ORIGINAL ARTICLE LIVER, PANCREAS, AND BILIARY TRACT

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 010;8: ORIGINAL ARTICLE LIVER, PANCREAS, AND BILIARY TRACT Danish Patients With Chronic Pancreatitis Have a Four-Fold Higher Mortality Rate Than the Danish Population CAMILLA NØJGAARD,* FLEMMING BENDTSEN,*, ULRIK BECKER,*, JENS RIKARDT ANDERSEN,, CLAUS HOLST, # and PETER MATZEN* *Department of Gastroenterology, Hvidovre Hospital, Copenhagen, Denmark; Faculty of Health Science, Department of Human Nutrition, University of Copenhagen, Copenhagen, Denmark; National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark; Nutrition Unit, Rigshospitalet, Copenhagen, Denmark; # Institute of Preventive Medicine, Center for Health and Society, Copenhagen, Denmark See related article, Dahlhoff M et al, on page 1585 in Gastroenterology. BACKGROUND & AIMS: We investigated mortality of patients with chronic pancreatitis (), compared with the Danish population and sought to determine whether clinical presentations of can be used in prognosis. We also investigated clinical factors associated with mortality and causes of death among these patients. METHODS: The Copenhagen Pancreatitis Study is a prospective study of patients admitted from 1977 to 198 to the 5 main hospitals in Copenhagen with a diagnosis of acute pancreatitis or. In 008, follow-up data were collected from these patients from the Danish Registries; this subcohort comprised 90 patients with probable (n 41) or definite (n 49). RESULTS: The mortality of patients with definite was 4-fold that of the Danish population and significantly higher than that of patients with probable (P.003; 95% confidence interval [CI], ); patients with probable had a - to 3-fold higher mortality rate than the population. In patients with definite, factors significantly associated with mortality included non-employment (P.015; 95% CI, ), and being underweight (P.00; 95% CI, ). Sex, alcohol use, smoking, single versus co-living, exocrine insufficiency, diabetes, pancreatic calcification, inheritance, painless, acute exacerbation of, or surgery for had no impact on survival. The most frequent causes of death were digestive diseases (19.5%), malignancies (19.5%), and cardiovascular diseases (11.3%). CONCLUSIONS: Danish patients with definite had a 4-fold higher mortality rate compared with the background population and a higher mortality rate than patients with probable. Being nonemployed or underweight had significant impact on survival. Keywords: Chronic Pancreatitis; Mortality; Prospective Cohort Study; Prognostic Factors. Chronic pancreatitis () is a complex disease with a high risk of complications. It is frequently caused by a high intake of alcohol and smoking, but often the etiology is unknown. 1 5 Knowledge of the natural history, course, and prognosis of the disease is limited, partly because diagnostic criteria have changed over the years, few diagnostic tools are available, and the gold standard biopsy is rarely an option. 6 Furthermore, studies of prognostic factors for patients with often are based on selected patients with complicated disease from tertiary centers 7 1 or are based on retrospective materials. 7,13 Only a few prospective cohort studies of nonselected patients have been published. 14,15 The Copenhagen Pancreatitis Study (S) is a large prospective cohort study of patients in the Copenhagen Municipality admitted with either acute pancreatitis (AP) or fulfilling precise diagnostic criteria. 16 The initial aim was to follow the course of the disease, and to evaluate the accuracy of new diagnostic and therapeutic methods. The S has the advantages of a prospective design, large size, geographic demarcation, and a 30-year follow-up period using record linkage to the Danish health registries. 17 The aims of this study were to investigate the following: (1) the mortality of patients compared with the background population; () whether the clinical presentations of (ie, presence of exocrine or endocrine insufficiency, pancreatic calcification, acute exacerbation in or painless ) have an impact on the prognosis; (3) the importance of other clinical factors associated with mortality; and (4) the causes of death in this patient population. Materials and Methods The Original Copenhagen Pancreatitis Study Cohort Patients residing in Copenhagen (population of 417,000) and admitted with diagnosed AP or were enrolled consecutively in the study from November 1977 to August 198. Abbreviations used in this paper: AP, acute pancreatitis; CI, confidence interval;, chronic pancreatitis; S, Copenhagen Pancreatitis Study; ER, endoscopic retrograde cholangiopancreaticography; ICD, International Classification of Diseases. 010 by the AGA Institute /10/$36.00 doi: /j.cgh

