Combinations of statins and fibrates may be increasingly prescribed to achieve lipid goals in high-risk patients and those with other cardiovascular risk factors, such as mixed dyslipidemia. The purpose of this retrospective cohort study was to compare rates of hospitalization for specific diagnoses in a cohort of new users of statins or fibrates, using claims data from a large United States health insurer. New users of statin, fibrate, or statin-fibrate therapy from 2004 to 2007 were identified; followed for hospitalization with rhabdomyolysis, renal impairment, hepatic injury, or pancreatitis; and confirmed by medical record review. Incidence rates (IRs) were compared across categories of fibrate or statin use, with adjusted IR ratios estimated using Poisson regression. A total of 584,784 patients initiated statins or fibrates. The IR of rhabdomyolysis in statins was 3.30 per 100,000 patient-years; the adjusted IR ratio for statin-fenofibrate combinations compared to statins alone was 3.75 (95% confidence interval 1.23 to 11.40). The IRs of renal impairment and pancreatitis in statins were 108.87 per 100,000 patient-years and 45.76 per 100,000 patient-years, respectively; the adjusted IR ratios for statin-fenofibrate combinations compared to statins alone were 1.47 (95% confidence interval 1.12 to 1.93) and 2.87 (95% confidence interval 2.05 to 4.02), respectively. The IR of hepatic injury with statins was 8.57 per 100,000 patient-years, with no risk difference between exposure groups. In conclusion, the risk for rhabdomyolysis was low, although higher in patients newly treated with statin-fibrate concurrent therapy than those treated with either as monotherapy. The risk for pancreatitis was higher in patients treated with fenofibrate, whether in combination with statins or alone.
The purpose of this retrospective pharmacoepidemiologic study was to compare the incidence of hospitalizations with rhabdomyolysis, myopathy, renal impairment, hepatic injury, and pancreatitis during periods of concurrent exposure to fenofibrate or gemfibrozil and statins with the incidence during periods of exposure to statins alone in a large United States insured population with real-world treatment patterns.
Methods
This was a retrospective cohort study using claims data from the Normative Health Informatics database, a large, multistate, geographically diverse population of managed care plan enrollees. The Normative Health Informatics database contains pharmacy data and data on medical services, procedures, and their accompanying diagnoses for >60 million current and past members from January 1993 through December 2008.
We constructed an inception cohort of new users of statins (atorvastatin, fluvastatin, lovastatin, pravastatin, rosuvastatin, or simvastatin), fibrates (fenofibrate or gemfibrozil), or both during the study period of January 1, 2004, to June 30, 2007. The date of the first dispensing during the study period with no dispensing of that drug during the previous 6 months (baseline period) was defined as the date of cohort entry (index date). Patients were required to be continuously enrolled for ≥6 months before the first dispensing, be ≥18 years of age, and have complete medical and pharmacy benefits in a commercial insurance plan. Statin initiators may have had baseline histories of fibrate use, and fibrate initiators may have had baseline histories of statin use.
Members of the cohort with any of the claims-based outcome diagnoses during the 6-month baseline period, or patients who had ever received cerivastatin or clofibrate, were excluded from the study.
Each day of follow-up was classified according to current exposure status to statins and fibrates on the basis of the date of dispensing and the days supplied for each medication. Days falling between the date of dispensing and the date of dispensing plus the number of days supply plus 15 days, inclusive, were considered as currently exposed to that drug. The classification of current statin exposure was made irrespective of the specific statin dispensed or dose of statin used. Follow-up began on the day after the index date and continued through the earliest of the date of health plan disenrollment, the date of each specific outcome, or December 31, 2007.
