Original Article
Sleep disorders increase the risk of venous thromboembolism in
individuals without sleep apnea: a nationwide population- based
cohort study in Taiwan
Wei-Sheng Chung a,b,c, Yung-Fu Chen b, Cheng-Li Lin d,e, Shih-Ni Chang f,Wu- Huei Hsu g,
Chia-Hung Kao g,h,*
1. Introduction
Insomnia is a common, worldwide complaint [1–3]; it is also a common sleep disorder (SD) that is characterized by difficulty falling and staying asleep. The association between SDs and comorbidities such as previous psychiatric disorders, circulatory diseases, and gastrointestinal diseases is evident [1]. Several studies have
reported that insomnia may predispose people to a risk of cardiovascular and cerebrovascular events [4–6].
Deep vein thrombosis (DVT) is the formation of blood clots in one or more of the deep veins, possibly causing a blood clot to break loose, travel through the bloodstream, and lodge in the lungs. Pulmonary thromboembolism (PE) is a potentially life-threatening
disorder involving embolic or thrombotic occlusion of the pulmonary arterial system. Combined, DVT and PE constitute venous
thromboembolism (VTE), which is a catastrophic disease and has a 30-day case fatality rate of 11–30% [7–9]. Atherothrombosis and VTE share common risk factors and pathophysiological characteristics of inflammation, endothelial injury, and hypercoagulability.
Multiple acquired risk factors are associated with VTE development.
The VTE incidence in immobilized post-cerebrovascular
accidents (CVAs) is relatively high [10]. Leg fractures and major
surgery are also critical risk factors for VTE [11], and the risk of VTE increases among people with spinal cord injuries as well [12]. In
addition, numerous cancers have also been shown to correlate with VTE [13,14].
A SD may drive inflammation and elevate the inflammatory mediators [15], and inflammation involves endothelial dysfunction and
coagulopathy [16]. Among people with VTE, the prevalence of obstructive sleep apnea appears to be higher than in the general
population [17]. However, SD is not a traditional risk factor for VTE, and studies investigating the possibility that SD may increase VTE risks have been few. A longitudinal, nationwide cohort study was conducted in Taiwan in order to investigate whether people with SDs other than sleep apnea had an increased risk of subsequent VTE development.
2. Methods
2.1. Data sources
The National Health Insurance (NHI) program was launched in
Taiwan in 1995, and since then has contracted 97% of medical providers and enrolled up to 99% of the Taiwanese population. The
National Health Research Institute (NHRI) is responsible for managing insurance claims data, which contains all medical claims from
1996 to 2011. The NHRI has established electronic datasets for administrative and research purposes. The National Health Insurance
Research Database (NHIRD) has been described in detail in previous studies [18,19]. The present study used a sub dataset of the NHIRD, which comprises one million randomly sampled beneficiaries enrolled in the NHI program in 2000, and involved collecting all of the records of the study participants from the period 1996–2011.
The diagnosis codes used in the NHRI are based on the International Classification of Diseases, Revision 9, Clinical Modification
(ICD-9-CM). The World Health Organization published
ICD-9-CM for clinical classification purposes as well as for death certificates [20]. The NHRI encrypts the personal information of people,
for privacy protection, and provides researchers with anonymous identification numbers associated with the relevant claims information, including gender, date of birth, registry of medical
services, and medication prescriptions. Individual consent is not required for accessing the NHIRD. The Institutional Review Board of
China Medical University in Central Taiwan exempted this study (CMU-REC-101-012).
2.2. Study participants
The present study consisted of a cohort of people with SDs and
a comparison cohortwithout SDs. Newly diagnosed peoplewith nonapnea
SD were identified in the database between January 1 1998
and December 31 2001 (ICD-9-CM 307.4 and 780.5). The index date was defined as the date of first diagnosis of SD. People with a history of VTE (ICD-9-CM 415.1 and 453.8), except for iatrogenic PE (ICD- 9-CM 415.11) or sleep apnea syndrome (ICD-9-CM 780.51, 780.53 and 780.57) before the index date, and those aged younger than 18 years were excluded. For each corresponding person with a SD, four controls without SD were selected using a systematic randomsampling method as a comparison cohort, and they were frequency
matched by age, gender, and index year, according to the same exclusion criteria. Finally, a total of 46,371 people with a SD and 92,742
controls without a SD were included in the present study.
2.3. Outcome measurement
The primary outcome, obtained from hospitalization records, was newly diagnosed DVT or PE. Each study participant was followed up until: a diagnosis of VTE was made, censorship for loss to follow-up, death, withdrawal from the database, the end of 2011 – whichever came first. Nearly all of the people with DVT and PE underwent comprehensive examinations before receiving intensive care. In Taiwan, the medical reimbursements and discharge notes of the patients were scrutinized in a peer-review process.
