Cholecystectomy Statin Use and Risk of Gallstone Disease Followed By

Statin Use and Risk of Gallstone Disease Followed by Cholecystectomy

Michael Bodmer; Yolanda B. Brauchli; Stephan Krähenbühl; et al. Online article and related content current as of November 11, 2010. JAMA. 2009;302(18):2001-2007 (doi:10.1001/jama.2009.1601)

http://jama.ama-assn.org/cgi/content/full/302/18/2001

Supplementary material eFigure and eTables http://jama.ama-assn.org/cgi/content/full/302/18/2001/DC1 Correction Contact me if this article is corrected. Citations This article has been cited 4 times. Contact me when this article is cited. Topic collections Nutritional and Metabolic Disorders; Lipids and Lipid Disorders; Surgery; Surgical Interventions; Hepatobiliary Surgery; Drug Therapy; Drug Therapy, Other Contact me when new articles are published in these topic areas. Related Letters Use of Statins and Gallstone Risk Ching-Sheng Hsu et al. JAMA. 2010;303(12):1146. Tetsuji Fujita. JAMA. 2010;303(12):1146.

Subscribe Email Alerts http://jama.com/subscribe http://jamaarchives.com/alerts

Permissions Reprints/E-prints [email protected] [email protected] http://pubs.ama-assn.org/misc/permissions.dtl

Downloaded from www.jama.com by guest on November 11, 2010 ORIGINAL CONTRIBUTION

Statin Use and Risk of Gallstone Disease Followed by Cholecystectomy

Michael Bodmer, MD, MSc Context Gallstone disease is a leading cause of morbidity in western countries and Yolanda B. Brauchli, PhD, MSc carries a high economic burden. Statins decrease hepatic cholesterol biosynthesis and Stephan Krähenbühl, MD, PhD may therefore lower the risk of cholesterol gallstones by reducing the cholesterol concentration in the bile. Data on this association in humans are scarce. Susan S. Jick, DSc Objective To study the association between the use of statins, fibrates, or other lipid- Christoph R. Meier, PhD, MSc lowering agents and the risk of incident gallstone disease followed by cholecystectomy. PPROXIMATELY 10% TO 20% Design, Setting, and Participants Case-control analysis using the UK-based Gen- of white adults in developed eral Practice Research Database. Incident patients between 1994 and 2008 and 4 con- countries have gallstones, and trols per each patient were identified and matched on age, sex, general practice, cal- gallstone disease is a leading endar time, and years of history in the database. The study population was 76% women causeA of gastrointestinal morbidity and and the mean (SD) age was 53.4 (15.0) years at the index date. Conditional logistic inpatient admission in western coun- regression was used to estimate the odds ratio (OR) of developing gallstones fol- tries.1,2 Colicky pain and complica- lowed by cholecystectomy in relation to exposure to lipid-lowering agents, stratified tions such as cholecystitis, choledo- by exposure timing and duration. The ORs and 95% confidence intervals (CIs) were adjusted for smoking, body mass index, ischemic heart disease, stroke, and estrogen cholithiasis, or pancreatitis occur in a use. substantial portion of gallstone carri- ers.3,4 Gallstone disease represents a se- Main Outcome Measure The adjusted OR (AOR) for developing gallstone dis- rious burden for health care systems ease followed by cholecystectomy in relation to exposure to lipid-lowering agents. worldwide; more than 700 000 chole- Results A total of 27 035 patients with cholecystectomy and 106 531 matched con- cystectomies are performed annually in trols were identified, including 2396 patients and 8868 controls who had statin use. the United States.5 Compared with nonuse, current statin use (last prescription recorded within 90 days before the first-time diagnosis of the disease) was 1.0% for patients and 0.8% for Gallstones are classified as either cho- controls (AOR, 1.10; 95% CI, 0.95-1.27) for 1 to 4 prescriptions; 2.6% vs 2.4% (AOR, lesterol (80%-90%) or pigment stones 0.85; 95% CI, 0.77-0.93) for 5 to 19 prescriptions, and 3.2% vs 3.7% (AOR, 0.64; (10%-20%). Cholesterol stones are 95% CI, 0.59-0.70) for 20 or more prescriptions. The AORs for current use of statins formed on the basis of cholesterol- defined as 20 or more prescriptions were similar (around 0.6) across age, sex, and body supersaturated bile,3,5 whereas pig- mass index categories, and across the statin class. ment stones consist mainly of polymer- Conclusion Long-term use of statins was associated with a decreased risk of gall- ized calcium bilirubinate.3 Risk factors stones followed by cholecystectomy. such as age, female sex, obesity, high- JAMA. 2009;302(18):2001-2007 www.jama.com carbohydrate and high-fat diet, estrogen-containing contraceptive use, and postmenopausal estrogen therapy pre- stone formation.7,8 In humans, some au- Author Affiliations: Division of Clinical Pharmacol- ogy and Toxicology (Drs Bodmer and Krähenbühl) and dispose to cholesterol gallstone forma- thors reported diminished gallstone for- Hospital Pharmacy (Dr Meier), University Hospital, 1,3-5 tion. mation, decreased cholesterol content Basel, Switzerland; Basel Pharmacoepidemiology Unit, Statins decrease hepatic cholesterol in bile, or gallstone dissolution,9-12 Division of Clinical Pharmacy and Epidemiology, De- partment of Pharmaceutical Sciences, University of biosynthesis and may thereby de- whereas others did not find evidence Basel, Basel, Switzerland (Drs Brauchli and Meier); and crease biliary cholesterol secretion,6 for such an effect.13-15 The authors of a Boston Collaborative Drug Surveillance Program, Bos- ton University School of Medicine, Lexington, Mas- consequently leading to diminished recent observational study reported a sachusetts (Drs Jick and Meier). cholesterol concentration in bile. Stud- slightly decreased frequency of chole- Corresponding Author: Christoph R. Meier, PhD, MSc, Basel Pharmacoepidemiology Unit, Hospital Phar- ies in animals have reported beneficial cystectomy in women with self- macy, University Hospital Basel, Spitalstrasse 26, effects of statins in preventing gall- reported long-term statin therapy.16 CH-4031 Basel, Switzerland ([email protected]).