The outcomes of interest for this study were the occurrence of any hospitalizations with rhabdomyolysis, myopathy, renal impairment, hepatic injury, or pancreatitis that were identified in the claims according to International Classification of Diseases, Ninth Revision, and Current Procedural Terminology procedure codes, listed in Table 1 , and confirmed by medical record review. The first hospitalization for each outcome between cohort entry and disenrollment or the end of the study period while exposed to statins and/or fibrates as determined by pharmacy claims was identified, and medical records were requested to confirm the outcome. Deidentified clinical information from the medical records was reviewed by clinical consultants, blinded as to the patients’ statin and fibrate drug exposure status, to adjudicate the outcomes, using the working definitions listed in Table 2 . We created an additional outcome: renal failure requiring renal replacement based on the subset of the confirmed renal impairment cases in which the patients also had claims codes for dialysis or renal transplantation up to 7 days before the confirmed renal impairment date or any time thereafter.
| Coding System | Code | Description |
|---|---|---|
| Rhabdomyolysis or myopathy | ||
| ICD-9 | 728.88 | Rhabdomyolysis |
| ICD-9 | 728.89 | Muscle/ligament disease not elsewhere classified |
| Renal impairment | ||
| ICD-9 | 584.x | Acute renal failure |
| ICD-9 | 586.x | Renal failure, unspecified |
| ICD-9 Proc | 39.95 | Hemodialysis |
| ICD-9 Proc | 54.98 | Peritoneal dialysis |
| CPT | 90918–90925 | End-stage renal disease services |
| CPT | 90935–90940 | Hemodialysis procedures |
| CPT | 90945–90999 | Miscellaneous dialysis procedures |
| Hepatic injury | ||
| ICD-9 | 570.x | Acute and subacute necrosis of liver |
| ICD-9 | 572.2 | Hepatic coma |
| ICD-9 | 572.4 | Hepatorenal syndrome |
| ICD-9 | 573.3 | Hepatitis, unspecified |
| ICD-9 | 573.4 | Hepatic infarction |
| ICD-9 | 573.8 | Other specified disorders of liver |
| ICD-9 | 573.9 | Unspecified disorder of liver |
| Pancreatitis | ||
| ICD-9 | 577.0 | Acute pancreatitis |
| ICD-9 | 577.x | Diseases of the pancreas |
| CPT | 48000 | Placement of drains, peripancreatic, for acute pancreatitis |
| CPT | 48001 | Placement of drains, peripancreatic, for acute pancreatitis; with cholecystostomy, gastrostomy, and jejunostomy |
| CPT | 48105 | Resection or debridement of pancreas |
| Condition | Definition ⁎ |
|---|---|
| Rhabdomyolysis † | (1) CK increase to ≥10 times the laboratory ULN, with concomitant muscle symptoms (e.g., weakness, aching, tenderness) and no obvious acute alternative cause (e.g., burns, crush injury), and (2) creatinine elevation to greater than or equal to the laboratory ULN or new clinical diagnosis of renal insufficiency or renal failure. Required hospitalization for ≥1 of the ICD-9 codes listed in Table 1 . |
| Myopathy † | (1) CK increase to ≥10 times the laboratory ULN, with concomitant muscle symptoms (e.g., weakness, aching, tenderness) and no obvious acute alternative cause (e.g., burns, crush injury), and (2) no creatinine elevation to greater than or equal to the laboratory ULN or new clinical diagnosis of renal insufficiency or renal failure. Required hospitalization for ≥1 of the ICD-9 codes listed in Table 1 . |
| Renal impairment | Abrupt (≤1 month) onset of evidence of kidney damage, that is, (1) if baseline creatinine available, at least a doubling of creatinine to greater than or equal to the laboratory ULN, or (2) if no baseline creatinine available, then creatinine increase to ≥2 times the laboratory ULN. Renal impairment associated with rhabdomyolysis was evaluated as rhabdomyolysis. Required hospitalization for ≥1 of the ICD-9 codes listed in Table 1 . |
| Renal impairment requiring renal replacement | Renal impairment (above) confirmed from the medical records of patients who also had claims for dialysis or renal transplantation within 7 days before the confirmed renal impairment date or any time thereafter. |
| Hepatic injury | Clinical diagnosis consistent with hepatic injury and no obvious acute alternative cause (e.g., infectious, chemical, obstructive, or alcoholic) and with (1) ALT ≥3 times the ULN, (2) bilirubin ≥3 mg/dl, or (3) prolongation of INR of ≥50%. Required hospitalization for ≥1 of the ICD-9 codes listed in Table 1 . |
| Pancreatitis | Clinical diagnosis of pancreatitis with (1) comparable symptoms (e.g., nausea, vomiting, and/or abdominal pain) or radiologic evidence of pancreatitis and (2) lipase ≥2 times the laboratory ULN regardless of amylase or, if no lipase available, amylase ≥2 times the laboratory ULN. Required hospitalization for ≥1 of the ICD-9 codes listed in Table 1 . |
⁎ For cases in which normal ranges for laboratory values were not available, normal laboratory values were assigned using published recommendation.
