HIV-1 protease inhibitors (PIs) contribute to hyperlipidemia in persons treated for HIV infection. There are potential drug-drug interactions between PIs and some statins, which are drugs frequently used to treat hyperlipidemia. We performed a retrospective cohort study using the TennCare program to determine prescribing rates of contraindicated combinations of PIs and statins in HIV-infected persons in Tennessee and to assess changes in prescribing after publication of treatment guidelines. Computerized files identified adult patients with antiretroviral prescriptions and overlapping prescriptions for PIs and statins from January 1, 1996 through June 30, 2002. A subset of these combinations was defined as contraindicated based on published guidelines. Changes in patterns of prescribing after publication of preliminary treatment guidelines were examined using a mixed-effects logistic regression model. There were 3448 persons who received PIs during the study period. The proportion of PI users receiving statins increased from 3.5% during January 1996 through December 2000 to 7.9% during January 2001 through June 2002 (P < 0.001). Contraindicated PI-statin use decreased from 42.0% of combinations to 20.8% during the same periods (P < 0.001). Use of PIs and statins has increased in the adult TennCare population. Contraindicated combinations have decreased but remain unacceptably high.
From the *Division of Infectious Diseases, Department of Medicine, Vanderbilt University School of Medicine, Nashville, TN; †Center for Health Services Research, Vanderbilt University School of Medicine, Nashville, TN; ‡Division of Pharmacoepidemiology, Department of Preventive Medicine, Vanderbilt University School of Medicine, Nashville, TN; §Department of Biostatistics, Vanderbilt University School of Medicine, Nashville, TN; ∥Comprehensive Care Center, Nashville, TN; and ¶Geriatric Research, Education, and Clinical Center, Nashville Veterans Administration Medical Center, Nashville, TN.
Received for publication June 3, 2004; accepted November 23, 2004.
Supported in part by a Vanderbilt Clinical Research Scholars award (K12 RR17697), the Agency for Healthcare Research and Quality, Centers for Education and Research on Therapeutics cooperative agreement (grant HS1-0384), and a cooperative agreement with the US Food and Drug Administration (FD-U-001641).
Presented in part at the 41st Annual Infectious Diseases Society of America Meeting, San Diego, CA, October 9-11, 2003.
Reprints: Todd Hulgan, Division of Infectious Diseases, Vanderbilt University School of Medicine, 345 24th Avenue North, Suite 105, Nashville, TN 37203 (e-mail: email@example.com).
HIV-1 protease inhibitors (PIs) as a component of highly active antiretroviral therapy (HAART) have dramatically reduced the morbidity and mortality associated with HIV infection.1 PIs are recommended agents for use in initial treatment regimens as well as in “salvage” regimens after treatment failure.2 These agents can cause dyslipidemia, however, including hyperlipidemia.3 One cohort study reported a 60% rate of hypercholesterolemia in persons receiving PIs.4 The multicenter prospective Swiss HIV Cohort Study found increased rates of hyperlipidemia in patients treated with PIs relative to those receiving non-PI-containing HAART.5 In another study, almost half of 133 patients receiving PIs met intervention criteria by the National Cholesterol Education Program (NCEP) II guidelines.6 Current guidelines advocate treatment of HIV-infected patients according to overall risk profile as outlined in the NCEP until data specific to this population are available.3,7 As a result, many patients are treated concomitantly with PIs and lipid-lowering agents, such as hydroxymethylglutaryl-coenzyme A (CoA) reductase inhibitors (statins).
Because of shared metabolism via the cytochrome P450 3A4 (CYP3A4) system, there are concerns about drug-drug interactions and an increased risk of skeletal muscle toxicity in patients taking PIs and statins in combination. This is a rare but well-recognized adverse effect of statins.8-10 There are case reports of acute myositis and rhabdomyolysis in patients receiving PIs and statins.11-14 Pharmacokinetic and clinical studies in human beings have shown that simvastatin levels are increased substantially in the presence of PIs in HIV-negative volunteers.8,15 Pravastatin, however, is metabolized to a much lesser extent by CYP3A4 and has been found to be relatively safe in combination with PIs.15-17 Based on pharmacokinetic data and limited clinical experience, atorvastatin can also be used with caution in combination with PIs.3,7 Little is known about the pharmacokinetics of other statins in combination with PIs. Fluvastatin is metabolized primarily CYP2C9 and has been studied in combination with PIs in a small number of HIV-infected persons,18,19 with no reports of significant interactions.
