Relative risk of functional dyspepsia in patients with sleep disturbance: a population-based cohort study

Increased prevalence of sleep disorders has been found in patients with functional dyspepsia; however, direction of causality remains unclear. Our aim was to compare the risk of incident functional dyspepsia between patients with and without sleep disturbance from a large population-based sample. Utilizing a nation-wide health insurance administrative dataset, we assembled an 11-year historic cohort study to compare subsequent incidence of diagnosed functional dyspepsia between adult patients with any diagnosis of sleep disturbance and age- and gender-matched controls. Hazard ratios adjusted for other relevant comorbidities and medications were calculated using Cox regression models. 45,310 patients with sleep disorder and 90,620 controls were compared. Patients with sleep apnea had a 3.3-fold (95% confidence interval: 2.82 ~ 3.89) increased hazard of functional dyspepsia compared with controls. This increased risk persisted regardless of previously diagnosed depression coexisted. Sleep disturbance was associated with an increased risk of subsequent functional dyspepsia. Potential mechanisms are discussed.

www.nature.com/scientificreports/ was significantly higher in both the apnea and non-apnea subgroups compared to the control cohort (18.9, 210, and 6.73 instances per 1000 person-years, respectively (Table 2)).

Risks of dyspepsia in patients with or without sleep disturbances stratified by comorbidities.
Hazard ratios were significantly elevated for both sleep disorder groups within almost all strata (age, gender, most comorbidities, and with or without sleeping pills) after adjustment (Table 2). Only patients with sleep apnea comorbid with depression, obesity, heart failure, and no benzodiazepine (BZD) use did not have significantly elevated risks of dyspepsia.

Risks of dyspepsia by subtypes of sleep disorders.
Significantly higher risks of dyspepsia were also found for all subtypes of non-apnea disturbance compared to controls (Table 3). Our sensitivity analysis, excluding incident dyspepsia within the 12 months after the index date, supported the robustness of our primary finding.
Further analyses on risks of dyspepsia categorized by sleep disturbances and comorbidities. Patients Table 4). Patients that had both sleep disturbance and obesity, heart failure, or no BZD use had higher risks of functional dyspepsia than those with comorbidities but without sleep disturbance (Table 4).

