Bowel movement frequency and risks of major vascular and non-vascular diseases: a population-based cohort study among Chinese adults

Objective The application of bowel movement frequency (BMF) in primary care is limited by the lack of solid evidence about the associations of BMF with health outcomes apart from Parkinson’s disease and colorectal cancer. We examined the prospective associations of BMF with major vascular and non-vascular diseases outside the digestive system. Design Population-based prospective cohort study. Setting The China Kadoorie Biobank in which participants from 10 geographically diverse areas across China were enrolled between 2004 and 2008. Participants 487 198 participants aged 30 to 79 years without cancer, heart disease or stroke at baseline were included and followed up for a median of 10 years. The usual BMF was self-reported once at baseline. Primary and secondary outcome measures Incident events of predefined major vascular and non-vascular diseases. Results In multivariable-adjusted analyses, participants having bowel movements ‘more than once a day’ had higher risks of ischaemic heart disease (IHD), heart failure, chronic obstructive pulmonary disease, type 2 diabetes mellitus and chronic kidney disease (CKD) when compared with the reference group (‘once a day’). The respective HRs (95% CIs) were 1.12 (1.09 to 1.16), 1.33 (1.22 to 1.46), 1.28 (1.22 to 1.36), 1.20 (1.15 to 1.26) and 1.15 (1.07 to 1.24). The lowest BMF (‘less than three times a week’) was also associated with higher risks of IHD, major coronary events, ischaemic stroke and CKD. The respective HRs were 1.07 (1.02 to 1.12), 1.22 (1.10 to 1.36), 1.11 (1.05 to 1.16) and 1.20 (1.07 to 1.35). Conclusion BMF was associated with future risks of multiple vascular and non-vascular diseases. The integration of BMF assessment and health counselling into primary care should be considered.

 Residual confounding may still exist, such as total energy intake and laxative use.

What is already known on this subject?
 Gastrointestinal symptomatology and gut microbiota have been related to human systemic diseases. However, solid epidemiological evidence is still lacking.
 There is no published prospective study that has examined the association between BMF and chronic diseases outside the digestive system other than Parkinson's disease and CVDs. The findings from four prospective studies conducted in American women and Japanese that had examined the association between BMF and CVDs were inconclusive.

What this study adds?
 BMF was associated with the risk of multiple vascular and nonvascular diseases outside the digestive system independent of traditional lifestyle and intermediate risk factors for common non-communicable diseases.
 Easily assessed BMF may be helpful to identify a change from usual pattern to abnormal bowel habits, one of the health signals that deserve attention.

INTRODUCTION
Properly functioning gastrointestinal tract plays an essential role in health. Diseases of the digestion system, such as cholelithiasis, non-alcoholic liver disease, and ulcerative colitis, have been prospectively associated with cardiovascular diseases (CVDs), [1][2][3] type 2 diabetes mellitus (T2DM), 4 and chronic kidney disease (CKD), 5 Also, recent advancements in gut microbiota have enhanced our understanding of the potential relationship of their composition, diversity, and metabolites to diseases. [6][7][8][9][10] Although gut transit time is a good indicator of gastrointestinal function, measuring it in a large epidemiology survey is quite difficult. 11 The frequency of bowel movement has been correlated with colonic transit time and may represent a simple quantifiable indicator of adequate colonic function. 12 Alternatively, bowel movement frequency (BMF) may serve as a reflection of lifestyle factors (e.g., diet and exercise) 13 14 and individual characteristics (e.g., gut microbiota). 15 16 A systematic review and meta-analysis that included four cohort studies and five casecontrol studies confirmed the association between constipation and subsequent diagnosis of Parkinson's disease. 17 For the association between BMF and CVDs, only four prospective studies were conducted in American women and Japanese, but findings remained inconclusive. 18-21 To our knowledge, there is no published prospective study that has examined the association between BMF and chronic diseases outside the digestive system other than Parkinson's disease and CVDs.
In the present study, we prospectively examined the associations of BMF with multiple vascular and nonvascular diseases outside the digestive system in the China Kadoorie Biobank (CKB) study of 0.5 million adults.

