The introduction of antiretroviral therapy (ART) has in many cases changed HIV from a high mortality infection to a chronic illness in the United States. With this evolution, cardiovascular disease (CVD) is now a leading cause of morbidity and mortality in this patient population. Patients with HIV have a 2-fold increased risk for myocardial infarction compared with those without HIV infection.1 Underlying causes for this increased risk include (1) metabolic alterations (eg, deposition of body fat, insulin resistance, high blood pressure, and abnormal lipid levels) resulting from the HIV itself, (2) dyslipidemia induced by antiretroviral medications, in particular protease inhibitors (PIs), and (3) increasing age and associated CVD risk factors of HIV-infected patients.1–3 Although much has been written on the prevalence, treatment, and control of CVD risk factors in the general population, there is relatively less information available on HIV-infected patients, especially treatment and control rates among patients with recognized HIV.
With the recognition that HIV-infected patients are at greater risk for CVD, it will be important to characterize the distribution of CVD risk factors and the success, or failure, of treatment and goal attainment in this population.
One model for care has been a specialized clinic with comprehensive and co-located services. The Spencer Cox Center delivers services through an interdisciplinary model of HIV care that addresses medical, psychological, social, emotional, environmental, and spiritual needs of patients through a patient-centered “one-stop shopping” model of care. Each patient is assigned a primary care provider (PCP) (ie, internist, physician assistant, nurse practitioner experienced in HIV/AIDS, or an infectious disease specialist), as well as a social worker, a mental health provider, a pharmacist, and an on-site pharmacy. Dental, pediatric, and obstetric and gynecologic services are also available, along with cardiology section where patients are referred by their PCP for any cardiovascular-related issue.
The objectives of this study were to: (1) determine the prevalence of low-density lipoprotein cholesterol (LDL-C) dyslipidemia and hypertension among HIV patients in an urban HIV/AIDS clinic that provides a comprehensive care, (2) describe the treatment and control of these risk factors, and (3) examine factors potentially associated with successful control of dyslipidemia and hypertension in these patients.
SUBJECTS AND METHODS
Study Design and Setting
We conducted a cross-sectional study using data from The Spencer Cox Center for Health Care, part of St. Luke's and Roosevelt Hospitals in New York City, and a designated New York State AIDS Center. This study was approved by the St. Luke's and Roosevelt Hospitals' Institutional Review Board.
Study patients were included if they were HIV positive and 20 years of age or older. Exclusion criteria were negative or unknown HIV status, transgendered patients, or known pregnancy at the time of data abstraction. Transgendered patients were excluded because their gender reassignment treatment was at various stages and could influence lipid levels and blood pressure. Patients with HIV-negative status (such as partners and family members of HIV-positive patients or those with risk factors for becoming infected) and patients with unknown status (patients with HIV testing pending) are also seen in the clinic but were excluded from this analysis. Of the 5232 patients in the clinic, 954 were excluded because they were younger than 20 years (301 patients), pregnant (35 patients), transgendered (60 patients), HIV-negative status (620 patients), and/or unknown HIV status (187 patients), where some patients fell into more than 1 category. A total of 4278 patients were therefore included in the analysis.
Patient information and notes from all clinic visits are maintained within the center's electronic medical records system that includes demographic information, service utilization, diagnoses, medications, and laboratory data. The most recent laboratory value or measurement in the medical record was abstracted. Each patient had 1 observation abstracted for each measure of interest. For patients at 2 of the 3 outpatient sites (N = 3270), data were collected in the calendar years 2008–2010. For patients at the remaining clinic site (N = 1008), data were collected from 2010 to 2011. Clinic format and protocols were similar in all 3 sites and did not vary during the study period.
Patient demographics, medical history, and medications were entered into each patient's electronic medical record and verified at each appointment. Information on date of birth was obtained from insurance and other documentations. Race, gender, and mode(s) of probable HIV transmission were based on patient's self-report. Translators were available for Spanish-speaking patients. Height, weight, and blood pressure were measured by trained clinic staff. Blood pressure was measured according to the clinic protocol with a Dinamap PRO 100 (Critikon; GE Healthcare, Buckinghamshire, United Kingdom) automated oscillometric device, with the patient seated and back and arm supported after resting for at least 5 minutes. Weight and height were measured with patients in stocking feet with removal of outer clothing. Body mass index was calculated as weight in kilograms divided by height in square meters.
