Antiviral Therapy 2014; 19:309–317 (doi: 10.3851/IMP2724)
Original article Risk of cardiovascular disease associated with HCV and HBV coinfection among antiretroviral-treated HIV-infected individuals Jennifer Gillis1, Marek Smieja2, Angela Cescon3, Sean B Rourke4,5,6, Ann N Burchell5,7, Curtis Cooper8†, Janet M Raboud1,7*†, the OHTN Cohort Study Group‡ Toronto General Research Institute, University Health Network, Toronto, ON, Canada Department of Pathology & Molecular Medicine, McMaster University, Hamilton, ON, Canada 3 British Columbia Centre for Excellence in HIV/AIDS, St Paul’s Hospital, Vancouver, BC, Canada 4 Department of Psychiatry, University of Toronto, Toronto, ON, Canada 5 Ontario HIV Treatment Network, Toronto, ON, Canada 6 St Michael’s Hospital, Toronto, ON, Canada 7 Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada 8 University of Ottawa, The Ottawa Hospital Research Institute, Ottawa, ON, Canada 1 2
*Corresponding author e-mail: [email protected] These authors contributed equally as senior authors ‡ A list of the OHTN Cohort Study Group members can be found via Additional file 1 †
Background: The increased risk for cardiovascular disease (CVD) in HIV is well established. Despite high prevalence of viral hepatitis coinfection with HIV, there are few studies on the risk of CVD amongst antiretroviral therapy (ART)-treated coinfected patients. Methods: Ontario HIV Treatment Network Cohort Study participants who initiated ART without prior CVD events were analysed. HBV and HCV coinfection were identified by serology and RNA test results. CVD was defined as any of: coronary artery disease including atherosclerosis, chronic ischaemic heart disease and arteriosclerotic vascular disease; myocardial infarction; congestive heart failure; cerebrovascular accident or stroke; coronary bypass; angioplasty; and sudden cardiac death. The impact of HBV and HCV coinfection on time to CVD was assessed using multivariable competing risk models accounting for left truncation between ART initiation and study enrolment.
Results: A total of 3,416 HIV-monoinfected, 432 HIV‑HBV- and 736 HIV–HCV-coinfected individuals were followed for a median (IQR) of 2.32 years (1.36– 8.02). Over the study period, 167 CVD events and 613 deaths were documented. After adjustment for age, gender, race, year initiating ART, weight and smoking status, HBV was not associated with time to CVD onset (aHR=1.05, 95% CI [0.63, 1.74]; P=0.86). There was an elevated risk of CVD for HCV-coinfected individuals, which approached statistical significance (aHR=1.44, 95% CI [0.97, 2.13]; P=0.07). Conclusions: Our results are consistent with a moderate increase of CVD among individuals with HIV–HCV coinfection relative to those with HIV infection alone, lending support to consideration of initiation of HCV antiviral treatment.
Introduction The risk for cardiovascular disease (CVD) in HIV-positive individuals has been widely studied [1–3]. Although complex, the pathophysiology appears to be influenced by high rates of traditional risk factors for CVD in HIV-positive populations (for example, smoking, alcohol and substance use) and the viral infection itself (for example, creating of a proinflammatory immune environment that may favour ©2014 International Medical Press 1359-6535 (print) 2040-2058 (online)
AVT-13-OA-2963_Gillis.indd 309
endothelial injury). The use of virologically suppressive antiretroviral therapy (ART) interrupts immunological decay and leads to normalization of the proinflammatory cytokine milieu, which is thought to ameliorate the risk of CVD, compared with untreated HIV infection [2]. However, these medications may also impact CVD risk through creation of abnormal lipid profiles, increased occurrence 309
06/06/2014 11:40:33
J Gillis et al.
