Author + information
- Received October 3, 2012
- Revision received January 30, 2013
- Accepted March 3, 2013
- Published online June 4, 2013.
- Jonathan C. Hsu, MD*,* (, )
- Yongmei Li, PhD†,
- Gregory M. Marcus, MD, MAS*,
- Priscilla Y. Hsue, MD‡,
- Rebecca Scherzer, PhD†,
- Carl Grunfeld, MD, PhD† and
- Michael G. Shlipak, MD, MPH†
- *Section of Cardiac Electrophysiology, Division of Cardiology, Department of Medicine, University of California, San Francisco, San Francisco, California
- †Department of Medicine, San Francisco Veterans Affairs Medical Center, University of California, San Francisco, San Francisco, California
- ‡Division of Cardiology, San Francisco General Hospital, San Francisco, California
- ↵*Reprint requests and correspondence:
Dr. Jonathan C. Hsu, Section of Cardiac Electrophysiology, Division of Cardiology, Department of Medicine, University of California, San Francisco, 500 Parnassus Avenue, MUE-434, Box 1354, San Francisco, California 94143.
Objectives The purpose of this study was to investigate the associations of traditional risk factors and longitudinal measures of human immunodeficiency virus (HIV) disease severity with risk of incident atrial fibrillation (AF) in a contemporary cohort of HIV-infected individuals.
Background Cardiovascular disease is common in HIV-infected persons; however, the most common cardiac arrhythmia, AF, has not been adequately studied in this population.
Methods We studied a national sample of 30,533 HIV-infected veterans followed in the Veterans Affairs HIV Clinical Case Registry from 1996 to 2011. We examined the independent associations of demographic characteristics, time-updated comorbidities, and time-updated clinical measurements including CD4+ cell count and viral load with the outcome of incident AF using proportional hazards regression for multivariable analysis.
Results Over a median follow-up of 6.8 years, 780 (2.6%) patients developed AF. After multivariable adjustment for traditional risk factors, a lower CD4+ cell count (<200 compared with >350 cells/mm3; hazard ratio [HR]: 1.4; 95% confidence interval [CI]: 1.1 to 1.8; p = 0.018) and higher viral load (>100,000 compared with <500 copies/ml; HR: 1.7; 95% CI: 1.2 to 2.4; p = 0.002) were independently associated with increased risk of incident AF. Additional risk factors independently associated with risk of AF included older age, White race, coronary artery disease, congestive heart failure, alcoholism, proteinuria, reduced kidney function, and hypothyroidism.
Conclusions In a large HIV-infected cohort, markers of HIV disease severity represented by low CD4+ cell count and high viral load, assessed by multiple time-updated measures, were independently associated with development of AF.
The human immunodeficiency virus (HIV) pandemic continues to be a major public health problem in the United States, affecting over 1.1 million people (1). Since the advent of highly active antiretroviral therapy, the natural history of HIV has drastically changed, as HIV mortality, acquired immune deficiency syndrome (AIDS), and AIDS-related hospitalizations have decreased substantially, leading to increased life expectancy (2). However, in the aging HIV-infected population, coronary artery disease, peripheral vascular disease, and congestive heart failure have emerged as growing cardiovascular health problems associated with significant morbidity and mortality (3,4).
The most common cardiac arrhythmia, atrial fibrillation (AF), is also associated with considerable morbidity such as stroke and heart failure (5), and has emerged as a growing problem in the aging U.S. population (6). HIV infection is known to be an independent risk factor for atherosclerosis and stroke (7,8), and similar mechanisms may be involved in the genesis of atherosclerotic cardiovascular disease and AF. To our knowledge, no previous study has examined the incidence of AF in HIV-infected persons despite shared risk factors associated with both atherosclerosis and AF. Additionally, no study has evaluated whether advanced or uncontrolled HIV infection is associated with the risk of developing AF.
We analyzed data from the Department of Veterans Affairs (VA) HIV Clinical Case Registry, a national registry of HIV-infected veterans that captures patient demographic and clinical information, healthcare utilization, and outcomes. In this large population of HIV-infected patients, we examined the incidence of AF to determine the burden of AF in HIV-infected persons and evaluate the strength of known AF risk factors in this population. In order to investigate a potential association of HIV disease severity with the risk of developing AF, we assessed whether a lower CD4+ cell count and a higher HIV RNA viral load were associated with an increased risk of developing AF independent of traditional risk factors.