2 April 010 MORTALITY IN CHRONIC PANCREATITIS 385 Table 1. The S Inclusion Criteria and Distribution of the Cohort by Criteria Probable AP S-0 Acute abdomen with pain in the upper half of the abdomen and S-amylase U/L Definite AP S-1 Acute abdomen with pain in the upper half of the abdomen and S-amylase 600 U/L S- Acute abdomen with pain in the upper half of the abdomen and acute inflamed pancreas at surgery for acute abdomen S-3 S-amylase 600 U/L and acute inflamed pancreas at surgery for acute abdomen cohort (n 90) Probable S-4 At least one earlier attack of AP and recurrent pain in the upper half of the abdomen 51 S-5 Diminished exocrine pancreatic secretion (Lundh test): duodenal amylase values 17 U/L S-6 Chronic inflamed pancreas at surgery for acute abdomen or at autopsy 36 Definite S-7 Pain in the upper half of the abdomen and diminished exocrine pancreatic secretion 85 (Lundh test) S-8 Pain in the upper half of the abdomen and pancreatic calcification 50 S-9 Steatorrhea ( 7 g fat/4 h; mean of stool collection for 3 days) and diminished exocrine pancreatic secretion (Lundh test) 66 The inclusion criteria (Table 1) were modified from the 1963 Marseille classification 18 (the internationally recommended diagnostic criteria at that time) and based on a combination of clinical history, pancreatic function tests, and biochemical, pathoanatomic, or radiologic findings except for findings from endoscopic retrograde cholangiopancreaticography (ER). Patients were excluded if the pancreas was macroscopically normal at surgery or if pancreatic cancer was diagnosed during the primary admission. A search of the National Patient Registry was made at the end of the inclusion period for patients admitted with a World Health Organization International Classification of Diseases (ICD), 8th edition code for AP or ( ) to complete the inclusion. The original S cohort comprised 67 patients 16 and was followed up until August 008. Retrospectively, patient hospital records were studied to ensure correct classification of the patients and for collection of data on smoking habits, previous admissions, and detailed descriptions of the ER when available. ER was described using the Cambridge criteria. 19 If pancreatic calcification was present, the patient had, by definition. Thirty patients were excluded secondarily for not fulfilling the selection criteria. Thus, the S cohort comprised 64 patients. Table. The Layer Score: A Score of 4 Points Is Diagnostic of 14 Points Pancreatic calcification Certain 4 Likely Typical histology Certain 4 Likely Exocrine pancreatic insufficiency (duodenal amylase output 17 U/L or stool fat 7 g/d) Pancreatic duct abnormalities on ER using 3 the Cambridge criteria Major clinical criteria: loss of weight 10 kg/1 months or upper abdominal pain or acute attack of AP Diabetes (fasting glucose 140 mg/dl) 1 During 1987 to 1988, patients with definite or probable (n 167) were asked to complete a mailed questionnaire concerning morbidity (admissions, abdominal pain, use of pain killers, development of diabetes); 71% (119 of 167) of patients answered this questionnaire. All patients fulfilling inclusion criteria S-4 to S-9 (Table 1) were assessed using the score developed by Layer et al, 14 referred to as the Layer score in this article (Table ). Patients With Chronic Pancreatitis A total of 90 patients were defined as having either probable or definite. Table 1 shows the distribution of patients by the inclusion criteria. was diagnosed at inclusion in 169 patients, and before inclusion in 117 patients: mean standard deviation years. In 4 patients, the exact date of diagnosis was not available. The Layer score (Table ) 0 frequently is used to diagnose and therefore was chosen as the basis of classification in probable and definite, replacing the original inclusion criteria described in Table 1. Table 3 illustrates the validity of the original S inclusion criteria according to the Layer score. Of