Study population
CKB is an ongoing large prospective cohort study of over 0.5 million adults from 10 geographically diverse sites across China. The baseline survey was conducted during In the current study, we firstly excluded participants with previously diagnosed cancer (n=2578), heart disease (n=15 472) or stroke (n=8884) at baseline, as well as those lost to follow-up soon after baseline (n=1) and those with missing body mass index (BMI; n=2), leaving 487 198 eligible participants. Some participants with disease of interest at baseline were further excluded when conducting analyses of particular disease outcome: (i) excluding participants who had a self-reported history of emphysema, bronchitis or pulmonary heart disease (n=13 288), who was spirometry-measured to have airflow obstruction (AFO) 24 at baseline (n=23 767) for the analysis of chronic obstructive pulmonary disease (COPD); (ii) excluding participants who had a self-reported history of diabetes (n=16 162) or screen-detected diabetes at baseline (n=14 138) for the analysis of T2DM; (iii) excluding participants who had a self-reported history of kidney disease (n=7575) for the analysis of CKD.

Patient and Public Involvement
There was no patient nor public involvement in this study.

Assessment of exposure
At the baseline survey, trained staff asked study participants about BMF: "About how often do you have bowel movements each week?" The response options were: more than once a day, once a day, once every 2-3 days, or less than 3 times a week. After completion of the baseline survey in July 2008, we randomly selected about 5% of the surviving participants for resurvey in the same year. To test the reproducibility of the BMF question, we included 1300 participants who completed the same questionnaire twice at an interval of few than 1.5 years (median 1.4 years). The linearly and quadratically weighted kappa coefficients 25 between the two questions were 0.40 and

Assessment of covariates
The baseline questionnaire collected covariate information: sociodemographic characteristics, lifestyle behavior, and personal and family medical history. Daily level of physical activity was calculated by multiplying the metabolic equivalent tasks (METs) value for a particular type of physical activity by hours spent on that activity per day and summing the MET-hours for all activities. 26 Height, body weight, and waist circumference (WC) were measured with calibrated instruments. BMI was calculated as weight in kilograms divided by height in meters squared. Prevalent diabetes at baseline was defined as self-reported diabetes diagnosed by a physician before the baseline survey or screen-detected diabetes, which was defined as 27 : random blood glucose≥7.0 mmol/L and fasting time≥8h; random blood glucose≥11.1 mmol/L and fasting time<8h; or fasting blood glucose≥ 7.0 mmol/L on subsequent testing. Forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC) were measured using a hand held Micro (MS01) Spirometer (CareFusion Corp). Pre-bronchodilator FEV1/FVC<0.7 was considered as screen-detected AFO 24 .

Assessment of outcomes
Information on disease incidence was identified by linking to local disease and death registries and national health insurance (HI) system, and by active follow-up. Until the end of 2016, nearly 97% of surviving participants had a successful linkage with HI data; the proportion was similar across 10 survey sites. Trained staff blinded to baseline information coded all diagnoses using the International Classification of Diseases, Tenth Revision (ICD-10).

Statistical analysis
We compared baseline characteristics across different BMF categories by using analysis of covariance for continuous variables and logistic regression for categorical variables, adjusting for age (years), sex and 10 survey sites as appropriate. We calculated person-years at risk from baseline until the diagnosis of the study outcome of interest, death, loss to follow-up, or 31 December 2016, whichever came first.
We used stratified Cox proportional hazards regression models to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs), with age as the underlying time scale and stratified by 5-year groups and 10 survey sites. The proportional hazards assumption for the Cox model was checked by log-log plots; no apparent violation was found. The covariates adjusted in the multivariable models were determined according to previous literatures and biological relevance and included sex (for the whole cohort analysis); level of education; occupation; household income; marital status; family history of certain diseases (for corresponding disease analysis only); smoking status; total physical activity level; alcohol consumption; intake frequency of fresh vegetables, fresh fruit, and red meat; BMI; waist circumference; and prevalent hypertension and diabetes at baseline (no adjustment for T2DM). We tested the linear trend of diseases risk across various metrics of BMF by modeling the levels of ordered categorical variables as a continuous variable in a separate model.
In the sensitivity analyses, we excluded cases which occurred during the first two years to reduce the possibility of reverse causation. We also further adjusted for the following factors one by one as potential confounders: intake frequency of rice, wheat, other staple food, and spicy food; tea consumption frequency; consumption of nutritional supplements; medications for cardiovascular diseases; medical history of peptic ulcer, cirrhosis or chronic hepatitis, and gallstone or gallbladder disease at baseline. To explore whether the associations between BMF and study outcomes were consistent across several baseline characteristics, we stratified the analyses by following variables: sex (men or women); age (<60 yr, ≥60 yr); obesity status defined jointly by BMI and WC (presence: BMI≥28.0 kg/m 2 or WC≥90.0cm (men)/≥85.0cm (women); absence); hypertension (presence or absence). The definition of obesity status was used because of the unique significance of abdominal obesity for the rapid growth of health risk in Asian adults. The tests for multiplicative interaction were performed using likelihood ratio tests by comparing models with and without crossproduct terms between the stratifying variable and BMF categories. P values are presented as unadjusted for multiple testing unless otherwise indicated. For testing of multiple primary outcomes, a Bonferroni-correction was applied to the significance level, α = 0.05 (two-sided), that divided 0.05 by the number of outcomes examined (5 vascular outcomes and 3 nonvascular outcomes).
Analyses were conducted with Stata 15.0.

