Association between irritable bowel syndrome and colorectal cancer: A nationwide population-based study.

Association between irritable bowel syndrome and colorectal

cancer: A nationwide

population-based study

☆

Cheng-Wen Hsiao

,Wen-Yen Huang

, Tao-Wei Ke

, Chih-Hsin Muo

,William

Tzu-Liang Chen

,

Fung-Chang Sung

, Chia-Hung Kao

⁎

a Division of Colon and Rectal Surgery, Department of Surgery, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan

b Department of Radiation Oncology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan c Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan

d Division of Colorectal Surgery, Department of Surgery, China Medical University Hospital, Taichung, Taiwan e Management Offce for Health Data, China Medical University Hospital, Taichung, Taiwan

f Graduate Institute of Clinical Medicine Science and School of Medicine, College of Medicine, China Medical University, Taichung, Taiwan

g Department of Nuclear Medicine and PET Center, China Medical University Hospital, Taichung, Taiwan h Department of Public Health, China Medical University and Hospital, Taichung, Taiwan

1. Introduction

Irritable bowel syndrome (IBS) and colorectal cancer (CRC) are major global health care problems. These diseases are common, and their total economic burden and sequelae are substantial. IBS is a chronic gastrointestinal syndrome characterized by relapsing abdominal pain and altered bowel habits in the absence of any organic abnormalities. It is the most commonly diagnosed gastrointestinal condition, and accounts for approximately 30% of all referrals to gastroenterologists [1,2]. A meta-analysis showed that the total prevalence is 11.2%, varying according to country from 1.1% to 45.0% and criteria used to defne IBS [3]. Patients with IBS are also associated with more frequent comorbidities, higher total medical costs, and substantially lower health-related

quality of life comparedwith the general population [4–9]. CRC is a universally lethal disease. It is the third most common cancer in men and

second in women worldwide, with over 1.2 million new cancer cases and 608,700 deaths estimated to occur annually [10].

Because IBS and CRC are common gastrointestinal problems,whether the risk of CRC increases after an IBS diagnosis is a public health concern. However, related studies are scanty and from the Western

to other ethnic groups. So far, data on Asian populations are lacking. In Taiwan, CRC is the second most common type of cancer for both men and women, with an incidence of 61.45 per 100,000 for men and 46.48 per 100,000 for women, according to the 2009 cancer report released by the Taiwan Department of Health. A previous survey showed

that the prevalence of IBS in Taiwan is 17.5%–22.1% [13]. Taiwan is a country with a high prevalence rate of IBS in Asia and is comparable to

those in developed countries in theWest [3,14,15].

A large medical database is available to researchers in Taiwan; that is, the Taiwan National Health Insurance Research Database (NHIRD), which is composed of de-identifedmedical claims from99% of the population of Taiwan (23 million people) and is contractedwith 97% of hospitals

and practitioners. It has been extensively used in epidemiologic

studies [16–18]. This provided an opportunity for us to examine the hypothesis that an IBS diagnosis is associatedwith an increased risk of subsequent CRC development by comparing age-, sex-, and time-matched

controls in a nationwide population-based cohort study. 2. Methods

2.1. Data sources

The Taiwan National Health Insurance (NHI) is a universal health program established in 1995. The data used in the Longitudinal Health Insurance Database (LHID) is a sub-data set of the NHIRD. The LHID randomly selected 1 million insured people fromthe NHIRD by using a systematic random sampling method and included approximately 5% of

the entire population. The details of this database have been published previously [18–20].

2.2. Study samples

The study cohort consisted of newly diagnosed IBS patients from

2000 to 2010 based on the International Classifcation Disease, 9th Revision, Clinical Modifcation (ICD-9-CM) codes 564.1 (n = 92,336). We

excluded 590 peoplewith a history of cancer (ICD-9-CM140-208) diagnosed before the index date (defned as the date of the frst diagnosis of

IBS) or with missing information on age or sex.

For each IBS case identifed in the study cohort,we randomly selected 2 insured participants in the same period as the comparison cohort (non-IBS cohort), matched with age (every 5 years), sex, and index year and month from people without a history of IBS.

of CRC (ICD-9-CM 153–154), December 31, 2010, or withdrawal from insurance.

Because of possible unmeasured or unknown confounders, we only selected 3 most common systemic diseases as the comorbidities. We

also identifed patients with the baseline comorbidities, including hypertension (ICD-9-CM 401–405), diabetes mellitus (ICD-9-CM 250),

and hyperlipidemia (ICD-9-CM 272). 2.3. Statistical analysis

Data analysis involved comparing distributions of age, sex, urbanization, and comorbidities between the IBS cohort and the non-IBS cohort,

whichwere examined using the chi-square test. We calculated the incidence density of CRC by using these variables and the IBS cohort to

non-IBS cohort rate ratio of CRC for each variable. For incidence density calculation, we calculated follow-up person-years for the participants until

CRC diagnosis, until 31 December, 2010 for thosewhowere uncensored, or the censoring date for other reasons, such as death, emigration, or termination of the insurance program.