Downloaded from www.jama.com by guest on November 11, 2010 STATIN USE, GALLSTONE DISEASE RISK, AND CHOLECYSTECTOMY

Two small studies reported conflict- demographics and characteristics (eg, Controls ing results; one found no association be- age, sex, height, weight, smoking sta- From the base population, 4 control pa- tween statin use and gallstone risk,17 tus), symptoms, medical diagnoses, re- tients were identified at random for each and the other reported a risk reduc- ferrals to consultants, and hospitaliza- patient with cholecystectomy, matched tion with limited statistical power.18 tions. Because physicians generate drug on calendar time (same index date), age Fibrates affect lipid metabolism prescriptions directly with the com- (same year of birth), sex, general prac- mainly by acting as synthetic ligands of puter using a coded drug dictionary, all tice, and number of years of active his- the peroxisome proliferator–activated recorded prescriptions include the tory in the GPRD prior to the index receptor ␣. Activation of peroxisome name of the preparation, route of ad- date. Controls had no recording of gall- proliferator–activated receptor ␣ af- ministration, dose of a single unit, num- stones, clinical complications sugges- fects numerous metabolic pathways in- ber of units prescribed, and intake regi- tive of gallstones, or cholecystectomy. cluding glucose and lipid metabolism, men (in most instances). The database The same exclusion criteria were ap- reducing level of plasma triglycerides has been described in detail else- plied to patients and controls. and glucose concentration, and increas- where,22-25 has been validated exten- ing high-density lipoprotein choles- sively,26-28 and has been the source for Exposure to Statins and Other terol levels. Studies in humans dem- numerous epidemiological studies pub- Lipid-Lowering Agents onstrated that short-term fibrate lished in peer-reviewed journals. From the computer record, exposure treatment is associated with decreased The study protocol was approved by was assessed to statins, fibrates, or other bile acid synthesis and biliary excre- the Independent Scientific Advisory lipid-lowering agents (anion-exchanger tion.19,20 In a randomized controlled Committee for Medicines and Health- resin, probucol, acipimox, niacin, fish trial, a higher incidence of cholecys- care Products Regulatory Agency Da- oil, or omega fatty acids) prior to the tectomy was noted in patients taking fi- tabase for Research. Informed consent index date for patients and controls. Pa- brates,21 and another study reported a by patients was not needed for this da- tients were classified as currently tak- higher prevalence of gallstones in pa- tabase study. ing medications if the last prescrip- tients taking fibrates.17 tion was recorded less than 90 days, or Therefore, currently available clini- Case Definition and Ascertainment as formerly taking medications if the last cal studies on the possible effect of stat- Patients aged 20 years or older with a prescription was recorded 90 or more ins on gallstone formation and the risk first-time diagnosis of gallstone dis- days prior to the index date. Medica- of cholecystectomy are limited by either ease or cholecystectomy between 1994 tion use was classified by duration of small sample size, lack of long-term fol- and 2008 were identified based on the use prior to the index date based on the low-up, sex restriction, or methodologi- Oxford Medical Information System number of prescriptions (statins: short- cal drawbacks. We conducted a large and Read codes. All patients were re- term, 1-4; medium-term, 5-19; or long- long-term observational study to ex- quired to have both a recorded diag- term, Ն20; fibrates and other lipid- plore the association between statin use nosis for gallstone disease or a compli- lowering drugs: short- to medium- and the risk of developing an incident cation thereof (such as bile duct term, 1-9; long-term, Ն10). Statin use diagnosis of gallstone disease fol- obstruction, cholecystitis, or cholan- was also classified according to the tab- lowed by cholecystectomy. gitis) and a record of cholecystectomy let dose. Finally, duration and timing within 2 years of the diagnosis or a cho- of use were combined into 1 exposure METHODS lecystectomy only. The date of the first- variable. A case-control analysis was con- time diagnosis of gallstone disease or ducted using the UK-based General the date of cholecystectomy in pa- Statistical Analysis Practice Research Database (GPRD). tients without a recording of a gall- Conditional logistic regression analy- The UK-based GPRD was established stone diagnosis is referred to as the in- ses were conducted using SAS statisti- around 1987 and encompasses data on dex date. Patients with less than 3 years cal software version 9.1 (SAS Institute approximately 5 million patients who of active history in the database prior Inc, Cary, North Carolina) to calcu- are registered with selected general to the index date, as well as those with late relative risk estimates as odds ra- practitioners. The general practition- a history of alcohol or drug abuse, can- tios (ORs) with 95% confidence inter- ers provide anonymized medical infor- cer (except non–melanoma skin can- vals (CIs) at a 2-sided P value of .05. mation for research purposes, which are cer), or human immunodeficiency vi- With an assumed prevalence of expo- recorded in a standard manner. Pa- rus prior to the index date were sure to statins of about 1%, the re- tients enrolled in the GPRD are repre- excluded. quired sample size was about 19 000 pa- sentative of the United Kingdom with A total of 300 potential patient pro- tients (and 4 times as many controls) regard to age, sex, geographic distri- files (selected at random) were re- to detect an OR of 0.75 at the signifi- bution, and annual turnover rate. The viewed to verify the validity of the pa- cance level of .05 with a power of information recorded includes patient tient selection criteria. greater than 90%.

