Prevalence of Cardio-metabolic Risk Factors in Hispanics Living with HIV

Abstract

Objective

Human Immunodeficiency Virus (HIV) infection and antiretroviral treatment are associated with metabolic and cardiovascular complications that resemble the metabolic syndrome (met-syndrome) and potentially increase the risk of diabetes and cardiovascular disease in this population. The purpose of this study was to determine the prevalence of met-syndrome and its individual components among Hispanics living with HIV in Puerto Rico (PR).

Methods

Data from 909 clinical records were extracted and the prevalence of met-syndrome determined using the NCEP-ATPIII criteria. Fisher's exact test was used to detect gender differences, and logistic regression to examine the effect of age, gender, smoking, years of HIV infection, antiretroviral therapy, and Hepatitis C co-infection.

Results

The prevalence of met-syndrome in our study group (35.4%) was higher than previously reported in the United States, but not higher than in the general population in PR. Females had a higher prevalence of met-syndrome (44.2%) than males (30.5%); mostly explained by high BMI and waist circumference. Age and gender were associated with the presence of met-syndrome.

Conclusion

Understanding ethnic and gender differences in the prevalence of metabolic risk factors is essential for the implementation of specific targeted interventions to prevent subsequent vascular morbidity and mortality in this population.

INTRODUCTION

The availability of highly active antiretroviral therapy (HAART) has significantly improved life expectancy for people living with Human Immunodeficiency Virus (HIV) infection. This clinical advance in therapy is associated with an increased risk of metabolic and cardiovascular complications such as visceral fat accumulation, dyslipidemia (i.e., high triglycerides, low HDL), insulin resistance, and elevated blood pressure^{1, 2}, all important components of the metabolic syndrome (met-syndrome). However, the potential synergy of metabolic complications in HIV infection, therapy, and race/ethnicity has not been evaluated.

Hispanics are disproportionately affected with diseases linked to the met-syndrome such as overweight/obesity, diabetes, and hypertension^{3, 4}, and the age-adjusted prevalence of met-syndrome among Hispanics in the United States (US) (40.6%) is higher compared with non-Hispanic Blacks (38.8%) and Whites (31.5%).⁵ Moreover, the age-adjusted prevalence of met-syndrome among Hispanics in Puerto Rico (PR) (38.1%)⁶ is similar to non-Hispanic Blacks in the US⁷, and higher than Hispanics in different Latin American countries.⁸ The prevalence of met-syndrome and cardio-metabolic risk factors among Hispanics living with HIV in PR is unknown, and little is known about the prevalence of these complications among Hispanics living with HIV in general. Therefore, the purpose of this study was to determine the prevalence of met-syndrome and the individual cardio-metabolic risk factors in Hispanic adults living with HIV in PR.

METHODS

Study Design

This cross-sectional study included all adult patients attending two HIV clinics and one HIV community-based alternative medicine program in San Juan, PR, between 2003 and 2007. These sites provide health care services to approximately 20% of the HIV/AIDS population living in PR⁹. Clinical records were reviewed and data extraction without personal identifiers completed by authorized personnel at each site. The consistency of data extraction and entry was checked by randomly selecting extraction forms, re-entering data, and comparing with the original file. Discrepancies were corrected after consultation with authorized personnel who verified the information with the original record in each site. Extraction forms excluded were those missing age, gender, height, weight, and two or more of the following: resting blood pressure, fasting glucose, triglyceride and HDL. The study was approved by the Institutional Review Board of the University of Puerto Rico, Universidad Central del Caribe, and the Mayo Clinic in Rochester, MN.

Study Outcomes

The most recent data available in each clinical record were used for analyses. Primary variables included: waist circumference, body mass index (BMI: kg/m²), fasting glucose, triglycerides and HDL, use of antihypertensive and lipid control medications, resting systolic and diastolic blood pressures, and diagnosis of diabetes. Waist circumference was used as an index of visceral fat and BMI as an index of obesity, both were collapsed into one criterion called body shape. Other variables included: gender, date of birth, date and age at HIV diagnosis, viral load, CD4 count, education, alcohol consumption, hepatitis C co-infection, smoking, and history of antiretroviral medications.

The National Cholesterol Education Program – Adult Treatment Panel III (NCEP-ATPIII) criteria were used to determine the proportion of participants having one or more of the cardio-metabolic risk factors^{10, 11}, and the proportion having the met-syndrome.

