Smoking and alcohol consumption and risk of incident diverticulitis in women

Abstract

Background & Aims

Much of what is known about the effects of alcohol and tobacco use on diverticular disease derives from studies of asymptomatic diverticulosis or complicated diverticulitis. We examined smoking and alcohol consumption and risk of incident diverticulitis in a large cohort of women.

Methods

We conducted a prospective study of 84,232 women in the Nurses’ Health Study II (NHS II) who were 39–52 years old and without known diverticulitis at baseline in 2003. Smoking was ascertained every 2 years and alcohol use every 4 years. We used Cox proportional hazards regression to estimate multivariable-adjusted hazards ratios (HR) and 95% confidence intervals (CI).

Results

During 1,139,660 person-years of follow up, we identified 3,018 incident cases of diverticulitis. After adjustment for other risk factors, current (HR 1.20; 95% CI, 1.04–1.39) and past smoking (HR 1.20; 95% CI, 1.11–1.30) were associated with increased risk of diverticulitis when compared to never smokers. Women who consumed ≥30 g/d of alcohol had a multivariate HR of 1.26 (95% CI, 1.05–1.50) when compared to women who did not drink. A joint analysis of smoking and alcohol found that individuals who ever smoked and consumed ≥15 g/d of alcohol were at highest risk of diverticulitis (multivariate HR 1.60; 95% CI, 1.16–2.21), compared to participants who never smoked and reported no alcohol use.

Conclusions

In this large prospective study of women, smoking and alcohol consumption were associated with an increased risk of incident diverticulitis. These data highlight additional modifiable risk factors for diverticulitis that may aid in prevention.

Graphical Abstract

Introduction

Diverticulitis is one of the most common gastrointestinal disorders seen in both the inpatient and outpatient clinical settings, resulting in substantial medical morbidity and healthcare costs¹. The increasing incidence of diverticulitis in recent decades may in part be due to diet and lifestyle factors^{2, 3}. Dietary fiber and red meat, physical inactivity, and obesity are wellestablished risk factors for diverticulitis^3–6. However, less is known about the effects of other modifiable lifestyle factors on risk of diverticulitis, including alcohol consumption and tobacco use.

Prior research on the effect of smoking on risk of diverticulitis has been conflicting or has focused on severe presentations of diverticulitis. Several studies have found an association between tobacco use and increased risk of colectomy and complicated diverticulitis^7–9, and one large prospective cohort study found that smoking was associated with an increased risk of developing symptomatic diverticular disease requiring hospitalization, as well as higher risk of associated abscess or perforation¹⁰. However, other studies have found no association including two prospective studies in men^11–13.

A number of studies have investigated the effect of alcohol on the prevalence of uncomplicated diverticulosis with varying results^14–17, but little is known about alcohol consumption and risk of diverticulitis. One study found an association between alcohol use disorder and hospitalization for diverticulitis¹⁸, while others studies of alcohol consumption have found no association^{13, 19}. We therefore prospectively evaluated the effects of tobacco use and alcohol consumption on risk of incident diverticulitis using the Nurses’ Health Study II (NHS II), a large, ongoing cohort study of women²⁰.

Methods

Study population

The NHS II is a prospective study comprised of 116,429 female registered nurses aged 25–42 years at cohort inception in 1989. Participants receive questionnaires on a biennial basis, querying about demographics, lifestyle factors, and medical history with a follow-up rate exceeding 90%. Every 4 years, participants provide dietary information.

Ascertainment of diverticulitis

The primary endpoint of this study was diverticulitis. Starting in 2015, participants were asked whether they had been diagnosed with diverticulitis requiring antibiotic therapy or hospitalization within the past two years. In 2017, this question was expanded to ask participants if they had ever had a diagnosis of diverticulitis, and if yes, to designate the year of diagnosis dating back to 2004. The 2017 questionnaires were used as the primary source for determining a diagnosis of diverticulitis, and the 2015 questionnaires were used if the 2017 questionnaire was not returned by participants (Supplementary Figure 1). We reviewed the medical records of a sample of 226 women reporting a diagnosis of diverticulitis on the 2017 questionnaire and self-report was confirmed in 92% of cases. Diverticulitis was defined as a CT or pathology report of diverticulitis or a provider diagnosis with a clinical presentation consistent with diverticulitis.