Downloaded from www.jama.com by guest on November 11, 2010 STATIN USE, GALLSTONE DISEASE RISK, AND CHOLECYSTECTOMY

The potential confounders age, sex, 10 mg of simvastatin became available lower the risk of developing gall- general practice, calendar time, and over-the-counter in August 2004.29 In stones followed by cholecystectomy, years of recorded history in the data- the fifth model, current long-term statin and because many, but not all statin pre- base were controlled for by matching use was compared with current short- scriptions, are issued for 28 days. and the ORs were further adjusted for term statin use to explore the associa- smoking status (none, current, former, tion of interest in a population diag- RESULTS or unknown) and body mass index nosed as having hypercholesterolemia; A total of 27 035 patients (n=9602 cho- (BMI, which was calculated as weight this analysis likely excluded patients lecystectomy only; n=17 433 cholecys- in kilograms divided by height in me- with normal cholesterol levels who may tectomy after a gallstone diagnosis or as- ters squared) (BMI categories: Ͻ18.5, have a different gallstone risk than pa- sociated complications) and 106 531 18.5-24.9, 25.0-29.9, 30-34.9, Ն35, or tients with hypercholesterolemia. The matched controls were identified unknown) in the multivariate model. final sensitivity analysis was con- (TABLE 1). The study population was The risk estimates were adjusted for a ducted taking another proxy for dura- 76% women and the mean (SD) age was history of ischemic heart disease, is- tion of statin use—the total number of 53.4 (15.0) years at the index date. The chemic stroke, transient ischemic statin tablets prescribed prior to the in- OR of developing an incident gallstone attack, and use of opposed or unop- dex date, which equals the total num- diagnosis with cholecystectomy sub- posed estrogens (current or past use ber of exposure days. This analysis was stantially increased with increasing BMI. [defined as last prescription Յ180 days performed to estimate more precisely In the study population, 11 264 before the index date or thereafter] of how much exposure time it takes to (2396 patients and 8868 controls) were 1-9 or Ն10 prescriptions). Further test- ing included whether other diseases (eg, liver cirrhosis, renal diseases, gastro- Table 1. Characteristics of Patients With Gallstone Disease Followed by Cholecystectomy and Controls intestinal tract ulcer, constipation, uri- No. (%) OR (95% CI) nary dysfunction, hypertension, hypo- Patients Controls thyroidism, inflammatory bowel a b disease, or diabetes mellitus) or other (n = 27 035) (n = 106 531) Crude Adjusted Age, mean (SD), y drugs (eg, thiazides, contraceptives) Ͻ40 5544 (20.5) 21 930 (20.5) confounded the main association of in- 40-59 11 619 (43.0) 46 143 (43.3) terest. Because this was not the case, Ն60 9872 (36.5) 38 458 (36.2) these parameters were not included in Sex the final model. Male 6379 (23.6) 25 068 (23.5) Various sensitivity analyses were con- Female 20 656 (76.4) 81 463 (76.5) ducted. In the first model, only pa- Body mass indexc 12.0-18.4 171 (0.6) 1760 (1.7) 0.58 (0.50-0.68) 0.58 (0.50-0.68) tients with abdominal pain recorded 18.5-24.9 6378 (23.6) 39 208 (36.8) 1 [Reference] 1 [Reference] within 90 days before the index date 25.0-29.9 8628 (31.9) 28 783 (27.0) 1.89 (1.82-1.96) 1.89 (1.82-1.96) were included to reduce the likeli- 30.0-34.9 4886 (18.1) 11 101 (10.4) 2.81 (2.69-2.94) 2.85 (2.72-2.97) hood of detection bias (ie, including pa- 35.0-60.0 2758 (10.2) 5281 (5.0) 3.32 (3.15-3.51) 3.47 (3.28-3.66) tients who had gallstones detected by Unknown 4214 (15.6) 20 398 (19.1) 1.21 (1.15-1.26) 1.36 (1.29-1.43) chance). In the second model, pa- Smoking tients with cholecystectomy in the ab- None 14 978 (55.4) 57 533 (54.0) 1 [Reference] 1 [Reference] sence of a recorded gallstone diagno- Current 5321 (19.7) 21 608 (20.3) 0.94 (0.91-0.97) 0.98 (0.95-1.02) sis were excluded. In the third model, Past 4481 (16.6) 15 825 (14.9) 1.09 (1.05-1.14) 1.02 (0.98-1.06) the index dates were shifted for all pa- Unknown 2255 (8.3) 11 565 (10.9) 0.68 (0.64-0.72) 0.77 (0.72-0.82) tients and their controls by 30, 90, and Ischemic heart disease 2129 (7.9) 6901 (6.5) 1.26 (1.20-1.33) 1.13 (1.07-1.19) 180 days backward in time because gall- Stroke or transient 611 (2.3) 2539 (2.4) 0.93 (0.85-1.02) 0.90 (0.82-0.99) stones may have been present prior to ischemic attack No. of prescriptions the actual first-time recording of the di- Opposed estrogens agnosis; this was done to rule out po- 1-9 555 (2.1) 1728 (1.6) 1.38 (1.25-1.52) 1.55 (1.40-1.72) tential protopathic bias (ie, the possi- Ն10 909 (3.4) 2445 (2.3) 1.64 (1.51-1.78) 2.00 (1.84-2.18) bility that general practitioners may Unopposed estrogens have stopped statin treatment in pa- 1-9 570 (2.1) 1267 (1.2) 2.05 (1.85-2.27) 2.07 (1.86-2.30) Ն tients due to abdominal pain). In the 10 1245 (4.6) 2565 (2.4) 2.30 (2.14-2.47) 2.55 (2.36-2.75) Abbreviations: CI, confidence interval; OR, odds ratio. fourth model, an analysis was re- aAdjusted for age, sex, general practice, and calendar time by matching. stricted to patients and controls whose bFurther adjusted for the variables listed in this table. cCalculated as weight in kilograms divided by height in meters squared. index date was before 2004 because

