Remember me
Youths living with HIV is a high-priority population within those living with HIV []. Youths living with HIV will need to manage their diagnosis for years longer than their adult peers, and negative health consequences associated with living with HIV may appear earlier in the life course [-]. Research shows significant gaps in the HIV treatment cascade in younger populations compared to older groups [-]. Youths living with HIV aged 18-24 years have lower uptake of HIV testing and lower rates of treatment initiation compared to older peers, which continues to present a significant challenge to epidemic control [-]. Notably, racial disparities are more pronounced in youths [,]. Youths who identify as Black and Latinx represent 54% and 25% of new HIV diagnoses (79% in aggregate), compared to those who identify as White and other races, accounting for 16% and 5% of new HIV diagnoses []. There is growing evidence to suggest that Black individuals are at a higher risk for cardiovascular disease (CVD) compared to individuals in other racial and ethnic groups []. Several factors contribute to this increased risk. One of the most significant is systemic racism and discrimination, which can result in chronic stress and inflammation that can damage the heart and blood vessels over time []. According to the American Heart Association, Black individuals are more likely to have hypertension and less likely to have it under control compared to other racial and ethnic groups []. Black individuals are also more likely to develop diabetes and are at a higher risk of dying from diabetes-related complications. These conditions are major risk factors for CVD and contribute to the increased risk seen in Black individuals.
Cardiovascular health disparities are well documented among older adults with HIV [-], but there is a dearth of research on cardiovascular health metrics and prevention strategies for youths living with HIV in the United States. Studies have shown that older adults with HIV have a higher risk of developing CVD compared to the general population, which is likely due to a combination of factors, including HIV-related inflammation, the use of antiretroviral therapy (ART), and lifestyle factors, such as smoking and poor diet [,,]. However, there is a lack of research on the cardiovascular health of youths living with HIV in the United States, despite the fact that this population is growing and faces unique challenges to their health. Research on cardiovascular health metrics is therefore critical to the overall understanding of the impact of HIV infection on behavioral factors and the relationship between HIV infection and cardiovascular risk. Research has suggested that indicators of CVD may appear earlier in individuals living with HIV, with some studies suggesting that these indicators can emerge as early as adolescence [,,], potentially due to a combination of factors. One factor is that HIV infection can cause chronic inflammation, which has been linked to the development of atherosclerosis, a key contributor to CVD [,]. Another factor is that some ARTs used to manage HIV infection may have adverse effects on lipid metabolism, leading to the accumulation of fatty deposits in the blood vessels and an increased risk of CVD [].
In addition to these factors, individuals living with HIV may also have a higher prevalence of traditional CVD risk factors, such as hypertension, diabetes, and smoking, which can further increase their risk of developing CVD []. Stigma and discrimination related to HIV infection can also contribute to poor mental health outcomes, which have been linked to an increased risk of CVD []. Given the potential for CVD indicators to emerge earlier in individuals living with HIV, it is important to develop effective prevention and treatment strategies targeted toward this population. This may include identifying and managing traditional CVD risk factors, optimizing ART to minimize potential negative effects on lipid metabolism, and addressing mental health concerns through counseling and other interventions. The American Heart Association’s Life’s Simple 7 (LS7) concept of ideal cardiovascular health is not well understood in youths living with HIV when compared to the Healthy People 2020-2030 goals [-]. A higher LS7 score indicates better cardiovascular health and is associated with a lower incidence of CVD [-]. Traditional assessments of individual cardiovascular risk factors are inadequate in capturing the overall risk posed by multiple factors at the population level []. It is important to consider risk factors that occur together [,] as cardiovascular risk factor profiles [-] and assess them using composite measures such as the Framingham Risk Score (FRS) []. The modified FRS is the most commonly used algorithm for cardiovascular risk assessment in the United States [,].
Reducing cardiovascular risk in youths living with HIV is a critical issue that requires attention. It is essential to conduct cardiovascular risk assessment in this population to optimize the early diagnosis and treatment of CVD during adolescence []. However, research focused on comorbidities in HIV has rarely explored the connection between detectable viral load (VL) and low cluster of differentiation 4 (CD4) lymphocyte counts with increased cardiovascular risk in a US-based population of youths living with HIV. Additionally, there are currently no published data on the prevalence of traditional cardiovascular risk factors, such as dyslipidemia, hypertension, diabetes, and smoking, among youths living with HIV aged 14-26 years, even though metabolic changes leading to atherosclerosis can begin early in life, and go undiagnosed for an extended period. Despite this, cardiac risk estimating algorithms like the FRS have not been applied to younger populations [,]. Therefore, the aim of this study was to develop cardiovascular risk profiles for a cohort of US-based youths living with HIV and compare their profiles based on VL and CD4 lymphocyte status.