The VA HIV Clinical Case Registry was created to allow VA monitoring of healthcare utilization for all HIV-infected veterans. The registry contains all demographic, clinical, laboratory, pharmacy, utilization, and vital status information entered into the VA electronic medical record for HIV-infected persons (9). We investigated the association between HIV severity indices (CD4+ cell count and viral load), patient demographics, comorbidities, and clinical measures with development of AF in this large, national sample of HIV-infected veterans. The Veterans Health Administration offers low-cost, comprehensive medical care to U.S. veterans with a broad, nationwide geographic scope (10).
We identified 30,820 patients diagnosed with HIV infection between January 1996 and January 2011 in the VA HIV Clinical Case Registry. Among these individuals, we excluded 90 patients with prevalent AF at entry and 197 patients who died within 1-month follow-up, leaving 30,533 patients for our analysis. Patients entered the cohort at the time of their first known documentation of HIV diagnosis and were censored at their first episode of AF, death, or last date of follow-up (January 2011).
The primary outcome was time from study entry to incident AF diagnosis, defined by database coding for first inpatient hospitalization or outpatient diagnosis of AF. The primary AF outcome comprised a diagnosis of AF or atrial flutter, specified by International Classification of Diseases-Ninth Edition (ICD-9) codes 427.31 and 427.32, respectively. Patients were classified as either having AF or atrial flutter based on the first arrhythmia diagnosed when censored. Previous studies have demonstrated a high sensitivity and specificity (95% and 99%, respectively) for the diagnosis of AF after extensive duplicative review of medical records to verify the diagnosis and confirm that the AF was new onset (11).
The primary predictors of interest were serological measures of HIV disease severity, including CD4+ cell count expressed in cells/mm3 and viral load expressed in RNA copies/ml. CD4+ cell count and viral load were treated as time-varying predictors, updated over time with the last value carried forward method. This method has been used in prior studies examining immunologic and virological predictors of HIV-related disease, and avoids misclassification due to reliance upon a single measurement at a single fixed point in time (12,13). For purposes of analysis, we divided CD4+ cell count and viral load into clinically meaningful ordinal categories, from low to high, as follows: CD4+ cell count <200, 200 to 350, or >350 cells/mm3 and viral load <500, 500 to 29,999, 30,000 to 100,000, and >100,000 RNA copies/ml.
We also investigated traditional demographic characteristics and comorbidities known to be associated with AF in our multivariate models to assess for their strength of association and to adjust for confounding in our multivariable models. Age was categorized by decades around the mean age of the cohort, as well as by both extremes of younger (<35 years) and older (≥65 years) patients. Male versus female sex was self-reported by veterans. Race was also self-reported, and categorized as white, black, or other (14). We identified comorbid illnesses using a combination of inpatient hospital discharge diagnoses, outpatient diagnoses, physician problem lists, procedures, and laboratory results (Online Appendix). We applied validated algorithms to define the following comorbidities: hypertension, diabetes, coronary artery disease, congestive heart failure, chronic lung disease, smoking, alcoholism, hyperthyroidism, and hypothyroidism (15–18). Clinical measures were also included in the analytical models and included body mass index, kidney function by estimated glomerular filtration rate (eGFR), and proteinuria dipstick. Baseline characteristics were defined if ever diagnosed before or at the time subjects entered the study, and all covariates were time-updated during the follow-up period.