of central obesity and metabolic syndromes, and potential effects on the endothelium [2–4]. Viral hepatitis coinfection is common in HIV with 30% HCV and 10% HBV coinfection rates globally [5]. In the context of HIV infection, HCV, and to a lesser extent HBV, have been shown to decrease hyperlipidaemia and hypercholesterolemia associated with ART [6–9]. In contrast to this positive impact on lipid profiles, HCV coinfection increases insulin resistance in HIV-positive individuals treated with ART [10,11] and likely increases the risk of type 2 diabetes [10]. Data on the association between HCV coinfection and subclinical CVD within HIV-positive populations are inconsistent. While the prevalence of carotid plaques has been shown to be higher among HCV-coinfected individuals than individuals infected with HIV alone [12], carotid intima-media thickness was not significantly increased in the presence of HCV coinfection [12–14]. Associations of HCV coinfection with endothelial function are mixed [11,14]. Clinical CVD outcomes in the context of HIV–viral hepatitis coinfection have been evaluated in few studies. Freiberg et al. [15] found a significant association between HCV coinfection and self-reported CVD in a cohort of HIV-positive individuals with past or current excess alcohol consumption. In a study of 19,424 HIV-infected individuals, HCV coinfection was found to increase the risk of cerebrovascular disease, but only a trend towards increased risk of acute myocardial infarction was identified [16]. The Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) cohort found no association between HCV or HBV coinfection and myocardial infarction or stroke [17]. Little data exist on the association of coinfection with HBV on the risk of CVD in the context of HIV infection. These studies demonstrate a lack of consensus on the association of viral hepatitis coinfection with the risk of CVD in HIV-positive individuals undergoing therapy. To address this uncertainty, we evaluated the impact of viral hepatitis coinfection on the incidence of new CVD in HIV-positive individuals treated with ART, adjusting for common CVD risk factors and competing causes of death.
lost to follow-up. The study follow-up period was from January 1995, the time of the first enrolment into the OCS, to January 2011, the time of the last follow-up. Data are collected from multiple sources including electronic medical charts and manual chart abstraction, linkages to laboratory databases and completion of questionnaires. Clinical and laboratory data are extracted every 6 months and include CD4+ T-cell counts, viral load measurements, hospitalizations and adverse events. HBV and HCV viral load, serology and genotype data were obtained from the Public Health Ontario Laboratories (PHOL), through which virtually all confirmatory HCV diagnostic testing in the province is conducted. Socio-behavioural and demographic data are collected from annually interviewer administered questionnaires.
Methods
Statistical methods
Study population and setting We studied participants enrolled in the Ontario HIV Treatment Network (OHTN) Cohort Study (OCS), a voluntary clinic-based cohort at 11 HIV care sites in Ontario, Canada [18]. The OCS was initiated in 2007 and includes new enrolees and participants from the HIV Ontario Observational Database (HOOD), the predecessor cohort study to the OCS, who consented to continue their enrolment or who had either died or were 310
AVT-13-OA-2963_Gillis.indd 310
Classification of hepatitis status Individuals were classified as having HBV if a positive laboratory test for HBV surface antigen or HBV viral load was ever present. Individuals were classified as HCV-infected if they were ever HCV antibody, viral load or genotype positive. All others were classified as HIV-monoinfected.
Outcome measures The outcome of interest was the time to first CVD event, which was defined as the number of years from ART initiation to the occurrence of any of coronary artery disease including atherosclerosis, chronic ischaemic heart disease and arteriosclerotic vascular disease; myocardial infarction; congestive heart failure; cerebrovascular accident or stroke; coronary bypass; angioplasty; and sudden cardiac death. Details about CVD events were abstracted from medical charts and classified according to ICD9 codes. For inclusion in the current analyses, participants had to have initiated ART and have not been diagnosed with CVD prior to ART initiation. Death from non-CVD causes was considered as a competing risk. Deaths for which cause was unknown were considered to be non-CVD-related since only 5% of deaths were CVD-related. Individuals who did not experience a CVD event or death from competing cause were censored on their last date of contact with the OCS.