The study analyzed deidentified electronic health record (EHR) data of adolescents for the Adolescent Medicine Trials Network protocols. The Adolescent Medicine Trials Network protocols were previously described in a publication [], and the participating sites included in the EHR extraction protocol [] provide HIV care to youths living with HIV. This care begins with a diagnosis, followed by linking them to an HIV care provider who can help manage their HIV on a regular basis.
Ethics ApprovalThe institutional review board (IRB) of Florida State University granted approval (STUDY00000549) for the analysis of deidentified data, which was carried out in compliance with the US Department of Health and Human Services 45 Code of Federal Regulations Part 46. The parent protocol did not involve an informed consent process, and a waiver of consent and HIPAA (Health Insurance Portability and Accountability Act) waiver were granted. To protect confidentiality, the data were deidentified and participants were identified only by a unique study ID. The data were analyzed in aggregate to compare the clinic site with individuals, and a centralized data extraction process was used at all sites. Each participating site received a signed statement confirming the IRB’s approval before the study began, and a reliance agreement was obtained from each site indicating their commitment to providing deidentified EHR data in accordance with the IRB’s approval.
The extracted EHR data included those who had standard of care and treatment visits at one of the participating sites between January 1, 2016, and December 31, 2016. The baseline data extraction cycle was launched in December 2017 and completed in April 2018. Once the data files were verified for each site, they were merged to produce a single analytic data set of 1093 youths living with HIV. To be included in these analyses, data on demographics, HIV biomarkers, vital statistics, cholesterol panel, and the International Statistical Classification of Disease and Related Health Problems, Tenth Revision (ICD-10), diagnosis codes were required. Data downloads from Baltimore, Maryland; Birmingham, Alabama; Los Angeles, California; Memphis, Tennessee; San Diego, California; Tampa, Florida; and Washington, District of Columbia; were included in the analyses. Because Brooklyn, New York; Miami, Florida; and Philadelphia, Pennsylvania; did not provide gender and systolic blood pressure data, these 3 sites were excluded from further analyses.
MeasuresDemographicsWe defined age as a continuous variable in years as reported by the site. Gender, race, and ethnicity definitions varied within a site’s data download. For gender, we constructed a binary variable of man and woman. Race and ethnicity were categorized into Black, Latinx, White, and other.
HIV and Immunologic BiomarkersEarliest VL values in “copies/mL” by date were used from each site. VL was log-transformed for analysis as a continuous outcome. The detectable VL threshold as reported by the site was coded as a dichotomous variable. Based on the guidelines for the use of antiretroviral agents in HIV-1 infected adults and adolescents, we defined immunocompromised as a CD4 T lymphocyte count of less than or equal to 200 [,]. CD4 lymphocyte count was divided by 100 to enhance the interpretability of the coefficients to correspond to a 100-cell increase rather than a 1-cell increase.
ART MedicationAll prescribed HIV medications were coded into a dichotomous variable, where 0=no prescribed medication and 1=at least one prescribed medication.
Substance Abuse DiagnosisWe used ICD-10 codes F1010, F1210, F1220, F1290, F1510, F1511, F1520, F1590, F17200, F17210, F1910, and F1920 to identify alcohol and drug dependence. Substance abuse was treated as a dichotomous variable.
BMIBMI was calculated based on weight and height (kg/m2). If multiple values for a given patient were present in EHR, values from the earliest date were used.
Cardiac Risk ScoresWe adopted the gender-specific cardiovascular risk algorithm developed by the Framingham Heart Study [] to construct a cardiovascular risk variable referred to as Cardiac Risk Score1 using clinic-based predictors that are routinely obtained in primary care and do not require laboratory testing. The variables required for constructing the Cardiac Risk Score1 include age, gender, systolic blood pressure, use of antihypertensive medication, smoking status, and diabetes status. As gender and systolic blood pressure data were required for the development of the Cardiac Risk Score1, those without these data were excluded from the first analytic sample.
