Secondary Logo

Journal Logo

Epidemiology

Human Immunodeficiency Virus Increases the Risk of Incident Heart Failure

Yen, Yung-Feng MD, MPH, PhD*,†,‡; Ko, Ming-Chung MD, PhD†,§; Yen, Muh-Yong MD*,‡; Hu, Bor-Shen MD*; Wang, Teng-Ho MD*; Chuang, Pei-Hung PhD; Lai, Hsin-Hao MD*; Chen, Chu-Chieh PhD; Deng, Chung-Yeh MD, ScD

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: March 1, 2019 - Volume 80 - Issue 3 - p 255-263
doi: 10.1097/QAI.0000000000001917

Abstract

INTRODUCTION

More than 36.7 million people have been infected with HIV worldwide.1 Moreover, 1.0 million people died of AIDS and other HIV-related illnesses worldwide in 2016.1 With the success of highly active antiretroviral therapy (HAART), people living with HIV/AIDS (PLWHA) are aging, and more chronic diseases (eg, cardiovascular diseases) are being diagnosed in this population.

Heart failure (HF) in PLWHA can cause significant morbidity and mortality. The pathogenesis of HF is multifactorial, including incident coronary artery disease (CAD),2 myocarditis,2 myocardial fibrosis,3 and cardiac steatosis.4 Previous studies showed that PLWHA had a higher risk of incident CAD.5 Cardiac magnetic resonance imaging and spectroscopy revealed that compared with HIV-uninfected individuals, PLWHA had a higher rate of myocardial fibrosis and cardiac steatosis.6 Despite some evidence suggesting that PLWHA have a higher risk of HF, the association between HIV infection and the development of HF has not been extensively studied. Two previous studies showed that veterans infected with HIV had a 1.2–1.8 fold higher risk of HF than individuals not infected with HIV.7,8 The higher risk of HF persisted among veterans who did not have a coronary heart disease before the incident HF event.7 In women with HF, HIV infection increased the risk of HF readmission.9 However, these previous studies on the association between HIV infection and HF enrolled either only veterans7,8 or female PLWHA.9

HAART plays an important role in improving the survival in PLWHA. Some antiretroviral drugs (eg, zidovudine) can cause mitochondrial damage and focal myocardial necrosis, possibly increasing the risk of HF development.11,12 However, studies determining the association between HAART and incident HF are limited. Alvi et al13 followed up 394 antiretroviral therapy–treated PLWHA who were hospitalized with HF and found that PLWHA receiving HAART with protease inhibitors have increased 30-day HF readmission. Therefore, we conducted a nationwide population-based cohort study to examine the association of HIV infection and HAART with the risk of incident HF during 2003–2014 in Taiwan.

METHODS

Data Source

This cohort study used 2003–2014 data from the Taiwan Centers for Disease Control HIV Surveillance Database. In Taiwan, medical professionals are required by law to report new HIV infections to TCDC within 24 hours of diagnosis. Reported HIV-infected cases in Taiwan are defined as having a positive HIV-1 Western blot or a positive polymerase chain reaction analysis results. Since 1997, all individuals in Taiwan infected with HIV have been offered HAART free of charge.14 This study was approved by the institutional review board of Taipei City Hospital (TCHIRB-10410122-E).

Study Subjects

This cohort study linked the Taiwan Centers for Disease Control and Prevention (CDC) HIV surveillance database to the Taiwan National Health Insurance Research Database. PLWHA who were aged 15 years or older were selected from the TCDC HIV Surveillance Database from 2003 through 2014. PLWHA who had a HF diagnosis (ICD-9-CM codes 428) before HIV diagnosis were excluded.

The control group was selected from the Taiwan National Health Insurance Research Database. In Taiwan, more than 99% of Taiwanese citizens have been covered by the National Health Insurance Program since 1995.15 The control group in this study was matched by age, sex, and date of enrollment (±7 days). Four controls were randomly selected for each HIV patient.16,17 Control subjects were excluded if (1) they had been reported to TCDC as HIV-infected cases or (2) they had received a diagnostic code for HF before inclusion into the study. The HIV patients and control groups were both followed up until a diagnosis of HF, death, or December 31, 2014. The death events were determined by the death certificate database of Taiwan. In Taiwan, it is regulated by the law that within 30 days after a patient dies, his or her death certificate must be registered by the physician in charge according to ICD-9-CM or ICD-10-CM codes. The cause-of-death coding in Taiwan is considered very accurate because trained medical registrars review and code all death certificates at the central office of the National Death Certification Registry.18