† Isolated myoglobinuria without concurrent increases in CK (irrespective of creatinine value) was not considered rhabdomyolysis. Myoglobinuria with concurrent increases in CK was assessed by criteria for rhabdomyolysis or myopathy.
The cohort members were categorized according to the initiating drug that qualified the patient for entry into the inception cohort: statin, fibrate, or both statin and fibrate. Information on the following key covariates was obtained from the claims during the 6-month baseline period before cohort entry: age, gender, geographic region, calendar year of cohort entry, and health services utilization variables, including number of visits to primary care physicians and number of co-morbidities, defined as the number of unique International Classification of Diseases, Ninth Revision, 3-digit codes received. Other covariates included history of diabetes diagnosis (International Classification of Diseases, Ninth Revision, code 250) or treatment, history of hypertension diagnosis or treatment, history of biliary disease, history of alcohol abuse, and exposure to contrast dye in the 30-day period before index date.
Confirmed cases were used to estimate the incidence rates (IRs) of each outcome according to patient-time in each study drug exposure category. Unconfirmed cases and potential cases that occurred during “nonexposed” time were censored at the date of the claims diagnosis. IRs were estimated by dividing the number of cases occurring within patient-time with a given drug exposure category by the total amount of patient-time (in patient-years) accrued in that category of drug exposure for the following exposure categories: statin only, fenofibrate only, gemfibrozil only, fenofibrate and statin concurrently, and gemfibrozil and statin concurrently. Because the number of events for some of the outcomes was small, the 95% confidence intervals (CIs) for the IRs were approximated using Byar’s formula.
Unadjusted IR ratios (IRRs) for each outcome were calculated for each exposure category relative to exposure to statins alone. Adjusted IRRs were calculated using Poisson regression to adjust for potentially confounding variables such as age group, gender, history of diabetes, history of hypertension, history of biliary disease, history of alcohol abuse, use of contrast dye, and health services utilization variables. Covariates were retained in the model at p <0.10.
All data analyses were performed using SAS version 9.1 (SAS Institute Inc., Cary, North Carolina). All analyses and reporting were done on appropriately deidentified data and only in aggregate form. Medical record abstraction was performed with permission from the medical directors of the health plans and individual health care providers. Under Health Insurance Portability and Accountability Act privacy regulations, we obtained privacy board approval to access protected health information and review medical records without obtaining individual written consent. We followed our standard operating procedures, which are consistent with the “Guidelines for Good Pharmacoepidemiology Practices.”
Results
We identified 584,784 patients who initiated statins and/or a fibrates during the study period of January 1, 2004, to June 30, 2007, and met all the other eligibility criteria. Of these, 86.9% were statin initiators, 12.5% were fibrate initiators, and 0.6% initiated statins and fibrates on the same day. Table 3 lists the baseline demographic and health care utilization characteristics of the study cohort during the 6-month baseline period by initiation drug. The fibrate initiators and combination initiators were somewhat younger, were more likely to be male, and had a higher proportion with histories of diabetes than the statin initiators. Among the statin initiators, 1.5% (n = 7,484) had a dispensing of fenofibrate, and 0.8% (n = 4,064) had a dispensing of gemfibrozil during the 6-month baseline period. Among the fibrate initiators, 23.0% (n = 16,877) had a dispensing of a statin during the baseline period.