We also constructed a second cardiovascular risk variable referred to as Cardiac Risk Score2 that used routinely obtained clinic-based predictors including laboratory testing. The variables required for constructing the Cardiac Risk Score2 include age, gender, systolic blood pressure, antihypertensive medication use, current smoking, diabetes, total cholesterol, and high-density-lipoprotein (HDL) cholesterol. As gender, total cholesterol, HDL cholesterol, and systolic blood pressure data were required for the development of the Cardiac Risk Score2, those without this data were excluded from the second analytic sample. The CONSORT (Consolidated Standards of Reporting Trials)-style diagram illustrates a comprehensive assessment of all missingness in the construction of Cardiac Risk Score1 and Cardiac Risk Score2 (). We assigned an age value of 21 years. Following the convention of the existing CVD prediction algorithm developed based on data obtained from the Framingham Heart Study, we constructed a separate gender-specific multivariable risk function algorithm for men versus women []. The syntax for both Cardiac Risk Score1 and Cardiac Risk Score2 can be found in the .
Figure 1. CONSORT (Consolidated Standards of Reporting Trials)-style diagram illustrating a comprehensive assessment of all missingness in the construction of Cardiac Risk Score1 and Cardiac Risk Score2. CD4: cluster of differentiation 4; HDL: high-density lipoprotein; VL: viral load. Data AnalysisWe included youths living with HIV who had health records of systolic blood pressure, antihypertensive medication use, current smoking, and diabetes status in 2016, in the first analytic sample for Cardiac Risk Score1 (n=813). Subsequently, we constructed a second analytic sample for Cardiac Risk Score2 (n= 398) by including total cholesterol and HDL cholesterol in addition to the clinic-based predictors mentioned above. As not all patients had records for these additional variables, the sample size of Cardiac Risk Score2 was reduced to 398.
We performed a descriptive analysis of the demographic and clinical characteristics of the sample, followed by bivariate analyses using the chi-square test and independent samples t test with detectable VL and low CD4 lymphocyte count as a dichotomous variable. Then, we ran 3 multivariable linear regression models with Cardiac Risk Score1 as a continuous outcome variable to determine the association between detectable VL and cardiovascular risk, impaired immune function (CD4 lymphocyte count ≤200) and cardiovascular risk, and the combined effect of detectable VL and low CD4 lymphocyte count with cardiovascular risk. We repeated these regression models with Cardiac Risk Score2 as the outcome variable. Covariates included age, race and ethnicity, gender, being prescribed ART medication, substance dependence, and BMI.
The linear regression coefficients were evaluated using unstandardized beta (B) as the point estimate, CI was 95% for the variability around that point estimate, and P value for statistical significance. Standardized beta (β) was used to estimate the effect size of independent variables on the dependent variable, categorized as small (.10), medium (.30), and large (.50) according to Cohen’s criteria. SPSS Statistics (version 25; IBM Corp) was used for all analyses.
provides a summary of the demographic and clinical characteristics of the analytic sample of the study, which includes 813 youths living with HIV. The mean age of the sample was 21 (SD 2.6) years. The sample was predominantly of racial and ethnic minorities, with 70.3% (570/811) of them identifying as Black and 10.9% (88/811) of them as Latinx. Additionally, the majority of the sample were men (563/813, 69.2%). The study also presents the prevalence of detectable VL and impaired immune health, indicating that 47.8% (366/766) of those with VL data had detectable VL and 8.6% (51/593) of those with CD4 lymphocyte data had a baseline count ≤200. The proportion of youths living with HIV who were currently taking ART was high at 88.6% (720/813). The mean BMI of the sample was 25 (SD 6.7) kg/m2, and the mean Cardiac Risk Scores 1 and 2 were 0.062 (SD 0.039) and 0.672 (SD 0.401), respectively.
Table 1. Demographic and clinical characteristics of the sample (N=813).CharacteristicsValuesRace and ethnicity (n=811)aART: antiretroviral therapy.
bVL: viral load.
cCD4: cluster of differentiation 4.
dATN: Adolescent Medicine Trials Network.
eCardiac Risk Score1: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, and antihypertensive medication use.
fCardiac Risk Score2: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, antihypertensive medication use, total cholesterol, and high-density-lipoprotein cholesterol.