the population, 41 had a Layer score of less than 4 (probable ), and 49 patients had a Layer score of 4 points or more (definite ). Classification of Etiologies In the S questionnaire, alcohol intake was divided into the following groups: group A, 0 g alcohol per day; group Table 3. Agreement Between the S Inclusion Criteria and Layer Score S inclusion criteria Layer score 4: probable Layer score 4: definite Agreement between the two kinds of assessments S-4: probable 38/51 13/ % probable S-5: probable 1/ 1/ 50.0% probable S-6: probable 0/36 36/36 0% probable S-7: definite 0/85 85/ % definite S-8: definite 0/50 50/ % definite S-9: definite /66 64/ % definite

3 386 NØJGAARD ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 4 B, 10 to 40 g alcohol per day; group C, 50 g or more of alcohol per day for less than 5 years; and group D, 50 g or more of alcohol per day for more than 5 years. In this context alcoholic was defined as patients with an alcohol consumption of 50 g or more of ethanol per day at inclusion irrespective of duration, and nonalcoholic was defined as patients with an alcohol consumption of less than 50 g ethanol per day. Hereditary was defined as patients with first-order relatives, who had AP or previous to the inclusion date. Idiopathic was defined as patients with an alcohol consumption of less than 50 g ethanol per day combined with no inheritance, no gallstone-induced, and no other etiologies (eg, hypercalcemia). Gallstonerelated AP was defined as patients with AP (S-0 to S-3), and an alcohol intake of less than 50 g ethanol per day combined with one or more of the following findings: gallstones in the biliary duct or gallbladder visualized during ultrasound, computed tomography, ER, cholecystography, surgery, or autopsy. Clinical Data of the Chronic Pancreatitis Cohort At the inclusion and follow-up visits, the following were noted: clinical history, physical signs, treatment, and laboratory test results. Patients suspected of having were subjected to additional examinations, that is, fasting glucose, glucose tolerance test, ultrasound of the pancreas, ER, abdominal radiograph, Lundh test meal with determination of amylase in the duodenal aspirate, and fat excretion in stool, when indicated. When gallstones were suspected, cholecystography, ultrasound, computed tomography, ER, or a combination of these was performed. Of the patients who had been diagnosed years before inclusion, only a few were re-examined for at inclusion. The Danish Registries In August 008, data from the S cohort were linked to the Causes of Death Registry and the National Patient Registry using each patient s unique personal identification number. The Causes of Death Registry contains information from all death certificates in Denmark since From this registry, date of death and cause of death during the follow-up period (November 1977 to August 008) were obtained. The National Patient Registry contains information about patients admitted to nonpsychiatric hospitals in Denmark since From this registry, dates and diagnoses of all admissions and discharges and the diagnoses from surgery during the follow-up period were obtained. The diagnoses from both the Causes of Death Registry and National Patient Registry were coded using the World Health Organization ICD 8th edition from January 1, 1977, to December 31, 1993, and the ICD 10th edition from January 1, Ethics The original S protocol was approved by the local ethics committees at the 5 hospitals in Copenhagen because no formal regional ethics committee was established at that time. The S follow-up period during 1987 to 1988 was approved by the Ethics Committee for Medical Research in Copenhagen (number V /88 b9). The register-based follow-up evaluation was approved by the Regional Committee for Biomedical Research Ethics in Copenhagen in May 008 (number 18483) and by the Danish Data Protection Agency ( ). Statistical Methods Age- and sex-specific mortality rates for the cohort were compared with the mortality in a matched-background population by calculating the standardized mortality ratio. Cox proportional hazard regression was used to test the association between clinical and social prognostic factors, and mortality among patients with. Only patients alive at the age of 35 (n 86) were included in the Cox regression analysis because so few died before