Highly active antiretroviral therapy (HAART) has dramatically improved survival of HIV-infected persons since its introduction in 1995. 1 With this breakthrough, however, have come significant adverse effects of the component medications, such as hyperlipidemia.
Although other individual antiretroviral agents have been associated with dyslipidemias, all currently available protease inhibitors (PIs) have been associated with marked alterations in lipid metabolism, resulting in elevated levels of triglycerides, total cholesterol, and low-density lipoprotein (LDL). 2–7 Average total cholesterol levels increase by 30% in patients exposed to PIs, 8 and approximately 30% have total cholesterol levels >240 mg/dL. 9–12 Furthermore, a growing number of reports describe an increase in cardiovascular events in patients exposed to PIs. 13–15 Although questions remain regarding a definitive causal association with cardiovascular events, the high prevalence of hyperlipidemia is disturbing because it is a treatable and iatrogenic complication of medications that are likely to remain the mainstay of HAART for the foreseeable future. Consequently, the Department of Health and Human Services Panel on Clinical Practices for Treatment of HIV Infection continues to recommend PIs as first-line therapy for HIV infection, recommending that providers check fasting cholesterol and triglyceride levels at baseline and every 3 to 4 months thereafter, 16 echoing the recommendations of other investigators. 2,17,18
The purpose of this study is to assess the level of adherence to lipid-screening guidelines among providers caring for HIV-infected veterans exposed to PIs and to identify individual and facility-level predictors of adherence to lipid-screening guidelines.
We assessed adherence to lipid-screening guidelines in a retrospective cohort of HIV-infected US veterans cared for at sites throughout the United States, including Puerto Rico. This study was part of a larger effort to evaluate and improve HIV clinical practice under the auspices of the Quality Enhancement Research Initiative for HIV (QUERI-HIV) program. 19
Study Setting and Subjects
The Veterans Administration (VA) is the largest single provider of HIV care in the United States, serving approximately 17,000 HIV-infected patients annually. 20,21 We examined data from all HIV-infected patients aged 19 years or older who received PIs for at least 6 consecutive months between July 1, 1999 and June 30, 2001. We chose this study period because the use of PIs was widespread during this time and the need to monitor lipid levels in patients taking PIs had become widely recognized in HIV specialty meetings and the medical literature by July 1999. 22–24 All VA facilities that treated at least 1 HIV-infected patient were included in the analysis.
Data regarding lipid testing, PI exposure, and individual patient characteristics were analyzed from the Immunology Case Registry (ICR). The ICR is a national database of all identified HIV-infected patients treated in the VA health care system. 21 The ICR database is managed by the VA Center for Quality Management (CQM), which links data using an encrypted social security number, thus protecting patient confidentiality while preventing duplication across sites when data are used for research purposes. Comorbidities, including history of prior hyperlipidemia, were defined on the basis of prior treatment of these conditions. A diagnosis of preexisting AIDS was based on the 1993 Centers for Disease Control and Prevention (CDC) case definition. 25
We obtained information regarding facility characteristics from a cross-sectional mailed survey of HIV lead physicians at 138 VA sites where HIV care was provided. Completed surveys were collected from 118 (86%) of these sites. 26 Lead physicians were asked to report information describing the organization of health care delivery for HIV-infected patients at their institution, including the presence of a quality improvement program, case management, and use of guidelines.
Protease Inhibitor Exposure
Exposure to a PI was defined as any prescription for indinavir, ritonavir, saquinavir, nelfinavir, or amprenavir filled at a VA pharmacy (data collected before commercial availability of lopinavir/ritonavir). Exposure was measured in months. A patient was considered to be exposed to PIs for an entire month if he or she received a prescription with enough pills to cover 50% of the days in a month.
Dependent Variables: Adherence to Lipid Guidelines
We estimated adherence to lipid guidelines by measuring the proportion of PI-exposed individuals who received at least 1 triglyceride or cholesterol laboratory test within the first 6 months of consecutive PI exposure. If a patient had more than 1 episode of PI exposure separated by a period of nonexposure, we included only the first episode for purposes of this analysis. We also measured the number of months from the first PI exposure to the first triglyceride or cholesterol laboratory test ordered among patients who received PIs for at least 6 consecutive months. Patients were censored from analysis following discontinuation of PI use.