Downloaded from www.jama.com by guest on November 11, 2010 STATIN USE, GALLSTONE DISEASE RISK, AND CHOLECYSTECTOMY taking statins, 1514 were taking fi- agents). Compared with nonuse of stat- creased AOR of 1.39 (95% CI, 1.12- brates, and 1038 were taking other ins, the AOR for 1 to 4 current pre- 1.72) for developing gallstones followed lipid-lowering agents. The majority of scriptions was 1.10 (95% CI, 0.95- by cholecystectomy. patients (87%) treated with lipid- 1.27), 0.85 (95% CI, 0.77-0.93) for 5 Because a comparison between pa- lowering agents were taking statins only to 19 current prescriptions, and 0.64 tients with statin use and nonuse not (ie, not in combination [concurrently (95% CI, 0.59-0.70) for 20 or more cur- only compares exposure to statins, but or subsequently] with other lipid- rent prescriptions regardless of timing also patients with hypercholesterol- lowering agents). Compared with the of use (TABLE 2). The findings were emia to patients without hypercholes- reference group of patients who did not similar for men and women and for terolemia, the analysis was restricted to take statins, the adjusted OR (AOR) of those younger than 60 years or for those patients with statin use (ie, patients with developing gallstones with cholecys- aged 60 years or older. The analysis also hypercholesterolemia). Among these tectomy was 0.78 (95% CI, 0.73-0.83) was stratified by individual statin, with patients, long-term statin use was com- for current statin use and 1.19 (95% CI, similar findings for all statins, with little pared with short-term statin use, yield- 1.07-1.32) for past statin use (ad- statistical power for some of the newer ing an AOR of 0.58 (95% CI, 0.50- justed in the multivariate analysis for compounds (TABLE 3). The AORs for 0.68). When this analysis was further age, sex, general practice, calendar time, current statin use of 20 or more pre- stratified by BMI categories, the AOR BMI, smoking history, a history of is- scriptions, stratified by dose, also were associated with long-term statin use chemic heart disease, stroke, transient assessed and the results are displayed compared with short-term use was 0.58 ischemic attack, and use of estrogens, in Table 3. Current long-term use of fi- (95% CI, 0.41-0.81) for normal weight fibrates, and other lipid-lowering brates was associated with a slightly in- (BMI, 20-24.9), 0.63 (95% CI, 0.48- 0.83) for overweight (BMI, 25-29.9), and 0.65 (95% CI, 0.50-0.86) for obese Table 2. Use of Statins and Risk of First-Time Gallstone Disease Followed by patients (BMI Ն30). Cholecystectomy Various sensitivity analyses (as de- No. (%) OR (95% CI) scribed in the “Methods” section) re- Patients Controls lated to statin use remained virtually (n = 27 035) (n = 106 531) Crudea Adjustedb unchanged in the subgroup of pa- Current statin use tients and their controls with abdomi- No 24 639 (91.1) 97 663 (91.7) 1 [Reference] 1 [Reference] nal pain before gallstone diagnosis with Yes 1832 (6.8) 7342 (6.9) 0.99 (0.94-1.05) 0.78 (0.73-0.83) cholecystectomy, in patients with or 1-4 Current prescriptions 277 (1.0) 832 (0.8) 1.32 (1.15-1.52) 1.10 (0.95-1.27) without recorded gallstone disease prior Men 89 (0.3) 312 (0.3) 1.13 (0.89-1.43) 0.99 (0.77-1.26) to the cholecystectomy, for the analy- Women 188 (0.7) 520 (0.5) 1.45 (1.22-1.71) 1.19 (0.99-1.42) ses in which the index dates were Age, y Ͻ60 108 (0.4) 284 (0.3) 1.55 (1.24-1.94) 1.18 (0.93-1.50) shifted by 30, 90, and 180 days, as well Ն60 169 (0.6) 548 (0.5) 1.20 (1.01-1.43) 1.06 (0.89-1.27) as for patients and their controls with 5-19 Current prescriptions 690 (2.6) 2550 (2.4) 1.07 (0.98-1.17) 0.85 (0.77-0.93) an index date before 2004. The analy- Men 256 (1.0) 959 (0.9) 1.07 (0.92-1.23) 0.89 (0.76-1.03) sis with the total number of tablets as Women 434 (1.6) 1591 (1.5) 1.08 (0.97-1.21) 0.84 (0.75-0.94) a proxy for days of exposure sug- Age, y gested that the risk of gallstone dis- Ͻ60 202 (0.7) 705 (0.7) 1.17 (1.00-1.38) 0.81 (0.69-0.96) ease followed by cholecystectomy starts Ն60 488 (1.8) 1845 (1.7) 1.03 (0.93-1.15) 0.88 (0.79-0.98) decreasing after about 1 to 1.5 years of Ն20 Current prescriptions 865 (3.2) 3960 (3.7) 0.86 (0.79-0.93) 0.64 (0.59-0.70) treatment with statins. Additional data Men 377 (1.4) 1636 (1.5) 0.90 (0.80-1.02) 0.70 (0.61-0.81) are available in the eFigure and eTables Women 488 (1.8) 2324 (2.2) 0.83 (0.75-0.92) 0.61 (0.55-0.69) 1 through 7 at http://www.jama.com. Age, y Ͻ60 186 (0.7) 736 (0.7) 1.04 (0.88-1.23) 0.67 (0.55-0.80) COMMENT Ն60 679 (2.5) 3224 (3.0) 0.81 (0.74-0.89) 0.65 (0.59-0.72) This large observational study pro- Past prescriptions 564 (2.1) 1526 (1.4) 1.46 (1.32-1.61) 1.19 (1.07-1.32) No. of prescriptions in past vides evidence that patients with long- 1-4 269 (1.0) 722 (0.7) 1.47 (1.27-1.69) 1.22 (1.05-1.41) term statin use have a reduced risk of 5-19 205 (0.8) 534 (0.5) 1.50 (1.27-1.77) 1.20 (1.01-1.42) gallstone disease followed by cholecys- Ն20 90 (0.3) 270 (0.3) 1.30 (1.02-1.65) 0.97 (0.75-1.25) tectomy compared with patients with- Abbreviations: CI, confidence interval; OR, odds ratio. out statin use. However, the OR was not aAdjusted for age, sex, general practice, and calendar time by matching. bFurther adjusted for body mass index (calculated as weight in kilograms divided by height in meters squared), smoking, decreased for patients with short-term history of ischemic heart disease, stroke, or transient ischemic attack, use of opposed or unopposed estrogens, fi- statin use but started to decrease after brates, and other lipid-lowering agents (anion-exchanger resin, probucol, acipimox, niacin, fish oil, or omega fatty acids). 5 prescriptions, reflecting approxi-