| Variable | Statin Initiators | Fibrate Initiators | Statin and Fibrate Initiators | Total |
|---|---|---|---|---|
| (n = 507,932 [86.9%]) | (n = 73,337 [12.5%]) | (n = 3,515 [0.6%]) | (n = 584,784 [100.0%]) | |
| Age group (years) | ||||
| 18–40 | 64,210 (12.6%) | 13,780 (18.8%) | 604 (17.2%) | 78,594 (13.4%) |
| 41–50 | 145,469 (28.6%) | 23,030 (31.4%) | 1,202 (34.2%) | 169,701 (29.0%) |
| 51–60 | 185,702 (36.6%) | 23,554 (32.1%) | 1,172 (33.3%) | 210,428 (36.0%) |
| 61–70 | 84,228 (16.6%) | 10,278 (14.0%) | 447 (12.7%) | 94,953 (16.2%) |
| ≥71 | 28,323 (5.6%) | 2,695 (3.7%) | 90 (2.6%) | 31,108 (5.3%) |
| Gender | ||||
| Female | 224,839 (44.3%) | 23,510 (32.1%) | 929 (26.4%) | 249,278 (42.6%) |
| Male | 283,093 (55.7%) | 49,827 (67.9%) | 2,586 (73.6%) | 335,506 (57.4%) |
| Census region | ||||
| Midwest | 126,510 (24.9%) | 17,318 (23.6%) | 808 (23.0%) | 144,636 (24.7%) |
| Northeast | 59,811 (11.8%) | 6,615 (9.0%) | 287 (8.2%) | 66,713 (11.4%) |
| South | 247,259 (48.7%) | 39,217 (53.5%) | 1,984 (56.4%) | 288,460 (49.3%) |
| West | 74,352 (14.6%) | 10,187 (13.9%) | 436 (12.4%) | 84,975 (14.5%) |
| Year of index date | ||||
| 2004 | 148,849 (29.3%) | 21,219 (28.9%) | 926 (26.3%) | 170,994 (29.2%) |
| 2005 | 147,870 (29.1%) | 21,722 (29.6%) | 1,028 (29.3%) | 170,620 (29.2%) |
| 2006 | 143,771 (28.3%) | 21,257 (29.0%) | 1,067 (30.4%) | 166,095 (28.4%) |
| 2007 | 67,442 (13.3%) | 9,139 (12.5%) | 494 (14.1%) | 77,075 (13.2%) |
| Medical history | ||||
| Diabetes | 103,522 (20.4%) | 18,797 (25.6%) | 1,083 (30.8%) | 123,402 (21.1%) |
| Hypertension | 278,597 (54.9%) | 43,147 (58.8%) | 2,112 (60.1%) | 323,856 (55.4%) |
| Biliary disease | 2,002 (0.4%) | 322 (0.4%) | 14 (0.4%) | 2,338 (0.4%) |
| Alcohol abuse | 1,287 (0.3%) | 189 (0.3%) | 14 (0.4%) | 1,490 (0.3%) |
| Contrast dye | 19,527 (3.8%) | 758 (1.0%) | 227 (6.5%) | 20,512 (3.5%) |
| Lipid-lowering medication use | ||||
| Statin | 0 (0.0%) | 16,877 (23.0%) | 0 (0.0%) | 16,877 (2.9%) |
| Fenofibrate | 7,484 (1.5%) | 0 (0.0%) | 0 (0.0%) | 7,484 (1.3%) |
| Gemfibrozil | 4,064 (0.8%) | 0 (0.0%) | 0 (0.0%) | 4,064 (0.7%) |
| Number of co-morbidities | ||||
| 0–7 | 287,873 (56.7%) | 41,669 (56.8%) | 2,074 (59.0%) | 331,616 (56.7%) |
| 8–22 | 208,232 (41.0%) | 30,009 (40.9%) | 1,369 (39.0%) | 239,610 (41.0%) |
| ≥23 | 11,827 (2.3%) | 1,659 (2.3%) | 72 (2.1%) | 13,558 (2.3%) |
| Number of primary care provider visits | ||||
| 0 | 89,020 (17.5%) | 10,601 (14.5%) | 824 (23.4%) | 100,445 (17.2%) |
| 1 | 122,800 (24.2%) | 16,735 (22.8%) | 785 (22.3%) | 140,320 (24.0%) |
| 2 | 124,306 (24.5%) | 19,308 (26.3%) | 769 (21.9%) | 144,383 (24.7%) |
| 3–6 | 151,967 (29.9%) | 23,717 (32.3%) | 982 (27.9%) | 176,666 (30.2%) |
| >6 | 19,839 (3.9%) | 2,976 (4.1%) | 155 (4.4%) | 22,970 (3.9%) |