Baseline characteristics of study participants
Distributions of sociodemographic and lifestyle factors and health status at baseline across four categories of BMF are displayed in Table 1. Of the 487 198 participants analyzed, the mean age was 51.5 years; 59.1% were women, and 43.1% resided in urban areas. Participants who defecated more frequently were more likely to be men and rural residents, have higher BMI and WC on average, and more likely to be hypertensive. Among men, consuming alcohol regularly was also related to frequent bowel movements.  stroke, 4204 heart failure, 11 054 COPD, 15 281 T2DM, and 6526 CKD. In the whole cohort, after multivariable adjustment, participants who reported having bowel movements "more than once a day" were associated with higher risks of IHD, heart failure, COPD, T2DM, and CKD when compared with the reference group ("once a day"). The corresponding HRs (95% CIs) were 1.12 (1.09 to 1.16), 1 Table 2). At the same time, the lowest BMF ("less than 3 times a week") was associated with increased risks of IHD, MCEs, ischaemic stroke, and CKD. The corresponding HRs (95% CIs) were 1.07 (1.02 to 1.12), 1.22 (1.10 to 1.36), 1.11 (1.05 to 1.16), and 1.20 (1.07 to 1.35), respectively (all P values < 0.00625). We observed linear trends between BMF and risk of certain outcomes, with a positive association for heart failure, COPD, and T2DM, and a negative association for MCEs and ischaemic stroke (all P values for trend <0.00625) ( Table 2). Results of stepwise adjusted models for each disease outcomes are presented in Supplementary Table 1. The association between BMF and T2DM was obviously attenuated after additional adjustment for BMI and waist circumference.

Sensitive analyses
In the sensitivity analyses, the associations of BMF with multiple disease outcomes did not change appreciably after excluding cases occurred in the first two years. The HRs (95% CIs) almost remained unchanged after additional adjustment for other potential confounders (data not shown).

Subgroup analyses
We found that BMF had a statistically significant multiplicative interaction with the obesity status for MCEs (P for interaction = 0.001). Compared to participants with BMF of once every 1-3 days, the increased risk of MCEs associated with lower BMF was only observed among obese participants ( Table 3). There was no heterogeneity between men and women in the association between BMF and most disease outcomes except ischaemic stroke (P for interaction = 0.006) (Supplementary Table 2). The associations between BMF and all study outcomes seemed to be consistent across  Table 3 and   Supplementary Table 4).

DISCUSSION
In this large prospective cohort study of Chinese adults, BMF was associated with the risk of multiple vascular and nonvascular diseases outside the digestive system independent of traditional lifestyle and intermediate risk factors for common noncommunicable diseases. Compared with BMF of once a day, constipation was associated with an increased risk of MCEs and ischaemic stroke; more frequent bowel movements were associated with an increased risk of heart failure, COPD, and T2DM. The increase in the risk of IHD and CKD was observed in both participants who were constipated, or those had frequent bowel movements. BMF was not associated with the risk of hemorrhagic stroke. Our findings persisted after exclusion of cases occurred in the first two years.