Downloaded from www.jama.com by guest on November 11, 2010 STATIN USE, GALLSTONE DISEASE RISK, AND CHOLECYSTECTOMY mately 1 to 1.5 years of treatment. The Table 3. Long-term Use of Statins and Risk of First-Time Gallstone Disease Followed by risk estimate was consistent across age Cholecystectomy and sex groups. Adjustment for impor- No. (%) OR (95% CI) tant risk factors1,18,30 for gallstone dis- Patients Controls ease did not materially alter the re- (n = 27 035) (n = 106 531) Crudea Adjustedb sults. Diabetes mellitus was not No statin use 24 639 (91.1) 97 663 (91.7) 1 [Reference] 1 [Reference] associated with a material risk alter- Ն20 Current prescriptions ation31 so it was not included in the fi- Atorvastatin 327 (1.2) 1404 (1.3) 0.92 (0.81-1.04) 0.66 (0.58-0.76) nal multivariate model. Ischemic heart 10 or 20 mg 232 (0.9) 990 (0.9) 0.69 (0.59-0.81) disease, ischemic stroke, and tran- 40 or 80 mg 95 (0.4) 414 (0.4) 0.60 (0.47-0.76) sient ischemic attack are important in- Fluvastatin 14 (0.1) 94 (0.1) 0.59 (0.33-1.03) 0.41 (0.23-0.73) dications for statin use, but the find- 20 mg 5 (0.02) 43 (0.04) 0.37 (0.14-0.95) ings were not altered after adjustment 40 or 80 mg 9 (0.03) 51 (0.05) 0.43 (0.21-0.90) for these comorbidities. Pravastatin 62 (0.2) 343 (0.3) 0.71 (0.54-0.94) 0.52 (0.39-0.68) The OR for gallstone disease fol- 10 or 20 mg 34 (0.1) 176 (0.2) 0.58 (0.40-0.85) lowed by cholecystectomy for pa- 40 mg 28 (0.1) 167 (0.2) 0.45 (0.30-0.68) tients with long-term statin use was ap- Rosuvastatin 32 (0.1) 134 (0.1) 0.93 (0.63-1.38) 0.68 (0.46-1.03) proximately 0.6 regardless of the 5 or 10 mg 25 (0.1) 97 (0.1) 0.73 (0.46-1.16) comparison group chosen (ie, [pre- 20 or 40 mg 7 (0.03) 37 (0.03) 0.54 (0.23-1.25) sumably] patients with normal choles- Simvastatin 422 (1.6) 1959 (1.8) 0.85 (0.76-0.95) 0.65 (0.58-0.73) terol levels or [presumably] patients 10 or 20 mg 277 (1.0) 1209 (1.1) 0.71 (0.62-0.82) with hypercholesterolemia and short- 40 or 80 mg 145 (0.5) 750 (0.7) 0.55 (0.45-0.66) All statins term statin use), suggesting that a pos- Low dose 573 (2.1) 2515 (2.4) 0.69 (0.62-0.77) sible effect of statins on gallstone for- High dose 284 (1.1) 1419 (1.3) 0.55 (0.48-0.63) mation may be largely independent of Abbreviations: CI, confidence interval; OR, odds ratio. the presence of hypercholesterolemia. aAdjusted for age, sex, general practice, and calendar time by matching. bFurther adjusted for body mass index (calculated as weight in kilograms divided by height in meters squared), smoking, In addition, correction of hypercholes- history of ischemic heart disease, stroke, or transient ischemic attack, use of opposed or unopposed estrogens, fi- terolemia in patients with statin use is brates, and other lipid-lowering agents (anion-exchanger resin, probucol, acipimox, niacin, fish oil, or omega fatty acids). achieved within weeks after initiating statin therapy; in comparison, the ob- current fibrate use, which is in accor- tential diagnostic or protopathic bias as served risk reduction of gallstone dis- dance with the literature.17,21 These find- an alternative explanation for the ob- ease followed by cholecystectomy was ings may be somewhat confounded by served association between statin use only evident among patients with long- indication because most patients take and the risk of gallstone disease with term statin use (Ͼ1 to 1.5 years), sug- fibrates to treat high levels of triglyc- cholecystectomy. gesting contributing factors other than erides and/or low levels of high- On the other hand, limitations in- hypercholesterolemia. In previous stud- density lipoprotein cholesterol, which cluded the possibility of some out- ies, total cholesterol and LDL choles- are both associated with gallstone for- come misclassification because origi- terol levels have been shown to be only mation.1,34,35 nal medical records were not used to weakly associated with gallstone for- Our study has several strengths. First, validate the gallstone diagnoses and/or mation, if at all.1,32,33 it is based on a large validated database cholecystectomy. However, in a previ- The observed risk reduction of gall- with documented high data quality and ous GPRD-based study, Gonza´lez- stone disease in patients with long- completeness. Second, it encompassed Pe´rez and Garcı´a-Rodriguez18 used a term statin use suggests a class effect more than 27 000 patients with gall- similar case definition and docu- for all statins, provided that the ob- stone disease followed by cholecystec- mented a high validity of gallstone di- served association is indeed causal. The tomy, yielding substantial statistical agnoses in the GPRD. They contacted stratification of current long-term use power even in stratified subgroup analy- general practitioners of a random of individual statins by dose further sug- ses. There also was long-term follow- sample of 263 patients and found a 90% gested a tendency toward a lower OR up for all patients, which enabled the as- confirmation rate for patients who un- for high-dose compared with low- sessment of long-term statin use, and derwent cholecystectomy and an 82% dose exposure. there were sufficient data for stratifica- confirmation rate for patients who did A substantially increased gallstone tion by individual statins. Third, de- not have cholecystectomy. Concern- risk with cholecystectomy was found tailed information on important comor- ing misclassification, patients in this for patients with high BMIs and for pa- bidities, concomitant drug therapies, and study also were required to have had a tients with estrogen use, as well as a BMI was available. Finally, sensitivity cholecystectomy, a well-defined clini- slightly increased risk for patients with analyses allowed the addressing of a po- cal end point mostly due to gallstone