that age. In univariate Cox proportional hazards models, patients were followed up from entry in the S until date of death or censoring. Censoring was either because of loss from follow-up, emigration, or the end of the follow-up period. In this study, no patients were lost from follow-up. The underlying time scale was age with delayed entry on inclusion in the S. The Student t test, Mann Whitney test, and chi-square test were used as appropriate to describe the baseline characteristics of the patients. The level of significance was set at 5% (P.05). SPSS (version 17.0; SPSS Inc, Chicago, IL), SAS (version 9.1.3; SAS Institute Inc, Cary, NC), and Stata (version 9.; StataCorp, College Station, TX) software were used. Results Patient Characteristics Chronic pancreatitis population: definite and probable. Table 4 shows the characteristics of the patients with probable (n 41) and definite (n 49) on inclusion in the S. Furthermore, probable patients had a significantly higher intake of alcohol during the days up to admission than patients with definite (7 of 4 vs 14 of 41; P.01), whereas ER investigations (144 of 09 vs 8 of 8; P.0001), and use of general analgesics (both opioids and nonopioids) in the treatment (96 of 10 vs 5 of 41; P.01) were more frequent in the definite patients. During the follow-up period, no significant differences in qualitative variables were found between the groups. Surgical procedures are summarized in Table 5; 74 patients underwent pancreatic surgery before or at inclusion and 33 patients underwent pancreatic surgery after inclusion. Patients with alcoholic chronic pancreatitis compared with nonalcoholic chronic pancreatitis. Table 4 shows the characteristics of alcoholic (n 18) and nonalcoholic (n 155) patients. For 7 patients, alcohol intake information was missing. Both patients with probable and definite were included in the following calculations. Alcoholic patients were divorced more frequently or living alone compared with nonalcoholics: 85 of 13 versus 70 of 143 (P.001). During the follow-up period, nonalcoholics were nonemployed more frequently (including unemployment, early retirement, and retirement) (40 of 83 vs 70 of 101; P.009), and more often were treated for diabetes compared with alcoholics (5 of 35 vs 10 of 38; P.01). These findings were generally the same whether AP was defined as more than 10 g alcohol per day or more than 50 g alcohol per day for more than 5 years.

4 April 010 MORTALITY IN CHRONIC PANCREATITIS 387 Table 4. Patient Characteristics on Inclusion in the S Definite a Probable a ( 50 g alcohol/d) Alcoholic Nonalcoholic ( 50 g alcohol/d) Number of patients b Mean age, y (SD) 51.0 (1.4) c 46.4 (13.6) 46.6 (10.4) c 53. (13.6) Sex, males/females 179/70 30/11 113/15 c 9/63 Daily intake of alcohol 50 g 143/49 (57.4%) 3/41 (56.1%) Tobacco g/d, mean (SD) 16.5 (9.9) 15.1 (15.) 18.4 (10.5) c 14.6 (10.9) Body mass index, mean (SD) 1.5 (3.7).9 (4.7) 1.6 (4.1) 1.7 (3.7) Prothrombin index (normal range, ), mean (SD) 1.18 (0.6) c 1.0 (0.30) 1.01 (0.31) 1.07 (0.30) S-albumin (normal range, mol/l), mean (SD) (106.0) c (89.0) (104.9) c (10.) Calcification 75/49 (30.1%) c 0 30/18 (3.4%) 4/155 (7.1%) Abdominal pain 174/49 (69.9%) 9/41 (70.1%) 95/18 (74.%) 105/155 (67.7%) Exocrine insufficiency 150/49 (60.%) c 3/41 (7.3%) 67/18 (5.3%) 84/155 (54.%) Diabetes 46/49 (18.5%) 4/41 (9.8%) 0/18 (15.6%) 8/155 (18.1%) Use of opioids 43/49 (17.3%) 8/41 (19.5%) 3/18 (18.0%) 17/155 (11.0%) AP in 49/49 (19.7%) 4/41 (9.8%) 8/18 (1.9%) 4/155 (15.5%) Pancreatic surgery before or at inclusion 74/49 (9.7%) c 1/41 (.4%) d 30/18 (3.4%) 43/155 (7.7%) Layer score, mean (SD) 7. (.8) c.1 (0.3) 6.3 (3.0) 6.6 (3.3) NOTE. Values represent number of patients (percentage) unless otherwise indicated. SD, standard deviation. a Defined using the Layer score and not the original inclusion criteria. b For 7 patients, alcohol intake data were missing. c P.05 with the Student t test, Mann Whitney test, or chi-square test, as appropriate, comparing definite with probable or alcoholic compared with nonalcoholic. d This patient had drainage of a pseudocyst because of previous AP without signs of. Etiology was presumed to be caused by alcohol in 18 patients (44.1%), inheritance in 16 patients (5.5%), hypercalcemia and hyperparathyroidism in 1 patient (0.3%), previous gallstone-associated AP in 1 