Outcome Variable

The outcome “new diagnosis of HF” was defined as the presence of an appropriate ICD-9-CM code (428).19 A person was considered to have a new onset of HF only if the condition occurred in an inpatient setting or 3 or more outpatient visits in Taiwan. The validation of the diagnosis of HF has previously been reported, with an accuracy of around 97.6%.19

Control Variables

Control variables included sociodemographics, comorbidities, HAART, AIDS status, and opportunistic infections (OIs) after the diagnosis of HIV. Sociodemographic variables included income level and urbanization. Income level was calculated from the average monthly income of the insured person and grouped into 3 levels: low [≤19,200 New Taiwan Dollars (NTD)], intermediate (19,201 NTD to <40,000 NTD), and high (≥40,000 NTD). Urbanization was categorized as urban and rural area. Comorbidities included CAD (ICD-9 code 410-414), dyslipidemia (ICD-9 code 272), diabetes (ICD-9 code 250), chronic kidney disease (CKD; ICD-9 code 580-587), hypertension (HTN; ICD-9 code 401-405), cancer (ICD-9 code 140-208), cerebral vascular disease (CVD; ICD-9 code 430–437), sleep apnea (ICD-9 code 780.51, 780.53, and 780.57), and hepatitis C virus (HCV) infection (ICD-9 code 070.41, 070.44, 070.51, and 070.54, V02.62). The OIs after the diagnosis of HIV included Mycobacterium tuberculosis infection (ICD-9 code 011-018), disseminated M. avium complex infection (ICD-9-CM code 0312), Pneumocystis jirovecii pneumonia (ICD-9 code 1363), Penicillium marneffei infection (ICD-9-CM code 1179), cytomegalovirus (ICD-9 code 078.5), candidiasis (ICD-9 code 112), herpes simplex (ICD-9 code 054), and herpes zoster (ICD-9 code 053). A person was considered to have a comorbidity or OI only if the condition occurred in an inpatient setting or 3 or more outpatient visits.20

Patients with HIV were considered to receive HAART if they received HAART before the new onset of HF. The classes of HAART included nucleoside reverse transcriptase inhibitors, nonnucleoside reverse transcriptase inhibitors, protease inhibitors, and integrase inhibitors. AIDS status was defined by a CD4 lymphocyte count of <200 cells/mm3 or by any other AIDS-defining condition in an individual with HIV infection.21

Statistical Analysis

First, the demographic data of the study subjects were analyzed. Continuous data are presented as mean ± SD, and a 2-sample t test was used to compare outcomes between groups. Categorical data were analyzed using the Pearson χ2 test where appropriate.

The incidence of HF per 100,000 person-years was calculated in patients with and without HIV infection. The relative hazards (RHs) of incident HF in PLWHA were estimated from Cox proportional hazards models as comparing with patients without HIV infection.

To determine the association between HIV infection and incident HF, a Cox proportional hazards model was used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) after adjusting for age, sex, and comorbidities. In these models, we used competing risk Cox proportional hazards regression to determine the risk of HF in study subjects, in which HF competes with the risk of death.22 Also, to evaluate the association between HAART and incident HF, a time-dependent Cox proportional hazards model was also used to identify risk factors for incident HF within PLWHA. In these models, HAART and AIDS status were regarded as time-dependent covariables,23 whereas other confounders such as age, sex, and comorbidities, which were collected at baseline, were considered as fixed covariates. Moreover, a time-dependent Cox proportional hazards model was used to determine the short-term and long-term effects of HAART on incident HF in PLWHA. Adjusted HRs (AHRs) with 95% CIs are reported to indicate the strength and direction of associations.

To examine the robustness of the main findings, subgroup and sensitivity analyses were conducted after stratifying study subjects by age, sex, and comorbidities. All data management and analyses were performed using SAS 9.4 statistical software (SAS Institute, Cary, NC).