Discussion
To our knowledge, this is the first study to investigate the risk of functional dyspepsia associated with sleep disturbances, taking advantage of a population-based dataset derived from a universal health insurance system. We found that risks of functional dyspepsia were significantly higher in patients with sleep disturbances, and that this applied to both the apnea and non-apnea subgroups. These associations remained significantly elevated, following a variety of adjustments and stratifications for demographic factors, comorbidities, and medication use. Further analysis showed that this risk was most elevated in the subgroup of patients with both sleep and depressive disorders. The associations of interest are consistent with previous cross-sectional and case control studies reporting disrupted or impaired quality of sleep among patients with dyspepsia 11,14,[19][20][21] . However, some studies have suggested stronger associations of dyspepsia with irritable bowel syndrome rather than functional dyspepsia [16][17][18] which we did not specifically investigate and would need further research. Our findings help to infer the direction of causation through the prospective design, although the opposite direction of causation (dyspepsia causing sleep disturbance) might also co-explain the cross-sectional relationship and would again need further specific investigation. It is possible that some cases of dyspepsia predate the onset of sleep disturbance despite being diagnosed later; however, this would predict an early divergence in incidence rates that was not sustained over follow-up; this is not apparent in the analysis of cumulative incidence displayed in Fig. 1. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Causal pathways between sleep disturbance and functional dyspepsia remain unclear, but plausible mechanisms include the chronic effects of sleep disturbance on gastrointestinal functions, including hypersensitivity to physiological stimuli, dysfunctions in autonomic nervous regulation, and/or alterations in neuroendocrine responses 20,22 . These will be considered and discussed in the following paragraphs.
First, visceral sensitivity may be altered due to disruptions of circadian rhythm in patients with sleep disturbances resulting in unpleasant abdominal sensations [11][12][13]16,23 . For example, Zhen et al. described bowel disturbances being significantly higher in rotating shift nurses compared to regular day nurses and that bowel symptom scores correlated with sleep disturbances positively 24 . Previous research has also shown that disruption of Stage 4 or slow-wave (deep) sleep, or the loss of rapid eye movement (REM) sleep, results in hyperalgesia or emergence of abdominal symptoms, such as nausea or cramping in healthy volunteers 15,[25][26][27] , and in patients with functional dyspepsia 12 or gastroesophageal reflux disease 28 . In addition, poor sleep quality (i.e. increased awakenings or altered sleep stages) has been found to be associated with gastrointestinal symptoms the next day among women with irritable bowel syndrome 29,30 . Other clinical studies have found that melatonin is helpful in reducing pain symptoms of irritable bowel syndrome without affecting colonic motility, and have suggested that this might be due to the restoration of circadian rhythms 16,31 . Although the focus has been more on irritable bowel syndrome than dyspepsia, it is plausible that adverse effects of poor sleep may influence both upper and lower gastrointestinal tracts 10 and that routine clinical diagnoses obtained from administrative databases may overlap across the two conditions. Second, gastric motility might be affected by autonomic dysregulation due to reduced sleep efficiency or increased proportions of REM sleep and intrusions during non-REM sleep 10,18,20,32 . These changes may indicate more nocturnal autonomic arousals 20,33 with similar autonomic effects to waking 20 , as evidenced by increased sympathetic activity 20 , elevated cortisol and adrenocorticotropic hormone, or diminished vagal output 33 , leading to decreased migrating motor complexes in the duodenum 19 . However, since previous research has only described self-reported gastrointestinal discomfort after poor sleep on the preceding night 10,19,20,29 , direct evidence is lacking on whether long-term dysregulation of the autonomic nervous system during sleep might induce visceral motor abnormalities leading to dyspepsia 34 .
Third, sleep disturbance has been found to be associated with alterations in brain regions such as the raphe nuclei, implicated in the release of serotonin, and regulation of circadian rhythm 35 , or the locus coeruleus, implicated in noradrenergic mediated physiological responses to stress and arousal 20,36 . These brain regions also receive afferent neuroendocrine signals from the enteric nervous system mediated by ascending serotonergic, cholinergic, or noradrenergic pathyways 20 , as well as having descending projections that terminate in the dorsal horn to regulate pain sensations. As a neurotransmitter in descending myenteric interneurons, serotonin is responsible for enhancing acetylcholine and calcitonin gene-related peptide release in prokinetic pathways 37 . Interference with such pathways might block the initiation of reflexes within the enteric nervous system and affect gut motility. However, over-stimulation of such pathways would also cause hypersensitivity and bowel discomfort such as nausea, pain, or bloating. Table 4. Further analyses on risks of dyspepsia categorized by with or without sleep disturbances and comorbidities. Rate # , incidence rate, per 1000 person-years; Adjusted HR † : mutually adjusted for age, gender, and co-morbidities of diabetes, hypertension, hyperlipidemia, anxiety, peptic ulcer disease, gastroesophageal reflux disease, stroke, ischemic heart disease, obesity, irritable bowel syndrome , cholecystitis , cholelithiasis and other disorders of biliary tract, alcoholism , heart failure , atrial fibrillation , chronic kidney disease , chronic obstructive pulmonary disease, asthma, fatigue and medication of benzodiazepine, zolpidem, aspirin, NSAIDs, steroids, proton pump inhibitors, and H2-receptor antagonistsin Cox proportional hazard regression. * P < 0.05, ** P < 0.01, *** P < 0.001. § p for interaction. www.nature.com/scientificreports/ Previous literature on dyspepsia has suggested that comorbidities from other organ systems and possible underlying psychological problems, such as social stressors or depression should be considered as etiological factors 14,21,38,39 . Our findings that the risk of dyspepsia was significantly higher in the sleep disorder groups within almost all comorbidities indicated that, in patients with comorbidities from multiple systems, having sleep disturbance is still key to the risk of their dyspepsia. Our findings of significant associations with sleep disturbance in both people with and without previous diagnoses of depression support the importance of including sleep disturbance alone in considering the etiology of dyspepsia, in addition to the other stressors. The risk of dyspepsia was not as significantly elevated in patients without sleep disturbance, and with only obesity, heart failure, or no BZD use implying that the risk of dyspepsia was influenced much by sleep disturbance than the other conditions. Although further elucidation of underlying mechanisms is still needed, we believe the chronic direct/indirect gastrointestinal effects of sleep disturbance may be implicated, including hypersensitivity to physiological stimuli, and/or dysfunctions in autonomic nervous regulation, and/or alterations in neuroendocrine responses.
Major strengths of this study included the use of a population-based dataset large enough to detect the association of interest, as well as the cohort design with a long observation period, and exposure/outcome measures derived from recorded clinical diagnoses rather than screening instruments. However, there are key limitations which need to be considered in interpreting the findings. First, measurement of comorbid physical disorders was limited to those for which ICD-9-CM codes had been listed, since these diagnoses and records were obtained from a database set up for insurance claims rather than for research purposes (although it is important to bear in mind that recorded diagnoses are required to justify treatments received). The diagnoses were clinician-assigned and might not generalize to research diagnoses. We therefore deliberately chose relatively broad and unambiguous diagnostic groups. Second, although we had access to abundant data on medical service utilizations, information was not present on certain other risk factors associated with dyspepsia, such as dietary habits, smoking 40 , body mass index 10,41 , and psychosocial stressors 39 . There were therefore still some residual confounders and some potential causal pathways that we were not able to explore. Third, the accuracy of the recorded diagnosis of functional dyspepsia needs to be considered, particularly when minor gastrointestinal sensations might be amplified by comorbid mental health conditions. Nevertheless, as we mentioned earlier, all insurance claims from the NHI are scrutinized on a regular basis according to the standardized diagnostic criteria in this study. Although our current study has compensating advantages of naturalistic observations from clinical environment, with regards to the above deficiencies, further well-designed long-term cohorts with diagnoses established by diagnostic interviews or laboratory examinations may help overcome these limitations.