Assessment of Smoking and Alcohol

We assessed smoking on biennial study questionnaires (cigarettes per day in the following categories: 0–1, 5–14, 15–24, 25–34, 35–44, 45+). We calculated pack-years of smoking by multiplying the number of packs smoked per day by the number of years of smoking reported, including previous smoking history.

Alcohol consumption was assessed every four years using a semi-quantitative food frequency questionnaire. Participants were asked to report their average intake of liquor, beer, white wine, and red wine in the previous year in one of nine categories ranging from “never or less than once per month” to “6 or more times per day.” This response was also converted into grams/day of total alcohol by multiplying the average ethanol content in each portion (e.g. 4-oz glass of wine) by the frequency of each beverage type. In a random sample of 173 Boston area participants who reported four 1-week dietary records performed every 3 months for 12 months, alcohol intake on the dietary records and the main study questionnaire were highly correlated.²¹.

Assessment of Covariates

We also assessed other important covariates including menopause status, menopausal hormone use, physical activity, aspirin use, and nonsteroidal anti-inflammatory drugs (NSAID) use. Body mass index (BMI) was calculated using participants’ reported height and weight. Dietary information was obtained every 4 years using a semi-quantitative food frequency questionnaire. Earlier studies in this cohort demonstrated the validity and reproducibility of the FFQs²², self-reported anthropometric measurements²³, and physical activity assessment²⁴.

Statistical analysis

We restricted the analysis to participants who returned the 2015 or 2017 questionnaires and began follow up with the 2003 questionnaire, after which we had data on time of diagnosis. We excluded individuals who reported diverticulitis prior to 2004, as the year of diagnosis was assessed beginning with this year, and those with missing information on smoking and alcohol consumption. Participant follow-up time accrued from the date of return of the study baseline questionnaire in 2003 to date of diagnosis of diverticulitis, death, or the end of follow-up (June 2017), whichever came first (Supplementary Figure 1). More than 99% of participants had data on alcohol and smoking.

We used Cox proportional hazard models to assess age-adjusted and multivariate hazard ratios (HR) and 95% confidence intervals (CI). Age-adjusted models included age in years and the calendar year of the questionnaire cycles in 2-year intervals. Consistent with prior studies on risk factors for diverticulitis, multivariable models adjusted for age, menopausal hormone status and hormone use, body mass index (BMI), physical activity, smoking status (when assessing the effects of alcohol use), alcohol intake (when assessing the effects of smoking), aspirin use, NSAID use, fiber intake, and red/processed meat intake^{5, 6, 25–30} . We updated smoking status, alcohol intake, and all covariates based on information from biennial questionnaires.

Smoking was categorized as never smoker, past smoker and current smoker. Participants were classified as past smokers if they reported a prior history of smoking and subsequently reported no smoking. We also analyzed pack-years of smoking (never smoking, 0.1–10.0, 10.1–20.0, > 20 pack years), and years since quitting smoking in past smokers (0–5.0, 5.1–10.0, 10.1–20.0, 20.1–30.0, > 30.1 years). Pack-years, as a measure of cumulative smoking exposure, were calculated based on smoking status on the biennial questionnaires (never smoker, past smoker, current smoker, and cigarettes smoked per day), as well as information from the prior cycle. For the analysis of alcohol intake, we created the following categories of grams of alcohol consumed per day: 0.1–4.9, 5.0–9.9, 10.0–14.9, 15.0–29.9, ≥30; non-drinkers (no intake) served as the referent group.

We utilized restricted cubic splines to test for a possible linear relationship between alcohol consumption and risk of diverticulitis stratified by smoking status (never vs ever)^{31, 32}. We also used restricted cubic splines to examine the relationship between types of alcohol consumed and diverticulitis risk. We truncated the population at the 99^th percentile of alcohol intake to reduce the influence of outliers. We also performed a joint analysis of smoking (never, past, current) and alcohol consumption (non-drinker, 0.1–14.9 g/d, ≥15 g/d).

To examine the relationship between smoking and alcohol and severity of diverticulitis, we also analyzed surgery for diverticulitis as a secondary outcome.