Bivariate Correlates of HIV and Immunologic BiomarkersDemographic and Clinical Correlatespresents findings on the demographic and clinical correlates of the study population; the table highlights several general trends observed in the data. First, patients who had detectable VL were more likely to be Black than those who had undetectable VL. The proportion of Black patients was 74% (271/366) among those with detectable VL, compared to 67% (268/400) among those with undetectable VL. This difference was statistically significant (χ²3=9.9; P<.05). Second, patients with detectable VL were more likely to be men than those without detectable VL. Specifically, 72.7% (266/366) of patients with detectable VL were men, compared to 65.2% (261/400) of patients without detectable VL. This difference was also statistically significant (χ²1=4.9; P=.03). Finally, patients with detectable VL were less likely to be on ART medication than those with undetectable VL. Specifically, 87.7% (321/366) of those with detectable VL were on ART medication, compared to 92.5% (370/400) of those with undetectable VL. This difference was statistically significant (χ²1=5.0; P=.03). These findings suggest that demographic and clinical factors are important correlates of VL in this population.
The 10-year risk of developing CVD was estimated using 2 cardiovascular risk assessment variables, Cardiac Risk Score1 and Cardiac Risk Score2. In bivariate analyses, we found significant associations between VL and cardiovascular risk as measured by both cardiac risk scores. Specifically, the results show that the association between VL and cardiovascular risk was statistically significant for Cardiac Risk Score1 (P=.001) and Cardiac Risk Score2 (P=.007). However, no significant differences were found between the demographic characteristics and the immunologic biomarker, CD4 lymphocyte count, as displayed in . When comparing patients who had a CD4 lymphocyte count of ≤200 with those who had a CD4 lymphocyte count of >200, no significant difference was found in Cardiac Risk Score1. However, the results show that with Cardiac Risk Score2, the risk of CVD was significantly increased in patients who had a CD4 lymphocyte count of ≤200 compared with those who had a CD4 lymphocyte count of >200 (P=.002). These findings suggest that VL is associated with an increased risk of CVD as measured by both Cardiac Risk Score1 and Cardiac Risk Score2, and that CD4 lymphocyte count may be an important factor in predicting cardiovascular risk using Cardiac Risk Score2.
Table 2. Bivariate analysis by detectable viral load (VL; N=766).CharacteristicsTotal sampleDetectable VL (n=366)Undetectable VL (n=400)F test (df)Chi-square (df)P valueRace and ethnicity (n=764), n (%)N/Aa9.881 (3).02aN/A: not applicable.
bART: antiretroviral therapy.
cCD4: cluster of differentiation 4.
dCardiac Risk Score1: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, and antihypertensive medication use.
eCardiac Risk Score2: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, antihypertensive medication use, total cholesterol, and high-density-lipoprotein cholesterol.
Table 3. Bivariate analysis by low cluster of differentiation 4 (CD4) lymphocyte count (N=593).CharacteristicsTotal sampleCD4 lymphocyte count≤200 (n=51)CD4 lymphocyte count>200 (n=542)F test (df)Chi-square (df)P valueRace and ethnicity (n=591), n (%)N/Aa6.548 (3).09aN/A: not applicable.
bART: antiretroviral therapy.
cVL: viral load.
dCardiac Risk Score1: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, and antihypertensive medication use.
eCardiac Risk Score2: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, antihypertensive medication use, total cholesterol, and high-density-lipoprotein cholesterol.
Multivariable Analyses of Cardiac Risk ScoresThe multivariable regression analysis of Cardiac Risk Score1 and Cardiac Risk Score2 scores is presented in and . For Cardiac Risk Score1, Model 1A showed a small effect of VL on increased cardiovascular risk (β=.067, SE 0.001; P=.008), while no significant association was found for CD4 lymphocyte count (Model 1B). Model 1C also demonstrated a significant positive association between VL and cardiovascular risk (P=.007), but no significant association was found for CD4 lymphocyte count. The findings of Model 1A and Model 1C were consistent.
For Cardiac Risk Score2, Model 2A showed a 38% increase in the likelihood of having cardiovascular risk for a 1000-point increase in VL (B=.038, 95% CI 0.011-0.066). Model 2B showed no significant association between CD4 lymphocyte count and cardiovascular risk. Model 2C demonstrated a significant positive association between VL and cardiovascular risk (P=.01), while that with the CD4 lymphocyte count was not significant (P=.54). The findings of Model 2A and Model 2C were consistent. The standardized beta value for the effect of VL on cardiovascular risk was only over .1 in both Cardiac Risk Score1 and Cardiac Risk Score2, indicating a small effect size.
Table 4. Multivariable linear regression Model 1 by Cardiac Risk Scores. Cardiac Risk Score1: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, and antihypertensive medication use.MultivariableModel 1A with VLa findingsModel 1B with CD4b lymphocyte findingsModel 1C with VL and CD4 lymphocyte findingsaVL: viral load.
bCD4: cluster of differentiation 4.
cART: antiretroviral therapy.
dN/A: not applicable.