Preliminary treatment guidelines published in November 2000 by the Adult AIDS Clinical Trial Group (AACTG) Cardiovascular Disease Focus Group recommended treatment of hyperlipidemia caused by PIs with pravastatin or atorvastatin with careful monitoring.20 Cerivastatin (no longer available in the United States) and fluvastatin were suggested as acceptable alternative agents, and lovastatin and simvastatin were to “be avoided.” In more recent guidelines from the International AIDS Society-USA metabolic complications panel7 as well as from the AACTG and HIV Medicine Association of the Infectious Diseases Society of America,3 pravastatin and atorvastatin remain the recommended first-line statins for treatment of hyperlipidemia associated with PIs. In the 2003 guidelines, lovastatin and simvastatin are considered to be “contraindicated”.3
As the HIV-infected population ages and PIs remain a cornerstone of HAART, the number of patients requiring treatment of dyslipidemia is likely to increase. Despite potentially dangerous drug-drug interactions, there are scant data regarding the prevalence of combined therapy with PIs and statins. We sought to characterize the use of PIs and statins in a large cohort of persons infected with HIV and to identify the frequency of use of contraindicated combinations (eg, simvastatin) in this population before and after preliminary treatment guidelines were published.
This was a retrospective cohort study to examine drug use trends between January 1, 1996 and June 30, 2002. We used the TennCare database to identify prescriptions for antiretroviral medications. The period before the publication of initial HIV hyperlipidemia treatment guidelines20 (January 1996-December 2000) was designated as the “preguideline” study period, and January 2001 until the end of the study was designated as the “postguideline” study period. The objectives of this study were to compare the prevalence of PI use, combination PI-statin use, and contraindicated PI-statin use before versus after publication of guidelines. The Bureau of TennCare, the Tennessee State Institutional Review Board (IRB), and the Vanderbilt University IRB approved the study, and informed subject consent was waived.
The population consisted of TennCare enrollees. TennCare is a modified state Medicaid program authorized in Tennessee by means of a waiver since 1994 that includes employed low-income persons who do not meet federal eligibility.21 An enrollment file indicated each person's periods of enrollment and demographic characteristics. Pharmacy files identified filled prescriptions. These data were used to define the study population and to determine exposure to study drugs. Automated pharmacy records are an excellent source of prescription medication data because these records are not subject to information bias22,23 and have concordance of more than 90% with patient self-reports of medication use.24-26 This database has been used extensively to address issues regarding medication exposure and drug toxicities in the TennCare population.
Persons included in the study were 18 years of age or older, had ≥91 days of continuous TennCare enrollment before and after the first antiretroviral prescription (to ensure availability of claims data), and had at least 2 filled prescriptions for any antiretroviral medication during the study period that were separated by ≥31 days but ≤91 days. We identified the date of the first of these 2 filled prescriptions as the date of entry into the study (t0). Subjects were considered to have exited the study if they lost TennCare enrollment for >30 days, reached the end of the study period, or died.
For the purposes of this study, HIV infection was defined by prescriptions for antiretroviral drugs. Persons receiving antiretroviral drugs for <30 days or receiving monotherapy with lamivudine or tenofovir disoproxil fumarate (presumably for treatment of chronic active hepatitis B) were not included in the study cohort.
Antiretroviral and Statin Use
Antiretroviral and lipid-lowering drugs prescribed in the study population were identified from computerized TennCare pharmacy files, which included information on drug dispensed, dose, units dispensed, and days of supply. Prescription information was collected and categorized by study quarter. Exposure was defined as days of individual drugs and drug classes per quarter. Combinations of antiretrovirals were defined by overlapping days of filled prescriptions within each quarter. For example, a person with any overlapping days of supply of 2 classes of antiretrovirals was considered to have combined use of those drug classes during that quarter. Combined use of PIs and statins was defined as overlapping days of filled prescriptions for those drugs within each quarter.