2.4. Exposure variables
In addition to a SD, the demographic characteristics such as gender, age, monthly income, occupation, medication of zolpidem and benzodiazepine (BZD), and comorbidities were analyzed. For other comorbid diseases, searches were conducted for a history of:
hypertension (ICD-9-CM 401–405); diabetes (ICD-9-CM 250); hyperlipidemia (ICD-9-CM 272); CVA (ICD-9-CM 430–438); heart
failure (ICD-9-CM 428); lower leg fracture or surgery (ICD-9-CM 820, 821, 823, 81.51, 81.52, 81.53, and 81.54); or malignancy (ICD-9-CM 140–208). For insurance premium estimation, the participants’
monthly incomes were classified into three groups with monthly pay of: < NTD15,000; NTD15,000–19,999; and ≥NTD20,000 (USD1.0 equals approximately NTD30 [New Taiwan Dollars]). White-collar workers were defined as people who perform professional, managerial or administrative work. Typically, a white-collar worker
executes his/her work in an office. Blue-collar workers were defined as people who perform manual labor, such as fishermen, farmers,
and industrial laborers. Other occupations primarily included were retired, unemployed, or low-income populations.
2.5. Statistical analysis
Data analysis was performed to compare the distributions of age, gender, and comorbidities between the SD and non-SD cohorts, which were examined using the Chi-squared test. The follow-up person-years were used for estimating the incidence density of VTE.
Univariable and multivariable Cox proportional hazards regression analyses were used to determine the effects of SDs on the risks of VTE, shown according to hazard ratio (HR) and 95% CI. The Kaplan–
Meier method was used to estimate the VTE-free survival curve of both cohorts, and the log-rank test was used to calculate whether significant differences existed.
All of the analyses were performed using SAS statistical software (version 9.2; SAS Institute, Inc., Cary, NC, USA), and the results were statistically significant when the two-tailed p-values were less than 0.05.
3. Results
3.1. Comparisons of the demographic characteristics and comorbidities between the people with sleep disorders and the comparison cohort
Table 1 displays the distribution of the demographic characteristics
and comorbidities of the people with SDs and the controls.
Among the people in the SD cohort, 63.5% were women and 47.4%
were aged younger than 49 years. The SD cohort exhibited a higher proportion of having a low income (26.1% vs 25.9%) and bluecollar jobs (39.8% vs 38.4%) than the comparison cohort. Compared with the people in the comparison cohort, people with SDs exhibited significantly higher rates of: hypertension (39.1% vs 24.5%,
p < 0.0001); diabetes (15.1% vs 10.1%, p < 0.0001); hyperlipidemia (20.4% vs 10.9%, p < 0.0001); CVAs (14.8% vs 7.44%, p < 0.0001); heart failure (2.43% vs 1.39%, p < 0.0001); lower leg fracture or surgery
(1.08% vs 0.64%, p < 0.0001); and cancer (4.74% vs 3.08%, p < 0.0001). 3.2. The incidence and hazard ratios of VTE for both cohorts by
demographic characteristics and comorbidity
Table 2 shows the incidence rate of VTE for both cohorts, and the HRs of the SD cohort compared with those of the non-SD cohort, according to gender, age, and comorbidity. The incidence rate of VTE
was 1.64-fold higher in the SD cohort than in the non-SD cohort (6.05 vs 3.70 per 10,000 person-years), with an adjusted HR of 1.79 (95% CI 1.49–2.16).
The gender-specific incidence shows that women with SDs have the highest risk of developing VTE (6.67 per 10,000 person-years) in both cohorts, yielding a SD to non-SD cohort HR of 2.19 (95% CI 1.74–2.74) among the women. The incidence of VTE increased with age for both cohorts and was higher for those with SD; the highest was in the oldest group. However, after this was adjusted for
covariates, the age-specific SD to non-SD cohort HRs were significantly higher for those aged ≦49 years, with an adjusted HR of 3.29
(95% CI 2.12–5.12), and 50–64 years, with an adjusted HR of 2.43 (95% CI 1.76–3.35), but was nonsignificant for the oldest group, with an adjusted HR of 1.11 (95% CI 0.84–1.47) (Table 2). Moreover, compared with the non-SD cohort, the SD cohort was associated with
a significantly higher risk of VTE, with an adjusted HR of 2.48 (95%
CI 1.74–3.54), in those who did not present with any comorbidity.
3.3. The interaction on the risk of VTE among sleep disorders, comorbidity and medication of zolpidem and benzodiazepine The interaction measurements between SDs and any comorbidity on the risk of VTE are shown in Table 3. Compared with those without SDs or comorbidities, the people without SDs with comorbidities exhibited a 2.09-fold risk of developing VTE (95% CI 1.62–2.70) and the highest risk was for those with both SDs and comorbidities (adjusted HR 3.37, 95% CI 2.52–4.51). Furthermore, Table 3 indicates that the risk of developing VTE in people with SDs
without comorbidities exhibited a relative decrease when they received treatment (adjusted HR 2.31, 95% CI 1.86–2.88 for people
receiving zolpidem treatment and adjusted HR 2.16, 95% CI 1.46–
3.20 for those receiving benzodiazepine treatment).