Patients were considered treated for dyslipidemia or hypertension if the prescription date and quantity of days supplied for the lipid-lowering or hypertension medication overlapped the date of the qualifying LDL-C and blood pressure measurements. Blood pressure treatment was defined as prescription of angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta blockers, alpha-2 receptor agonists, or diuretics. Dyslipidemia treatment was defined as prescription of statins, fibrates, fish oil, intestinal absorption inhibitors, niacin, or bile acid sequestrants. As this was a cross-sectional study, we were not able to gather information on the timing of therapy initiation.
Definition of Key Variables
Blood lipids were measured on samples drawn in hospital laboratories. Not all specimens were taken in the fasting state; however, it has been shown that LDL-C calculated by the Friedewald formula and HDL-C change only minimally in response to food intake.4 LDL-C was calculated by the Friedewald formula5 unless triglycerides were >400 mg/dL, rendering the formula inaccurate. In this case, the most recent lipid profile was searched for where LDL-C could be calculated and that profile was used. Total cholesterol, HDL-C, and triglycerides were measured according to the standard laboratory protocols (Centers for Disease Control and Prevention, Atlanta, GA).
All patients were classified according to criteria from the National Cholesterol Education Program (NCEP) Adult Treatment Panel III and the Framingham risk score6 for a calculated 10-year risk of coronary heart disease (CHD) of <10%, 10%–20%, or >20% and assigned to 1–3 risk categories (low, intermediate, and high) as follows7:where CHD equivalent is defined as having at least one among the following conditions: diabetes, peripheral artery disease, ischemic stroke, carotid artery disease, and abdominal aortic aneurysm.
1. Low risk: 0–1 risk factor + 10-year risk of <10%;
2. Intermediate risk: ≥2 risk factors + 10-year risk of <10% or 10%–20%;
3. High risk: CHD or CHD equivalent or 10-year risk of >20%.
Patients who were not taking lipid medications were classified as having dyslipidemia if LDL-C was ≥190 mg/dL for low-risk patients; ≥160 mg/dL for intermediate-risk patients; and ≥130 mg/dL for high-risk patients. Goal for LDL-C was defined as <160 mg/dL for low risk, <130 mg/dL for intermediate risk, and <100 mg/dL for high risk. In the presence of slightly different versions and/or applications of the Framingham score, we used the version at http://www.framinghamheartstudy.or/risk/coronary.html, so that the score was derived for patients who are 30–74 years old and without overt CHD at the baseline examination.
Treatment was defined as taking medication for dyslipidemia. Goal attainment was defined as being at the NCEP Adult Treatment Panel III goal for patients' respective risk category.
Hypertension was defined as being prescribed antihypertensive medications or having a reading of ≥140 mm Hg systolic or ≥90 mm Hg diastolic. Patients on medication for blood pressure were considered treated and at goal if they had a blood pressure reading <140/90 mm Hg.
Blood pressure categories of stages I (blood pressure ≥140 mm Hg systolic and ≥90 mm Hg diastolic) and II (blood pressure ≥160 mm Hg systolic and ≥100 mm Hg diastolic) were combined to form a hypertension outcome consisting of a “not at goal” or uncontrolled level for comparison with the combined levels of normal and prehypertension (“at goal” or controlled).8
Overweight and Obesity, Cigarette Smoking, and Diabetes
Overweight and obesity were classified according to the guidelines established by the National Heart, Lung, and Blood Institute, the National Institutes of Health9 using the following body mass index cutoffs: underweight <18.5, normal 18.5–24.9, overweight 25–29.9, obesity I 30.0–34.9, obesity II 35.0–39.9, and morbidly obese ≥40. Cigarette smoking was defined by chart notation of tobacco use at the time of data abstraction. Patients were classified with diabetes using the standard definition: fasting plasma glucose ≥126 mg/dL or casual plasma glucose ≥200 mg/dL or on insulin or oral hypoglycemic agents.