Use of PIs and statins was categorized as use of any PI in combination with the following statins during a given study quarter: pravastatin or atorvastatin (recommended combination), fluvastatin or cerivastatin (alternative), or simvastatin or lovastatin (contraindicated). These categories were based on published treatment guidelines.3,7,20 Although cerivastatin was withdrawn from the market in 2001 because of reports of rhabdomyolysis, there was no evidence of excessive risk in combination with PIs before its withdrawal, and it was considered an “alternative” statin in the 2000 recommendations.20 There was only 1 episode of prescribing of cerivastatin in combination with a PI during the study period, and this was included as an alternative combination for analysis purposes. Persons could have been considered as having recommended and contraindicated PI-statin use in the same quarter if there were overlapping prescription days of both categories.
We estimated quarterly prevalence (January-March, April-June, July-September, October-December) of PI use by dividing the number of persons receiving PIs by the number receiving antiretroviral therapy (ART) in each quarter. Combination PI-statin prevalence was estimated by dividing the number of persons receiving PI-statins by the number receiving PIs quarterly. Contraindicated PI-statin prevalence was estimated by dividing the number of persons receiving contraindicated PI-statin combinations by the number receiving any PI-statin combinations quarterly. Because of the limited number of quarters before versus after guidelines, PI and statin use were collapsed to summary indicators before and after guidelines for each person as follows: if they received any PI, any PI-statin combination, or any contraindicated PI-statin combination, they were classified as such. To account for persons receiving PI and statin treatments before and after guidelines, mixed-effects logistic regression assuming an exchangeable covariance structure was used to compare prevalences of PI and statin use. Data management and statistical analyses were performed using SAS, version 8.2 (SAS Institute, Cary, NC).
A total of 5392 persons received prescriptions for antiretroviral drugs during the study period (Table 1). These persons had a median age of 36 years, 28% were female, 48% were African American, and 86% resided in urban areas of the state. Persons were enrolled in TennCare primarily because of disability (57%), lack of other health insurance (34%), as Aid to Families with Dependent Children (9%), or because of advanced age (<1%). They were followed in the cohort for a median of 34 months, and most subjects (67%) were followed from entry until the end of the study period. Almost 16% lost enrollment and almost 18% died during the study period.
Use of PIs increased from a rate of 29% of all ART-treated persons during 1996 to 44% during the first 2 quarters of 2002 (Fig. 1). Overall, 3448 persons (64% of all ART-treated persons) used any PI during the study period. Proportional PI use decreased from 64% of antiretroviral drug users during the preguideline period to 51% during the postguideline period (P < 0.001; Table 2). This decline in PI use coincided with increased use of nonnucleoside reverse transcriptase inhibitors (NNRTIs; see Fig. 1).
Combined PI-statin use increased during the postguideline period (Fig. 2). Of all persons who received PIs, 202 (5.9%) had 1 or more overlapping prescriptions for PIs and statins. This proportion increased from 3.5% during the preguideline period to almost 8% during the postguideline period (P < 0.001). Use of contraindicated statins in combination with PIs decreased from 42.0% of total combined PI-statin use to 20.8% during these periods (P < 0.001). Twenty-three persons who filled prescriptions for combined PI-statins in more than 1 quarter received contraindicated and noncontraindicated combinations during the study period. Most of these prescriptions (78%) were changed from contraindicated to recommended or alternative combinations.
We identified a large cohort of persons with exposure to PIs. Just more than 200 of these persons were exposed to combined therapy with PIs and statins. Our data show that rates of combined use of statins with PIs were increased in the TennCare population during 2001 through June 2002 compared with the previous 4 years.
Use of combinations of PIs and statins with the greatest potential for adverse effects decreased significantly in the latter period after preliminary guidelines were published. The relative rate of inappropriate combination therapy remained high, however, with 1 in every 5 persons treated with a PI and statin between January 2001 and June 2002 receiving a contraindicated combination. Potential reasons for continued prescribing of these drug combinations include lack of knowledge of potential drug-drug interactions and/or practice guidelines, the relative rarity of toxicities, the limited in vivo data characterizing these interactions and their effects, and other factors that contribute to guideline nonadherence by health care providers.27,28
A possible confounder of prescribing is the effect of formulary restrictions during the study. During the course of the study, more than 10 managed care organizations (each with its own formulary, which was modified frequently) were part of the TennCare program; thus, the impact of formulary restrictions cannot be quantified. Throughout the study, however, prior authorization for nonformulary drugs was available and could typically be obtained with submission of a 1-page telefaxed request form. This should have kept formulary restrictions from significantly limiting the use of more appropriate drug combinations, especially in the setting of potential drug-drug interactions.