patient (0.3%), and the remaining 144 patients (49.8%) were classified as idiopathic. Ninety-one of 49 (36.5%) patients with definite had been hospitalized previously with an attack of AP. Survival and Prognostic Factors The standardized mortality ratio for patients with definite was increased by a factor of 4.3 to 4.5 compared with the background population, whereas the mortality rate for patients with probable was.1 to.8 times the expected mortality rate in the background population (Table 6). Cox regression analysis confirmed that the probable group had a significantly better survival rate than the definite group (P Table 5. Surgical Procedures in the Patient Population Surgical procedures Number of procedures before S or at inclusion, n 81 in 74 individuals Number of procedures during follow-up evaluation, n 41 in 33 individuals Surgeries used for AP in Peritoneal drainage with/without necrosectomy Surgeries used for Subtotal pancreatectomy (including 8 Whipple) Pancreaticogastrostomy or 1 19 pancreaticojejunostomy Combinations/others 5 6 Surgeries used for complications Drainage of pseudocyst 18 4 of AP/ Drainage of pancreatic abscess 1 Biliary bypass 11 7 Gastroenteroanastomosis 3 Combinations/others 3 Others Explorative laparotomy 16 1 Underwent surgery for other reasons (cholecystectomy) if was found preoperative

5 388 NØJGAARD ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 4 Table 6. Standardized Mortality Ratio: A Comparison Between the Cohort and the Danish Normal Population 36 Probable Definite Sex Observed death Expected death Standardized mortality ratio 95% CI Male Female Male Female ; 95% confidence interval [CI], ; Figure 1). Therefore, probable patients were excluded from the following survival analysis. Cox regression analysis showed a significantly better survival rate for definite patients who were employed compared with patients without a job (P.015; 95% CI, ), which was even more pronounced by exclusion of patients older than 60 years at entry (P.0003; 95% CI, ). Definite patients who were overweight (body mass index, 5) had a significantly better survival rate than underweight patients (body mass index, 0) (P.016; 95% CI, ). The analysis was repeated on the 169 patients who were diagnosed at inclusion as having identical significant prognostic factors as mentioned earlier (data not shown). The following factors had no influence on survival in definite patients (P.05): sex, alcohol, smoking, single/co-living, exocrine insufficiency, diabetes, pancreatic calcification, inheritance, painless, acute exacerbation in, and surgery for. Smoking habits were not registered at inclusion but could be retrieved in 133 of 147 accessible patient records. Causes of Death During the 30-year follow-up period, 66 patients (91.7%) in the cohort died. Table 7 summarizes the cause of death for these patients. Most patients died of digestive diseases (19.5%), malignancies (19.5%), and cardiovascular diseases (11.3%); 0.8% committed suicide, and 10 patients (3.8%) died of pancreatic cancer. Cause-specific analyses of patients with definite who died of digestive diseases showed abdominal pain Figure 1. Kaplan Meier survival curves for patients older than 35 years of age with probable and definite (P.003). Table 7. Causes of Death for the Patient Population Causes of death Number of deaths in the population Percentage of deaths in the total Danish population in 006 Digestive diseases 5 (19.5%) 5.3 Chronic pancreatitis 1 (7.9%) Alcoholic liver diseases including 4 (9.0%) 1.6 cirrhosis Malignancy 5 (19.5%) 8.3 Pancreatic cancer 10 (3.8%) 1.5 Cardiovascular diseases 30 (11.3%) 19.1 Diabetes 0 (7.5%).3 Respiratory diseases 19 (7.1%) 9.6 Senile decay including stroke and 17 (6.3%) dementia Mental illness (not dementia) 16 (6.0%) Accident 13 (4.9%) 3.6 Infectious diseases (including 7 (.6%) 1.4 tuberculosis) Suicide (0.8%) 1. Diseases in the urinary tract or 4 (1.5%) 1.9 gynecologic diseases Other not clearly defined causes 31 (11.7%) 3. Patients died within the past 3 (1.1%) years a All 66 55,13 a The Danish National Health Service has not yet received information on the cause of death for 3 patients. (P.009; 95% CI, ), and non-employment for patients younger than 60 years of age (P.018; 95% CI, ) as significant prognostic factors. For patients with definite who died of cardiovascular diseases, there