Poor sleep has been associated with fatigue, depression, accidents and injuries, suicide, post-traumatic stress disorder, increased disability and mortality rates, decreased quality of life, altered immune function, higher medical care costs, and the development of several major chronic diseases including heart disease, stroke, diabetes, and cancer. Few studies, if any, have examined relationships between diagnosed sleep disorders and cancer incidence in Veterans.
This study utilized pre-existing data from the VA's Informatics and Computing Infrastructure (VINCI) to address these specific aims: 1). Characterize trends in sleep disorder prevalence among US Veterans, and; 2). Conduct a retrospective cohort study to test the hypothesis that sleep disorder diagnoses are associated with an increased risk for cancer of the prostate, breast, colorectum, or total cancer.
For Aim 1, de-identified data for US Veterans seeking medical care (N=9,798,034, FY2000-FY2010) were accessed. Cases were defined as patients with at least two outpatient diagnoses at least 30 days apart for: sleep disturbances (ICD-9 codes: 780.50-59), nonorganic sleep disorders (307.40-49), organic insomnia (327.00-09), organic hypersomnia (327.10-19), organic sleep apnea (327.20-29), or circadian rhythm sleep disorders (327.30-39). Annual prevalence rates were summarized by: age, gender, race, income, combat exposure, service period, health factors (body mass index [BMI], tobacco use), and comorbid disease (post-traumatic stress disorder or PTSD, cardiovascular disease, stroke, diabetes, hypertension, depression). Some trends were characterized via further stratification of the above categories. For Aim 2, de-identified VINCI data for Veterans in the Southeast Network (VISN-7) were analyzed (N=663,869, 1999-2010). Sleep disorder cases were defined as described above, and primary sleep disorder diagnoses concurrent with, or subsequent to a cancer diagnosis were excluded from the analysis. Cancer cases were defined as patients with a primary tumor (any stage) identified in the VA Tumor Registry, excluding benign or in situ tumors. The relationship between prior sleep disorder diagnosis and subsequent cancer diagnosis was summarized as the hazard ratio (with 95% confidence intervals) using time-dependent Cox proportional hazards regression analyses, after adjustment for the effects of age, gender, state of residence, and marital status. Analyses were performed among all sleep disorder patients, and separately among subgroups of patients with sleep apnea, insomnia, or other diagnoses. To evaluate the role of sleep disorder severity on cancer risk, Veteran patients were grouped according to the total number of prescriptions, surgeries, and clinical procedures they received in relation to a sleep disorder (treatment index).
For Aim 1, a five-fold increase in the prevalence of all sleep disorder diagnoses was observed among Veteran patients during the 10-year study period (FY2000: ~1.0%, FY2010: 4.9%). Overall, apneas and insomnias were the first (59%) and second (32%) most common sleep disorders diagnosed, respectively. Apnea prevalence among Veteran patients increased from 0.5% in FY2000 to 3.2% in FY2010, while insomnia increased from 0.3% in FY2000 to 1.2% in FY2010. Each year, male Veterans were more likely to be diagnosed with any sleep disorder or with apnea, whereas female Veterans were more likely to receive an insomnia diagnosis. Starting in FY2000, Veterans diagnosed with depression were 3.5 times more likely to develop apnea and 3.8 times more likely to develop insomnia compared to those without depression. These differences persisted across the study period. In FY2000, Veterans with a PTSD diagnosis were 2.9 times more likely to receive an apnea diagnosis and 3.6 times more likely to be diagnosed with insomnia relative to those without PTSD. This difference persisted for sleep apnea diagnoses in FY2010, and an even larger disparity between PTSD and non-PTSD Veterans was observed for insomnia diagnoses in FY2010; Veterans with PTSD were 4.6 times more likely to be diagnosed with insomnia compared to non-PTSD patients. Veterans with combat experience were 30% more likely to have a diagnosis of either apnea or insomnia than those without combat experience. Persian Gulf Veterans experienced the largest increase in apnea prevalence over time; rates were 8.4 times higher in FY2010 compared to those in FY2000. Veterans who participated in Operation Enduring Freedom or Iraqi Freedom (OEF/OIF) experienced the largest increases in insomnia prevalence; a 30-fold increase between FY2002 (0.08%) and FY2010 (2.4%).
For Aim 2, sleep disorder diagnoses were identified among 56,055 (8%) of the eligible Veteran patients seeking care within VISN-7 during the study period. Sleep apnea (46%) and insomnia (40%) were the most common diagnoses among patients with sleep disorders. There were 18,138 cancer diagnoses in the study population (42% prostate, 12% colorectal, 1% female breast, 46% other). A complex relationship between sleep disorders and cancer incidence among Veteran patients was observed during the 11-year follow-up period, with an apparent reduction in the hazard ratio for cancer among sleep disordered patients, generally within eight or nine years of their sleep disorder diagnosis, and a ~10%-40% increase in the cancer hazard ratio after ten or more years of follow-up among patients with sleep disorders. This pattern was observed among those with any sleep disorder when examining risks for total cancer, but was inconsistent among strata of sleep disorder or cancer subtype. Increased cancer risks after a ~10-year latency tended to be most common among those with sleep apnea and a colorectal cancer diagnosis. Analyses examining potential dose-response using the treatment index as a surrogate measure of sleep disorder severity suggested an increase in cancer risk with increasing sleep disorder treatment, and possible effect modification by race or co-morbid disease.
Consistent with our preliminary data, the Aim 1 investigation identified a striking increase in sleep disorder prevalence among US Veterans over a 10-year span. The greatest increases occurred among Veterans with co-morbid disease (PTSD, depression), and among those participating in the Persian Gulf and OEF/OIF conflicts. These trends emphasize the need to better understand their potential impact on cancer or other chronic disease risks among Veterans.
None at this time.
Aging, Older Veterans' Health and Care, Health Systems
Epidemiology, Research Infrastructure
Risk Adjustment, Risk Factors, Statistical Methods