There are several limitations of our study. We used antiretroviral prescriptions as a surrogate for HIV infection. Although it is possible that this resulted in misclassification, we believe this to be unlikely. Combination ART is only indicated for treatment of HIV infection and has no other use except as postexposure prophylaxis (PEP) after possible exposure to HIV. Recommended PEP treatment is generally for 28 days; thus, it is highly unlikely that a patient with a second filled prescription >30 days after the first would be incorrectly identified as having HIV. Furthermore, in the case of a health care-related exposure, PEP is provided by the employer; thus, it would not appear in our pharmacy database. Persons receiving monotherapy with either of the 2 antiretroviral medications used to treat chronic active hepatitis B infection in HIV-negative individuals were also excluded. We therefore believe that the use of multiple antiretroviral prescriptions is a highly sensitive and specific method of identifying HIV-infected persons receiving treatment in the TennCare program. Obviously, this method does not allow for identification of untreated persons with HIV infection, but this population was not necessary to address our question.
As with any study of this type, there are limitations in generalizability. A recent study examined prescribing patterns in the California Medicaid HIV-infected population and found a 6-fold increase in lipid-lowering drug use in PI-treated persons from 1996 to 2000, from 1.7% to 10.6%.29 Although that study did not characterize specific drug combinations, overall rates of PI-statin prescribing in California were comparable with our results, and simvastatin and lovastatin were the third and fourth most frequently prescribed statins in their HIV population. We believe our findings corroborate these data describing PI and statin use in a demographically and geographically distinct patient population.
We acknowledge that the use of guideline publication as the point of transition between study periods is artificial. It is likely that there were changes during the study period that influenced prescribing practice and were unrelated to the guidelines. Nevertheless, the initial guidelines in 2000 were the first time that emerging data and treatment recommendations regarding PI-statin interactions were made widely available within the HIV health care community. From a public health perspective, the most important question is not the magnitude of effect of the guidelines, per se. Rather, it is the extent to which prescribing of a potentially harmful drug combination decreased over time, regardless of what influenced these changes.
Our results are also limited by the fact that TennCare enrollees make up only a proportion, albeit a substantial proportion, of HIV-infected persons in Tennessee. TennCare currently includes 1,400,000 covered lives (26% of state's population). As of 1997, 43% of persons in Tennessee infected with HIV and 54% of those with AIDS had enrollment in TennCare programs for at least 1 day. Thirty-four percent and 44%, respectively, were enrolled for more than 320 days.30
Knowledge about the epidemiology of PI-associated hyperlipidemia and the role of ART and PIs in contributing to cardiovascular disease in HIV-infected persons is evolving. Early studies attempting to find an association between treatment with PIs, lipid abnormalities, and adverse cardiovascular events in patients with HIV reported conflicting results.31-34 More recent data have shown a modest but statistically significant increased risk of myocardial infarction in persons on combination ART35 and in those with ≥18 months of PI exposure36 as well as an increased risk of any first cardiovascular event in patients aged 35 to 65 years with more than 1 year of PI exposure.37 At least some of this excess risk is believed to be a result of PI-associated hyperlipidemia. Additionally, a preliminary report from the first large-scale, randomized, controlled trial of lipid-lowering therapy in HIV-infected subjects, AACTG study 5087, suggested that single-agent therapy was not effective in achieving NCEP guidelines in HIV-infected patients with mixed hyperlipidemia.38 Only patients on dual therapy with pravastatin and fenofibrate were able to meet composite NECP goals. Recently approved39,40 and future PIs may have more favorable effects on lipids, but as the HIV-infected population lives longer and ages along with the general population, use of lipid-lowering drugs in persons receiving PIs is likely to increase.
Current recommendations are that persons on HIV therapy be routinely screened for dyslipidemia and treated according to NCEP guidelines for non-HIV-infected persons.3 Based on estimates of prevalence and incidence of hyperlipidemia caused by PIs, the rates of statin use among PI users in the TennCare population seems to be low. This may be a result of low screening rates and/or low treatment rates. Korthuis et al41 have recently reported lipid screening rates of <60% among HIV infected veterans receiving PIs. Our study was not designed to address questions of provider adherence to screening guidelines and/or appropriate responses to such screening. Future studies should examine these rates in TennCare and other cohorts.