Comparison with other studies and potential mechanism
Several prospective studies have examined the association of BMF with CVDs. [18][19][20][21] Women's Health Initiative observational study (WHI) followed 73 047 postmenopausal women for 6.4 years on average and had asked women to self-rate the severity of constipation at baseline. 18 In this study, women with severe constipation had a 23% higher risk of death from a composite of cardiovascular events (HR=1.23; 95% CI: 1.03-1.47) compared with women with no constipation. Separate analysis for each type of cardiovascular events was limited due to a small number of events.
Findings from the Nurse's Health Study (NHS) of 86 289 women aged 30-55 years followed-up for nearly 30 years showed that BMF was not associated with the incidence of coronary heart disease and stroke and cardiovascular mortality. 21 Compared with women with daily bowel movements, those with BMF > 1 time/day showed a modest increase in the risk of total mortality (HR=1.10; 95% CI: 1.06-1.15).
The Ohsaki cohort, 45 112 Japanese aged 40-79 years with a follow-up of 13.3 years, has shown that the risk of overall CVD mortality increased in participants reporting BMF of 1 time/2-3 days (HR=1.21; 95%: 1.08-1.35) and ≤1 time/4 days (HR=1.39; 95% CI: 1.06-1.81) compared with those reporting ≥1 time/day. 19 The respective HRs (95% CI) for death from ischaemic stroke were 1.27 (1.00-1.61) and 1.97 (1.21-3.21). The study was underpowered for analyses of IHD and other subtypes of stroke mortality and found a null association. Another similar study based on the Japan Collaborative Cohort (JACC) of 72 014 Japanese aged 40-79 years did not observe statistically significant association between BMF and death from coronary heart disease, ischaemic and haemorrhagic stroke, but was too underpowered to allow a firm conclusion. 20 However, this study indicated that laxative use, a potential reflection of severe constipation, was associated with increased risk of dying from coronary heart disease and ischaemic stroke.
In the present study, we associated constipation with increased risk of MCEs and ischaemic stroke, in line with most previous findings. With regard to the associations between BMF and other outcomes, to our knowledge, no report has yet been made in published prospective studies. Previous studies linked constipation or high frequency of bowel movements to the alterations of composition, diversity, and functions of gut microbiota. 16 28 29 Altered gut microbiota has been suggested to interact with the host through both metabolism dependent pathways and metabolism independent processes and contribute to the development and progression of diseases such as atherosclerosis, heart failure, CKD, and T2DM. 30 Similarly, the gastrointestinal tract and respiratory tract, although separate organs, are part of a shared mucosal immune system termed the gut-lung axis. 31 The microbiota and metabolites in the gut and the lungs can modulate systemic and local immunity and influence the pathogenesis of respiratory diseases. 31 Also, slow transit constipation is accompanied by reduced fermentation efficacy of complex carbohydrates, which diminishes the production of the shortchain fatty acid, an energy source for colonocytes, leading to oxidative stress. 32 Oxidative stress has also been related, with different degrees of importance, to the onset and progression of atherosclerosis, cardiovascular diseases, neurological diseases, and kidney disease. 33 Despite several biologically plausible mechanisms underlying these associations have been suggested, further research is warranted to Several established risk factors for chronic diseases, such as unhealthy diets, physical inactivity, and obesity may also affect BMF. 13 14 However, in the present study, the relative risk estimates for multiple disease outcomes, except T2DM, did not change appreciably after adjusting for these lifestyle factors. It indicates that the associations between BMF and multiple diseases were independent of traditional risk factors. Besides, most of the associations between BMF and multiple disease outcomes were consistent across baseline subgroups defined by sex, age, obesity status, and hypertension status in the present analyses. The significant increase in the risk of MCEs associated with constipation was only observed among obese participants. We assume that both obesity-and constipation-related oxidative stress act synergistically to increase the MCEs risk. It is still unclear what contributed to the sex difference in the association between constipation and ischaemic stroke.

Strengths and limitations
To our knowledge, this is the first prospective study examining the associations of BMF with a range of disease outcomes outside the digestive system collected through linkage to multiple data systems. The present study has a large sample size and longterm follow-up, and hence was well powered for analyses of cardiovascular subtypes and other outcomes. Participants included both men and women geographically spread across urban and rural areas of China. Prospective design and long-term follow-up minimize the potential for reverse causation. A detailed collection of lifestyle factors and other covariates allowed adjustment for several potential confounders.
Some limitations deserve attention. First, BMF was self-reported once at baseline, which may cause nondifferential misclassification and potentially attenuate true associations. Second, although we have controlled for several potential confounders, residual confounding may still exist. For example, we failed to adjust for total energy intake and laxative use. However, the prevalence of laxative use was as low as 5% in the Chinese population, 34   Cox regression models were stratified by 5-year groups and ten survey sites, with age as the underlying time scale.
* Combined group including "once a day" and "once every 2-3 days". † A Bonferroni-corrected threshold was used, =0.00625.   Multivariable models adjusted for same set of covariates as Table 2, except for the stratified variable of interest.