RESULTS

Participant Selection

A total of 25,061 individuals with HIV infection were reported to Taiwan CDC during the period from January 1, 2003, through December 31, 2014. After excluding those younger than 15 years (n = 33), those with antecedent HF (n = 324), and those with incomplete data (n = 551), the remaining 24,153 PLWHA were included in the analysis (see Figure S1, Supplemental Digital Content, https://links.lww.com/QAI/B248). Another 96,612 subjects without HIV infection, matched for age and sex, were randomly selected for the control group. The overall mean (SD) age was 32.6 (9.9) years, and 93.83% of the subjects in the PLWHA were males. Mean (SD) follow-up time was 5.83 (3.38) years in the PLWHA and 5.84 (3.37) years in the control group. The demographic characteristics and comorbidities of the 2 groups are shown in Table 1. There were no significant differences in age or sex between groups. Compared with control subjects, PLWHA at baseline had a significantly higher proportion of CAD, CKD, malignancy, CVD, sleep apnea, and HCV infection, but had a significantly lower proportion of dyslipidemia and HTN.

T1
TABLE 1.:
Characteristics of the PLWHA and Matched Controls

Incident Rate of HF

During the study follow-up period, 641 individuals had a new onset of HF, including 192 (0.79%) PLWHA and 449 (0.46%) controls. The incidence rate of HF per 100,000 person-years was 136.34 in the PLWHA and 79.54 in the control group (P < 0.001). The RHs of incident HF was 1.63 [95% CI: 1.37 to 1.93] between PLWHA and the control group. Time to diagnosis of incident HF was significantly shorter in PLWHA than in those without HIV infection (P < 0.001, the log-rank test; Fig. 1).

F1
FIGURE 1.:
Kaplan–Meier curves for time to diagnosis of incident HF in patients with and without HIV infection.

Association of HIV Infection With Incident HF

A Cox proportional hazards model was used to identify independent risk factors for HF. After adjusting for age, sex, and comorbidities, HIV infection significantly increased the risk of incident HF (AHR 1.52; 95% CI: 1.27 to 1.82) (Table 2). Other risk factors associated with incident HF included older age, CAD, diabetes, CKD, HTN, and CVD.

T2
TABLE 2.:
Univariate and Multivariate Analyses for the Factors Associated With Incident HF Among Subjects With and Without HIV Infection

Subgroup and Sensitivity Analysis for the Association Between HIV Infection and HF

This study conducted the subgroup analysis to evaluate the association between HIV and HF, after adjustment for patient demographics and comorbidities. HIV infection was significantly associated with a higher risk of HF in all patients' subgroups, except those who were aged 50 years or older and those with high income (see Figure S2, Supplemental Digital Content, https://links.lww.com/QAI/B248). Among the subgroup subjects, female PLWHA had the highest risk of incident HF, followed by those living in urban areas.

Association of HAART With Incident HF Within PLWHA

The incidence rate of HF was 163.01 in PLWHA receiving HAART and 108.62 per 100,000 person-years in PLWHA not receiving HAART (P < 0.001) (Table 3). The univariate analysis showed that PLWHA receiving HAART had a significantly higher risk of incident HF than those not receiving HAART (HR 1.68; 95% CI: 1.24 to 2.28). After adjusting for age, sex, comorbidities, and OIs after the HIV diagnosis, a time-dependent Cox proportional hazards model showed that the RHs of incident HF were 1.35 (95% CI: 0.92 to 1.98) between PLWHA receiving HAART and those not.

T3
TABLE 3.:
Univariate and Multivariate Analyses for the Factors Associated With Incident HF Among PLWHA
table3-a
TABLE 3-A.:
Univariate and Multivariate Analyses for the Factors Associated With Incident HF Among PLWHA

This study evaluated the association between different classes of HAART and the risk of incident HF. After adjusting for age, sex, comorbidities, and OIs after the HIV diagnosis, a time-dependent Cox proportional hazards model showed that none of the classes of antiretroviral drugs were significantly associated with the risk of incident HF (see Table 1, Supplemental Digital Content, https://links.lww.com/QAI/B248).