was a trend toward inheritance of as a prognostic factor (P.054; 95% CI, ). For patients who died of malignancies, no factors were identified as significant. Discussion Mortality, Prognostic Factors, and Causes of Death In a retrospective study, Levy et al 13 observed a higher mortality rate in patients compared with a matched French population. Miyake et al 1 found a higher mortality rate in alcoholic patients compared with an age- and sex-matched population but no difference in nonalcoholic patients. Overall, we found a - to 3-fold higher mortality rate in probable patients, and a 4-fold higher mortality rate in definite patients compared with the mortality rate in a standardized background population. For definite patients, a Cox regression analysis showed that non-employment and low body mass index were the only factors having a significant impact on mortality. Alcohol and smoking had no individual prognostic influence on mortality; this is surprising because both have been shown to be important etiologic factors for the development of. 1,11,0, 5 The strength of this statement may be hampered, however, because smoking habits were not registered in the original S questionnaire. At the time the S was initiated, smoking habits were not thought to be important,

6 April 010 MORTALITY IN CHRONIC PANCREATITIS 389 but we compensated for this by adding retrospective information from patient records. Some of the patients in this study were diagnosed with several years before inclusion, and they may have changed their lifestyle (eg, reduced their alcohol intake 3 ), thereby confounding the prognostic analysis. However, the lack of prognostic information for alcohol intake, when exclusively analyzing patients who were included in the study on diagnosis of, confirms that alcohol had no prognostic influence in this study. Non-employment as a prognostic factor is well known in social medicine 6 and in other diseases. 7 Lankisch et al 15 described the socioeconomic factors in patients; they found an increase in unemployment and retirement after diagnosis, but the association with mortality has not been explored previously. In patients younger than 60 years at entry, unemployment led to a significantly higher mortality rate that could not be explained by any other factor such as sex, alcohol consumption, smoking habits, presence of pain, exocrine insufficiency, endocrine insufficiency, or pancreatic surgery. Non-employment in these patients may be caused by the severity and duration of their disease, and thereby absence from work; it also may be attributable, however, to other individual risk factors such as level of education, level of income, or psychiatric comorbidity, 6 factors that were not recorded in this study. A higher mortality rate observed in underweight patients (body mass index, 0) hypothetically may be caused by the severity of the disease. In light of our data, the clinical recommendations on the management of these patients would be to focus on social support and nutritional treatment, as well as alcohol, although it is not possible to estimate a low limit of daily alcohol intake protecting against or a high limit leading to. 0 In this subcohort, 5% had surgery for before or at inclusion, but we found no significant difference in survival between patients who underwent pancreatic surgery compared with patients who did not undergo surgery. We observed a high frequency of liver diseases including cirrhosis. This is in agreement with the findings of Apte and Wilson, 8 who found a frequency of synchronous cirrhosis and alcoholic of 40% to 50% at autopsy, and with Pitchumoni et al, 9 who found that 9% of alcoholics with cirrhosis also had diffuse pancreatic fibrosis. However, Aparisi et al 30 and Ammann et al 11 found a rare coincidence of alcoholic and cirrhosis. In 198, Thorsgaard et al 7 found a high frequency of suicide in patients ( of 6; 7.7%). Also, Mullhaupt et al 10 described a high incidence of suicide, especially in idiopathic juvenile (14.3%), whereas in alcoholic the frequency was.%. In this study, the frequency of suicide was surprisingly low (0.8%), and actually lower than in the Danish nonmatched background population (1985,.4%; 006, 1.%; source: Danish National Health Service, public statistics, 006). This may be explained partly by being less severe in this cohort compared with the selected patient cohorts. 