Data Management and Statistical Analyses

To include all participants and avoid biases, missing values were imputed using the Markov Chain Monte Carlo multiple-imputation method. Then, a sensitivity analysis was conducted including 280 patients with complete datasets. Descriptive statistics were performed for socio-demographic and research variables (i.e., means, standard deviations, and proportions). Fisher's exact test was used to identify gender differences in the prevalence of met-syndrome, visceral fat/obesity, dyslipidemia, high fasting glucose, and hypertension; and differences between our study group and the general population in PR using published data.⁶ The effect of age, sex, hepatitis C, and type of antiretroviral therapy was tested using logistic regression analysis. An alpha ≤0.05 was used for statistical significance using SPSS Statistical Software (Release 18, SPSS Inc., Chicago, III).

RESULTS

Study Participants

A total of 909 data extraction forms were received from the collaborating sites, and 13 did not meet the inclusion criteria (Figure 1). From the remaining 897 records, 574 (64%) were males (a gender distribution closely resembling that of the HIV/AIDS epidemic in PR), with a mean age of 44.7±10.1 yrs. All participants were 21 years of age or older, non pregnant, with no history of illicit drug use/abuse, and without current AIDS diagnosis. An algorithm including the number of records reviewed, extraction forms excluded, and number of missing values for each cardio-metabolic risk factor by gender is presented in Figure 1.

Figure 1.

Algorithm of Clinical Records Reviewed and Number of Missing Values for Each Cardio-Metabolic Risk Factor

Cardio-metabolic Variables

Socio-demographic and clinical characteristics organized by gender are presented in Table 1. No gender differences were observed for age, age at HIV infection, CD4, viral load, proportion with hepatitis C co-infection, and proportion using antiretroviral therapy. There were significant differences in the females reporting less formal education and males having a higher prevalence of smoking and alcohol abuse.

Table 1.

General Characteristics of Study Participants

Variable	All (n=897)	Females (n=323)	Males (n=574)	P-value
Age (yrs)	44.7 (10.1)	45.2 (10.1)	44.4 (10.1)	0.19
Age at HIV (yrs)	37.9 (10.7)	38.2 (10.9)	37.7 (10.5)	0.70
CD4 (cells/μL)	473 (322)	486 (326)	467 (320)	0.55
<200	19%	16%	20%	0.22
201–500	41%	44%	40%
>500	40%	40%	40%
HIV RNA (copies/ml)
Non detectable	51%	48%	52%	0.27
<1,000	13%	14%	12%
1,000 – 30,000	21%	23%	19%
>30,000	16%	14%	17%
Education ≥ High School (%)	56	48	60	<0.001
Smokers (%)	50	44	53	0.01
Alcohol abuse (%)	39	27	45	<0.001
Hepatitis C (%)	21	17	23	0.06
HAART-Naïve (%)	10	10	9	0.75
PI Only (%)	3	2	3	0.82
NRTI/NNRTI Only (%)	34	38	32	0.07
HAART (%)	45	42	46	0.31

Data are given as mean (SD) except when noted. HAART= highly active antiretroviral therapy, PI= protease inhibitors, NRTI/NNRTI= nucleoside and non-nucleoside reverse transcriptase inhibitors

The prevalence of cardio-metabolic variables organized by gender is presented in Table 2. Females had higher waist circumference, BMI, and HDL; and lower triglycerides and systolic blood pressure than males. Mean fasting glucose, and systolic and diastolic blood pressures were within normal values for both males and females. However, mean triglyceride, BMI and waist circumferences were higher than normal values in both males and females.

Table 2.

Cardio-Metabolic Characteristics of Study Sample

Variable	All	Females	Males	P-value
Fasting Glucose (mg/dL)	96.1 (28.2) (n=825)	95.0 (29.0) (n=296)	96.7 (27.8) (n=529)	0.11
<100	76%	77%	75%	0.69
100 – 124	16%	15%	17%
>124	8%	9%	8%
Body Shape [BMI (kg/m2)]	26.3 (5.4) (n=674)	27.2 (6.1) (n=236)	25.9 (4.9) (n=438)	0.007
<18.5	4%	5%	4%	0.03
18.5 – 24.9	41%	35%	45%
25.0 – 29.9	35%	35%	35%
>29.9	20%	25%	17%
Body Shape [Waist Circumference (cm)]	94.3 (5.6) (n=434)	96.6 (17.8) (n=164)	93.0 (13.9) (n=270)	0.03
Triglycerides (mg/dL)	193.8 (162.5) (n=799)	163.8 (99.0) (n=290)	210.9 (187.4) (n=509)	0.003
<150	50%	57%	46%	0.003
150 – 199	20%	19%	20%
200 – 500	26%	22%	28%
>500	4%	2%	6%
HDL (mg/dL)	43.7 (14.2) (n=374)	48.0 (15.9) (n=137)	41.2 (12.4) (n=237)	<0.001
<41	47%	34%	54%	<0.001
41 – 50	27%	25%	27%
51 – 60	16%	24%	11%
>60	10%	17%	7%
Blood Pressure (mmHg)	(n=838)	(n=307)	(n=531)
<120/80	43%	49%	39%	0.004
120–139/80–89	40%	36%	43%
140–159/90–99	14%	10%	16%
>160/100	3%	5%	3%
Systolic BP (mmHg)	120.5 (16.2)	118.1 (17.1)	121.8 (15.6)	<0.001
Diastolic BP (mmHg)	75.5 (10.6)	74.7 (10.9)	75.9 (10.4)	0.10
Hypertension (%)	18.5 (n=686)	18.0 (n=245)	18.8 (n=441)	0.78
Diabetes (%)	9.8 (n=519)	11.6 (n=190)	8.8 (n=329)	0.31