Short-Term and Long-Term Effects of HAART on Incident HF

A time-dependent Cox proportional hazards model was used to determine the short-term and long-term effects of HAART on incident HF among PLWHA. The incidence rates of HF per 100,000 person-years were 108.62, 200.74, 160.66, and 149.67 in PLWHA not receiving HAART, those receiving HAART within 1 year, those receiving HAART between 1 and 2 years, and after 2 years of treatment, respectively (Table 4). As the duration of HAART increased, the risk of HF decreased (P = 0.014). After adjusting for age, sex, comorbidities, and OIs after the HIV diagnosis, compared with PLWHA not receiving HAART, the RH of incident HF was 1.63 (95% CI: 0.95 to 2.80) in PLWHA receiving HAART within 1 year of treatment, 1.73 (95% CI: 0.96 to 3.10) in PLWHA receiving HAART between 1 and 2 years, and 1.08 (95% CI: 0.69 to 1.68) in PLWHA receiving HAART after 2 years of treatment.

T4
TABLE 4.:
Short-Term and Long-Term Effects of HAART on Incident HF

DISCUSSION

This nationwide cohort study found that PLWHA had a significantly higher risk of developing incident HF compared with controls, after adjusting for demographic data and comorbidities.

Our study revealed robust associations between HIV infection and incident HF after stratifying patients by age, sex, and comorbidities. HIV infection significantly increased the risk of incident HF in all patients' subgroups, except those who were aged 50 years or older and those with high income.

This study found that PLWHA had a higher risk of HF than individuals not infected with HIV. A previous study showed that PLWHA had a higher prevalence of HF than individuals not infected with HIV (7.2% vs. 4.4%).24 However, longitudinal studies that examine the association of HIV infection with the subsequent development of HF are limited. An early report from the Veterans Aging Cohort Study followed up 8486 male veterans (2391 patients with HIV and 6095 patients without HIV) and found that the HR for developing HF in patients infected with HIV was 1.81 (95% CI: 1.39 to 2.36).7 The increased risk of HF in PLWHA persisted among those without CAD.7 Another report from the Veterans Aging Cohort Study enrolled 98,015 veterans (31,523 patients with HIV and 66,492 patients without HIV) and showed that veterans infected with HIV had an increased risk of HF with reduced ejection fraction (HR, 1.61; 95% CI: 1.40 to 1.86) or preserved ejection fraction (HR, 1.21; 95% CI: 1.03 to 1.41).8 Janjua et al9 revealed that compared with HIV-uninfected individuals, HIV infection increased the risk of HF readmission (HR, 2.58; 95 CI: 1.55 to 4.29) in female PLWHA. This study followed up 24,153 PLWHA and found that PLWHA had 1.5-fold higher risk of incident HF than patients not infected with HIV. The findings of our study suggest that HIV infection was an independent risk factor for incident HF.

HIV-related myocardial damage, fibrosis, and cardiac steatosis may account for the higher risk of incident HF in PLWHA. In vitro studies of human and rat cardiomyocytes, HIV can enter myocytes directly through pathways independent of a C–C chemokine receptor type 5.10,25 The invasion of HIV into cardiac myocytes could cause myocardial inflammation and cytokine release.26 HIV-related proinflammatory cytokines (eg, interleukin-1β and tumor necrosis factor-alpha) can promote expression of inducible nitric oxide synthase in cardiomyocytes.27 High concentrations of nitric oxide and tumor necrosis factor-alpha could induce cardiomyocyte apoptosis27 and lead to depressed heart function.27

HIV-associated myocardial fibrosis may also account for the higher risk of incident HF in PLWHA. Myocardial fibrosis represents previous myocarditis and can cause cardiac dysfunction.28 A previous magnetic resonance imaging study showed that 76% of asymptomatic HIV-infected patients had myocardial fibrosis, which was higher than 13% of HIV-uninfected patients.6 Myocardial fibrosis can reduce ventricular compliance and lead to the progression to HF.3

Cardiac steatosis may also explain the higher risk of incident HF in PLWHA. A previous study found that HIV-infected individuals had 47% higher median myocardial lipid levels than HIV-uninfected individuals.6 Animal studies found that lipid over storage in cardiac myocytes can produce lipotoxic intermediates that cause apoptosis, which can lead to the development of HF.4,29

HAART is essential for improving the outcomes in PLWHA. However, studies determining the association between HAART and the development of HF are limited. Alvi et al13 followed up 394 PLWHA and reported that PLWHA with HF who were receiving a protease inhibitor had increased 30-day HF readmission. Our study found that none of the classes of HAART were significantly associated with the risk of incident HF in PLWHA. A previous study showed that PLWHA with a well-controlled viral load had a lower risk of incident HF.7 Our study evaluated the short-term and long-term effects of HAART on incident HF and found that the incident rates of HF decreased as the duration of HAART increased.