Demographic and clinical characteristics were summarized using frequencies and percentages for categorical variables and medians and IQR for continuous variables. The demographic and clinical characteristics of HIV–HBV and HIV–HCV-coinfected individuals were compared with HIV-monoinfected individuals using c2 tests and Wilcoxon rank-sum tests, as appropriate. Incidence rates of CVD events and death were calculated for HIV-monoinfected, HIV–HBV and HIV– HCV-coinfected individuals. ©2014 International Medical Press
06/06/2014 11:40:33
HIV–viral hepatitis coinfection and risk of CVD
The time origin of interest, ART initiation, occurred prior to cohort enrolment for some individuals. While death is necessarily left truncated between the time origin and enrolment into the cohort, the availability of retrospective administrative data on comorbid conditions provided data on CVD events during this period. To our knowledge, no methods are currently available to handle differential left truncation in the context of time to event regression modelling. As such, data were left truncated between the time of ART initiation and enrolment in the OCS for those individuals who initiated ART prior to enrolment, and CVD events that occurred prior to enrolment were excluded from the analysis. Predictors of progression to CVD were explored using cumulative incidence function plots and univariable Fine and Gray models [19] accounting for the competing risk of non-CVD death. A multivariable Fine and Gray model was used to examine the impact of HBV and HCV coinfection after adjusting for factors known a priori to be associated with progression to CVD including age, sex, race, weight centered on average weight of males and females accordingly, baseline smoking status and year of ART initiation.
Imputation of missing smoking status at ART initiation Multiple imputation methods were used to impute missing smoking status at ART initiation [20]. Univariable cumulative logit models for smoking status and univariable logistic regression models were used to determine which covariates were significantly associated with baseline smoking status and missingness, respectively. Those covariates that were significant in the univariable left-truncated Fine and Gray models were also considered for inclusion in the imputation model. The final imputation model included HBV and HCV status, age, sex, race, risk factors for HIV acquisition, education, year of ART initiation, baseline regimen, previous CVD-related events (diabetes diagnosis, hypertension, dyslipidaemia, angina and peripheral vascular disease), CVD event or death and OCS site.
Results Of the 5,644 individuals who were enrolled in the OCS as of December 2010, 4,501 had initiated ART with no prior history of CVD and came from sites providing sufficient baseline demographic and clinical data. During the study period, 3,852/4,501 (86%) of participants were tested at least once for HCV. Of the 720 participants with at least one positive antibody result, 367 (51%) had confirmatory RNA testing. An additional 16 participants had only RNA-positive results, for a total of 736 HCV-positive individuals. 85% of the participants were tested at least once during the Antiviral Therapy 19.3
AVT-13-OA-2963_Gillis.indd 311
study period for HBV. In total, 349 participants were HIV–HBV-coinfected, 653 were HIV–HCV-coinfected and 83 were HIV–HBV–HCV tri-infected. Demographic and clinical characteristics at ART initiation of HIV–HBV and HIV–HCV-coinfected individuals were compared with HIV-monoinfected individuals (Table 1). In Table 1, tri-infected individuals were included in both of the HIV–HBV- and HIV– HCV-infected groups. HIV–HBV- and HIV–HCVcoinfected individuals were more likely to have been infected with HIV through injection drug use, had lower education and income, were smokers at initiation of ART, and had been diagnosed with HIV longer than HIV-monoinfected individuals. Smoking status at ART initiation was imputed for 692 (15%) of the study population. The proportion of individuals with missing smoking data was similar by coinfection status and CVD outcome (data not shown). A total of 167 CVD events and 613 non-CVD deaths were documented during the study period after enrolment into the OCS with a median (IQR) time of followup of 2.32 years (1.36–8.02). An additional 46 CVD events were excluded from the analysis because they occurred after ART initiation but prior to enrolment. The numbers of CVD events excluded from the analysis because they occurred prior to enrolment were 39, 4 and 4 for HIV, HIV–HCV and HIV–HBV, which was 25%, 10% and 18% of the CVD events for each group, respectively. Since the proportion of events excluded from the analysis due to left truncation was slightly higher among HIV-monoinfected individuals, exclusion of these events may have biased our findings slightly towards the null. Of the 46 patients with events that were excluded due to left-truncation, 4 had a subsequent CVD event during the study period. CVD event type is tabulated by coinfection status in Table 2. 25% and 16% of CVD events were attributed to myocardial infarction and cerebrovascular accident (stroke), respectively. 11% of the CVD events were cardiac procedures of coronary bypass and angioplasty. Coronary artery disease, congestive heart failure and sudden cardiac death respectively contributed 29%, 12% and 9% of events. The incidence (95% CI) of CVD (events per 1,000 person-years follow-up) was 8.70 (4.68, 12.7) for HIV– HBV, 9.62 (6.38, 12.9) for HIV–HCV-coinfected individuals and 7.59 (6.22, 8.96) for HIV-monoinfected individuals. The incidence (95% CI) of the competing risk, death (events per 1,000 person-years followup), was 40.3 (31.6, 48.9) for HIV–HBV, 34.9 (28.7, 41.1) for HIV–HCV-coinfected individuals and 27.3 (24.6, 29.9) for HIV-monoinfected individuals. Figure 1 depicts the cumulative incidence functions of CVD events and non-CVD death for HIV–HBV- and HIV– HCV-coinfected individuals. 311
06/06/2014 11:40:33
J Gillis et al.