The distributions of demographic and clinical variables were analyzed descriptively: continuous variables were summarized as mean, standard deviation, and range. Categorical variables were summarized using frequency and percentage. Of note, we indicate total sample size used for specific calculations whenever informative because proportions of missing data varied, which is common in similar clinical settings. In each analysis, we included non-missing data in the included/relevant variable(s). Prevalence, treatment, and control of risk factors were of primary interest in this study, so they were determined in the overall population and within subgroups formed by risk status, gender, and race/ethnicity. Continuous variables were discretized in certain analyses using widely accepted thresholds. For example, the variable of viral load was analyzed as a dichotomized variable with cutpoint at 200 copies per milliliter, where a viral load of <200 copies per milliliter indicates good suppression.
Simple and multiple logistic regression models were fitted to examine demographic and clinical variables potentially associated with the outcomes of (1) at LDL-C goal and (2) at hypertension goal. For each analysis, unadjusted and adjusted odds ratios and the corresponding 95% confidence interval and P-value were reported.
As a sensitivity analysis, multiple imputations (with 10 imputations) were used to assess the potential association of missing data with outcomes. SAS version 9.3 (SAS Institute, Cary, NC) was used for statistical analyses.
Demographic and Clinical Characteristics
Table 1 summarizes the demographic and clinical characteristics of the 4278 patients included in the analysis. The patient population was primarily male (75%), with most being 40–49 years old (36%) or 50–59 years old (29%), and ethnically reflective of the New York City metropolitan area: Black (45%), Hispanic (35%), and White (14%). Patients varied in exposure risk to HIV and included men who have sex with men (42%), heterosexual sexual contact (34%), and past intravenous drug use (8.8%). Most resided in 2 of the 5 boroughs of New York City, Manhattan (45%), or Bronx (24%). Most of the patients were medically insured through Medicaid (60%).
Relatively few patients (2.7%) had pre-existing CHD and 7.5% had a CHD equivalent, with diabetic patients comprising 7.3% of the overall population. Most patients were either overweight (31%) or obese (30%), and 42% smoked tobacco. The overall prevalence of LDL-C dyslipidemia was 35%, and the prevalence of hypertension was 43%. Most patients were classified as low risk (73%) or intermediate risk (23%) by NCEP and Framingham risk score.
Variables related to HIV are also listed in Table 1. Eighty-eight percent of patients were on ART. Most of those (66%) on any ART were on regimens that contained a PI. For those on therapy, 67% had viral load <200 copies per milliliter. Mean CD4 count for all patients was 468 cells per cubic millimeter.
Prevalence, Treatment, and Control of LDL-C Dyslipidemia
Results for LDL-C dyslipidemia are summarized in Tables 2 and 3 and Figure 1. Of the 4278 patients included, 838 did not have an available LDL-C value. The prevalence of LDL-C dyslipidemia was 37% in men and 31% in women. When divided by racial/ethnic groups, the prevalence of LDL-C dyslipidemia was 30% in Blacks, 40% in Hispanics, and 37% in Whites.
Ninety percent of patients with dyslipidemia were treated, and treatment rates were high irrespective of subgroups defined by risk category, gender, and racial/ethnic group. Treatment rates were >90% in patients with CHD or CHD equivalent.
Overall, 75% of patients with LDL-C dyslipidemia who were treated were at goal. Men and women had similar attainment of goals (76% vs. 74%), and Hispanic (78%) and White (78%) were higher than Black (70%). Attainment of goal was greatest (90%) in those categorized as low CVD risk, followed by those at intermediate risk (75%) and high risk (56%). Consistent with this, attainment of goal was lower in those with CHD (57%) or CHD equivalent (62%). Of note, those at higher risk have the most stringent LDL-C goal (<100 mg/dL).
In adjusted analysis, factors associated with attaining goal included younger age, no cigarette smoking, and not having CHD or CHD equivalent (P < 0.01). These observed associations were not substantially changed in adjusted analyses (Table 3).
Prevalence, Treatment, and Control of Hypertension
Results for hypertension are shown in Tables 2 and 3 and Figure 2. Of the 4278 patients included, 152 did not have a blood pressure value. Overall, 43% of patients had hypertension. The prevalence of hypertension was similar among men and women (43% for both) and racial/ethnic groups (Black 47%, Hispanic 40%, White 43%, and Other 23%).