We used descriptive statistics to summarize patient and facility characteristics and applied χ2 tests for categoric variables and t tests for continuous variables to test their crude (unadjusted) associations with lipid screening within 6 months of exposure to PIs.
We hypothesized that lipid screening would depend on predisposing patient characteristics (age, gender, HIV exposure group, and race/ethnicity), medical need (comorbidities such as a history of AIDS, cardiovascular or cerebrovascular disease, diabetes, hypertension, smoking, or preexisting hyperlipidemia), and enabling characteristics (facility traits such as urban location and the presence of HIV case management, guidelines, and quality improvement programs) according to the behavioral model of access to care. 27,28
To test these hypotheses, we fit 3 logistic regression models to predict the likelihood of receiving a lipid-screening test by 6 months according to predisposing patient characteristics alone (model A), patient characteristics plus comorbidities (model B), and predisposing patient characteristics plus comorbidities plus facility characteristics (model C). Models were adjusted for within-facility clustering by Huber-White correction. 29 Models were assessed for goodness of fit with a Hosmer-Lemeshow test. 30 Odds ratios from logistic regression were converted to relative risk (RR), because lipid screening was a common outcome. 31 Finally, a Kaplan-Meier survival analysis of time to first lipid-screening test was fit to all patients with at least 6 months of consecutive PI exposure. We performed multiple hot deck imputation for missing patient characteristics and comorbidities (race = 1%, HIV risk = 12% missing) but not for facility characteristics, because the latter were not missing at random and the primary unit of analysis was the individual patient. Imputation did not alter the direction or significance of results, so nonimputed results are reported here.
There were 18,765 HIV-infected persons who received care in the VA system between July 1, 1999 and June 30, 2001. Of these, 4065 received PIs for at least 6 consecutive months during this period.
PI-exposed patients had a mean age of 46.8 years and had a mean of 5.2 outpatient clinic visits in the year preceding their first onset of PI exposure during the study period. Eighty-five percent of patients had been exposed to PIs previously, with a mean of 19 months of cumulative PI exposure before study baseline. They were predominantly male and non-Latino white, with a history of AIDS at the time of first VA enrollment (Table 1). The prevalence of cardiac risk factors reported at the time of enrollment for HIV care at the VA was low; 12% of patients had a history of hyperlipidemia. The majority of patients were cared for at large urban facilities with HIV-specific practice guidelines and case management. The median number of HIV-infected enrollees per facility was 255 (standard deviation [SD] = 287).
Crude (unadjusted) associations between selected factors and the proportion screened for lipids within 6 months of PI use are shown in Table 2 for all patients receiving PIs for 6 months. Of these, 2395 (59%) received lipid screening during the first 6 months of PI use. Unadjusted lipid-screening rates were lowest among persons with unknown HIV risk, the youngest age group, and those cared for in nonurban areas.
Multivariate associations of patient predisposing characteristics and lipid screening are reported in model A in Table 3. Persons of younger age category compared with those of older age category and persons reporting intravenous drug use (IVDU), heterosexual contact, or unknown HIV risk exposure compared with men who have sex with men were less likely to receive lipid screening. In model B in Table 3, persons with a prior history of hyperlipidemia were more likely to receive lipid screening than average, whereas other comorbidities appeared to be only weakly associated with lipid screening. When facility characteristics were added to this model, the inverse association between IVDU and unknown HIV risk and lipid screening persisted, and management in an urban location was positively associated with lipid screening (see model C in Table 3). Age was positively associated with lipid screening in all models.
A Kaplan-Meier survival analysis of time to first lipid-screening test is shown in Figure 1. We find that 59% of PI-exposed persons received lipid screening by 6 months. Adjusting for all other model C covariates, HIV-infected persons cared for in large urban facilities were more likely to receive lipid screening after the first 6 months of PI exposure than persons cared for in nonurban facilities (RR = 1.3, range: 1.0–1.5). Inclusion of facility size did not change the direction or significance of RR for urban location (data not shown). It took 16 months for 75% of the population to receive lipid screening.