Conclusions
Using a large, population-based dataset, we found a raised risk of functional dyspepsia in patients with sleep disturbance, both for apnea and non-apnea subgroups. Underlying mechanisms still need further investigation; however, our findings do indicate that sleep disturbance should be identified more systematically and clinical advice/management considered to reduce the risk of functional gastrointestinal disorders.  42 . The NHIRD contains and encrypts all healthcare claim data, including diagnostic codes, medical institutions, outpatient and inpatient orders, prescriptions, and expenditures for healthcare from Taiwan's National Health Insurance (NHI) program launched in 1995: a single-payer insurance program with near-universal coverage of more than 99% of Taiwan's residents 42 . Personal information in NHIRD cannot be identified after encryption, and use of the NHIRD must be authorized and audited by the Taiwan Bureau of NHI. Diagnoses are coded by physicians according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), and are reviewed by the NHI Review Committee every three months to audit the appropriateness of diagnoses and treatments claimed by clinicians. Related to our study, the accuracy and validity of gastrointestinal or ulcer-related diagnoses from the NHIRD have been found to be acceptable when comparing to diagnoses from the medical charts 43 . The study described here was approved by the Institutional Review Board of China Medical University in central Taiwan (CMUH-104-REC2-115). All methods were carried out in accordance with relevant regulations and guidelines. Due to the anonymity of the NHIRD, informed consent is not required from patients providing data.

Methods
Sample ascertainment. For  www.nature.com/scientificreports/ applying the same exclusion criteria. All exposed and control patients were followed from the index date until the date of receiving a diagnosis of functional dyspepsia, withdrawal from the NHI program, or the end of 2011.
Outcome and covariates. The primary outcome of our study was dyspepsia (536.8) incorporated from inpatient and outpatient records. Such diagnosis of functional dyspepsia was made by clinicians according to the Rome III and IV criteria: i.e., defining dyspepsia as any combination of four symptoms, including postprandial fullness, early satiety, epigastric pain, and epigastric burning severe enough to affect daily activities and lasted at least 3 days per week over 3 months; with an onset of 6 months in advance. . For inclusion eligibility (as independent or dependent variables or covariates), any disease diagnosis needed to appear in at least three outpatient visits or one hospitalization. Medication that might be associated with both sleep and dyspepsia was also extracted, namely: benzodiazepines (BZD), zolpidem, aspirin, non-steroidal anti-inflammatory drugs (NSAIDs) 45 , steroids, proton pump inhibitors, and h2-receptor antagonists.
Statistical analysis. Demographic and clinical characteristics of patients with sleep disturbance diagnoses (apnea subgroup and non-apnea subgroup) and controls were compared using chi-squared tests for categorical variables, and student's t-tests for continuous variables. We computed the incidence rate (per 1000 personyears) of dyspepsia for each cohort. Cox proportional hazards models were used to assess the risk of subsequent dyspepsia associated with exposure status, and hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated. Multivariate models were used to adjust for age, sex, comorbidities, and medication. We also compared the cumulative incidence of dyspepsia between the exposed and control cohorts using the Kaplan-Meier method, and tested their differences by log-rank tests. Additive interactions between comorbidities and sleep disturbance as exposures were also explored. Finally, to reduce the influence of undiagnosed dyspepsia occurring prior to sleep disturbance, a sensitivity analysis excluded dyspepsia in the 12 months after index date. All statistical analyses were performed using SAS (Version 9.3; SAS Institute, Cary, NC, USA) and the significance level was set at 0.05 for two-tailed tests.