7,10 It remains to be established whether or not suicide is associated with because patients with alcohol as an etiology may have a significantly higher suicide rate. 31 The incidence of pancreatic cancer in this population (3.8%) was almost equal to the findings in the multicenter study by Lowenfels et al 3 in 1993 (56 of 015;.8%), of which the present data were a part, and higher than that observed by Pedrazzoli et al 33 (1.%) and Talamini et al 5 (1.9%). Pancreatic cancer is also a more frequent cause of death in patients compared with the Danish background population (1.5%; source: Danish National Health Service, public statistics, 006). This confirms the carcinogenic effect of, whereas the difference between the studies can be explained by the difference in follow-up time. Moreover, we could not confirm the high incidence of death caused by extrapancreatic cancer, previously found by other studies. 7,1,34 In other large cohorts of the percentage of idiopathic was lower (Ammann et al, 11 4%; Lankisch et al, 35 8%) and the frequency of initial AP was lower (Ammann et al, 8 8%). These differences in frequency might be caused by national variation or misclassified patients because of undisclosed alcoholic patients in the idiopathic group. Strengths and Limitations of the Study The S represents a long follow-up evaluation of patients to date and the follow-up evaluation was complete. Patients were included prospectively and well-characterized clinically. We compared the prognosis of the cohort with the general population. It would have been interesting to compare our patient population with a matched population with respect to all other parameters apart from developing pancreatitis. The present trial was not planned with inclusion of a matched patient for every included patient with pancreatitis and we found no other matched Danish populations apart from the one presented in the article. The S inclusion criteria from 1977 did not include ER findings, but were based primarily on clinical data and thereby were more practically useful compared with any other classification system at that time. The Layer score was applied in the present study to validate the diagnostic accuracy and to classify probable or definite ; 5% of patients with an inclusion criterion of S-4 and all patients with S-6 would have been misclassified as having probable if the Layer score had not been used (Table 3). Conclusions Patients with definite had a 4-fold higher mortality rate than the background population and patients with a suspicion of had twice the mortality rate compared with the background population. Unlike alcohol and smoking, both non-employment and being underweight had a significant impact on survival in these patients. In the future, more attention should be given to social support and nutritional treatment in patients. References 1. Maisonneuve P, Lowenfels AB, Müllhaupt B, et al. Cigarette smoking accelerates progression of alcoholic chronic pancreatitis. Gut 005;54: Lowenfels AB, Maisonneuve P, Cavallini G, et al. Prognosis of chronic pancreatitis: an international multicenter study. International Pancreatitis Study Group. Am J Gastroenterol 1994;89: Levy P, Mathurin P, Roqueplo A, et al. A multidimensional casecontrol study of dietary, alcohol, and tobacco habits in alcoholic men with chronic pancreatitis. Pancreas 1995;10: Lin Y, Tamakoshi A, Hayakawa T, et al. Associations of alcohol drinking and nutrient intake with chronic pancreatitis: findings from a case-control study in Japan. Am J Gastroenterol 001;96: 6 67.

7 390 NØJGAARD ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No Dufour MC, Adamson MD. The epidemiology of alcohol-induced pancreatitis. Pancreas 003;7: Etemad B, Whitcomb D. Chronic pancreatitis: diagnosis, classification, and new genetic developments. Gastroenterology 001; 10: Thorsgaard PN, Nyboe AB, Pedersen G, et al. Chronic pancreatitis in Copenhagen. A retrospective study of 64 consecutive patients. Scand J Gastroenterol 198;17: Ammann RW, Hammer B, Fumagalli I. Chronic pancreatitis in Zurich, Clinical findings and follow-up studies of 10 cases. Digestion 1973;9: Andersen B, Pedersen N, Scheel J, et al. Incidence of alcoholic chronic pancreatitis in Copenhagen. Scand J Gastroenterol 198; 17: Mullhaupt B, Truninger K, Ammann R. Impact of etiology