Original Article
Long-term risk of pancreatitis and diabetes after
cholecystectomy in
patients with cholelithiasis but no pancreatitis history: A
13-year
follow-up study
Ming-Shian Tsai
a, Cheng-Li Lin
b,c, Yao-Chun Hsu
d, Hui-Ming Lee
a, Chia-Hung
Kao
e,f,g,⁎
1. Introduction
Cholelithiasis, or gallstones, is associated with pancreatitis in nearly
one-third of patients experiencing pancreatitis [1–3]. Although the
incidence of biliary pancreatitis has increased in the previous 2 decades, no reductionwas observed in the case-fatality rate in patientswith pancreatitis
[3,4]. Therefore, preventing the occurrence of biliary pancreatitis and its associated complications is imperative. Because themigration of gallstones has been regarded as a crucial factor causing recurrent pancreatitis, cholecystectomy is generally recommended in healthy
patients experiencing biliary pancreatitis [5–7].
Although an association between gallstones and pancreatitis has been recognized formore than 100 years, the effect of cholecystectomy on this risk has not been extensively studied.Moreover, it is controversialwhether cholecystectomy can reduce the risks of pancreatitis and probably subsequent diabetes in patients with cholelithiasis who did not previously
experience pancreatitis. Studies have shown that patientswith gallstones smaller than 5 mm in diameter have a signifcantly increased risk of
developing acute biliary pancreatitis [8,9]. Cholecystectomy appeared to
facilitate reducing the risk of pancreatitis only in these patients [8].
However, a study that enrolled patients who were diagnosed with gallstones between 1950 and 1970 reported that cholecystectomy reduced the risk of pancreatitis in patientswith cholelithiasis, regardless
of whether previous pancreatitis had occurred [10]. However, laparoscopic cholecystectomy (LC) was
not universally applied during
the study period.Moreover, because the overall incidence of pancreatitis
reported in the study was low [10], performing cholecystectomy to
prevent pancreatitis was suggested only if a patient had previously experienced pancreatitis.
In consideration of the great advancement of minimally invasive surgery and the increased incidence of biliary pancreatitis during the past 2 decades, reappraising whether cholecystectomy could reduce the risk of biliary pancreatitis in patients who have not previously experienced pancreatitis is essential. In addition to conducting this reappraisal, we aimed to determinewhether cholecystectomy is associated with reduced risk of subsequent diabetes because it is well established
that pancreatitis may lead to the onset of diabetes [11]. In this 13-year
follow-up study,we compared the risks of pancreatitis and diabetes in patients with gallstone diseases but no history of pancreatitis or diabetes by
analyzing a broadly representative population-based cohort from Taiwan's National Health Insurance Research Database (NHIRD). 2. Methods
2.1. Data source
Taiwan launched a single-payer National Health Insurance (NHI) program on March 1, 1995. The NHI program covers more than 99% of
the approximately 23.74 million residents of Taiwan (http://www.nhi.
gov.tw/english/index.aspx). For research purposes, the Taiwan National Health Research Institutes manages and releases the NHIRD annually. The data used in this study originated from the Longitudinal Health Insurance Database 2000 (LHID2000),which is a subset of the NHIRD that
contains all claims data of one million benefciaries from 1996 to 2011. No signifcant differences exist in age, sex, or health care costs between the sampled group and all enrollees in the NHI program. The LHID2000 provides encrypted patient identifcation numbers; records on patients' sex, date of birth, and dates of admission and discharge; ICD-9-CM
(International Classifcation of Diseases, Ninth Revision, ClinicalModifcation) codes of diagnoses and procedures; and details on prescriptions
and costs covered by NHI. This study was exempted from full ethical review (IRB permit number: CMU-REC-101-012). Informed consent was not required because the datasets were devoid of identifable personal information.