Overall, 75% of patients with hypertension were treated. Women had higher treatments rates (82%) than men (73%). Hispanics (77%) and Blacks (77%) had higher treatment rates than Whites (65%). Unlike treatment for LDL-C dyslipidemia, those in higher risk groups had higher treatment rates (70%, 74%, and 87% for low, intermediate, and high risk, respectively). Treatment rates were highest (>92%) in those with CHD or CHD equivalents.
Attainment of goal for blood pressure, among those who were treated, was low for all groups. Only 45% of women and 42% of men attained goal. Rates of attaining blood pressure goal for Hispanics (49%) and Whites (42%) were higher than those for Blacks (39%). Goal attainment rate was 54% of those at high risk, 59% at intermediate risk, and 56% of those at low risk. Goal attainment was 65% and 58% for those with CHD or CHD equivalent, respectively.
Factors associated with success at attaining blood pressure goal included being on ART and having a lower viral load. Consistent results were observed in unadjusted and adjusted regression (Table 3).
The main results were fairly consistent before and after imputation. For example, the prevalence, treatment, and control rates of LDL-C dyslipidemia were, respectively, 29%, 86%, and 76% on average with imputation (vs. 35%, 90%, and 75% without imputation) and the counterparts of hypertension were, respectively, 42%, 74%, and 57% on average with imputation (vs. 43%, 75%, and 57% without imputation).
Previous studies have provided information on the prevalence of CVD risk factors in patients with HIV infection, but few have described results from clinical settings regarding how well these risk factors are treated and controlled. In this study, we show that in a population of outpatients infected with HIV, treatment rates for LDL-C dyslipidemia were high, but that a significant percentage of patients were not adequately controlled to the recommended target LDL-C goals. The treatment rate for hypertension was 75%, but control was poor, with 57% of the patients at goal blood pressure.
Unique to this study is that the population represents patients in a specialized HIV clinic and not patients recruited into a clinical trial, thus reflecting a real-world setting. Furthermore, the population is diverse, including approximately 25% women, an age range from 20 to 87 years, varied risk sources for HIV, and a diverse race/ethnicity reflective of a large urban population.
There is limited information from other populations of HIV-infected patients. In the HIV-HEART Study, an ongoing, prospective multi-center study of 803 patients (82% male, 88% White), 21% were hypertensive. Treatment rates for hypertension stratified by the Framingham risk score were 40% in low-risk group, 40% in intermediate-risk group, and 83% in high-risk groups. Dyslipidemia was prevalent among 24% (defined as elevated total cholesterol), and of those with known LDL-C values and of moderate or high risk, there were more than 50% and 30%, respectively, who were untreated. Not all patients had LDL-C measurement, and patients were enrolled from many centers in Germany and not one clinic where all patients received similar care.10 In a study of 542 patients in a Norwegian HIV clinic, 32% were hypertensive. Only 26 of these patients had been identified as having hypertension, and of these, 9 were on treatment and at goal.11
Our results for LDL-C dyslipidemia showed good treatment compared with previous studies, except for those patients who were at higher risk for CVD according to the established guidelines. We used current population guidelines and a risk stratification algorithm to determine cutpoints for goal attainment. Recent primary care guidelines for HIV providers have endorsed this approach.12 It is unclear if traditional risk stratification and prediction formulas apply to patients with HIV and therefore accurately determine LDL-C goals. For example, Parra et al13 and Friis-Moller et al3 found that the Framingham risk score underestimated the presence of subclinical atherosclerosis and manifest disease in HIV-infected patients.
Despite the growing understanding that patients infected with HIV are at higher risk compared with non-HIV patients, current guidelines for LDL-C do not recommend more stringent goals for HIV patients. No validated CVD risk calculator exists specific to patients infected with HIV. Such a risk calculator should consider HIV-specific factors such as class(es) of ART as potential contributors of increased cardiovascular risk. For our study, we did not have information on specific ART medications and classes other than the category “PIs” that are known to exacerbate dyslipidemia. In our regression analysis, being on ART was not significantly associated with success to attain goal for LDL-C.
A lower goal for LDL-C would increase the prevalence of LDL-C dyslipidemia in our patient population and decrease the number attaining goal. In this study, the lowest attainment of goal (57% or 62%, respectively) were in those with CHD or CHD equivalents and those in the highest risk groups (56%), where the LDL-C goal was <100 mg/dL. The reasons for this include the more stringent target and limitations for use of lipid-lowering medications (eg, adverse effects, interaction with ART, and coverage by Medicaid).