on the painful early stage of chronic pancreatitis: a long-term prospective study. Z Gastroenterol 005;43: Ammann RW, Akovbiantz A, Largiader F, et al. Course and outcome of chronic pancreatitis. Longitudinal study of a mixed medical-surgical series of 45 patients. Gastroenterology 1984;86: Cavallini G, Frulloni L, Pederzoli P, et al. Long-term follow-up of patients with chronic pancreatitis in Italy. Scand J Gastroenterol 1998;33: Levy P, Milan C, Pignon J, et al. Mortality factors associated with chronic pancreatitis. Unidimensional and multidimensional analysis of a medical-surgical series of 40 patients. Gastroenterology 1989;96: Layer P, Yamamoto H, Kalthoff L, et al. The different courses of early- and late-onset idiopathic and alcoholic chronic pancreatitis. Gastroenterology 1994;107: Lankisch P, Lohr-Happe A, Otto J, et al. Natural course in chronic pancreatitis. Pain, exocrine and endocrine pancreatic insufficiency and prognosis of the disease. Digestion 1993;54: Copenhagen pancreatitis study. An interim report from a prospective epidemiological multicentre study. Scand J Gastroenterol 1981;16: Andersen TF, Madsen M, Jørgensen J, et al. The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull 1999;46: Sarles H. Pancreatitis symposium, Marseilles Basel: S. Karger, Axon A, Classen M, Cotton P, et al. Pancreatography in chronic pancreatitis: international definitions. Gut 1984;5: Lankisch M, Imoto M, Layer P, et al. The effect of small amounts of alcohol on the clinical course of chronic pancreatitis. Mayo Clin Proc 001;76: Miyake H, Harada H, Ochi K, et al. Prognosis and prognostic factors in chronic pancreatitis. Dig Dis Sci 1989;34: Bourliere M, Barthet M, Berthezene P, et al. Is tobacco a risk factor for chronic pancreatitis and alcoholic cirrhosis? Gut 1991; 3: Cavallini G, Talamini G, Vaona B, et al. Effect of alcohol and smoking on pancreatic lithogenesis in the course of chronic pancreatitis. Pancreas 1994;9: Imoto M, Dimagno E. Cigarette smoking increases the risk of pancreatic calcification in late-onset but not early-onset idiopathic chronic pancreatitis. Pancreas 000;1: Talamini G, Bassi C, Falconi M, et al. Alcohol and smoking as risk factors in chronic pancreatitis and pancreatic cancer. Dig Dis Sci 1999;44: Lundin A, Lundberg I, Hallsten L, et al. Unemployment and mortality a longitudinal prospective study on selection and causation in 49,31 Swedish middle aged men. J Epidemiol Community Health 010;64: Carlsen K, Høybye MT, Dalton SO, et al. Social inequality and incidence of and survival from breast cancer in a populationbased study in Denmark, Eur J Cancer 008;44: Apte M, Wilson J. Alcohol-induced pancreatic injury. Best Pract Res Clin Gastroenterol 003;17: Pitchumoni CS, Glasser M, Saran RM, et al. Pancreatic fibrosis in chronic alcoholics and nonalcoholics without clinical pancreatitis. Am J Gastroenterol 1984;79: Aparisi L, Sabater L, Del-Olmo J, et al. Does an association exist between chronic pancreatitis and liver cirrhosis in alcoholic subjects? World J Gastroenterol 008;14: Flensborg-Madsen T, Knop J, Mortensen EL, et al. Alcohol use disorders increase the risk of completed suicide irrespective of other psychiatric disorders. A longitudinal cohort study. Psychiatry Res 009;167: Lowenfels A, Maisonneuve P, Cavallini G, et al. Pancreatitis and the risk of pancreatic cancer. International Pancreatitis Study Group. N Engl J Med 1993;38: Pedrazzoli S, Pasquali C, Guzzinati S, et al. Survival rates and cause of death in 174 patients with chronic pancreatitis. J Gastrointest Surg 008;1: Hansen T, Laursen M, Christensen E, et al. Chronic pancreatitis and extrapancreatic cancer: a retrospective study among 181 patients with chronic pancreatitis. Int J Pancreatol 1995;18: Lankisch PG, Assmus C, Maisonneuve P, et al. Epidemiology of pancreatic diseases in Lüneburg County. A study in a defined German population. Pancreatology 00;: Juul S. An introduction to Stata for health researchers. Oxan, UK: Taylor & Francis, 008. Reprint requests Address requests for reprints to: Camilla Nøjgaard, MD, Hvidovre Hospital, Kettegård Allé 30, 650 Hvidovre, Denmark. mille@dadlnet.dk; fax: (45) Conflicts of interest The authors disclose no conflicts. Funding This study was supported by the Hvidovre Hospital Research Foundation, Solvay Pharma, the YKL-40 Foundation, and the Foundation of 1870.