| Number of specialist visits | ||||
| 0 | 309,759 (61.0%) | 46,038 (62.8%) | 2,171 (61.8%) | 357,968 (61.2%) |
| 1 | 82,268 (16.2%) | 11,549 (15.8%) | 511 (14.5%) | 94,328 (16.1%) |
| 2 | 45,548 (9.0%) | 6,422 (8.8%) | 333 (9.5%) | 52,303 (8.9%) |
| 3–6 | 54,844 (10.8%) | 7,429 (10.1%) | 395 (11.2%) | 62,668 (10.7%) |
| >6 | 15,513 (3.1%) | 1,899 (2.6%) | 105 (3.0%) | 17,517 (3.0%) |
| Number of laboratory tests | ||||
| 0–1 | 119,567 (23.5%) | 12,738 (17.4%) | 979 (27.9%) | 133,284 (22.8%) |
| 2–4 | 125,652 (24.7%) | 19,001 (25.9%) | 753 (21.4%) | 145,406 (24.9%) |
| 5–10 | 170,542 (33.6%) | 26,642 (36.3%) | 1,097 (31.2%) | 198,281 (33.9%) |
| 11–20 | 67,979 (13.4%) | 11,057 (15.1%) | 464 (13.2%) | 79,500 (13.6%) |
| >20 | 24,192 (4.8%) | 3,899 (5.3%) | 222 (6.3%) | 28,313 (4.8%) |
| Number of surgical procedures | ||||
| 0 | 194,032 (38.2%) | 26,370 (36.0%) | 1,649 (46.9%) | 222,051 (38.0%) |
| 1 | 157,798 (31.1%) | 22,310 (30.4%) | 971 (27.6%) | 181,079 (31.0%) |
| 2 | 83,372 (16.4%) | 12,937 (17.6%) | 492 (14.0%) | 96,801 (16.6%) |
| 3–6 | 66,314 (13.1%) | 10,621 (14.5%) | 383 (10.9%) | 77,318 (13.2%) |
| >6 | 6,416 (1.3%) | 1,099 (1.5%) | 20 (0.6%) | 7,535 (1.3%) |
| Number of prescriptions | ||||
| 0–2 | 113,754 (22.4%) | 13,055 (17.8%) | 345 (9.8%) | 127,154 (21.7%) |
| 3–5 | 192,644 (37.9%) | 25,149 (34.3%) | 1,343 (38.2%) | 219,136 (37.5%) |
| 6–11 | 163,490 (32.2%) | 26,796 (36.5%) | 1,439 (40.9%) | 191,725 (32.8%) |
| >11 | 38,044 (7.5%) | 8,337 (11.4%) | 388 (11.0%) | 46,769 (8.0%) |
| Total costs | ||||
| <$750 | 167,870 (33.1%) | 21,567 (29.4%) | 1,115 (31.7%) | 190,552 (32.6%) |
| $750–$1,500 | 111,866 (22.0%) | 16,319 (22.3%) | 823 (23.4%) | 129,008 (22.1%) |
| $1,501–$3,000 | 98,024 (19.3%) | 15,801 (21.6%) | 598 (17.0%) | 114,423 (19.6%) |
| $3,001–$10,000 | 88,814 (17.5%) | 14,930 (20.4%) | 600 (17.1%) | 104,344 (17.8%) |
| >$10,000 | 41,358 (8.1%) | 4,720 (6.4%) | 379 (10.8%) | 46,457 (7.9%) |
More than 80% (n = 484,345) of the cohort were exposed to only statins during the follow-up period, 5.6% (n = 32,769) were exposed to only fenofibrate during the follow-up period, and 1.7% (n = 9,986) were exposed to only gemfibrozil during the follow-up period ( Table 4 ). The remaining 9.9% (n = 57,684) received various combinations of statins, fenofibrate, and gemfibrozil, either sequentially or concurrently. Of those, 46,568 patients were exposed to statins and fibrates (either fenofibrate or gemfibrozil) on the same day during the follow-up period: 36,319 patients received statins and fenofibrate concurrently, 7,967 patients received statins and gemfibrozil concurrently, and 2,282 patients received fenofibrate and gemfibrozil during the follow-up period with concomitant treatment with statins.