2.2. Sampled patients
All patientswho had a history of cholelithiasis (ICD-9-CM code 574) between 2000 and 2010 were included in the study population. Patients with cholecystectomy (ICD-9-CM procedure codes 51.22 and 51.23) were assigned to the cholecystectomy cohort. Patients who had a history of pancreatitis (ICD-9-CM codes 577.0 and 577.1) or diabetes (ICD-9-CM
code 250), those who were younger than 20 years, and those for whom complete information was lacking were excluded. The date of the cholecystectomy was used as the index date. The comparison cohort was randomly selected from the remaining patients who had cholelithiasis
but did not undergo cholecystectomy. For each patient in the cholecystectomy cohort, 2 comparison patients without a history of diabetes or
pancreatitis were identifed and frequency-matched according to age (within 5-year spans), sex, and the year of the index date.
2.3. Outcome and comorbidities
All patients were followed up until they received a new diagnosis of pancreatitis or diabetes during the follow-up period. Each patient was monitored from the index date until he or she was diagnosed with pancreatitis or diabetes or until he or she was censored because of loss to follow-up, death,withdrawal frominsurance, or the end of December 31, 2011. Patients with claims records showing a history of hypertension (ICD-9-CM codes 401–405), hyperlipidemia (ICD-9-CM code
272), hepatitis B (ICD-9-CM codes 070.2, 070.3, and V02.61), hepatitis C (ICD-9-CM codes 070.41, 070.44, 070.51, 070.54, V02.62, and 070.7), cirrhosis of the liver (ICD-9-CM codes 571.5 and 571.6), or alcoholrelated illness (ICD-9-CM codes 291, 303, 305, 571.0, 571.1, 571.3,
790.3, and V11.3, including alcoholic psychoses, alcohol dependence syndrome, alcohol abuse, alcoholic fatty liver, acute alcoholic hepatitis, alcoholic cirrhosis and alcoholic liver damage) identifed at the baseline were considered to have comorbidities.We also considered inspection performed using choledochoscopy.
2.4. Statistical analysis
The demographic characteristics and comorbidities of patients with cholelithiasis who underwent cholecystectomy and those who did not undergo cholecystectomy, including age (20–49 years, 50–64 years, and ≥65 years), sex, and comorbidities, were compared using a chisquare test. We used a Student's t test for continuous variables. Age-,
sex-, and comorbidity-specifc incidence densities (1000 personyears) of pancreatitis and diabetes were estimated. Univariable and
multivariable Cox proportional hazards regression models were used to estimate the hazard ratios (HRs) and 95% confdence intervals (CIs) for pancreatitis and diabetes in patients with cholelithiasis who underwent cholecystectomy compared with those in the comparison cohort. The multivariable models were simultaneously adjusted for
age; sex; comorbid hypertension, hyperlipidemia, hepatitis B, hepatitis C, cirrhosis of the liver, and alcoholic liver disease; and inspection through choledochoscopy. The cumulative incidence of pancreatitis and diabetes in both cohorts was assessed using the Kaplan–Meier method, and the differences between the curves were evaluated using a log-rank test. All statistical analyses were performed using SAS statistical software (Version 9.3 for Windows; SAS Institute, Inc., Cary, NC,
USA). A 2-tailed p value b .05 was considered statistically signifcant. 3. Results
The cholecystectomy cohort comprised 4467 persons and the comparison
cohort comprised 8823 persons (Table 1). In the present
study, 42.5% of the patients were younger than 49 years and 57.2% were female. The mean ages for the cholecystectomy and comparison
cohortswere 54.3 (SD=16.0) and 54.5 (SD=16.0) years, respectively. Compared with the comparison cohort, inspection through
choledochoscopy was more prevalent in the cholecystectomy cohort at the baseline (p b .001).
During themean follow-up periods of 5.71 years for the cholecystectomy cohort and 5.89 years for the comparisons cohort, the overall incidence density of pancreatitis was signifcantly lower in patients with
cholecystectomy than it was in the comparison cohort (1.96 vs 3.75 per 1000 person-years, crude HR = 0.52, 95% CI = 0.38–0.71), with
an adjusted HR of 0.49 (95% CI = 0.36–0.68) (Table 2 & Fig. 1A).