3.4. Probability free of VTE for people with or without sleep disorders during follow-up
Fig. 1 shows the Kaplan–Meier curve of VTE-free survival for those
with SDs and those without SDs. The results indicate that the VTEfree rate was significantly lower for those with SDs than for the non- SD cohort (log-rank p < 0.0001). 4. Discussion
Previous studies have reported that 25% of adults in Taiwan experience insomnia and that the prevalence of insomnia increases
as age increases [3,21], which is consistent with studies conducted in other countries [2,22,23]. Several cohort studies have
indicated that insomnia symptoms are associated with an increased risk of cardiovascular diseases [24–27]. However, these
studies have not objectively assessed sleep apnea and, thus, the contribution of non-apnea SD to the increased risk of cardiovascular
diseases could not be controlled for. Recent studies have determined that non-apnea SD increases the risk of subsequent acute
coronary syndrome and ischemic stroke development [4,24,28].
The relationship between SDs and VTE has yet to be investigated.
To the best of the authors’ knowledge, this is the first study
to investigate whether non-apnea SD increases the risk of developing VTE, by applying a longitudinal population-based cohort study in an Asian population. The present study determined that people with non-apnea SDs exhibit a 47% increased risk of VTE development, compared with the general population, after adjusting for age, gender, and comorbidities.
The mechanisms underlying the epidemiological association
between non-apnea SD and VTE remain unclear. One possible theory is that non-apnea SD may contribute to systemic inflammation and vascular vulnerability. Numerous studies have reported that SDs are associated with inflammatory responses [29–31]. Inflammation may initiate clotting, decrease the activity of the natural anticoagulant mechanisms, and impair the fibrinolytic system [32]. In addition, inflammation causes widespread endothelial dysfunction [33].
When the interactions between inflammation-coagulation and inflammation-endothelial dysfunction overwhelm the natural
defense systems, catastrophic eventsmay occur, such as those manifested in VTE.
The people with non-apnea SDs exhibited a higher prevalence
of comorbidities and coexistent conditions associated with the development of VTE than those in the comparison cohort. Nonapnea
SD remained an independent risk factor for developing VTE after covariates were controlled for.
Most of the people in the non-apnea SD cohort in the present
study were women. Numerous studies have observed a female preponderance of insomnia [34]. After adjusting for age and other
comorbidities in the present study, it was observed that women with
non-apnea SDs exhibited a 1.76-fold increased risk of developing VTE compared with those without SDs. In addition, the incidence rate of VTE increased as age increased for both cohorts. However, after gender and comorbidities were adjusted for, the risk of VTE was the highest for young adults among the people with nonapnea SD, suggesting that SDs exert the highest effect on the health of the young adults.
The people with non-apnea SDs exhibited significantly higher risks of subsequent VTE development compared with the controls, regardless of the existence of comorbidities. Moreover, the
people with non-apnea SDs and any comorbidity exhibited multiplicative risks of developing VTE, compared with the controls
without any comorbidity.
A significant difference was observed in the VTE occurrence between the non-apnea SD cohort and comparison cohort during follow-up. Maintaining an adequate sleep duration and quality by practicing optimal sleep habits and treating SDs may reduce the risk of VTE and increase wellness.
Several limitations must be considered when interpreting these findings.
First, the NHIRD does not provide detailed lifestyle information, such as body mass index, physical activity levels, and family history, all of which are potential confounding factors in the present study.
However, these factors may be randomly distributed in these two large cohorts. Second, the lack of data on objective sleep measurements or other mental health conditions that are highly comorbid with SDs, and on medication history, may be a critical limitation. Finally, the lack of drug-treatment data, such as those on hormone-replacement therapy and the use of contraceptives and anticoagulant drugs, may have influenced the primary outcomes of the present study.
The present study is the first to provide epidemiologic data with which to address the association between non-apnea SD and VTE. The
strength of the present study is that it provides a nationwide, populationbased- cohort longitudinal study for the Asian population with which
to investigate the relationship between non-apnea SD and the increased risk of subsequent VTE events. Moreover, the study population
was obtained from physicians’ diagnoses. The findings of the present study may benefit from additional analyses in future studies regarding which specific SDs contribute to VTE incidence.
In summary, this study was determined that people with nonapnea SDs are at a higher risk of developing VTE. Because the number
of people with non-apnea SDs is progressively increasing, the enhancement of sleep-disorder management may be important for
decreasing VTE events.