Statistical analyses were performed using SAS software (version 9.4, SAS Institute Inc., NC). Two-sided P values of <0.05 were considered statistically significant. The study was approved by the Institutional Review Boards of the Harvard T.H. Chan School of Public Health and Partners Healthcare.

Results

Baseline characteristics

Among the 84,232 eligible women and 1,139,660 person-years of follow-up, we identified 3,018 cases of diverticulitis. Baseline participant characteristics in 2003 are reported in Table 1. In general, women who were never smokers reported less alcohol intake and were more likely to be premenopausal. Never and past smokers reported more physical activity than current smokers and higher calorie and fiber intake. Never smokers also tended to report lower usage of aspirin and NSAIDs than past and current smokers. On average, current smokers were less likely to use multivitamins and oral contraceptives and were more likely to consume red meat (Table 1). Generally, women who reported moderate or heavy alcohol use were more likely to be current or past smokers, and to use aspirin and NSAIDs (Table 1).

Table 1.

Characteristics of participants in the Nurses’ Health Study II at baseline in 2003 according to smoking status and alcohol consumption

	Smoking			Alcohol
	Never (n=55515)	Past (n=22068)	Current (n=6649)	None (n=31233)	Moderate (n=43769)	Heavy (n=9230)
Age, years	48.0 (4.7)	49.2 (4.5)	48.7 (4.5)	48.4 (4.7)	48.3 (4.6)	48.9 (4.5)
Smoking pack-years	0 (0)	10.3 (8.4)	23.4 (13.7)	3.9 (9.3)	4.8 (9.1)	6.7 (10.1)
Current smoker, %	-	-	-	6.4	8.2	11.8
Past smoker, %	-	-	-	18.4	29	39
Premenopausal, %	68.0	65.7	58.5	65.4	67.1	67.9
Current postmenopausal hormone use, %	39.6	38.5	36.6	38.9	39.4	36.7
Body mass index, kg/m2	26.9 (6.3)	27.2 (6.4)	26.8 (5.8)	28.2 (7)	26.5 (5.9)	25.0 (4.6)
Physical activity, MET-hours/week	20.2 (25.6)	22.4 (28.6)	18.8 (26.8)	17.4 (24.2)	21.9 (27)	25.8 (30.1)
Alcohol intake, g/d	4.1 (7.5)	7.1 (10.3)	7.3 (12)	-	-	-
Total beer, serving/d	-	-	-	0 (0)	0.1 (0.2)	0.4 (0.8)
Total wine, serving/d	-	-	-	0 (0)	0.2 (0.2)	1.1 (0.9)
Liquor, serving/d	-	-	-	0 (0)	0 (0.1)	0.3 (0.6)
Calorie intake, kcal/d	1848 (577)	1850 (570)	1816 (593)	1795 (583)	1847 (570)	2018 (562)
Fiber intake, g/d	20.3 (6.2)	20.3 (6.3)	17.5 (5.7)	20.2 (6.8)	20.2 (5.9)	18.8 (5.6)
Red/processed meat intake, serving/d	0.9 (0.7)	0.9 (0.7)	1.1 (0.7)	0.9 (0.7)	0.9 (0.6)	0.9 (0.6)
Alternate Healthy Eating Index	50.3 (12.1)	52.1 (12.1)	46.3 (11.3)	48.7 (12.4)	51.2 (12.0)	53.0 (11.5)
Aspirin use, %	22.6	24.1	25.6	22.5	23.2	25.2
NSAID use, %	42.7	46.6	45.5	40.9	45.1	48.8
Multivitamin use, %	61.4	61.1	51.5	59.1	61.3	62
Oral contraceptive use, %	7.5	7.1	2.7	5.8	7.6	8.8
Hypertension, %	16.0	16.0	15.4	18.9	14.3	14.0
Diabetes, %	2.7	3	2.6	4.4	2.0	1.0
Hypercholesterolemia, %	18.9	19.1	21.1	21.7	18.2	15.1