This nationwide cohort study has several strengths. First, unlike previous cohort studies that included only veterans,7,8 women,9 or HIV-infected individuals from a single center,13 our study was conducted in a general population from a nationally representative sample and thus has greater generalizability. Second, this nationwide population-based study traced all PLWHA and control patients with minimized referral bias because all medical care was covered by Taiwan National Health Insurance. Third, the timing of HAART in all patients was collected and HAART was regarded as a time-dependent variable in the analysis. Longitudinal studies that failed to account for changes in exposure during the study period will not be able to precisely estimate the effect of exposure on outcomes.30

This study nevertheless has some limitations. First, because of data limitation, we had no information on some important variables (eg, smoking, illicit drug use, and obesity) in the research subjects. Second, when a new HIV-infected individual was reported to Taiwan CDC, viral loads and CD4 counts—the index of advanced stage of HIV infection—were not required to be reported to the HIV Surveillance System. However, Taiwan CDC enforces medical professionals to report HIV-infected individuals' AIDS status—an indicator of immune status—within 24 hours of AIDS diagnosis. The multivariable analysis in our study controlled for AIDS status among all PLWHA and treated AIDS status as a time-dependent variable to mimic the real scenario regarding the immune status of HIV-infected individuals. Third, the diagnosis of HF relied on administrative claims data recorded by physicians or hospitals, and the outcome of HF may have been misclassified. However, a previous study found that the accuracy of the Taiwan NHIRD in recording HF diagnoses was high (97.6%).19 Moreover, there is no reason to suspect that the validity of claims data would differ with a patient's HIV status. This nondifferential misclassification of outcome would bias the results toward a null association. Finally, to strengthen the diagnostic accuracy of incident HF, this study excluded HIV-infected individuals with prior HF in the analysis. However, some of the excluded HIV-infected individuals with prior HF may have already had the HIV infection before the first diagnosis of HF, which would underestimate the association between HIV infection and the development of HF in this study.

In conclusion, this nationwide long-term cohort study found a link between HIV infection and HF. HIV infection was found to be an independent risk factor for incident HF, after adjusting for demographic data and comorbidities. Because HF could cause significant morbidity in PLWHA, our study suggests that clinicians need to be aware of the higher risk of HF in this population.

ACKNOWLEDGMENTS

The authors are grateful for statistical consultation at the Biostatistical Consultation Centre, National Yang-Ming University, Taipei, Taiwan.