| Drug Exposure During Follow-Up ⁎ | Number of Patients | Statins During Baseline | Fenofibrate During Baseline | Gemfibrozil During Baseline |
|---|---|---|---|---|
| Statins only | 484,345 | 0 (0%) | 1,225 (0.3%) | 811 (0.2%) |
| Fenofibrate only | 32,769 | 1,036 (3.2%) | 0 (0%) | 0 (0%) |
| Gemfibrozil only | 9,986 | 308 (3.1%) | 0 (0%) | 0 (0%) |
| Statins and fenofibrate concurrently | 36,319 | 12,438 (34.3%) | 5,911 (16.3%) | 256 (0.7%) |
| Statins and gemfibrozil concurrently | 7,967 | 2,070 (26.0%) | 60 (0.8%) | 2,402 (30.2%) |
| Fenofibrate and gemfibrozil sequentially with concurrent statins | 2,282 | 566 (24.8%) | 141 (6.2%) | 379 (16.6%) |
| Total | 584,784 | 16,877 (2.9%) | 7,484 (1.3%) | 4,064 (0.7%) |
Table 4 lists how many patients had exposure to statins, fenofibrate, or gemfibrozil during the 6-month baseline period by drug exposure during the follow-up period. Less than 1% of the group that received only statins during the follow-up period had a dispensing of a fenofibrate or gemfibrozil during the 6-month baseline period ( Table 4 ). More than 1/3 of those in the concurrent statin and fenofibrate group had a dispensing of a statin during the 6-month baseline period, and 16% had a dispensing of fenofibrate during the 6-month baseline period.
We identified a total of 5,250 claims-based events from 4,886 unique patients using all follow-up time for the entire cohort. Restricting to those events that occurred during “exposed” time, meaning exposure to a statin and/or fibrate on the date of the event, we evaluated 3,116 potential events from 2,930 unique patients. After claims profile review, 3,065 potential events were pursued for medical record review and case adjudication. We successfully retrieved the medical records for 2,136 cases (70%), and 1,027 events (48%) were confirmed.
Including only the patient-time occurring during exposure to our primary exposure groups, 85% of the follow-up time was exposed to statins only, 7% to fenofibrate only, 2% to gemfibrozil only, 5% to statins and fenofibrate, and 1% to statin and gemfibrozil.
The incidence of each outcome event by exposure category is listed in Table 5 . There were 22 confirmed cases of rhabdomyolysis, 11 cases of myopathy, 638 cases of renal impairment, 48 cases of hepatic injury, and 308 cases of pancreatitis. Further restricting the renal impairment events to those requiring dialysis or transplant, we had 146 confirmed cases of renal failure requiring renal replacement therapy. The IR of hospitalized rhabdomyolysis ranged from 2.78 per 100,000 patient-years in patients treated with fenofibrate alone to 20.70 per 100,000 patient-years in patients treated with statins and gemfibrozil, with an IR of 15.00 per 100,000 patient-years in patients treated with statins and fenofibrate. The IRs and 95% CIs for all outcomes are listed in Table 5 .