We initially compared baseline characteristics of participants who eventually did and did not develop incident AF. Normally distributed continuous variables were expressed as means and standard deviations and were compared by unpaired t tests, whereas continuous variables not normally distributed were expressed as medians and interquartile ranges (IQR), and were compared by the Wilcoxon rank sum test. Categorical variables were expressed as percentages and were compared by the chi-square test. We compared incidence rates with 95% confidence intervals (CI) of AF, stratified by age and race and were presented as number of events per 1,000 person-years. We obtained estimates of relative risk using a time-to-event analysis with multivariable Cox proportional hazards regression models including demographic and traditional risk factors, as well as markers of HIV severity (CD4+ cell count and viral load). To determine which covariates to include in the final multivariate models, likely predictors and confounders were specified a priori and included for face validity. Other candidate covariates were included in the multivariate models if they were associated with either the primary predictors of interest or the outcome of AF from general clinical knowledge and data from prior studies. In our final model, we adjusted for demographic characteristics, markers of HIV disease severity (CD4+ cell count and HIV RNA viral load), comorbidities (hypertension, diabetes, coronary artery disease, congestive heart failure, chronic lung disease, smoking, alcoholism, hyperthyroidism, and hypothyroidism), and clinical measures (body mass index, eGFR, and proteinuria). We assessed the proportional hazards assumption using the Schoenfeld test and by comparing plots of log (−log[survival]) versus log of survival time. A missing indicator variable was included to retain observations for missing data in our final multivariable models, unless this caused violation of the proportional hazards assumption. Analyses were performed using Stata version 11.0 (StataCorp, College Station, Texas).
Of the 30,533 HIV-infected veterans who comprised the study cohort, we identified 780 (2.6%) patients who developed AF over a median duration of HIV infection and follow-up of 6.8 years (IQR: 2.9 to 11.2), which amounted to a total of 214,012 person-years of observation. These events comprised 641 diagnoses of AF, and 139 of atrial flutter. There were 60 patients who had both AF and atrial flutter during follow-up. Among patients who developed AF, 68% (534 of 780) were hospitalized for an alternative reason before the diagnostic AF event. There were 7,163 total deaths during the follow-up period. The overall incidence rate of AF in our cohort was 3.6 events per 1,000 person-years (95% CI: 3.4 to 3.9); however, we also stratified overall AF incidence rates by age and race (Fig. 1). The crude incidence of AF increased with age, and was 15-fold higher in patients ≥65 years of age (15.5 events per 1,000 person-years; 95% CI: 13.0 to 18.5) compared with those younger than 35 years of age (1.1 events per 1,000 person-years; 95% CI: 0.8 to 1.5). Among patients age <45 years, blacks and whites had similar rates of AF. By contrast, among those age ≥45 years, blacks had consistently lower rates of AF.
At baseline, patients who went on to develop incident AF were older, and more likely to be white and to have traditional AF risk factors, including hypertension, diabetes, coronary artery disease, heart failure, overweight/obesity, reduced eGFR, and proteinuria (Table 1). Among those diagnosed with AF, the CHADS2 score at diagnosis was 0 in 32.3%, 1 in 37.2%, 2 in 18.3%, 3 in 9.2%, 4 in 2.4%, and 5 in 0.5%. Throughout follow-up, we analyzed a median of 31 (IQR: 0 to 73) measurements of CD4+ cell count per subject, and a median of 57 (IQR: 13 to 109) total measurements of HIV viral load per subject. Baseline median CD4+ cell count and viral load were similar among veterans who did and did not develop AF.
Association between markers of HIV disease severity with incident AF
In this HIV-infected cohort, overall, there was a significant association between worsened HIV disease severity as measured by time-updated CD4+ cell count and HIV RNA viral load and the risk of AF. Lower CD4+ cell count (<200 cells/mm3 compared with >350 cells/mm3) and higher HIV RNA viral load (>100,000 copies/ml compared with <500 copies/ml) were independently associated with 40% and 70% higher risks of AF, respectively, in models that adjusted for demographic characteristics, time-updated comorbidities, and time-updated clinical measurements (Table 2). We performed a subanalysis stratified by age at baseline (<50 years and ≥50 years). This analysis suggested that a higher HIV RNA viral load (>100,000 copies/ml compared with <500 copies/ml) was associated with an increased risk of AF in both younger (hazard ratio [HR]: 1.9; 95% CI: 1.2 to 3.1; p = 0.007) and older (HR: 1.6; 95% CI: 0.9 to 2.6; p = 0.081) patients. Although the association appeared to be somewhat weaker in the older age group, there was no statistically significant difference in the association of HIV viral load with AF in the 2 age groups (test for interaction: p = 0.817). Similarly, we observed the association of low CD4+ count (<200 cells/mm3 compared with >350 cells/mm3) with AF to be equivalent in both younger (HR: 1.3; 95% CI: 0.9 to 2.0; p = 0.158) and older (HR: 1.4; 95% CI: 1.0 to 2.0; p = 0.062) participants (test for interaction: p = 0.935).