Data are given as mean (SD) except when noted. BMI= body mass index, HDL= high density lipoprotein, BP= blood pressure.

High fasting glucose (≥100 mg/dl) was observed in 24% of participants, 35% were overweight (BMI = 25–29.9 kg/m²), 20% were obese (BMI ≥30 kg/m²), 50% had high triglycerides (≥150 mg/dl), 74% had low HDL levels (≤50 mg/dl), and 58% had blood pressure levels considered in the pre-hypertensive/hypertensive category (>120/80 mmHg). No gender differences were observed in the proportion of patients diagnosed with diabetes or hypertension, conditions that developed after HIV diagnosis.

Metabolic Syndrome

The prevalence of met-syndrome using actual and imputed values were not different (40.0% vs. 35.4%, respectively; P=0.28). The prevalence of each component of the met-syndrome from actual and imputed values were also not different (i.e., elevated fasting glucose: 15.9% vs. 17.4%; elevated BMI and waist circumference: 36.4% vs. 39.7%; elevated triglycerides: 50.2% vs. 52.7%; low HDL: 54.0% vs. 51.2%; and elevated blood pressure: 49.6% vs. 48.3%; P>0.05 for all). Because of these findings, the following results are based on the imputed values.

Gender differences in the prevalence of met-syndrome and its individual components are presented in Figure 2. Females had a higher prevalence of met-syndrome compared to males (44% vs. 31%, respectively, P=0.04), and body shape was the only individual component significantly higher in women compared to men, indicating that women had higher abdominal or general obesity than men. Logistic regression also confirmed that males were less likely to have the met-syndrome compared to females (OR: 0.61, 95% CI: 0.42–0.92; P=0.02).

Figure 2.

Prevalence of Metabolic Syndrome and Individual Components by Gender

Older HIV patients (age ≥60 years) were more likely to have met-syndrome than those younger than 30 years of age (OR: 1.04, 95% CI: 1.01–1.08; P=0.01), and the prevalence was higher in females, particularly in the younger and older age group (Table 3). After adjusting for age and gender, factors such as smoking, alcohol abuse, education level, age at HIV infection, antiretroviral therapy, and co-infection with hepatitis C did not influence the prevalence of met-syndrome.

Table 3.

Proportion of Hispanic Adults Living with HIV Meeting the Criteria for Metabolic Syndrome Organized by Age Group

Age Group (yrs)	All (n=897)	Females (n=323)	Males (n=574)
< 30 (n=60)	17.2%	29.5%	11.5%
30–39 (n=195)	28.5%	38.2%	23.3%
40–49 (n=381)	37.1%	44.4%	33.1%
50–59 (n=201)	38.4%	45.6%	34.3%
≥ 60 (n=60)	55.3%	63.9%	47.8%

DISCUSSION

This is the first study to report the prevalence of cardio-metabolic risk factors and met-syndrome among Hispanics living with HIV in PR. The 35.4% prevalence of met-syndrome was high compared to the 24–26% prevalence reported among adults living with HIV in the US^{12, 13}, 18% in Australia¹⁴, and 17% in Spain¹⁵. However, the crude prevalence of the met-syndrome was not higher in the group of adults living with HIV compared to the general population in PR⁶ (35.4% vs. 43.3%, respectively, P=0.155). Also, the prevalence of individual components such as HDL, blood pressure, or obesity and central obesity were not different between our study group and the general population in PR⁶ (51.2% vs. 46.0%,P=0.27; 48.3% vs. 46.1%, P=0.44; 39.7% vs. 49.0%, P=0.13, respectively). These results are consistent with previous reports^{12, 13}, and suggest that environmental, socio-cultural and/or genetic factors are as influential as HIV infection and its treatment in explaining the risk of cardio-metabolic complications among Hispanics living with HIV in PR.