We calculated hazard ratios (HRs) and 95% confdence intervals (CIs) by using the Cox proportional-hazards model to examine the risks of CRC for patients with and without IBS. We also analyzed the risk of CRC by stratifying the observation period into less than 2 years and more than 2 years. Kaplan–Meier analysis and the log-rank test were used to calculate the CRC-free proportions for the risk of developing CRC in IBS patients and in non-IBS patients.

All analyses were performed using SAS statistical software (version 9.1 for Windows; SAS Institute, Inc., Cary, NC, USA), and the results were considered statistically signifcant when two-tailed p values were less than .05.

3. Results

3.1. Characteristics of the study participants

Table 1 shows a comparison of the distribution of sociodemographic characteristics and baseline comorbidity status between the 2 cohorts. The mean (±SD) age of the IBS cohort was 44.6 (±19.7) years, and that for the non-IBS cohort was 44.4 (±19.7) years. In the IBS cohort, 3.64% of patients had undergone a colonoscopy and/or sigmoidoscopy before the diagnosis of IBS, whereas only 0.88% of the non-IBS cohort had undergone these procedures (p b .0001). Among the 275,238 patients, the IBS cohort was more susceptible to comorbidities, including

hypertension (25.2% vs 21.7%, p b .0001), hyperlipidemia (15.6% vs 11.7%, p b .0001), and diabetes mellitus (9.47% vs 7.97%, p b .0001). 3.2. Risk and relative risk of CRC

Table 2 shows the CRC incidence densities in follow-up person-years for the 2 cohorts and the IBS to the non-IBS cohort rate ratios. Overall, the follow-up results showed a higher CRC incidence in the IBS cohort than in the non-IBS cohort (16.79 vs 6.34 per 10,000 person-years; IRR = 2.65, 95% CI = 2.39–2.93). Kaplan–Meier survival analysis showed that the CRC-free rate was 2% lower in IBS patients than in non-IBS controls; however, it was signifcant in the follow-up period (log-rank p b .0001) (Fig. 1). The increased risk was noted during the frst 2 years, particularly within the initial 3 months. The incidence rate of CRC in the IBS cohort was 2.89-fold higher than in the comparison cohort (20.76 vs 7.19 per 10,000 person-years) for male patients and 2.41-fold for female patients. In both the IBS and non-IBS cohorts, the incidence densities of CRC were highest in the oldest group (49.72 vs 20.01 per 10,000 person-years), whereas the IRR of CRC was highest in patients aged 30–44 years (IRR = 3.93, 95% CI = 2.69–5.73)

(Table 2).

The study period-stratifed analysis for incidence and IRR of CRC showed that the risks were signifcantly greater within 2 years (IRR = 10.4, 95% CI = 8.40–12.8 for the frst year; IRR = 2.01, 95%

CI = 1.48–2.72 for the second year) (Table 2). Cox's proportional regression model indicated that, compared with

the non-IBS cohort within 2 years, the HR of CRC was signifcantly increased in patients with IBS (HR = 6.72, 95% CI = 5.70–7.92) (Table 3). However, when the study period was more than 2 years, the relationship was not signifcant (HR = 1.08, 95% CI = 0.93–1.26). The HRs of CRC measured by comorbidities were 0.78 for patients

with hypertension, 0.99 for those with hyperlipidemia, and 1.13 for patients with diabetes mellitus within 2 years (Table 3).

Table 4 shows the distributions of colonoscopy/sigmoidoscopy after the index date between the IBS and non-IBS cohort by study duration (within 2 years and over 2 years). The IBS cohort was most likely to have a colonoscopy/sigmoidoscopy within 2 years (13.0% vs 2.28%), and the IBS cohort was most likely to have polyps within 2 years (4.16% vs 0.88%).

This is the largest study to examine the association between IBS and the subsequent risks of CRC. Using the nationwide database, we included 91,746 patients in the IBS cohort and 183,492 patients in the comparison cohort.We found an increased risk of CRC in the initial 2 years after a diagnosis of IBS. Thereafter, no association was observed between IBS and CRC.

The increased risk during the frst 2 years can be attributed to certain factors. Misclassifcationmust be considered frst; that is, IBS-like symptoms were reported before the diagnosis of CRC. IBS and CRC have clinical presentations on a pathophysiologic spectrumof diseases because of the considerable overlap between symptoms in patients with IBS and CRC. Both diseases share a symptomcomplex with altered bowel habits and abdominal pain. Thus, it is diffcult to distinguish CRC fromIBS at the initial intervention, especially formild disease activity. The risk of CRC in the IBS cohort increased during the initial 2 years, particularly in the frst 3 months. This may have occurred because CRC was initially misclassifed as IBS; however, subsequent examinations showed CRC. Before a diagnosis of IBS, it is crucial to exclude any organic colorectal problems. However, it is generally accepted that a colonoscopy/ sigmoidoscopy is unnecessary for all patients before making a diagnosis of IBS. Some experts also suggest that a symptom-based diagnosis, such as that based on the Rome III criteria, without an “alarm” symptom is suffcient for the diagnosis of IBS, and that no further examinations are required. Most guidelines recommend that a colonoscopy be reserved for patients over the age of 50 years and/or the presence of