Table 5. Multivariable linear regression Model 2 by Cardiac Risk Scores. Cardiac Risk Score2: defined as the risk score for patients with systolic blood pressure, smoking, diabetes, antihypertensive medication use, total cholesterol, and high-density-lipoprotein cholesterol.MultivariableModel 2A with VLa findingsModel 2B with CD4b lymphocyte findingsModel 2C with VL and CD4 lymphocyte findingsaVL: viral load.
bCD4: cluster of differentiation 4.
cART: antiretroviral therapy.
dN/A: Not applicable.
This study was among the first to examine the creation of cardiovascular risk profiles and assess their connections with HIV biomarkers in youths living with HIV aged 14-26 years. The results showed that detectable VL was statistically significantly linked to cardiovascular risk, even after adjusting for demographic and clinical factors, underscoring the need for integrating cardiovascular health and HIV care in regular clinical practice. The study also revealed that higher plasma VL was associated with a slight yet statistically significant increase in cardiovascular risk, regardless of CD4 lymphocyte count, ART exposure, and other demographic and clinical factors. These findings contribute to our understanding of the broad spectrum of the VL-cardiovascular risk relationship. In contrast, CD4 lymphocyte count did not demonstrate a connection to cardiovascular risk in any of the adjusted models.
There is currently no universally accepted method for evaluating cardiovascular risk and conveying this risk to youths living with HIV. The conventional approach of expressing increased risk of CVD as a probability of an event over the subsequent 10 years may discourage patients from adhering to healthier lifestyles and preventive care []. Using average age as a constant factor (the same for everyone) in our risk scores could pave the way for the development of an EHR-integrated monitoring device that assesses cardiovascular risk in young HIV-positive populations and enables effective risk communication to youths living with HIV. This approach could also have significant implications for clinical decision-making regarding treatment thresholds in youths living with HIV. Since age is a key factor in predicting absolute risk [], the use of our risk scores could help classify youths living with HIV who are at risk of cardiovascular health issues. Our study provides a foundation for future research that can further investigate this approach and apply it to HIV-negative youth populations. As we obtained similar results from both risk scores, the simpler version could serve as a cost-effective means of assessing high-risk HIV-positive youth for cardiovascular health conditions using readily available clinic-based predictors. However, it is crucial for researchers to periodically assess potential biases that may not have been apparent in this study.
Previous studies have indicated that the FRS is not effective in categorizing lifetime risk in younger individuals [] and have recommended relative risk estimates instead of age-dependent absolute risk estimates for those with low short-term risk []. A study that examined the ability of the FRS and the Adult Treatment Panel III to predict long-term risk for coronary heart disease death in young men aged 18-39 years found that neither method identified individuals under 30 years of age as high risk despite significant risk factor burden []. In this study, age was kept constant, meaning all individuals aged 18-29 years were given the risk estimate of a 30-year-old []. However, using continuous risk scores in our study could offer advantages over arbitrary classifications of high cardiovascular risk. In a cohort of youths living with HIV with lower event rates than the original Framingham cohort, identifying only high-risk individuals based on a >20% absolute risk in 10 years, despite significant risk factor burden, may not be the most effective strategy for estimating and communicating cardiovascular risk to younger individuals.
The FRS, on the other hand, has its own set of advantages. Given its strengths, we opted to adapt the FRS as a tool to assess CVD risk among youths living with HIV, despite the existence of other risk assessment algorithms. FRS was developed using a large, community-based cohort of individuals from Framingham, Massachusetts []. This population was representative of the US population at the time the study was conducted, which is important because HIV-positive individuals in the United States may have different risk profiles compared to those in other regions. FRS is based on traditional CVD risk factors, such as age, gender, blood pressure, total cholesterol, HDL cholesterol, and smoking status []. These risk factors are commonly assessed in clinical settings and are readily available in EHRs, making FRS a feasible tool for use in routine clinical care for youths living with HIV. FRS is a simple tool that is easy to use and understand, which may be particularly important for health care providers who may not have extensive training in CVD risk assessment. Other risk assessment algorithms may also be appropriate depending on the specific characteristics of the population being assessed. Nonetheless, there is still a need for further research to establish a definitive and widely accepted standard for assessing the risk of CVD in young adults living with HIV.
Despite the use of a multiclinic sample, there are potential limitations to the current findings. First, the study design was cross-sectional, preventing the es
Comments (0)