Because we were unable to assess clinical events caused by the use of contraindicated PI-statin combinations, we cannot comment on the clinical implications of such use. Even with limited data on the clinical implications, however, the available evidence provides a rationale for the most appropriate clinical practice. Statins known to have more favorable pharmacokinetics are available and effective. Thus, any use of drugs within the same class having even the theoretic potential for dangerous interactions should be unnecessary, especially when the intervention is for primary prevention. Although prescribing in the TennCare population seems to have improved in response to practice guidelines, continued education of HIV health care providers about the appropriate use of PI-statin combinations is needed.
The authors acknowledge the Vanderbilt-Meharry Center for AIDS Research (P30 AI054999) for providing academic support for collaborative efforts.
1. Palella FJ, Jr, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med
3. Dube MP, Stein JH, Aberg JA, et al. Guidelines for the evaluation and management of dyslipidemia in human immunodeficiency virus (HIV)-infected adults receiving antiretroviral therapy: recommendations of the HIV Medical Association of the Infectious Disease Society of America and the Adult AIDS Clinical Trials Group. Clin Infect Dis
4. Tsiodras S, Mantzoros C, Hammer S, et al. Effects of protease inhibitors on hyperglycemia, hyperlipidemia, and lipodystrophy: a 5-year cohort study. Arch Intern Med
5. Fellay J, Boubaker K, Ledergerber B, et al. Prevalence of adverse events associated with potent antiretroviral treatment: Swiss HIV Cohort Study. Lancet
6. Henry K, Melroe H, Huebesch J, et al. Atorvastatin and gemfibrozil for protease-inhibitor-related lipid abnormalities. Lancet
7. Schambelan M, Benson CA, Carr A, et al. Management of metabolic complications associated with antiretroviral therapy for HIV-1 infection: recommendations of an International AIDS Society-USA panel. J Acquir Immune Defic Syndr
8. Hsyu PH, Schultz-Smith MD, Lillibridge JH, et al. Pharmacokinetic interactions between nelfinavir and 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors atorvastatin and simvastatin. Antimicrob Agents Chemother
9. Omar MA, Wilson JP, Cox TS. Rhabdomyolysis and HMG-CoA reductase inhibitors. Ann Pharmacother
10. Vu D, Murty M, McMorran M. Statins: rhabdomyolysis and myopathy. Can Med Assoc J
. 2002;166:85-86, 90-81.
11. Cheng CH, Miller C, Lowe C, et al. Rhabdomyolysis due to probable interaction between simvastatin and ritonavir. Am J Health Syst Pharm
12. Hare CB, Vu MP, Grunfeld C, et al. Simvastatin-nelfinavir interaction implicated in rhabdomyolysis and death. Clin Infect Dis
13. Martin C, Hoffman V, Berggren R. Rhabdomyolysis in a patient receiving simvastatin concurrently with highly active antiretroviral therapy. Presented at: 40th Interscience Conference on Antimicrobial Chemotherapy; 2000; Toronto.
