk of individual components of the composite CVD outcome, AF, CKD, and all-cause mortality was no longer significant. This secondary analysis included participants with incident/newly diagnosed type 2 diabetes; the difference in findings is therefore likely to be driven by the fact that these individuals had had type 2 diabetes for a shorter period of time at the end of follow-up compared with the primary analysis cohort (which included participants with prevalent diabetes). Another possible reason is that patients in the prevalent OSA group might have received better vascular prevention compared with the control group before the incident diagnosis of type 2 diabetes.

Routinely collected data may be subject to incorrect, inconsistent, or incomplete recording. However, type 2 diabetes, CVD, and CKD are part of the U.K. QOF, which is linked to practice funding, and recording quality is therefore expected to be high. The QOF indicator set for diabetes includes requirements to measure and record smoking status, BMI, blood pressure, HbA1c, ACR, and serum creatinine as well as to perform and record annual foot assessments and retinal screening; this information is therefore likely to be well recorded for patients included in the cohort. After adjusting for patient demographics, the prevalence of major chronic diseases and death rates in THIN are similar to national rates (38).

Ethnicity is poorly recorded in THIN and was therefore not available for all patients in the cohort. Previous analysis has shown that the prevalence of OSA in THIN may be lower than expected rates on the basis of existing literature, suggesting that OSA may be underdiagnosed or underreported in U.K. primary care data (7); it is possible, therefore, that patients diagnosed with OSA in routine care represent a more severe or symptomatic phenotype, and hence our results may not be representative of all patients with type 2 diabetes and OSA. Underdiagnosis of OSA in primary care may have resulted in patients with undiagnosed OSA being included in the unexposed group, leading to an underestimation of the strength of the observed associations.

The presence of detection bias cannot be completely ruled out; however, we believe that it is unlikely to have had a major impact. All patients with type 2 diabetes are reviewed at least annually in England and assessed for vascular complications as part of the QOF; therefore, surveillance bias is unlikely to have had a major impact on our results. This is supported by the lack of increased risk of retinopathy and by evidence from previous studies in which screen-detected OSA was associated with vascular complications in patients with type 2 diabetes.

The study included a large sample size from a data set that is generalizable to the U.K. population. This is the first population-based cohort study to assess the association of OSA diagnosed after type 2 diabetes with cardiovascular end points, including composite CVD, AF, IHD, stroke/TIA, and HF, and with microvascular and renal outcomes. The large sample size and characterization of the study population allowed us to adjust for a large number of variables that might affect the associations between OSA and CVD in patients with type 2 diabetes. In addition, the matching design further strengthened the methodology of our study.

In conclusion, patients with type 2 diabetes who go on to develop comorbid OSA are at increased risk of incident CVD, including IHD, stroke/TIA, HF, and AF, as well as at increased risk of PN, DFD, CKD, albuminuria, and all-cause mortality. Physicians need to recognize that patients with type 2 diabetes who develop OSA are a high-risk population and that strategies to detect OSA and prevent vascular complications should be implemented. RCTs examining the impact of CPAP on cardiovascular and microvascular outcomes in patients with OSA and type 2 diabetes are needed.

A.A.T. and K.N. are joint senior authors.

This article contains supplementary material online at https://doi.org/10.2337/figshare.12014598.

This article is featured in a podcast available at https://www.diabetesjournals.org/content/diabetes-core-update-podcasts.

See accompanying article, p. 1859.

Acknowledgments. THIN is a registered trademark of Cegedim SA in the U.K. and other countries. Reference made to the THIN database is intended to be descriptive of the data asset licensed by IQVIA.

Funding. A.A.T. is a clinician scientist supported by the National Institute for Health Research in the U.K. (CS-2013-13-029).

The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the National Institute for Health Research, or the Department of Health.