REFERENCES

1. UNAIDS. Joint United Nations Program on HIV/AIDS and World Health Organization. UNAIDS Report on the Global AIDS Epidemic; 2017. Available at: http://www.unaids.org/. Accessed May 15, 2018.
2. Remick J, Georgiopoulou V, Marti C, et al. Heart failure in patients with human immunodeficiency virus infection: epidemiology, pathophysiology, treatment, and future research. Circulation. 2014;129:1781–1789.
3. Travers JG, Kamal FA, Robbins J, et al. Cardiac fibrosis: the fibroblast awakens. Circ Res. 2016;118:1021–1040.
4. Christoffersen C, Bollano E, Lindegaard ML, et al. Cardiac lipid accumulation associated with diastolic dysfunction in obese mice. Endocrinology. 2003;144:3483–3490.
5. Grunfeld C, Delaney JA, Wanke C, et al. Preclinical atherosclerosis due to HIV infection: carotid intima-medial thickness measurements from the FRAM study. AIDS. 2009;23:1841–1849.
6. Holloway CJ, Ntusi N, Suttie J, et al. Comprehensive cardiac magnetic resonance imaging and spectroscopy reveal a high burden of myocardial disease in HIV patients. Circulation. 2013;128:814–822.
7. Butt AA, Chang CC, Kuller L, et al. Risk of heart failure with human immunodeficiency virus in the absence of prior diagnosis of coronary heart disease. Arch Intern Med. 2011;171:737–743.
8. Freiberg MS, Chang CH, Skanderson M, et al. Association between HIV infection and the risk of heart failure with reduced ejection fraction and preserved ejection fraction in the antiretroviral therapy era: results from the veterans aging cohort study. JAMA Cardiol. 2017;2:536–546.
9. Janjua SA, Triant VA, Addison D, et al. HIV infection and heart failure outcomes in women. J Am Coll Cardiol. 2017;69:107–108.
10. Fiala M, Popik W, Qiao JH, et al. HIV-1 induces cardiomyopathy by cardiomyocyte invasion and gp120, Tat, and cytokine apoptotic signaling. Cardiovasc Toxicol. 2004;4:97–107.
11. Currie PF, Boon NA. Immunopathogenesis of HIV-related heart muscle disease: current perspectives. AIDS. 2003;17:S21–S28.
12. Lewis W, Copeland WC, Day BJ. Mitochondrial DNA depletion, oxidative stress, and mutation: mechanisms of dysfunction from nucleoside reverse transcriptase inhibitors. Lab Invest. 2001;81:777–790.
13. Alvi RM, Neilan AM, Tariq N, et al. Protease inhibitors and cardiovascular outcomes in patients with HIV and heart failure. J Am Coll Cardiol. 2018;72:518–530.
14. Chen YM, Kuo SH. HIV-1 in Taiwan. Lancet. 2007;369:623–625.
15. Cheng TM. Taiwan's new national health insurance program: genesis and experience so far. Health Aff. 2003;22:61–76.
16. Wacholder S, McLaughlin JK, Silverman DT, et al. Selection of controls in case-control studies. I. Principles. Am J Epidemiol. 1992;135:1019–1028.
17. Beaumont JJ, Steenland K, Minton A, et al. A computer program for incidence density sampling of controls in case-control studies nested within occupational cohort studies. Am J Epidemiol. 1989;129:212–219.
18. Lu TH, Lee MC, Chou MC. Accuracy of cause-of-death coding in Taiwan: types of miscoding and effects on mortality statistics. Int J Epidemiol. 2000;29:336–343.
19. Lin YS, Chen TH, Chi CC, et al. Different implications of heart failure, ischemic stroke, and mortality between nonvalvular atrial fibrillation and atrial flutter-a view from a national cohort study. J Am Heart Assoc. 2017;6:e006406.
20. Yen YF, Chung MS, Hu HY, et al. Association of pulmonary tuberculosis and ethambutol with incident depressive disorder: a nationwide, population-based cohort study. J Clin Psychiatry. 2015;76:e505–e511.
21. Schneider E, Whitmore S, Glynn KM, et al. Revised surveillance case definitions for HIV infection among adults, adolescents, and children aged <18 months and for HIV infection and AIDS among children aged 18 months to <13 years—United States, 2008. MMWR Recomm Rep. 2008;57:1–12.
22. Russo S, Lopalco L. Is autoimmunity a component of natural immunity to HIV? Curr HIV Res. 2006;4:177–190.
23. Anderson AM, Fountain JA, Green SB, et al. Human immunodeficiency virus-associated cytomegalovirus infection with multiple small vessel cerebral infarcts in the setting of early immune reconstitution. J Neurovirol. 2010;16:179–184.
24. Al-Kindi SG, ElAmm C, Ginwalla M, et al. Heart failure in patients with human immunodeficiency virus infection: epidemiology and management disparities. Int J Cardiol. 2016;218:43–46.
25. Twu C, Liu NQ, Popik W, et al. Cardiomyocytes undergo apoptosis in human immunodeficiency virus cardiomyopathy through mitochondrion- and death receptor-controlled pathways. Proc Natl Acad Sci U S A. 2002;99:14386–14391.
26. Magnani JW, Dec GW. Myocarditis: current trends in diagnosis and treatment. Circulation. 2006;113:876–890.
27. Monsuez JJ, Escaut L, Teicher E, et al. Cytokines in HIV-associated cardiomyopathy. Int J Cardiol. 2007;120:150–157.
28. Friedrich MG, Sechtem U, Schulz-Menger J, et al. Cardiovascular magnetic resonance in myocarditis: a JACC White Paper. J Am Coll Cardiol. 2009;53:1475–1487.
29. Sharma S, Adrogue JV, Golfman L, et al. Intramyocardial lipid accumulation in the failing human heart resembles the lipotoxic rat heart. FASEB J. 2004;18:1692–1700.
30. Collett D. Modelling Survival Data in Medical Research. 2nd ed. London: Chapman & Hall; 2003.
Keywords:

heart failure; human immunodeficiency virus; highly active antiretroviral treatment

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.