For alcohol use none: 0 g/d; moderate; ≤1 4g/d; heavy: >15 g/d

Smoking

Both past and current smoking were associated with an increased risk of incident diverticulitis (Table 2). After adjustment for other known and potential confounders, the multivariate HR for current smoking was 1.20 (95% CI; 1.04–1.39), and for past smoking was HR 1.20 (95% CI, 1.11 – 1.30) when compared to never smoking. When smoking history was assessed in pack-years, we did not observe a dose-response relationship (Table 2). Compared to never smokers, the multivariate HR was 1.22 (95% CI, 1.11–1.34) for women with a 0.1 – 10.0 pack-year history, 1.14 (95% CI, 1.01–1.29) for 10.1–20.0 pack-years, and 1.22 (95% CI, 1.08–1.37) for >20.0 pack-years. Using current smoking as a reference, there was no difference in risk of diverticulitis when stratified by the number of years since quitting smoking. For example, the multivariate HR was 1.15 (95% CI, 0.91–1.44) for 0–5 years since quitting smoking and 0.94 (95% CI, 0.79–1.12) for more than 30 years since quitting. In the analysis restricted to participants undergoing surgery for diverticulitis, the magnitude of the association between smoking and diverticulitis was strengthened (multivariate HR 1.48; 95% CI 1.04–2.12 for current and 1.46; 95% CI 1.20–1.78 for past vs never smokers) (Supplementary Table 1).

Table 2.

Smoking status, pack-years and risk of incident diverticulitis

	Smoking status
	Never smoker	Past smoker	Current smoker
Cases	1801	1014	203
Person-years	749649	319260	70751
Age-adjusted, HR (95% CI)	1 (ref)	1.23 (1.14, 1.33)	1.27 (1.10, 1.47)
MV-adjusted, HR (95% CI)	1 (ref)	1.20 (1.11, 1.30)	1.20 (1.04, 1.39)

	Smoking pack-years
	Never smoker	0.1–10.0	10.1–20.0	>20.0
Cases	1818	570	297	333
Person-years	754960	192318	98596	93786
Age-adjusted, HR (95% CI)	1 (ref)	1.21 (1.11, 1.33)	1.19 (1.06, 1.35)	1.33 (1.18, 1.49)
MV-adjusted, HR (95% CI)	1 (ref)	1.22 (1.11, 1.34)	1.14 (1.01, 1.29)	1.22 (1.08, 1.37)

MV (multivariable) model was adjusted for age, menopausal hormone status and hormone use, body mass index, alcohol intake, physical activity, aspirin use, NSAID use, fiber intake, and red/processed meat intake. Reference case numbers vary due to a small number of participants who provided smoking status but not pack-years.

Alcohol

We also observed a positive association between alcohol consumption (g/d) and risk of diverticulitis that was largely driven by women with the highest consumption. After controlling for other potential risk factors, women who consumed 30 or more g/d of alcohol (more than 2 drinks/d) had an HR of 1.26 (95% CI, 1.05–1.50; P trend .04) when compared to individuals who did not consume alcohol (Table 3). Cubic splines analysis suggested that the relationship was stronger among never smokers than ever smokers (P for linearity=0.004 and P for linearity=.15, respectively) (Figure 1). In a cubic splines analysis of types of alcohol consumed (servings/d), liquor and white wine, but not red wine or beer, demonstrated a positive association with risk of diverticulitis (P for linearity<.001, P for linearity=.04, P for linearity=.5, and P for linearity=.49, respectively; P values for non-linearity all >.05; Figure 2). After limiting the analysis to individuals undergoing surgery for diverticulitis, the association between alcohol and diverticulitis was no longer significant (multivariate HR 1.01; 95% CI, 0.62–1.67 for individuals consuming at least 30 g/d compared to non-drinkers) (Supplementary Table 2).

Table 3.

Alcohol consumption and risk of incident diverticulitis

	Alcohol consumption (g/d)						P-trend
	0	0.1–4.9	5.0–9.9	10.0–14.9	15.0–29.9	≥30.0
Cases	1049	955	383	229	259	143
Person-years	410115	353459	150526	91877	92750	40933
Age-adjusted model	1 (ref)	1.11 (1.02, 1.21)	1.00 (0.89, 1.13)	0.95 (0.82, 1.10)	1.04 (0.90, 1.19)	1.25 (1.05, 1.48)	.17
MV-adjusted model	1 (ref)	1.12 (1.02, 1.22)	1.06 (0.94, 1.19)	1.02 (0.89, 1.19)	1.11 (0.97, 1.28)	1.26 (1.05, 1.50)	.04

MV model was adjusted for age, menopausal hormone status and hormone use, body mass index, smoking status, physical activity, aspirin use, NSAID use, fiber intake, and red/processed meat intake.