Six in 10 veterans taking PIs receive recommended lipid screening. This is comparable to lipid-screening rates among diabetic Medicare fee-for-service enrollees, 60% of whom received lipid screening in 2000 through 2001, but lower than the 89% lipid-screening rate observed in diabetic veterans. 32 In this study, the VA consistently achieved higher performance measure scores than Medicare fee-for-service programs. Other performance measures were consistently higher in the VA than in Medicare fee-for-service programs. It is likely, therefore, that lipid-screening rates for HIV-infected persons are lower than 59% outside the VA health care system. Our data suggest that there remains room for improved screening for iatrogenic hyperlipidemia in HIV-infected persons.
The strongest association with lipid screening was previously known conditions that affect lipid levels, particularly diabetes and prior hyperlipidemia. Associations between other individual patient characteristics and less adherence to lipid-screening guidelines, such as IVDU, were slightly attenuated when comorbid conditions and facility characteristics were added to the model. This suggests that the impact of individual patient characteristics on a provider's decision to perform lipid screening may be determined by patient comorbidities and characteristics of the facility where the patient is seen. Competing needs of patients with comorbid conditions are likely to have an adverse impact on a provider's ability to address nonurgent issues such as lipid screening, as previously demonstrated in other VA populations where the likelihood of preventive counseling decreased with increasing visit acuity. 33 This is particularly germane to HIV care, where visit acuity is often high and current management guidelines do not address comorbid conditions. Despite relatively higher screening rates, lipid screening in patients with advancing age, diabetes, and prior hyperlipidemia did not exceed 71%. Interventions to improve quality of care should seek to increase lipid screening further in these particularly high-risk patients.
Patients with an unknown HIV risk factor were less likely to receive lipid screening in all models. Although other authors 34 caution that stratification on missing data status can produce biased effect estimates, this unknown category was primarily based on direct patient responses rather than being missing from the data set. We believe that nondisclosure of HIV risk group is a marker for patients who are less engaged with the health care system. Matthews et al 35 reported a similar prevalence of “unknown” HIV risk group in a non-VA clinical database. Unknown HIV risk group predicted increased mortality in this cohort.
The 22% of our population (888 persons) cared for at nonurban facilities were 30% less likely to receive lipid screening than those cared for in urban areas, even after adjusting for the presence of case management, guidelines, and quality improvement programs. This association persisted after controlling for the number of HIV patients treated by each facility. Although the numbers of HIV-infected patients seen at rural facilities are a small part of the US epidemic, these facilities have limited resources, less-experienced providers, and a lower proportion of eligible persons receiving HAART or Pneumocystis carinii pneumonia (PCP) prophylaxis compared with persons receiving care in urban areas. 36,37
Although we believe our data draw attention to the need for improved lipid screening for HIV patients, attempts to improve lipid screening should not detract from attempts to address gaps in overall quality of care. Measures specifically targeting lipid screening should be incorporated into broad interventions that comprehensively address critical HIV performance measures (eg, the proportion of eligible persons receiving HAART, PCP prophylaxis, and vaccinations). Facility-level interventions such as improved clinical information systems and provider decision support are most likely to achieve these improvements in quality, as proposed by the Chronic Care Model for improving outcomes in persons with chronic illness. 38,39
Our study is limited by the absence of health outcomes, such as the occurrence of myocardial infarction following the onset of PI use. Long-term data regarding cardiovascular risk in persons exposed to HAART continue to be collected, and several studies of relatively brief duration suggest a small but clinically significant increase in cardiac events among persons taking HAART. 13–15 Two large clinical cohorts, however, showed no association between PI use and cardiac events. 40,41 One of these, a retrospective outcomes study of HIV-infected veterans, showed an unchanged cardiovascular disease rate since PIs became widely available but also demonstrated an increased risk of disease in persons with preexisting risk factors. 41 This and clinical reasoning led the authors to recommend risk factor reduction and therefore screening in this population. This evidence, together with the iatrogenic nature of PI-associated hyperlipidemia, has led to continued recommendations for lipid screening with treatment as in non–HIV-infected patients. 2,16–19,42
A second limitation is that generalization of these results to HIV-infected persons receiving care outside the VA may be limited, given that our study population of VA enrollees is older, more predominantly male, and has a greater prevalence of minority race/ethnicity groups than a national probability sample of HIV-infected persons in the United States. 21
In summary, lipid-screening rates are suboptimal among HIV-infected veterans receiving PIs. No subgroup had rates above 71%, so widespread provider and patient education efforts are warranted. Systemic interventions to improve overall HIV quality of care should also address lipid screening, particularly among patients with unknown or IVDU HIV risk and those cared for in nonurban areas.