Fig. 1A shows the cumulative pancreatitis incidence curve for the 2 cohorts and that the cholecystectomy incidence curve is signifcantly
lower than that for comparison cohort (log-rank test, p b .001). We further performed age- and sex-specifc analyses for the risk of
pancreatitis in both cohorts (Table 2). The incidence of pancreatitis increased
with age in both cohorts. Patients who underwent cholecystectomy and were aged 49 years or younger had a lower risk for
pancreatitis (crude HR = 0.46, 95% CI = 0.28–0.77), with an adjusted HR of 0.43 (95% CI = 0.26–0.73). The corresponding adjusted HRs showed that patients older than 65 years had a signifcantly lower risk for pancreatitis (adjusted HR = 0.50, 95% CI = 0.30–0.85). The incidence of pancreatitiswas higher in men than inwomen in both cohorts.
However, the pancreatitis risk associated with cholecystectomy was similar in women (adjusted HR = 0.51, 95% CI = 0.32–0.80) and men (adjusted HR= 0.49, 95% CI= 0.32–0.76).
Next,we analyzed the association between cholecystectomy and the risk of pancreatitis and stratifed the patients according to comorbidities. The results showed that the risk of pancreatitis was signifcantly
lower in patients without comorbidities (adjusted HR = 0.38, 95% CI = 0.22–0.64) and those with hypertension (adjusted HR = 0.54,
95% CI= 0.34–0.85) (Table 2).
On the other hand, the overall incidence of diabetes was not signifcantly lower in the cholecystectomy cohort than in the comparison cohort (11.5 vs 12.9 per 1000 person-years), with an adjusted HR of 0.92
(95% CI = 0.80–1.06) (Table 3). Among patients with cirrhosis of the
liver, the risk of diabetes was signifcantly higher in the cholecystectomy cohort than in the comparison cohort (adjusted HR = 2.31; 95%
CI = 1.03–5.17). The Kaplan–Meier curve showed that the cumulative incidence of diabetes did not differ between the cholecystectomycohort
and the comparison cohorts (Fig. 1B, log-rank test, p = .12).
4. Discussion
This nationwide cohort study demonstrated that cholecystectomy for cholelithiasis was associated with a reduced risk of pancreatitis, but not diabetes, in patients with no history of pancreatitis. Our results showed that the cholecystectomy group exhibited an adjusted HR of 0.49 (95% CI = 0.36–0.68) for pancreatitis compared with the control group, after we accounted for mortality as the competing cause of risk and adjusted for multiple known confounding factors. However, no signifcant difference in the risk of diabeteswas observed between the control
and cholecystectomygroups. According to our results, reduced risks of pancreatitis indicate a potential beneft that could be neglectedwhen clinicians consider surgical treatment for patients with cholelithiasis, especially when the patients have no history of pancreatitis. Further prospective studies investigating the associations found in our study are therefore warranted.
In the present study, the absolute difference in the incidence of pancreatitis between the 2 cohorts was 1.95 per 1000 person-years
(Table 2). In otherwords, cholecystectomyin 100 cholelithiasis patients without pancreatitis history should be able to prevent 1 patient from experiencing pancreatitis for every approximately 5-year follow-up.
When the follow-up period is extended to 25 years, the number of cholecystectomy needed to prevent 1 patient from pancreatitis will decrease
iatrogenic biliary tract injury ranged from0.4 to 0.5% during laparoscopic
cholecystectomy, [12–16] indicating 1 in 200 patientsmay encounter
this severe complication. It is also reported that aged patients undergoing cholecystectomyhadmore blood loss, but not other surgical complications,
than younger population [17]. Therefore, in patients with life
expectancy of at least 2–3 decades, e.g.middle-aged population,we believe that the beneft of cholecystectomy in prevent pancreatitis may
outweigh its potential hazards, specifcally biliary tract injury. However, it needs more cost/effectiveness analyses to validate our hypothesis. Alcohol-related illness, defned by ICD-9-CM codes 291, 303, 305, 571.0, 571.1, 571.3, 790.3, and V11, was signifcantly less prevalent in the cholecystectomy group than the non-cholecystectomy group. Patients with alcohol-related diseases are generally less healthy and therefore not suitable for surgical intervention, which was one possible
explanation of this observation. The differences in alcohol consumption between the two groups could potentially bias our results, since alcohol
intake is a well-known risk factor of pancreatitis [18]. However,
cholecystectomy was still associated with decreased risk of pancreatitis after adjustment of alcohol-related illness and in patients without this
comorbidity (Table 2). Consequently, we believed the reduced risk
of pancreatitis in the cholecystectomy group could not be explained simply by less alcohol use.