Figure 1.

Restricted cubic spline analysis of alcohol consumption and risk of incident diverticulitis in all participants (A), never smokers (B) and ever smokers (C)

Figure 2.

Restricted cubic spline analysis of subtypes of alcohol consumption and risk of incident diverticulitis

We examined the joint association of smoking and alcohol consumption on risk of diverticulitis (Figure 3). The risk increased both with increasing alcohol and increasing smoking. Among non-smokers, the HR associated with high alcohol was 1.24, and among non-drinkers, the HR associated with smoking was 1.22. If one had both behaviors, the HR was 1.60, relative to avoiding both behaviors. Thus, the risk associated with alcohol and smoking was multiplicative, but there was no statistically significant multiplicative interaction (synergy) (P=.22).

Figure 3.

Joint analysis of smoking (current, former, never) and alcohol consumption (0g/d, 0.1–14.9g/d, ≥15g/day) and risk of incident diverticulitis

Discussion

In this large prospective study of women, both past and current smoking were associated with an increased risk of diverticulitis when compared to never-smoking, although no dose-dependent relationship was seen. The association of smoking with diverticulitis appeared to be longstanding, as there was no difference in risk when past smokers were stratified according to time since quitting. The association between smoking and diverticulitis was strengthened in an analysis of women who underwent surgery for diverticulitis. Alcohol consumption was associated with an increased risk of diverticulitis, with the greatest risk seen in women who consumed an average of two or more alcoholic beverages per day. Liquor had the strongest association with diverticulitis when intake was stratified by type of alcohol. Joint analysis revealed that participants who smoked tobacco and consumed more than one alcoholic beverage a day had the highest risk of incident diverticulitis.

Several studies have examined the role of smoking on diverticular disease; however most of these studies assessed hospitalization for symptomatic diverticular disease rather than diverticulitis specifically^{8, 16, 33} or looked at severe outcomes of diverticulitis such as colectomy or perforation^{9, 10, 19}. These studies have consistently found that smoking increases the risk of severe diverticulitis. The current study extends these prior findings by demonstrating that smoking increases risk for the spectrum of diverticulitis outcomes including severe disease requiring surgery. Identifying risk factors for less severe manifestations of diverticulitis is clinically relevant given the high burden of disease seen in the emergency department and ambulatory settings³⁴. Indeed, nearly 90% of patients present with uncomplicated diverticulitis and most are managed outside the hospital³⁵. In addition, in comparison to prior studies, we were able to control for a more comprehensive list of known or suspected confounders including diet, lifestyle, and medication use. Furthermore, prior studies examined smoking at baseline and not cumulative smoking exposure (pack-years of smoking).

Very few prior studies have evaluated the role of alcohol on diverticulitis, with many prior studies looking at diverticulosis or symptomatic diverticular disease as an outcome rather than diverticulitis^{13, 16, 33}. In addition, alcohol use has largely been classified as regular vs non-regular use or heavy consumption has been identified using administrative codes for alcohol use disorder^{11, 15}. In our study, we ascertained daily alcohol intake using a validated food frequency questionnaire so that we were able to study both modest and high intake, assess for a dose-response relationship, update consumption over follow-up rather than relying on baseline assessment, and examine type of alcohol. Also unique to the current analysis, we examined the joint association between smoking and alcohol. In this analysis, we found that participants with the highest risk of incident diverticulitis were those who were current smokers and consumed one or more drinks of alcohol daily.