In exploratory analysis of patients who only developed AF (excluding those with atrial flutter), we found a similar association between markers of HIV disease severity and development of AF. After adjustment, a higher HIV RNA viral load (>100,000 copies/ml compared with <500 copies/ml) was significantly associated with an increased risk of AF (HR: 1.6; 95% CI: 1.1 to 2.4; p = 0.011), and a lower CD4+ cell count (<200 cells/mm3 compared with >350 cells/mm3) had a nonsignificant association with increased AF risk (HR: 1.3; 95% CI: 0.9 to 1.7; p = 0.106).
Because patients with worsened HIV disease severity may more often utilize healthcare, we performed exploratory analyses that adjusted for the number of annual outpatient visits as well as number of annual inpatient hospitalizations. After adjustment for the number of annual outpatient visits in the original multivariable model, both a higher HIV RNA viral load (>100,000 copies/ml compared with <500 copies/ml; HR: 1.8; 95% CI: 1.3 to 2.5; p = 0.001), and a lower CD4+ cell count (<200 cells/mm3 compared with >350 cells/mm3; HR: 1.3; 95% CI: 1.0 to 1.8; p = 0.030) remained associated with an increased risk of AF. After adjustment for the number of annual inpatient hospitalizations in the original multivariable model, a higher HIV RNA viral load (>100,000 copies/ml compared with <500 copies/ml) continued to be associated with an increased risk of AF (HR: 1.4; 95% CI: 1.0 to 2.0; p = 0.036), whereas the association of AF with a lower CD4+ cell count (<200 cells/mm3 compared with >350 cells/mm3) was attenuated somewhat and was no longer statistically significant (HR: 1.2; 95% CI: 0.9 to 1.6; p = 0.181). In a sensitivity analysis, we investigated the association between HIV disease severity and the risk of AF after excluding events from patients diagnosed with a single, isolated inpatient AF episode (n = 115 exclusive of 780 with AF). After multivariable adjustment, a higher HIV RNA viral load (>100,000 copies/ml compared with <500 copies/ml) continued to be associated with an increased risk of AF (HR: 1.7; 95% CI: 1.1 to 2.4; p = 0.009), whereas a lower CD4+ cell count (<200 cells/mm3 compared with >350 cells/mm3) showed little association with AF risk (HR: 1.1; 95% CI: 0.8 to 1.5; p = 0.547).
Risk factors associated with incident AF
In demographic adjusted analyses, we found that markers of HIV disease severity, comorbidities, and clinical measures were all significantly associated with incident AF (Table 2). In models that adjusted for demographic characteristics, time-updated comorbidities, and time-updated clinical measurements, there was a progressive increase in the incidence of AF with increasing age decades; compared with those age <35 years, those ≥65 years had an approximate 8-fold increase in the adjusted risk of AF. Overall, blacks had an approximate 40% reduction in the adjusted risk of AF compared with whites. As seen in Table 2, HIV-infected persons with cardiovascular comorbidities including coronary artery disease and congestive heart failure had adjusted 2.4-fold and 4.8-fold increased risk of AF, respectively. Additional risk factors associated with an increased risk of AF after multivariable adjustment included alcoholism, hypothyroidism, kidney disease, and proteinuria.
In a large, national sample of HIV-infected veterans receiving care in the Veterans Health Administration, we observed a significant and graded association between markers of HIV severity and incident AF. Importantly, both a lower CD4+ cell count and a higher viral load assessed by multiple, time-updated measures provided complementary and independent prognostic information, with an overall pattern of worsened HIV disease severity corresponding with an increased risk of incident AF. It is of note that CD4+ cell count and viral load were similar at baseline in those who did and did not go on to develop AF, suggesting a more direct role of HIV severity in the development of AF. Nonetheless, traditional AF risk factors including increased age, white race, coronary artery disease, congestive heart failure, alcoholism, proteinuria, and reduced kidney function were also strongly and independently associated with increased AF risk in the setting of HIV. To our knowledge, this is the first and only study to evaluate AF incidence in HIV-infected individuals, and to establish a biologically plausible association of markers of HIV severity with incident AF. Our results are clinically meaningful because they are the first to explore incidence rates of AF in the HIV-infected, and the first to suggest that HIV infection severity may be directly linked with the risk of developing AF.
