Customize your JAMA Network experience by selecting one or more topics from the list below.
Murphy MJ, Sheng X, MacDonald TM, Wei L. Hypertriglyceridemia and Acute Pancreatitis. JAMA Intern Med. 2013;173(2):162–164. doi:10.1001/2013.jamainternmed.477
Author Affiliations: Department of Biochemical Medicine (Dr Murphy) and Medicines Monitoring Unit (Drs Sheng, MacDonald, and Wei), Ninewells Hospital and Medical School, University of Dundee, Dundee, Scotland.
Acute pancreatitis (AP) is common and potentially serious.1 Common causes are gallstones and alcohol abuse; other causes include medications, common bile duct obstruction, trauma, and hypertriglyceridemia.2 Although the association between hypertriglyceridemia and AP is well established, estimates of risk are based on case series and studies of high-risk groups.3-5 The risk of AP from hypertriglyceridemia in the general population is not well characterized. We report results from a cohort study using record-linkage methods to estimate the risk and relative burden of AP in patients with differing degrees of hypertriglyceridemia.
We used databases in the Medicines Monitoring Unit of our institution that included dispensed community prescriptions, hospital discharge data, regional laboratory data, and other data. The study population consisted of residents of Tayside, Scotland, who were registered with a primary care physician between 1993 and 2007 and remained resident in Tayside or died during the study period. Study subjects had at least 1 serum triglyceride measurement between 1993 and 2007, entered the study on the date of their highest triglyceride measurement during the study period, and were categorized by triglyceride concentration into 1 of the following 3 cohorts: 149 mg/dL or lower (group 1); 150 to 499 mg/dL (group 2); or 500 mg/dL or higher (group 3). (To convert triglycerides to millimoles per liter, multiply by 0.0113.)
The primary study outcome was incident AP during follow-up (hospitalization with a primary diagnosis of AP or serum amylase activity of 300 U/L or higher during follow-up [reference interval, 0-100 U/L]). Data were summarized as mean (SD), or number of subjects (percentage).
A Cox regression model was constructed to adjust for potential confounders; data were expressed as hazard ratios (HRs) with 95% CIs. Covariates were age at study entry; sex; socioeconomic status; concentrations of total and high-density lipoprotein (HDL) cholesterol; comorbidities of gallstones, other biliary disease, diabetes, alcohol-related liver disease, alcoholic cirrhosis, alcoholic hepatitis,6 alcohol hospitalization, chronic pancreatitis, and renal failure; and use during follow-up of gastrointestinal drugs, diuretics, lipid-regulating drugs, analgesics, sodium valproate, antibacterial drugs, corticosteroids, estrogens and hormone therapy, and musculoskeletal and joint disease drugs. The Scottish Index of Multiple Deprivation7 was used as a measure of socioeconomic status. Population-attributable risks (PARs) were calculated for each triglyceride group and other AP risk factors. Sensitivity analyses were performed by (1) excluding subjects with hospitalization for gallstones, chronic pancreatitis, renal failure, alcohol morbidities, other biliary disease, and not adjusting for concentrations of total and HDL cholesterol; (2) using the date of first, or lowest, measured triglyceride concentration as the entry date; and (3) using average triglyceride concentration during follow-up for categorization.
There were 31 740 subjects in group 1, 31 887 in group 2, 3642 in group 3. There were 116 AP events in group 1, 178 in group 2, and 37 in group 3 during the 127 473, 143 495, and 14 935 person-years of follow-up, respectively. Crude incidence of AP was 0.91 per 1000 person-years (95% CI, 0.76-1.09) in group 1, 1.24 (95% CI, 1.07-1.44) in group 2, and 2.48 (95% CI, 1.79-3.42) in group 3. The proportion of incident AP cases exposed to moderate hypertriglyceridemia (150-499 mg/dL) was 0.54; for severe hypertriglyceridemia (≥500 mg/dL), it was 0.11. Corresponding PARs were 18.37% and 7.74%.
The Table lists results of univariate and multivariate Cox regression analysis for incident AP admission. There was a significant dose-response relationship between triglyceride concentration and incident AP (adjusted HR, 1.04 [95% CI, 1.02-1.05]). Compared with group 1, the adjusted HR for AP was 1.50 (95% CI, 1.14-1.97) for group 2 and 3.20 (95% CI, 1.99-5.16) for group 3. People who were older, had gallstones, renal failure, chronic pancreatitis, or other biliary disease had a higher risk of incident AP. Use of statins, antibacterial drugs, diuretics, or musculoskeletal joint drugs was associated with a lower risk. The risk of incident AP increased by 4% for every 100-mg/dL increase in triglyceride concentration (after adjustment for covariates and removal of patients hospitalized for gallstones, chronic pancreatitis, alcohol-related morbidities, renal failure, and other biliary disease). The HR for AP associated with severe hypertriglyceridemia (≥500 mg/dL) was higher than the HR associated with moderate hypertriglyceridemia (150-499 mg/dL). A much greater proportion of AP cases was exposed to moderate rather than severe hypertriglyceridemia, explaining why the PAR of AP attributable to moderate hypertriglyceridemia was 18.37% compared with 7.74%.
This study was population-based, with long follow-up (15 years) and low migration. We validated the 82% of cases where records were retrievable and confirmed the accuracy of diagnosis in 95.3%. We adjusted for an extensive range of confounders and performed a wide range of sensitivity analyses to test the robustness of the relationship between triglyceride concentration and incident AP. However, unmeasured confounders may have influenced the results. Nevertheless, observational studies are the only realistic approach to study this association given the low incidence of AP (about 1 per 1000 person-years of follow-up). Our findings that statins were associated with reduced risk of incident AP are consistent with the results of a recent meta-analysis of 28 randomized controlled trials of lipid-modifying drugs.8
Correspondence: Dr Murphy, Department of Biochemical Medicine, Ninewells Hospital and Medical School, Dundee DD1 9SY, Scotland (email@example.com).
Accepted for Publication: August 6, 2012.
Published Online: November 26, 2012. doi:10.1001/2013.jamainternmed.477
Author Contributions: Dr Murphy had full access to all of the data in the study and takes responsibility for the data in the study and the accuracy of the data analysis. Study concept and design: Murphy and MacDonald. Acquisition of data: MacDonald and Wei. Analysis and interpretation of data: Murphy, Sheng, MacDonald, and Wei. Drafting of the manuscript: Murphy, Sheng, and Wei. Critical revision of the manuscript for important intellectual content: Murphy, MacDonald, and Wei. Statistical analysis: Sheng and Wei. Study supervision: MacDonald and Wei.
Conflict of Interest Disclosures: None reported.
Additional Information: All authors have completed the Unified Competing Interest form (available on request from the corresponding author).