Downloaded from www.jama.com by guest on November 11, 2010 STATIN USE, GALLSTONE DISEASE RISK, AND CHOLECYSTECTOMY disease. In addition, outcome or expo- because of lack of understanding or lated to this article. We have conducted numerous studies using the UK-based General Practice Research Da- sure misclassification in observational information. Therefore, poor socio- tabase in the past, and we are presently working on studies tends to drive the relative risk economic status could introduce a several projects. Some, but not all of the studies are industry-sponsored. This particular study was not estimates toward the null, unless mis- bias toward a decreased risk of gall- funded. Our group has been conducting research for classification is differentially distrib- stone disease among patients with the pharmaceutical industry for many years, and these uted (eg, preferentially affecting 1 sub- statin use. However, such bias should collaborations involve numerous companies including Pfizer, Novartis, Bristol-Myers Squibb, AstraZen- group of patients), which is unlikely in lead to a reduced gallstone risk for all eca, and Merck Sharp & Dohme. Drs Brauchli, Jick, this context. statin exposure groups, and not only and Meier reported conducting research related to pso- riasis in the past for MerckSerono, Switzerland. Another limitation is that the analy- for those with long-term statin use. In Additional Information: The eFigure and eTables 1 ses were not adjusted for potentially addition, cases and controls were through 7 are available at http://www.jama.com. relevant lifestyle parameters altering matched on general practice that may, the risk of gallstone disease such as to some degree, control for socioeco- REFERENCES 36,37 physical activity, high-fat or high- nomic status because patients from 1. Shaffer EA. Gallstone disease: epidemiology of carbohydrate diet,1 intake of fruits the same neighborhood tend to con- gallbladder stone disease. Best Pract Res Clin 38 Gastroenterol. 2006;20(6):981-996. and vegetables, or coffee consump- sult the same general practitioner. 2. Sylwestrowicz TA, Shaffer EA. Gallbladder func- 39 tion. Use of thiazide diuretics was Although a large number of con- tion during gallstone dissolution: effect of bile acid associated with a moderately in- founding factors were considered and therapy in patients with gallstones. Gastroenterology. 30 1988;95(3):740-748. creased risk of gallbladder disease ; other influences were attempted to be 3. Lammert F, Miquel JF. Gallstone disease: from genes however, inclusion of thiazide diuret- kept to a minimum, a statement about to evidence-based therapy. J Hepatol. 2008;48 (suppl 1):S124-S135. ics in this analysis did not alter the the causality between statins and gall- 4. Portincasa P, Moschetta A, Palasciano G. Choles- main association of interest. Physical stone disease with cholecystectomy terol gallstone disease. Lancet. 2006;368(9531): 230-239. activity was primarily shown to be cannot be made due to the observa- 5. Marschall HU, Einarsson C. Gallstone disease. J In- protective in patients with normal tional nature of the data. tern Med. 2007;261(6):529-542. weight or in those moderately over- In summary, this large observa- 6. Kallien G, Lange K, Stange EF, Scheibner J. The pravastatin-induced decrease of biliary cholesterol se- 36,37 weight, while there was a strong tional study provides evidence that cretion is not directly related to an inhibition of cho- risk reduction associated with statin long-term use of statins is associated lesterol synthesis in humans. Hepatology. 1999; 30(1):14-20. use in obese patients with BMIs of 30 with a decreased risk of developing a 7. Abedin MZ, Narins SC, Park EH, Smith PR, Kirkwood or higher (AOR, 0.65; 95% CI, 0.50- diagnosis of gallstone disease requir- KS. Lovastatin alters biliary lipid composition and dis- solves gallstones: a long-term study in prairie dogs. 0.86) who are the least likely to ing cholecystectomy. Our findings may Dig Dis Sci. 2002;47(10):2192-2210. engage in intense physical activity. A be of clinical relevance given that gall- 8. Davis KG, Wertin TM, Schriver JP. The use of simvastatin for the prevention of gallstones in the lithogenic diet high in fat and carbohydrates stone disease represents a major bur- prairie dog model. Obes Surg. 2003;13(6):865-868. with low intake of fruits and veg- den for health care systems. 