Although there are no previously reported studies examining the association of HIV infection and AF, other studies have found that HIV-infected individuals have higher than expected risk for a number of age-related conditions, including cardiovascular disease, certain types of cancer, osteoporosis, and kidney disease. Potential mechanisms may involve changes to the adaptive immune system seen in the very old (“immunosenescence”) and may potentially be related to persistent inflammation (19,20). Because both advanced age and inflammation have been implicated in an increased risk for the development of AF in previous studies, an association of HIV and AF could share similar mechanisms (21,22). In our study, the associations of HIV viral load and CD4+ cell count with AF were somewhat attenuated after controlling for the number of hospitalizations, suggesting that the association is potentially mediated by the severity of illness and its attendant inflammatory state.
Overall, in an HIV-infected population, our study found that older age predicted a consistently rising risk of AF development. In Table 3, we compare age-stratified crude incidence rates of AF in our HIV-infected cohort with previously published reports in large epidemiological studies from Manitoba (23), the Atherosclerosis Risk in Communities study (24), Framingham (25), Rotterdam (26), and Olmsted County, Minnesota (27). Our HIV cohort had higher representation of the younger age strata (patient age <65 years) due to the demographics of HIV-infected veterans. Crude incidence rates of AF were much higher among these younger HIV-infected patients in our study compared with community-living individuals without HIV reported in these other community-based studies. Our rates were even more striking because the comparison studies included fewer persons of African descent, who have lower AF risk, whereas blacks comprised approximately one-half of our cohort. These findings are suggestive, but not conclusive, of an increased incidence of AF in HIV-infected individuals, including younger patients, which can only be fully evaluated in future studies that include a non–HIV-infected comparison group.
In our study, both CD4+ cell count and HIV RNA viral load were independently associated with an increased risk of AF. These associations were only modestly attenuated after adjusting for traditional risk factors, suggesting a strong association that was not overtly confounded by the traditional risk factors included in our multivariate models. Low CD4+ cell counts seen in HIV-infected persons may have similarities to age-associated immunologic changes seen in HIV-uninfected older patients marked by defects in T-cell regenerative potential and loss of immunoregulatory function (28). In previous studies, a low CD4+ T-cell count has been associated with early-onset age-associated diseases, potentially by a contribution of immunosenescence, persistent immunodeficiency, and inflammation (19). A high HIV RNA viral load has been associated with persistently high levels of inflammation, as defined by levels of inflammatory cytokines such as interleukin-6, C-reactive protein, and fibrinogen, among others (29,30). Many of these inflammatory markers decline with antiretroviral therapy, suggesting that active HIV replication is either directly or indirectly responsible for this inflammatory response. Because similar inflammatory markers have been associated with AF (22,31), it is possible that viral replication itself may be associated with increased AF risk, potentially through an inflammatory mechanism that has yet to be elucidated. In general, our findings merit future studies investigating specific mechanisms by which HIV infection may lead to development of AF in affected individuals.
Our findings raise important questions for the clinical management of AF in the HIV-infected population. Although our findings suggest that HIV-infected individuals may be at higher risk of AF, particularly those with severe HIV disease, we do not yet know whether AF is associated with stroke, heart failure, or death in these populations. Many previous studies have shown a marked reduction in the risk of stroke with prescription of oral anticoagulation in AF patients with specific risk factors (32,33). These studies, in general, did not specifically include HIV patients; therefore, whether the same risk factors predispose HIV-infected patients to stroke, and whether oral anticoagulation reduces the risk of stroke in this population, remain to be investigated.