14. Mah Ming JB, Gill MJ. Drug-induced rhabdomyolysis after concomitant use of clarithromycin, atorvastatin, and lopinavir/ritonavir in a patient with HIV. AIDS Patient Care STDS
15. Fichtenbaum CJ, Gerber JG, Rosenkranz SL, et al. Pharmacokinetic interactions between protease inhibitors and statins in HIV seronegative volunteers: ACTG Study A5047. AIDS
16. Baldini F, Di Giambenedetto S, Cingolani A, et al. Efficacy and tolerability of pravastatin for the treatment of HIV-1 protease inhibitor-associated hyperlipidaemia: a pilot study. AIDS
17. Moyle GJ, Buss NE, Gazzard BG. Pravastatin does not alter protease inhibitor exposure or virologic efficacy during a 24-week period of therapy. J Acquir Immune Defic Syndr
18. Doser N, Kubli S, Telenti A, et al. Efficacy and safety of fluvastatin in hyperlipidemic protease inhibitor-treated HIV-infected patients. AIDS
19. Calza L, Manfredi R, Chiodo F. Statins and fibrates for the treatment of hyperlipidaemia in HIV-infected patients receiving HAART. AIDS
20. Dube MP, Sprecher D, Henry WK, et al. Preliminary guidelines for the evaluation and management of dyslipidemia in adults infected with human immunodeficiency virus and receiving antiretroviral therapy: recommendations of the Adult AIDS Clinical Trial Group Cardiovascular Disease Focus Group. Clin Infect Dis
21. Mirvis DM, Chang CF, Hall CJ, et al. TennCare-health system reform for Tennessee. JAMA
22. Ray WA, Griffin MR. Use of Medicaid data for pharmacoepidemiology. Am J Epidemiol
23. Strom BL, Carson JL. Use of automated databases for pharmacoepidemiology research. Epidemiol Rev
24. West SL, Savitz DA, Koch G, et al. Recall accuracy for prescription medications: self-report compared with database information. Am J Epidemiol
25. Leister KA, Edwards WA, Christensen DB, et al. A comparison of patient drug regimens as viewed by the physician, pharmacist and patient. Med Care
26. Johnson RE, Vollmer WM. Comparing sources of drug data about the elderly. J Am Geriatr Soc
27. Cabana MD, Rand CS, Powe NR, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA
28. Halm EA, Atlas SJ, Borowsky LH, et al. Understanding physician adherence with a pneumonia practice guideline: effects of patient, system, and physician factors. Arch Intern Med
29. Stein JH, Wu Y, Kawabata H, et al. Increased use of lipid-lowering therapy in patients receiving human immunodeficiency virus protease inhibitors. Am J Cardiol
30. Bailey JE, Van Brunt DL, Raffanti SP, et al. Improvements in access to care for HIV and AIDS in a statewide Medicaid managed care system. Am J Manag Care
31. Barbaro G. HIV infection, antiretroviral therapy and cardiovascular risk. J Cardiovasc Risk
32. Rickerts V, Brodt H, Staszewski S, et al. Incidence of myocardial infarctions in HIV-infected patients between 1983 and 1998: the Frankfurt HIV-cohort study. Eur J Med Res
33. Coplan PM, Nikas AA, Leavitt RY, et al. Indinavir did not increase the short-term risk of adverse cardiovascular events relative to nucleoside reverse transcriptase inhibitor therapy in four phase III clinical trials. AIDS
34. Depairon M, Chessex S, Sudre P, et al. Premature atherosclerosis in HIV-infected individuals-focus on protease inhibitor therapy. AIDS
35. Friis-Moller N, Sabin CA, Weber R, et al. Combination antiretroviral therapy and the risk of myocardial infarction. N Engl J Med
36. Mary-Krause M, Cotte L, Simon A, et al. Increased risk of myocardial infarction with duration of protease inhibitor therapy in HIV-infected men. AIDS
37. Iloeje U, Yuan Y, L'Italien G, et al. Protease inhibitor exposure time and risk of cardiovascular disease in HIV-infected patients. Presented at: 11th Conference on Retroviruses and Opportunistic Infections; 2004; San Francisco.
38. Aberg J, Zackin R, Evans S, et al. A prospective, multicenter, randomized trial comparing the efficacy and safety of fenofibrate versus pravastatin in HIV-infected subjects with lipid abnormalities: ACTG 5087. Presented at: 40th Annual Meeting of the Infectious Diseases Society of America; 2002; Chicago.
39. Haas DW, Zala C, Schrader S, et al. Therapy with atazanavir plus saquinavir in patients failing highly active antiretroviral therapy: a randomized comparative pilot trial. AIDS
40. Sanne I, Piliero P, Squires K, et al. Results of a phase 2 clinical trial at 48 weeks (AI424-007): a dose-ranging, safety, and efficacy comparative trial of atazanavir at three doses in combination with didanosine and stavudine in antiretroviral-naive subjects. J Acquir Immune Defic Syndr
41. Korthuis PT, Asch SM, Anaya HD, et al. Lipid screening in HIV-infected veterans. J Acquir Immune Defic Syndr
Keywords:© 2005 Lippincott Williams & Wilkins, Inc.
HIV-1; protease inhibitors; hydroxymethylglutaryl-coenzyme A reductase inhibitors; Medicaid; pharmacoepidemiology