In this study, we observed that both past and current smokers had similar levels of risk when compared to non-smokers. Additionally, we did not find a dose-response relationship between smoking and risk of diverticulitis. Although a study of perforated diverticulitis observed a dose-response relationship between cigarette use at baseline and perforation¹⁰, similar to our findings, other studies have found that any history of smoking was associated with an increased risk of surgical intervention such as colectomy or postoperative recurrence of diverticulitis^{7, 9}. Notably, studies of coronary artery disease risk have also not observed a dose-response with smoking. A meta-analysis of five large studies found a nonlinear relationship between the number of cigarettes smoked per day and the relative risk of disease; it was suggested that the dose-response characteristics of the risk relationship are less steep at higher doses, with a low threshold for harm³⁶. These findings suggest that tobacco exposure, even in low doses, may have long-lasting effects on the body. Studies have demonstrated that cigarette smoking causes a chronic inflammatory state, characterized by increased counts of circulating leukocytes, C-reactive protein, and other acute-phase reactants, independent of years since quitting smoking^{36, 37}. The systemic inflammation induced by tobacco may help explain our findings. In previous work, we have shown that pre-diagnostic plasma levels of markers of inflammation were associated with incident diverticulitis³⁸. Another possible explanation for smoking’s long-lasting impact on diverticulitis risk could be an association between cigarette smoking and the development of diverticulosis, which has been demonstrated in other studies^{14, 15}. As the development of diverticulosis occurs before diverticulitis, smoking may influence diverticulitis risk in two ways: increasing the risk of diverticulosis, and then of subsequent diverticulitis.

In our analysis of alcohol consumption, we found an increased risk of diverticulitis in women in the highest category of consumption (more than 2 drinks per day). As noted above, prior studies of alcohol and diverticulitis have examined alcohol as a categorical/binary variable. While the etiopathogenesis of diverticulitis is incompletely understood, alcohol consumption is known to impact prostaglandin synthesis in the gut. In one study, prostaglandin synthesis was decreased in patients with alcohol use disorder when compared to those without this disorder³⁷. Prostaglandins help protect the gastrointestinal mucosa, and impaired synthesis results in compromised mucosal integrity and increased bacterial translocation^{39, 40}. Another mechanism that could help explain our findings for both alcohol consumption and smoking is the role both of these exposures play in the generation of advanced glycation end-products (AGEs). AGEs are a class of end products derived from nonenzymatic glycation that have the potential to produce reactive oxygen and nitrogen species, leading to oxidative stress, inflammation, and apoptosis. Advanced glycation end products have also been associated with decreased gut microbial diversity, impaired gut barrier function, increased proinflammatory cytokines, and intestinal epithelial damage⁴¹. Lastly, our prior work linked markers of systemic inflammation to increased risk of diverticulitis³⁸. Previous randomized trials have demonstrated that red and white wine, but not gin, reduced markers of systemic inflammation, likely due to both ethanol and polyphenols components of wine^42–45 . This is consistent with our findings that liquor, but not red wine, is associated with increased risk of diverticulitis.

We acknowledge several limitations of our study. Diverticulitis diagnoses were self-reported; however, review of a subset of medical records confirmed the diagnosis in more than 90% of cases establishing the validity of self-report in this population. Additionally, our study participants were female nurses, thus it is possible that our findings may not be generalizable to men or other populations. Nonetheless, the results of other studies have generally been consistent with ours. It is also possible that a dose-response relationship for smoking may have been detected in a population with higher smoking intensity, but current smokers had an average of 23 smoking pack-years which is above the threshold used to identify patients who may benefit from lung cancer screening. Lastly, while we were able to adjust for a broad range of risk factors for diverticulitis, the observational nature of our study makes it possible that residual confounding may have impacted our results.

In summary, in this large prospective study of women, we found that both past and current smoking were associated with an increased risk of diverticulitis. We also found that heavy alcohol consumption was associated with increased risk. A joint analysis of smoking and alcohol found that individuals who smoked and consumed more than 1 alcoholic beverage per day were at highest risk. As there are currently no medical means to prevent diverticulitis other than dietary and lifestyle interventions, counseling patients about the avoidance of smoking and alcohol may help lower the risk of developing diverticulitis.

What You Need to Know.

Background:

Although smoking has been shown to increase risk of asymptomatic diverticulosis and complicated diverticulitis, less is known about how these behaviors impact uncomplicated diverticulitis.

Findings:

Current and past smoking and heavy alcohol consumption, particularly liquor, increase the risk of diverticulitis. Participants who smoked tobacco and consumed one or more alcoholic beverages a day had the highest risk of incident diverticulitis.

Implications for patient care:

Counseling patients about the avoidance of smoking and alcohol may help lower the risk of developing diverticulitis.