9. Chapman BA, Burt MJ, Chisholm RJ, Allan RB, Yeo etables and low physical activity36,37 is KH, Ross AG. Dissolution of gallstones with simva- 1 Author Contributions: Dr Meier had full access to all statin, an HMG CoA reductase inhibitor. Dig Dis Sci. markedly associated with obesity and of the data in the study and takes responsibility for 1998;43(2):349-353. an increased risk of having gallstones. the integrity of the data and the accuracy of the data 10. Porsch-Ozc¸ürümez M, Hardt PD, Schnell-Kretschmer analysis. Drs Bodmer and Brauchli contributed equally H, et al. Effects of fluvastatin on biliary lipids in subjects Because statin use is more frequent in to this article. with an elevated cholesterol saturation index. Eur J Clin overweight and obese patients (75% Study concept and design: Bodmer, Brauchli, Meier. Pharmacol. 2001;56(12):873-879. Acquisition of data: Bodmer, Brauchli, Jick, Meier. 11. Smith JL, Roach PD, Wittenberg LN, et al. Ef- of all patients with statin use in the Analysis and interpretation of data: Bodmer, Brauchli, fects of simvastatin on hepatic cholesterol metabo- study population had a BMI Ն25), a Krähenbühl, Meier. lism, bile lithogenicity and bile acid hydrophobicity in higher gallstone risk would be Drafting of the manuscript: Bodmer, Brauchli. patients with gallstones. J Gastroenterol Hepatol. 2000; Critical revision of the manuscript for important in- 15(8):871-879. expected to be associated with statin tellectual content: Krähenbühl, Jick, Meier. 12. Wilson IR, Hurrell MA, Pattinson NR, Chapman use. However, the opposite was true, Statistical analysis: Brauchli. BA. The effect of simvastatin and bezafibrate on bile Administrative, technical, or material support: composition and gall-bladder emptying in female non– suggesting that statins also are protec- Krähenbühl, Jick, Meier. insulin-dependent diabetics. J Gastroenterol Hepatol. tive in overweight and obese patients. Study supervision: Krähenbühl, Meier. 1994;9(5):447-451. Financial Disclosures: The Division of Clinical Phar- 13. Miettinen TE, Kiviluoto T, Taavitsainen M, Vuoristo Furthermore, there was no control macology and Toxicology, University Hospital Basel M, Miettinen TA. Cholesterol metabolism and serum for socioeconomic status. In theory, (Drs Bodmer and Krähenbühl), Basel Pharmacoepi- and biliary noncholesterol sterols in gallstone pa- demiology Unit, University of Basel (Drs Brauchli and tients during simvastatin and ursodeoxycholic acid patients with low socioeconomic sta- Meier), and the Boston Collaborative Drug Surveil- treatments. Hepatology. 1998;27(3):649-655. tus may have a higher gallstone risk lance Program (Drs Jick and Meier) reported having 14. Sharma BC, Agarwal DK, Baijal SS, Saraswat VA. due to obesity, poor diet habits, and a research collaboration with various companies in the Pravastatin has no effect on bile lipid composition, pharmaceutical industry. Regarding drug companies nucleation time, and gallbladder motility in persons lack of physical activity. Although producing and selling statins, the Basel Pharmacoepi- with normal levels of cholesterol. J Clin Gastroenterol. socioeconomic status was not found demiology Unit as well as the Boston Collaborative 1997;25(2):433-436. Drug Surveillance Program had or currently have a re- 15. Smit JW, van Erpecum KJ, Renooij W, et al. The to be linked to statin prescribing in search collaboration with AstraZeneca and Novartis. effects of the 3-hydroxy, 3-methylglutaryl coen- England,40 it may be possible that The Division of Clinical Pharmacology and Toxicol- zyme A reductase inhibitor pravastatin on bile com- ogy at the University Hospital Basel currently re- position and nucleation of cholesterol crystals in cho- patients with poor socioeconomic sta- ceives an unconditional grant from AstraZeneca. All lesterol gallstone disease. Hepatology. 1995;21 tus are less likely to take statins of these funded research collaborations are unre- (6):1523-1529.