First, our cohort comprised HIV-infected individuals only, without a specific noninfected comparison group. The VA Clinical Case Registry is a well-developed disease registry, but it does not include uninfected controls. However, this is the first study to examine AF incidence in an HIV-infected cohort, and should be considered an important step in defining a possible link of HIV-infection with risk of AF. Future studies should include uninfected controls in order to enable a direct comparison of AF risk. Second, our findings may not be generalizable to specific populations poorly represented in our cohort, such as women, nonveterans, or HIV-infected persons without access to medical care. Third, although we adjusted for a number of potential clinical factors associated with both HIV-infection severity and AF, residual confounding is still possible. In particular, patients who went on to develop AF had more comorbid conditions, and we cannot exclude the possibility that additional unknown confounding variables not included in multivariable adjustment may account for some or all of the associations we found. Fourth, veterans with more severe HIV infection may be more likely to seek medical care or have screening tests such as an electrocardiogram. Therefore, we cannot rule out the possibility that detection bias accounted for some or all of the association we found between lower CD4+ cell count and higher viral load with higher risk of AF. Fifth, although previous studies have confirmed excellent sensitivity and specificity using ICD-9 codes for AF diagnoses (11), we were not able to confirm AF or atrial flutter diagnoses in our study by over-reading of electrocardiograms or telemetry strips because this information was not available in the VA Clinical Case Registry. Finally, we did not study antiretroviral therapy as a predictor of incident AF in this HIV-infected cohort. We focused on markers of HIV-infection severity as the first step in suggesting a link between HIV-infection and AF, which may or may not reflect antiretroviral treatment. Future studies should investigate the complex interplay of specific drug treatment for HIV and subsequent risk of incident AF.
In a large, national sample of HIV-infected persons, markers of HIV severity, including CD4+ cell count and HIV viral load were associated with development of AF even after adjustment for traditional risk factors. These findings suggest a potential biological link between HIV infection and risk of incident AF that should be the focus of additional studies comparing AF incidence and clinical outcomes in the HIV infected versus the HIV uninfected. In HIV-infected individuals, future research should evaluate mechanisms by which HIV disease severity increases AF risk, whether AF is associated with an increased risk of stroke, heart failure, and death, and whether or not oral anticoagulation improves outcomes in this population.
For criteria to define outcomes and comorbid illnesses based on previous published work, please see the online version of this article.
This research was supported by a grant from the National Institutes of Health, University of California, San Francisco–Gladstone Institute of Virology & Immunology Center for AIDS Research, P30-AI027763. Dr. Marcus receives speaker’s fees from St. Jude Medical; is a consultant for InCarda; and receives research support from Baylis; Medical, Gilead, and SentreHeart Inc. Dr. Hsue is an advisor for Gilead and Pfizer; and has received research support from Tobira.
All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acquired immune deficiency syndrome
- atrial fibrillation
- body mass index
- confidence interval
- estimated glomerular filtration rate
- human immunodeficiency virus
- hazard ratio
- interquartile range
- Veterans Affairs
- Received October 3, 2012.
- Revision received January 30, 2013.
- Accepted March 3, 2013.
- American College of Cardiology Foundation
- Saves M.,
- Chene G.,
- Ducimetiere P.,
- et al.
- De Castro S.,
- d’Amati G.,
- Gallo P.,
- et al.
- Dries D.L.,
- Exner D.V.,
- Gersh B.J.,
- Domanski M.J.,
- Waclawiw M.A.,
- Stevenson L.W.
- Lloyd-Jones D.M.,
- Wang T.J.,
- Leip E.P.,
- et al.
- Bruyand M.,
- Thiebaut R.,
- Lawson-Ayayi S.,
- et al.
- Choi A.I.,
- Li Y.,
- Deeks S.G.,
- Grunfeld C.,
- Volberding P.A.,
- Shlipak M.G.
- Miller D.R.,
- Safford M.M.,
- Pogach L.M.
- Chung M.K.,
- Martin D.O.,
- Sprecher D.,
- et al.
- Aviles R.J.,
- Martin D.O.,
- Apperson-Hansen C.,
- et al.
- Heeringa J.,
- van der Kuip D.A.,
- Hofman A.,
- et al.
- Miyasaka Y.,
- Barnes M.E.,
- Gersh B.J.,
- et al.
- Neuhaus J.,
- Jacobs D.R. Jr..,
- Baker J.V.,
- et al.