1. Palella Jr, FJ, 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. 1998; 338:853–860.
2. Carr A, Samaras K, Thorisdottir A, et al. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study. Lancet. 1999; 353:2093–2099.
3. Periard D, Telenti A, Sudre P, et al. Atherogenic dyslipidemia in HIV-infected individuals treated with protease inhibitors. The Swiss HIV Cohort Study. Circulation. 1999; 100:700–705.
4. Walli RK, Michl GM, Segerer S, et al. Dyslipidemia and insulin resistance in HIV-infected patients treated with reverse transcriptase inhibitors alone and in combination with protease inhibitors [abstract 645]. Presented at the Sixth Conference on Retroviruses and Opportunistic Infections, Chicago, IL, January/February 1999.
5. Riddle TM, Kuhel DG, Woollett LA, et al. HIV protease inhibitor therapy increases hepatic lipoprotein production via stabilization of activated sterol regulatory element-binding protein-1 (SREBP-1) in the nucleus [abstract 659]. Presented at the Eighth Conference on Retroviruses and Opportunistic Infections, Chicago, IL, February 2001.
6. Petit JM, Duong M, Florentin E, et al. Down expression of LDL receptor in HIV-infected patients on HAART: quantification by flow cytometry [abstract 661]. Presented at the Eighth Conference on Retroviruses and Opportunistic Infections, Chicago, IL, February 2001.
7. Carr A, Samaras K, Chisholm DJ, et al. Pathogenesis of HIV-1 protease inhibitor-associated peripheral lipodystrophy, hyperlipidaemia, and insulin resistance. Lancet. 1998; 351:1881–1883.
8. Passalaris JF, Sepkowitz KA, Glesby MJ. Coronary artery disease and human immunodeficiency virus infection. Clin Infect Dis. 2000; 31:787–797.
9. Purnell JQ, Zambon A, Knopp H, et al. Effect of ritonavir on lipids and post-heparin lipase activities in normal subjects. AIDS. 2000; 13:51–57.
10. Behrens G, Dejam A, Schmidt H, et al. Impaired glucose tolerance, beta cell function and lipid metabolism in HIV patients under treatment with protease inhibitors. AIDS. 1999; 13(Suppl):F63–F70.
11. Mulligan K, Grunfeld C, Tai VW, et al. Hyperlipidemia and insulin resistance are induced by protease inhibitors independent of changes in body composition in patients with HIV infection. J Acquir Immune Defic Syndr. 2000; 23:35–43.
12. Keruly JC, Mehta S, Chaisson RE, et al. Incidence of and factors associated with the development of hypercholesterolemia and hyperglycemia in HIV-infected patients using a protease inhibitor [abstract I-95]. In: Programs and Abstracts of the 38th Interscience Conference on Antimicrobial Agents and Chemotherapy (San Diego, CA). Washington, DC: American Society for Microbiology; 1998:392.
13. Holmberg SD, Moorman AC, Tong T, et al. HIV Outpatient Study [abstract 941]. Presented at the Ninth Conference on Retroviruses and Opportunistic Infections, Seattle, WA, February 2002.
14. Friis-Moller N, Weber R, D'Arminio Monforte A, et al. Exposure to HAART is associated with an increased risk of myocardial infarction: the D:A:D study. AIDS. 2003; 17:1179–1193.
15. Moore RD, Keruly JC, Lucas G. Increasing incidence of cardiovascular disease in HIV-infected persons in care [abstract 132]. Presented at the 10th Conference on Retroviruses and Opportunistic Infections, Boston, MA, February 2003.
16. Panel on Clinical Practices for Treatment of HIV Infection. Guidelines for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents.
Department of Health and Human Services and the Henry J. Kaiser Family Foundation; Washington, DC, 2001. Available at: http://www.hivatis.org/guidelines
17. 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. 2000; 31:1216–1224.