Duality of Interest. A.S., G.N.T., and K.N. received funding from AstraZeneca (RSBD20464). W.H. reports personal fees and nonfinancial support from Novo Nordisk, Eli Lilly, AstraZeneca, Boehringer Ingelheim, and Napp Pharmaceuticals. A.A.T. reports personal fees and nonfinancial support from Novo Nordisk, Eli Lilly, AstraZeneca, and Boehringer Ingelheim; personal fees from Janssen; and nonfinancial support from Impeto Medical, ResMed, and Aptiva. K.N. reports fees from Sanofi and Boehringer Ingelheim outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author Contributions. N.J.A., A.S., and K.N. designed and performed the analysis. N.J.A. and A.A.T. wrote the initial draft of the manuscript. N.J.A., A.A.T., and K.N. conceived the idea for the study. All authors reviewed and revised the manuscript and agreed to the submission of the final manuscript. N.J.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Rawshani
A
,
Rawshani
A
,
Franzén
S
, et al
.
Mortality and cardiovascular disease in type 1 and type 2 diabetes
.
N Engl J Med
2017
;
376
:
1407
1418
2.
Cusick
M
,
Meleth
AD
,
Agrón
E
, et al.;
Early Treatment Diabetc Retinopathy Study Research Group
.
Associations of mortality and diabetes complications in patients with type 1 and type 2 diabetes: early treatment diabetic retinopathy study report no. 27
.
Diabetes Care
2005
;
28
:
617
625
3.
U.K. Prospective Diabetes Study Group
.
Quality of life in type 2 diabetic patients is affected by complications but not by intensive policies to improve blood glucose or blood pressure control (UKPDS 37)
.
Diabetes Care
1999
;
22
:
1125
1136
4.
Davies
MJ
,
D’Alessio
DA
,
Fradkin
J
, et al
.
Management of hyperglycemia in type 2 diabetes, 2018. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD)
.
Diabetes Care
2018
;
41
:
2669
2701
5.
Gregg
EW
,
Li
Y
,
Wang
J
, et al
.
Changes in diabetes-related complications in the United States, 1990-2010
.
N Engl J Med
2014
;
370
:
1514
1523
6.
Rawshani
A
,
Rawshani
A
,
Franzén
S
, et al
.
Risk factors, mortality, and cardiovascular outcomes in patients with type 2 diabetes
.
N Engl J Med
2018
;
379
:
633
644
7.
Subramanian
A
,
Adderley
NJ
,
Tracy
A
, et al
.
Risk of incident obstructive sleep apnea among patients with type 2 diabetes
.
Diabetes Care
2019
;
42
:
954
963
8.
Lavrentaki
A
,
Ali
A
,
Cooper
B
,
Tahrani
A
.
Mechanisms of disease: the endocrinology of obstructive sleep apnoea
.
Eur J Endocrinol
2018
;
180
:
R91
R125
9.
Tahrani
AA
,
Ali
A
.
Obstructive sleep apnoea and type 2 diabetes
.
Eur Endocrinol
2014
;
10
:
43
50
10.
Marin
JM
,
Carrizo
SJ
,
Vicente
E
,
Agusti
AG
.
Long-term cardiovascular outcomes in men with obstructive sleep apnoea-hypopnoea with or without treatment with continuous positive airway pressure: an observational study
.
Lancet
2005
;
365
:
1046
1053
11.
Parra
O
,
Sánchez-Armengol
Á
,
Capote
F
, et al
.
Efficacy of continuous positive airway pressure treatment on 5-year survival in patients with ischaemic stroke and obstructive sleep apnea: a randomized controlled trial
.
J Sleep Res
2015
;
24
:
47
53
12.
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
1078
13.
Punjabi
NM
,
Caffo
BS
,
Goodwin
JL
, et al
.
Sleep-disordered breathing and mortality: a prospective cohort study
.
PLoS Med
2009
;
6
:
e1000132
14.
Peker
Y
,
Hedner
J
,
Norum
J
,
Kraiczi
H
,
Carlson
J
.
Increased incidence of cardiovascular disease in middle-aged men with obstructive sleep apnea: a 7-year follow-up
.
Am J Respir Crit Care Med
2002
;
166
:
159
165
15.
Yaggi
HK
,
Concato
J
,
Kernan
WN
,
Lichtman
JH
,
Brass
LM
,
Mohsenin
V
.
Obstructive sleep apnea as a risk factor for stroke and death
.
N Engl J Med
2005
;
353
:
2034
2041
16.
Tahrani
AA
,
Ali
A
,
Raymond
NT
, et al
.
Obstructive sleep apnea and diabetic neuropathy: a novel association in patients with type 2 diabetes
.
Am J Respir Crit Care Med
2012
;
186
:
434
441
17.
Altaf
QA
,
Dodson
P
,
Ali
A
, et al
.
Obstructive sleep apnea and retinopathy in patients with type 2 diabetes. A longitudinal study
.
Am J Respir Crit Care Med
2017
;
196
:
892
900
18.
Tahrani
AA
.
Obstructive sleep apnoea in diabetes: does it matter
?
Diab Vasc Dis Res
2017
;
14
:
454
462
19.
Tahrani
AA
,
Ali
A
,
Raymond
NT
, et al
.