Downloaded from www.jama.com by guest on November 11, 2010 STATIN USE, GALLSTONE DISEASE RISK, AND CHOLECYSTECTOMY

16. Tsai CJ, Leitzmann MF, Willett WC, Giovannucci 23. Jick H. A database worth saving. Lancet. 1997; 32. Petitti DB, Friedman GD, Klatsky AL. Association EL. Statin use and the risk of cholecystectomy in 350(9084):1045-1046. of a history of gallbladder disease with a reduced con- women. Gastroenterology. 2009;136(5):1593- 24. Walley T, Mantgani A. The UK General Practice centration of high-density-lipoprotein cholesterol. 1600. Research Database. Lancet. 1997;350(9084):1097- N Engl J Med. 1981;304(23):1396-1398. 17. Caroli-Bosc FX, Le Gall P, Pugliese P, et al; Gen- 1099. 33. Thijs C, Knipschild P, Brombacher P. Serum lipids eral Practitioners’ Group of Vidauban. Role of 25. Wood L, Martinez C. The general practice re- and gallstones: a case-control study. Gastroenterology. fibrates and HMG-CoA reductase inhibitors in search database: role in pharmacovigilance. Drug Saf. 1990;99(3):843-849. gallstone formation: epidemiological study in an un- 2004;27(12):871-881. 34. Fruchart JC, Brewer HB Jr, Leitersdorf E; Fibrate selected population. Dig Dis Sci. 2001;46(3):540- 26. Jick H, Jick SS, Derby LE. Validation of informa- Consensus Group. Consensus for the use of fibrates 544. tion recorded on general practitioner–based comput- in the treatment of dyslipoproteinemia and coronary 18. Gonza´lez-Pe´rez A, Garcı´a-Rodriguez LA. Gall- erised data resource in the United Kingdom. BMJ. 1991; heart disease. Am J Cardiol. 1998;81(7):912-917. bladder disease in the general population: associa- 302(6779):766-768. 35. Sun H, Tang H, Jiang S, et al. Gender and meta- tion with cardiovascular morbidity and therapy. 27. Jick H, Terris B, Derby LE. Further validation of bolic differences of gallstone diseases. World J Pharmacoepidemiol Drug Saf. 2007;16(5):524- information recorded on a general practitioner Gastroenterol. 2009;15(15):1886-1891. 531. based computerized data resource in the United 36. Leitzmann MF, Giovannucci EL, Rimm EB, et al. 19. Post SM, Duez H, Gervois PP, Staels B, Kuipers Kingdom. Pharmacoepidemiol Drug Saf. 1992; The relation of physical activity to risk for sympto- F, Princen HM. Fibrates suppress bile acid synthesis 1:347-349. matic gallstone disease in men. Ann Intern Med. 1998; via peroxisome proliferator-activated receptor-alpha- 28. Jick SS, Kaye JA, Vasilakis-Scaramozza C, et al. 128(6):417-425. mediated downregulation of cholesterol 7 alpha- Validity of the general practice research database. 37. Leitzmann MF, Rimm EB, Willett WC, et al. Rec- hydroxylase and sterol 27-hydroxylase expression. Ar- Pharmacotherapy. 2003;23(5):686-689. reational physical activity and the risk of cholecystec- terioscler Thromb Vasc Biol. 2001;21(11):1840- 29. Filion KB, Delaney JA, Brophy JM, Ernst P, Suissa tomy in women. N Engl J Med. 1999;341(11): 1845. S. The impact of over-the-counter simvastatin on the 777-784. 20. Sta˚hlberg D, Reihner E, Rudling M, Berglund L, number of statin prescriptions in the United King- 38. Tsai CJ, Leitzmann MF, Willett WC, Giovannucci Einarsson K, Angelin B. Influence of bezafibrate on he- dom: a view from the General Practice Research EL. Fruit and vegetable consumption and risk of cho- patic cholesterol metabolism in gallstone patients: re- Database. Pharmacoepidemiol Drug Saf. 2007; lecystectomy in women. Am J Med. 2006;119 duced activity of cholesterol 7 alpha-hydroxylase. 16(1):1-4. (9):760-767. Hepatology. 1995;21(4):1025-1030. 30. Leitzmann MF, Tsai CJ, Stampfer MJ, Willett WC, 39. Leitzmann MF, Willett WC, Rimm EB, et al. A pro- 21. Davidson MH, Armani A, McKenney JM, Jacobson Giovannucci E. Thiazide diuretics and the risk of gall- spective study of coffee consumption and the risk of TA. Safety considerations with fibrate therapy. Am J bladder disease requiring surgery in women. Arch In- symptomatic gallstone disease in men. JAMA. 1999; Cardiol. 2007;99(6A):3C-18C. tern Med. 2005;165(5):567-573. 281(22):2106-2112. 22. Garcı´a-RodriguezLA, Perez-Gutthann S. Use of 31. Pagliarulo M, Fornari F, Fraquelli M, et al. Gall- 40. Reid FD, Cook DG, Whincup PH. Use of statins the UK General Practice Research Database for stone disease and related risk factors in a large co- in the secondary prevention of coronary heart dis- pharmacoepidemiology. Br J Clin Pharmacol. 1998; hort of diabetic patients. Dig Liver Dis. 2004;36 ease: is treatment equitable? Heart. 2002;88(1): 45(5):419-425. (2):130-134. 15-19.

The new individual must work out the whole prob- lem of science, letters and theology for himself; can owe his fathers nothing. —Ralph Waldo Emerson (1803-1882)

Downloaded from www.jama.com by guest on November 11, 2010