An epidemiological study og drug-induced acute pancreatitis utilizing a Swedish case-control network

Sammanfattning: Drugs have rarely been considered as a possible cause of acute pancreatitis in clinical practice. Here, we have investigated various risk factors for this disease, and in particular drugs, within a nation-wide, case-control surveillance network for studies on drug-induced diseases. We have described and evaluated this case-control network and its quality assurance and control procedures and we present the distribution of etiologies of acute pancreatitis in Sweden. We also document the extent to which drugs, diseases, smoking and alcohol are risk factors for acute pancreatitis. This study was a population-based, case-control study encompassing 1.4 million inhabitants aged 20-85 from four areas in Sweden conducted between January 1, 1995 and May 31, 1998. The patients were hospitalized in departments of surgery for their first episode of acute pancreatitis and had no previous history of known bile stone disease. Control subjects were recruited continuously in a random fashion from the general study population. Information was obtained from medical records and through telephone interviews. Overall, there were 2,453 potential patients and 2,245 potential controls among whom 779 and 2,128, respectively, were eligible for inclusion. 462 patients and 1,781 controls were ultimately included in the study. The expert evaluators found evidence for biliary stones in 50% of the patients and for alcohol use in 23%, but in 29% neither of these factors was present. In all, other factors (e.g., drugs) may have contributed to the development of acute pancreatitis in 52% of the cases. A multivariable analysis, defining exposure as any use within 30 days of admission to the hospital, revealed that the adjusted odds ratio (OR) for H 2 antagonists was 2.4 (95% CI 1.2-4.8), for proton pump inhibitors 2.1 (1.2-3.4), for non-steroidal anti-inflammatory drugs derived from acetic acid 2.3 (1.3-4.0) and for antibacterial agents designed for systemic use 1.9 (1.1-3.2). Inflammatory bowel disease exhibited an OR of 3.4 (1.5-7.9), while the corresponding value for other gastrointestinal tract disorders, including ulcer, 1.5 (1.1-1.9). Smoking was significantly associated with acute pancreatitis OR 1.7 (1.2-2.1) and for those smoking more than 20 cigarettes per day, the OR was 4.0 (2.2-7.5). Alcohol in moderate amounts did not reach statistical significance, but for those drinking more than 420g of alcohol each week the OR was 4.1 (2.2-7.5). Six percent of the patients and three percent of the control subjects had prevalent diabetes. Eleven percent of the patients and seven percent of the controls had a BMI >30. The relative risk for a hospital stay exceeding 14 days or for treatment in an intensive care unit was 2.5 (1.1-5.4) for patients with a BMI >30, compared to those with a BMI of 20-25. Among antidiabetic drugs, only the use of glibenclamide increased the risk for acute pancreatitis, demonstrating an adjusted OR of 2.5 (1.1-5.9). Current use of H 2 antagonists and proton pump inhibitors had an adjusted OR of 4.9 (1.6-15.0) and 3.2 (1.4-7.4), respectively. For both of these groups, the OR values were higher for patients taking one DDD/day or more than for those taking lower daily doses, but the confidence limits were broad. We found that presence of gastritis and/or gastro esophageal reflux disease, within the12 months preceding hospitalization had an increased adjusted OR of 1.9 (1.2-3.0). Inflammatory bowel disease was also significantly associated with acute pancreatitis, with an adjusted OR of 5.1 (2.0-13.0) We found variations among the monitors. One deviated significantly from the others. This monitors data were excluded. Seven monitors participated from the start, of this study, 5 of whom interviewed both patients and control subjects, while 2 interviewed controls only. Seven monitors were added during the course of the study, 3 interviewing patients and controls and 4 interviewing only controls. Differences between these groups of monitors had little impact on the observed odds ratios. A pronounced enhancement of the sensitivity of the monitors was observed with time but had little impact on the odds ratios. Intensive training of interviewers in the beginning of a study is of importance for their sensitivity over time.