Obstructive sleep apnea and diabetic nephropathy: a cohort study
.
Diabetes Care
2013
;
36
:
3718
3725
20.
Daly
B
,
Toulis
KA
,
Thomas
N
, et al
.
Increased risk of ischemic heart disease, hypertension, and type 2 diabetes in women with previous gestational diabetes mellitus, a target group in general practice for preventive interventions: a population-based cohort study
.
PLoS Med
2018
;
15
:
e1002488
21.
Adderley
NJ
,
Nirantharakumar
K
,
Marshall
T
.
Risk of stroke and transient ischaemic attack in patients with a diagnosis of resolved atrial fibrillation: retrospective cohort studies
.
BMJ
2018
;
361
:
k1717
22.
Zhang
Y
,
Peloquin
CE
,
Dubreuil
M
, et al
.
Sleep apnea and the risk of incident gout: a population‐based, body mass index–matched cohort study
.
Arthritis Rheumatol
2015
;
67
:
3298
3302
23.
Maguire
A
,
Blak
BT
,
Thompson
M
.
The importance of defining periods of complete mortality reporting for research using automated data from primary care
.
Pharmacoepidemiol Drug Saf
2009
;
18
:
76
83
24.
Lévesque
LE
,
Hanley
JA
,
Kezouh
A
,
Suissa
S
.
Problem of immortal time bias in cohort studies: example using statins for preventing progression of diabetes
.
BMJ
2010
;
340
:
b5087
25.
Charlson
ME
,
Pompei
P
,
Ales
KL
,
MacKenzie
CR
.
A new method of classifying prognostic comorbidity in longitudinal studies: development and validation
.
J Chronic Dis
1987
;
40
:
373
383
26.
Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease
[article online].
Kidney Int Suppl
2013
;3:
3
150
. Available from https://kdigo.org/wp-content/uploads/2017/02/KDIGO_2012_CKD_GL.pdf. Accessed 8 October 2019
27.
General Practitioners Committee
,
NHS Employers and NHS England
.
2016/17 General Medical Services (GMS) contract Quality and Outcomes Framework (QOF) Guidance for GMS contract 2016/17
. Leeds, NHS Employers,
2016
.
28.
Levey
AS
,
Stevens
LA
,
Schmid
CH
, et al.;
CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration)
.
A new equation to estimate glomerular filtration rate
.
Ann Intern Med
2009
;
150
:
604
612
29.
Rice
TB
,
Foster
GD
,
Sanders
MH
, et al.;
Sleep AHEAD Research Group
.
The relationship between obstructive sleep apnea and self-reported stroke or coronary heart disease in overweight and obese adults with type 2 diabetes mellitus
.
Sleep (Basel)
2012
;
35
:
1293
1298
30.
Seicean
S
,
Strohl
KP
,
Seicean
A
,
Gibby
C
,
Marwick
TH
.
Sleep disordered breathing as a risk of cardiac events in subjects with diabetes mellitus and normal exercise echocardiographic findings
.
Am J Cardiol
2013
;
111
:
1214
1220
31.
Koo
CY
,
Drager
LF
,
Sethi
R
, et al.;
Sleep and Stent Study Investigators
.
Obstructive sleep apnea and diabetes independently add to cardiovascular risk after coronary revascularization
.
Diabetes Care
2018
;
41
:
e12
e14
32.
Goudis
CA
,
Ketikoglou
DG
.
Obstructive sleep and atrial fibrillation: pathophysiological mechanisms and therapeutic implications
.
Int J Cardiol
2017
;
230
:
293
300
33.
Bavishi
AA
,
Kaplan
RM
,
Peigh
G
, et al
.
Patient characteristics as predictors of recurrence of atrial fibrillation following cryoballoon ablation
.
Pacing Clin Electrophysiol
2019
;
42
:
694
704
34.
A Tahrani
A
.
Obstructive sleep apnoea and vascular disease in patients with type 2 diabetes
.
Eur Endocrinol
2015
;
11
:
81
89
35.
Jullian-Desayes
I
,
Joyeux-Faure
M
,
Tamisier
R
, et al
.
Impact of obstructive sleep apnea treatment by continuous positive airway pressure on cardiometabolic biomarkers: a systematic review from sham CPAP randomized controlled trials
.
Sleep Med Rev
2015
;
21
:
23
38
36.
McEvoy
RD
,
Antic
NA
,
Heeley
E
, et al.;
SAVE Investigators and Coordinators
.
CPAP for prevention of cardiovascular events in obstructive sleep apnea
.
N Engl J Med
2016
;
375
:
919
931
37.
Craig
S
,
Kylintireas
I
,
Kohler
M
, et al
.
Effect of CPAP on cardiac function in minimally symptomatic patients with OSA: results from a subset of the MOSAIC randomized trial
.
J Clin Sleep Med
2015
;
11
:
967
973
38.
Blak
BT
,
Thompson
M
,
Dattani
H
,
Bourke
A
.
Generalisability of The Health Improvement Network (THIN) database: demographics, chronic disease prevalence and mortality rates
.
Inform Prim Care
2011
;
19
:
251
255
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.