18. 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. 2002; 31:257–275.
19. Rubenstein LV, Mittman BS, Yano EM, et al. From understanding health care provider behavior to improving health care: the QUERI framework for quality improvement. Med Care. 2000; 38:I129–I141.
20. Rahman A, Deyton LR, Goetz M, et al. Inversion of inpatient/outpatient HIV service utilization: impact of improved therapies, clinician education and case management in the US Department of Veterans Affairs [abstract 443/42429]. Presented at the 12th World AIDS Conference, Geneva, June–July 1998.
21. Backus L, Mole L, Chang S, et al. The Immunology Case Registry. J Clin Epidemiol. 2001; 54(Suppl):S12–S15.
22. Churchill DR, Pym AS, Back DJ, et al. Hyperlipidaemia following treatment with protease inhibitors in patients with HIV-1 infection. Br J Clin Pharmacol. 1998; 46:518–519.
23. Gallet B, Pulik M, Genet P, et al. Vascular complications associated with use of HIV protease inhibitors. Lancet. 1998; 351:1958–1959.
24. Behrens G, Schmidt H, Meyer D, et al. Vascular complications associated with use of HIV protease inhibitors. Lancet. 1998; 351:1958.
25. Centers for Disease Control and Prevention. 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR. 1992; 41:RR17.
26. Yano EM, Phillips B, McWorter V, et al. VHA Survey of HIV/AIDS Programs and Practices, Results of a National Survey. HIV/AIDS Quality Enhancement Research Initiative (QUERI).
HSR&D section, VA San Diego Healthcare System, San Diego, CA. 2001.
27. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995; 36:1–10.
28. Andersen RM, Bozzette S, Shapiro M, et al. Access of vulnerable groups to antiretroviral therapy among persons in care for HIV disease in the United States. Health Serv Res. 2000; 35:389–416.
29. Huber PJ. The behavior of maximum likelihood estimates under nonstandard conditions. In: Lecam LM, Neyman J, eds. Proceedings of the Fifth Berkeley Symposium in Mathematical Statistics and Probability. Berkeley: University of California Press; 1967:221–223.
30. Hosmer DW, Lemeshow S. Applied Logistic Regression.
New York: John Wiley & Sons; 1989.
31. Zhang J, Yu KF. What's the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998; 280:1690–1691.
32. Jha AK, Perlin JB, Kizer KW, et al. Effect of the transformation of the Veterans Affairs Health Care System on the quality of care. N Engl J Med. 2003; 348:2218–2227.
33. Chernof B, Sherman SE, Lanto AB, et al. Health habit counseling amidst competing demands: effects of patient health habits and visit characteristics. Med Care. 1999; 37:738–747.
34. Greenland S, Finkle WD. A critical look at methods for handling missing covariates in epidemiologic regression analyses. Am J Epidemiol. 1995; 142:1255–1264.
35. Mathews WC, Barber E. Rising mortality rates among recent entrants to care for HIV infection. [abstract 911]. Presented at the 10th Conference on Retroviruses and Opportunistic Infections, Boston, MA, February 2003.
36. Cohn SE, Berk ML, Berry SH, et al. The care of HIV-infected adults in rural areas of the United States. J Acquir Immune Defic Syndr. 2001; 28:385–392.
37. Graham RP, Forrester ML, Wysong JA, et al. HIV/AIDS in the rural United States: epidemiology and health services delivery. Med Care Res Rev. 1995; 52:435–452.
38. Hecht FM, Wilson IB, Wu AW, et al. Optimizing care for persons with HIV infection. SGIM AIDS Task Force. Ann Intern Med. 1999; 131:136–143.
39. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness. JAMA. 2002; 288:1775–1779.
40. Klein D, Hurley L, Quesenberry CP, et al. Do protease inhibitors increase the risk for coronary heart disease in patients with HIV-1 infection? J Acquir Immune Defic Syndr. 2002; 30:471–477.
41. Bozzette SA, Ake CF, Tam HK, et al. Cardiovascular and cerebrovascular events in patients treated for human immunodeficiency virus infection. N Engl J Med. 2003; 348:702–710.
42. Grundy S, Becker D, Clark LT, et al. Detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). Circulation. 2002; 106:3145–3421.