A study proposed that several local factors, including small gallbladder stones, enlarged cystic ducts, properly sized impacted stones, and a functioning common channel between the biliary tract and pancreatic
duct, predispose patients to gallstone pancreatitis [9]. Because these factors
also contribute to the development of cholangitis in patients with
cholelithiasis and justify the indication for cholecystectomy, the cholecystectomy group may have had more of these factors compared with
the control group. The proportion of patients in the cholecystectomy
group who underwent endoscopic retrograde cholangiographic procedureswas signifcantly higher than that in the comparison group during
the same hospitalization period (Table 1), supporting our arguments.
Because we were unable to adjust for these predisposing factors, the real effects of cholecystectomy on the risk of pancreatitis may have been underestimated in this study.
It should be noted that the incidence of pancreatitis, but not diabetes, was slightly higher in the cholecystectomy group than in the control
group within the frst year after cholecystectomy (Fig. 1). Because the elevated risk of pancreatitis disappeared approximately 2 years after cholecystectomy, we believe that this elevation is likely related to the procedure. This fndingmay be explained by gallstone migration during the surgical procedure as well as the higher percentage of endoscopic retrograde cholangiographic procedures in the cholecystectomy group
[19].
Diabetes is a long-term concern of patients with chronic and/or
acute pancreatitis [20–23]. Moreover, a higher prevalence of gallstone
diseaseswas reported in patientswith diabetes comparedwith controls
[24]. Although we found that cholecystectomy is associated with a reduced
risk of pancreatitis, we failed to detect any difference in newly
onset diabetes risks between the cholecystectomy and comparison cohorts, except in cirrhotic patients. Cholecystectomy appeared to be unrelated
to the risk of diabetes, at least in the patientswithout a history of gallstone pancreatitis. However, this fnding did not exclude the
possibility that cholecystectomy in patients with gallstone pancreatitis may reduce the risk of diabetes. Further study is required to clarify the effects of cholecystectomy in this group of patients.
Although our data showed that cholecystectomy is associated with decreased risk of pancreatitis in the patients without previous pancreatitis attack, we still could not obtain the conclusion regarding whether
cholecystectomy is really helpful for reducing the pancreatitis risk. After all, the present study was not a randomized controlled trial, and there existed too many factors which may infuence whether a patient would undergo cholecystectomy or not. Selection bias could not be completely avoided, even afterwe consideredmany possible confounding factors in the present study. We should keep this point in mind
when we interpret the results.
The strength of this study is that it was a nationwide observational
study that demonstrated a reduced risk of pancreatitis after cholecystectomy for cholelithiasis in patientswithout previous pancreatitis. Furthermore, the completeness of the NHIRD enabled us to control for
multiple confounding factors. However, this study had limitations. First, the NHIRD does not provide detailed information on tobacco use, body mass index, physical activity level, socioeconomic status, and family history, all of which are potential confounding factors. In particular, the lack of information regarding alcohol consumption amount may
have biased the results [18].Moreover, we could not determinewhether there existed differences in the severity of pancreatitis between the two groups, due to lack of details regarding pancreatitis severity in NHIRD. Second, we were unable to validate the diagnoses of cholelithiasis, pancreatitis, and diabetes by performing chart review. Theoretically, the diagnoses of cholelithiasis, cholecystectomy, pancreatitis, and diabetes were reliable because patient diagnoses are strictly audited for the purpose of reimbursement. Finally, we determined the occurrence of pancreatitis by investigating hospitalized patients. We may
therefore have overlooked patients with severe or mild cases who did not seek hospital care. Although these limitations could underestimate or overestimate the actual incidence of pancreatitis and diabetes, they should not bias the results of the present study because of the accessibility
and high coverage rate of universal health insurance in Taiwan [25].
In summary, this study demonstrated that cholecystectomy for cholelithiasis is associatedwith a reduced risk of pancreatitis, but not diabetes,