According to the CDC¹⁶, behavioral risk factors linked to cardio-metabolic complications, such as poor nutrition and low physical activity, are highly prevalent among Hispanics in PR compared to Hispanics and non-Hispanic Whites in the US (physical inactivity: 49% vs. 33.2% and 23.8%, respectively; consumption of fruits and vegetables: 14.3% vs. 20.4% and 23.5%, respectively). Hispanic adults living with HIV in PR are likely to have similar risk behaviors and therefore, similar risk of cardio-metabolic dysfunction as the general population in PR. However, we observed important differences in the prevalence of individual components of the met-syndrome between our study group and the general population in PR⁶. The two most important factors driving the prevalence of met-syndrome in the general population in PR⁶ were elevated fasting glucose (49.8%) and abdominal obesity (49.0%)⁶, while, similar to previous studies¹⁷, the two most influential factors in our study group were hypertriglyceridemia (52.7%) and low HDL levels (51.2%).

A lower prevalence of elevated fasting glucose and a higher prevalence of elevated triglycerides were the most important differences observed between our study group and the general population in PR (17.4% vs. 49.8%, P<0.01; and 52.7% vs. 31.3%, P<0.01; respectively). Mondy et al.¹³ reported similar results when they compared adults living with HIV and the general population in the US, suggesting that different approaches might be necessary for the management and control of cardio-metabolic complications in adults living with HIV compared to the general population.

Although obesity was not among the components of the met-syndrome with the highest prevalence in our study group, it has become an important health problem among females and males living with HIV. Before the introduction of HAART, elevated BMI was associated with a slower progression from HIV infection to AIDS; whereas now, elevated BMI is associated with high cholesterol, triglycerides, glucose, and insulin resistance.^{18, 19}

Males and females in the general population in PR did not differ in their prevalence of met-syndrome⁶; however, females in our study group had a higher prevalence of met-syndrome compared to males. Obesity/waist circumference was the only individual component with a higher prevalence among females compared to males, and similar to Mondy et al.¹³, we observed a higher mean BMI and waist circumference, and lower triglyceride levels among females compared to males. These differences suggest that gender-specific approaches might be needed for the prevention and clinical care of cardio-metabolic dysfunction in adults living with HIV.

The met-syndrome is known to increase the risk of cardiovascular disease, and the risk is higher as the number of individual components increases, particularly in females.²⁰ In the present study, the prevalence of at least one cardio-metabolic disorder was 92% (females = 95%, males = 91%) compared to 85% in the general population in PR⁶; and the prevalence of meeting all five criteria was 2% in our study group compared to 7% in the general population. Elevated triglycerides (53%) and low HDL levels (51%) were the components with the highest prevalence; high fasting glucose the component with the lowest prevalence (17%), and elevated resting blood pressure the third most influential component with a prevalence of 48.3%. This is higher than the prevalence of hypertension (17–21%) previously reported among adults living with HIV in the US.²¹ The presence of all mentioned factors combined with persistent levels of immune activation may explain the high frequency of cardiovascular events observed in this population.

Although many metabolic abnormalities in adults living with HIV have been attributed to antiretroviral therapy^{12, 14, 22, 23}, in the present study, specific antiretroviral classes were not related to met-syndrome after adjusting for age and sex. Compared to adults living with HIV that were naïve or were not taking antiretrovirals for at least two years, the odds of having met-syndrome in those taking HAART was not significant (OR:1.17, 95%CI; 0.65–2.15, P=0.62). This observation differs from Bergersen et al.²⁴ who reported a higher prevalence of met-syndrome in HAART treated compared with HAART-naïve patients, but are in agreement with Sobieszczyk et al.²² who reported no differences in the prevalence between HIV infected women naïve to antiretrovirals with those taking HAART.

Some important limitations in our study need to be addressed. Missing data is a known problem when using data extraction from clinical records, and using imputation methods is not always a valid remedy. However, when we compared the prevalence of met-syndrome and its individual components using imputation and actual values, no significant differences were detected. Another limitation in our data extraction was the lack of information regarding physical activity and nutritional behaviors, and the inconsistency in recording follow-up evaluations and metabolic related variables such as waist circumference, HDL, LDL, lipodystrophy and microalbuminuria. However, this is the largest dataset that has evaluated the prevalence of met-syndrome in a Hispanic HIV infected population.

In summary, the prevalence of met-syndrome among Hispanic adults living with HIV in PR was higher than previously reported in the US but not different than the general population in PR. The two most prevalent components of the met-syndrome in Hispanics with HIV (i.e., elevated triglycerides, low HDL) were different from the general population in PR (i.e. elevated fasting glucose, waist circumference), but elevated resting blood pressure was the third most prevalent component in both populations. We suggest that social and cultural influences might be as strong as the HIV infection itself and antiretroviral therapy in explaining the development of cardio-metabolic dysfunction in Hispanics living with HIV in PR. Understanding ethnic and gender differences in the prevalence of cardio-metabolic risk factors is essential for the implementation of specific targeted interventions to prevent subsequent vascular morbidity and mortality in this population.