Study design 4
Present key elements of study design early in the paper 4, 5   12 Results: After adjustment for potential confounders, participants having bowel 13 movements "more than once a day" had higher risks of ischaemic heart disease, heart 14 failure, chronic obstructive pulmonary disease, type 2 diabetes mellitus, and chronic 15 kidney disease when compared with the reference group ("once a day"

11
In the sensitivity analyses, we excluded cases which occurred during the first two 12 years to reduce the possibility of reverse causation. We also further adjusted for the 13 following factors one by one as potential confounders: intake frequency of rice, 14 wheat, other staple food and spicy food; tea consumption frequency; consumption of 15 nutritional supplements (fish oil/cod liver oil, vitamins, calcium/iron/zinc, ginseng 16 and other herbal products); medications for cardiovascular diseases (aspirin, ACE-I, 17 beta-blocker, statins, diuretics and Ca 2+ antagonist); medical history of peptic ulcer, 18 cirrhosis or chronic hepatitis, and gallstone or gallbladder disease at baseline.

19
To explore whether the associations between BMF and study outcomes were 20 consistent across several baseline characteristics, we stratified the analyses by 21 following variables: sex (men or women); age (<60 yr, ≥60 yr); obesity status defined Sensitive analyses 5 In the sensitivity analyses, the associations of BMF with multiple disease outcomes 6 did not change appreciably after excluding cases occurred in the first two years. The 7 HRs (95% CIs) almost remained unchanged after additional adjustment for other 8 potential confounders (data not shown).

Subgroup analyses
10 We found that BMF had a statistically significant multiplicative interaction with the 11 obesity status for MCEs (P for interaction = 0.001). Compared to participants with 12 BMF of once every 1-3 days, the increased risk of MCEs associated with lower BMF 13 was only observed among obese participants ( Table 3). There was no heterogeneity 14 between men and women in the association between BMF and most disease outcomes 15 except ischaemic stroke (P for interaction = 0.006) (Supplementary Table 2 17 not observe statistically significant association between BMF and death from 18 coronary heart disease, ischaemic and haemorrhagic stroke, but was too 19 underpowered to allow a firm conclusion. 21 However, this study indicated that 20 laxative use, a potential reflection of severe constipation, was associated with 21 increased risk of dying from coronary heart disease and ischaemic stroke.    Cox regression models were stratified by 5-year groups and ten survey sites, with age as the underlying time scale. Multivariable models were adjusted for sex (men or women); level of education (no formal school, primary school, middle school,  Multivariable models adjusted for the same set of covariates as Table 2, except for BMI (kg/m2) and waist circumference (cm). Obesity status was defined according to BMI and WC, with "yes" as BMI≥28.0 kg/m 2 or WC≥90.0 cm (men)/≥85.0 cm (women).
* Combined group including "once a day" and "once every 2-3 days". † A Bonferroni-corrected threshold was used, =0.00625.   Multivariable models adjusted for same set of covariates as Table 2, except for the stratified variable of interest.

13
In sensitivity analyses, we excluded cases that occurred during the first two years 14 of follow-up. We also further adjusted for the following factors one by one as 15 potential confounders: intake frequency of rice, wheat, other staple food, spicy food, 16 and tea; consumption of nutritional supplements (fish oil/cod liver oil, vitamins, 17 calcium/iron/zinc, ginseng and other herbal products); medications for CVDs (aspirin, 18 ACE-I, beta-blocker, statins, diuretics and Ca 2+ antagonist); medical history of kidney 19 disease, peptic ulcer, cirrhosis or chronic hepatitis, and gallstone or gallbladder 20 disease at baseline.

Subgroup analyses
12 There was a significant interaction between BMF and obesity status on the risk of 13 MCEs (P for interaction = 0.001). Compared to participants with BMF of once every 14 1-3 days, the increased risk of MCEs associated with lower BMF was only observed 15 among obese participants ( Table 3). There was no heterogeneity by sex in the 16 association between BMF and most disease outcomes except for ischaemic stroke (P 17 for interaction = 0.006) (Supplementary Table 2 14 The study was underpowered for analyses of IHD and stroke mortality of other 15 subtypes. Another similar study based on the Japan Collaborative Cohort (JACC) of 16 72 014 Japanese aged 40-79 years did not observe significant associations of BMF 17 with deaths from coronary heart disease, ischaemic and haemorrhagic stroke, but it 18 was also underpowered to allow a firm conclusion. 20 However, this study indicated 19 that laxative use, a potential reflection of severe constipation, was associated with 20 increased risks of deaths from coronary heart disease and ischaemic stroke.
* Combined group including "once a day" and "once every 2-3 days". † A Bonferroni-corrected threshold was used, =0.00625.   Multivariable models adjusted for same set of covariates as Table 2, except for the stratified variable of interest.
* Combined group including "once a day" and "once every 2-3 days". † A Bonferroni-corrected threshold was used, =0.00625.   Multivariable models adjusted for same set of covariates as Table 2, except for the stratified variable of interest.