“alarmsigns,” such as fever, weight loss, blood in stools, anemia, abnormal physical or laboratory results, and a family history of infammatory bowel disease or colon cancer [21–24]. Although we did not observe an association between IBS and CRC in the long term, a considerable fraction of IBS patientswere subsequently diagnosedwith CRC, particularly in the frst 2 years. Some of them were CRC patients initially

misdiagnosed to have IBS; however, we did not determine the number of these patients. The symptom-based diagnosis of IBS may lead to missing CRC, which requires quite different management than that for

IBS. In this study, only 3.64% of patients underwent a colonoscopy/sigmoidoscopy before the diagnosis of IBS. Thus,we recommend that physicians

carefully assess IBS patients, and endoscopic exclusion may be

of expansion of using colonoscopy/sigmoidoscopy and the indications for these examinations can be used for further studies.

Another possible explanation for the increased risk in the frst

2 years is the “surveillance effect” because the casesmay have hadmedical examinations to a greater extent than the controls. In the analysis of the distribution of colonoscopy and/or sigmoidoscopy after the index date between the IBS and non-IBS groups, a signifcantly higher frequency of colonoscopy and/or sigmoidoscopy was observed in the IBS

cohort. In addition, more patients in the IBS cohort were diagnosed with colonic polyps. Therefore, it is possible that more participants in

the IBS cohort were diagnosed with CRC because of the tendency of clinicians to order diagnostic tests for IBS symptoms.

The resultwas consistentwith previous studies fromWestern countries. Nørgaard et al. performed a retrospective cohort study to examine

the risk of CRC among patients with IBS using data fromthe Danish National Registry of Patients and the Danish Cancer Registry from 1977 to

2008 [12]. They included 57,851 IBS patients and found that the standardized incidence ratios were 8.42 (95% CI: 6.48–10.75) for colon cancer

and 4.81 (95% CI: 2.85–7.60) for rectal cancer in the frst 3 months after IBS diagnosis. However, the risk of CRC decreased 1–10 years after an IBS diagnosis. They indicated that the increased risks in the frst 3 months may be explained by diagnostic confusion caused by overlapping symptomatology. Similarly, Rodriguez et al. conducted a population-based study to examine the incidence of subsequent CRC and infammatory bowel disease in patients with IBS and functional dyspepsia, and compared them with the general population by using the General Practice Research Database in England and Wales [11]. With a mean follow-up time of 36 months, the cumulative risk of a benign and malignant colorectal tumor was approximately 1% during the frst year after a diagnosis of IBS. However, the risk in IBS patients did not differ signifcantly from that in the general population after the frst year. Although these studieswere conducted inWestern countries, the results obtained from Taiwan were consistent with their results. Large cohort studies from other countries and ethnic groups are necessary to reach a frm conclusion.

In the IBS cohort, patients had more colonoscopies and/or sigmoidoscopies before a diagnosis of IBS. There might be a concern that they

However, we thought it was less likely to signifcantly affect our result because only a small number of patients (3.64%) in the IBS cohort had endoscopies before a diagnosis of IBS. Moreover, if it actually had an effect on the risk of following CRC, it should be an underestimation in IBS cohort. The risk might be even higher in the IBS cohort in initial 2 years, but thiswas less likely to affect the risk after 2 years. Hence, our fnding was still true: an increased risk of CRC in the initial 2 years but no association thereafter.

The large sample size from the nationwide population-based data set strengthened the statistical power of the examination of associations between IBS and CRC. The patients presented a wide range of demographic characteristics, which enabled us to perform stratifed analyses

according to age, sex, and comorbidities. The study population was effectively defned, and the databasewas generally accurate. The accuracy

and high validity of diagnoses identifed in the NHIRD also have been demonstrated in the previous studies [25–27]. NHIRD included primary care, secondary care, tertiary care, home and community care, and urgent care. In addition, computerized data registries andwidespread coverage of the residents and health providers of the nationwide database

allowed us to trace nearly all medical service-use histories for the patients and reduce the likelihood of loss to follow-up.

This study had some limitations. First, the NHIRD does not provide detailed information on patients, such as their smoking habits, alcohol consumption, body mass index, physical activity, and family history of cancer, which are risk factors for CRC. We were unable to contact patients directly to obtain more information because of the anonymity ensured by the identifcation numbers. Second, the results derived from a cohort study are generally of a lowermethodological quality than those from randomized trials because a cohort study design is subject to several biases related to adjustment for confounders. Despite the meticulous study design with an adequate control of confounding factors, a

crucial limitation was that bias may occur because of possible unmeasured or unknown confounders.

In conclusion, IBS is not associated with the long-termdevelopment of CRC in Taiwan. The increased risk of CRC in the frst 2 years may be

attributed to misclassifcation. Physicians must carefully assess IBS patients and consider endoscopic examinations for more patients to exclude