Because there is scarce information on hypertension treatment and goal attainment in HIV-infected patients, comparisons with data from the general population may be informative. A report from the National Health and Nutrition Examination Survey of people in the United States aged 18 years of age and older from 2001 to 2010 showed that 77% of patients with hypertension were taking any antihypertensive drug and of those treated, 60% were controlled.14 Data from the New York City Health and Nutrition Examination Survey found that, among residents aged 20 years and older, 26% of subjects had hypertension, 73% were treated, and of those treated, 65% were at goal. A lack of routine place of medical care was most strongly associated with poor control. Among nonelderly adults with treated hypertension, Blacks had a 4-fold higher odds than Whites of having uncontrolled hypertension.15
The treatment rates for hypertension in our study were relatively high. There are several reasons for this. There were 371 (8.7%) patients seen in the cardiology section of the clinic in addition to being followed by their PCP. The treatment rate was higher among those seen by a cardiologist (92%) compared with those who were only seen by their PCP (72%), whereas goal attainment was similar (56% vs. 57%). Because of the small number and interpretation confounded by indication, it is difficult to draw conclusions regarding the effect of provider status on dyslipidemia management. There could also be a heightened awareness among PCPs, as the clinic holds weekly conferences with several each year conducted by the cardiologist. Finally, we feel that the model of a specialized clinic for patients with HIV that includes co-located medical and support services may optimize treatment of CVD risk factors.
Although treatment rates for hypertension were relatively good, attainment of goal was low. A reason for this might be the influence of ART, in particular PI.16,17 In a study of all HIV-infected patients at 2 naval medical centers in the United States, hypertension was common (31%) among the 707 mostly male patients and did not differ between those receiving and those not receiving ART.18 However, in the Multicenter AIDS Cohort Study of 5578 participants, prolonged use of ART was significantly associated with a higher prevalence of systolic hypertension.16 In our study, 88% of patients were taking at least 1 ART medication and 66% of those on ART were on a regimen that included a PI. For hypertension, being on ART was associated with attaining blood pressure goal. Having a high viral load was also associated with not attaining blood pressure goal, whereas CD4 count did not show significant association. The significance and interpretation of these findings is unclear, as these analyses are exploratory, and the observed findings warrant further validation and elucidation.
Unlike many studies investigating CVD risk factors in patients infected with HIV, our study provided information on prevalence, treatment, and control of 2 important cardiovascular risk factors, dyslipidemia and hypertension. Our study population consisted of patients in a HIV clinic where all patients received standardized care and support without regard for ability to pay, which may minimize some confounding and heterogeneity issues in access to care. Also, the study sample is large and diverse in gender, age, race/ethnicity, and risk sources/status, which may make subgroups analyses meaningful.
There are several limitations to this study. Because of the cross-sectional design, results should be interpreted as prevalence (rather than risk or prospective event rate) and correlation (not causation or prospective association). Data abstraction from 2 of the 3 clinics was performed at a different calendar time than that from the third clinic. The proportion of missing data was varied but we addressed this issue by a standard statistical technique. Also, patients are from a single institution in an urban area; thus, representativeness or generalizability may be an issue. Although we did have information on the use of PIs, we did not have information on other classes of ART and specific PIs or on adherence to medical regimens. Finally, not all values of LDL-C were from fasting samples.
This study showed that control of LDL-C dyslipidemia among patients with HIV was good overall except for those at high risk, whereas control of high blood pressure was poor. It is still unclear if and how ART may influence CVD risk factors and make control more difficult or easier.
This report offers new information on management and goal attainment of CVD risk factors in patients living with HIV and insight into the HIV care model that includes expertise in the management of comorbidities, such as CVD risk factors, in addition to primary care for HIV. As CVD is becoming a leading cause of morbidity and mortality in these patients, our findings could provide a basis for further investigation regarding optimal treatment regimens and may support the development of guidelines specific to this population of patients.
The authors thank David Ferris, MD, and Rituparna Pati, MD, and the staff, providers, and patients in the Spencer Cox Center for Care, St. Luke's and Roosevelt Hospitals.