Table 1. Demographic and clinical characteristics at initiation of antiretroviral therapy by viral hepatitis coinfection statusa Variables
HIV (n=3,416)
HIV–HBV (n=432)
P-value
HIV–HCV (n=736)
Demographics Age, years 36 (31–43) 36 (31–42) 0.61 36 (31–42) Sex Male 2,887 (85) 405 (94)
Original article Risk of cardiovascular disease associated with HCV and HBV coinfection among antiretroviral-treated HIV-infected individuals Jennifer Gillis1, Marek Smieja2, Angela Cescon3, Sean B Rourke4,5,6, Ann N Burchell5,7, Curtis Cooper8†, Janet M Raboud1,7*†, the OHTN Cohort Study Group‡ Toronto General Research Institute, University Health Network, Toronto, ON, Canada Department of Pathology & Molecular Medicine, McMaster University, Hamilton, ON, Canada 3 British Columbia Centre for Excellence in HIV/AIDS, St Paul’s Hospital, Vancouver, BC, Canada 4 Department of Psychiatry, University of Toronto, Toronto, ON, Canada 5 Ontario HIV Treatment Network, Toronto, ON, Canada 6 St Michael’s Hospital, Toronto, ON, Canada 7 Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada 8 University of Ottawa, The Ottawa Hospital Research Institute, Ottawa, ON, Canada 1 2
*Corresponding author e-mail: [email protected] These authors contributed equally as senior authors ‡ A list of the OHTN Cohort Study Group members can be found via Additional file 1 †
Background: The increased risk for cardiovascular disease (CVD) in HIV is well established. Despite high prevalence of viral hepatitis coinfection with HIV, there are few studies on the risk of CVD amongst antiretroviral therapy (ART)-treated coinfected patients. Methods: Ontario HIV Treatment Network Cohort Study participants who initiated ART without prior CVD events were analysed. HBV and HCV coinfection were identified by serology and RNA test results. CVD was defined as any of: coronary artery disease including atherosclerosis, chronic ischaemic heart disease and arteriosclerotic vascular disease; myocardial infarction; congestive heart failure; cerebrovascular accident or stroke; coronary bypass; angioplasty; and sudden cardiac death. The impact of HBV and HCV coinfection on time to CVD was assessed using multivariable competing risk models accounting for left truncation between ART initiation and study enrolment.
Results: A total of 3,416 HIV-monoinfected, 432 HIV‑HBV- and 736 HIV–HCV-coinfected individuals were followed for a median (IQR) of 2.32 years (1.36– 8.02). Over the study period, 167 CVD events and 613 deaths were documented. After adjustment for age, gender, race, year initiating ART, weight and smoking status, HBV was not associated with time to CVD onset (aHR=1.05, 95% CI [0.63, 1.74]; P=0.86). There was an elevated risk of CVD for HCV-coinfected individuals, which approached statistical significance (aHR=1.44, 95% CI [0.97, 2.13]; P=0.07). Conclusions: Our results are consistent with a moderate increase of CVD among individuals with HIV–HCV coinfection relative to those with HIV infection alone, lending support to consideration of initiation of HCV antiviral treatment.
Introduction The risk for cardiovascular disease (CVD) in HIV-positive individuals has been widely studied [1–3]. Although complex, the pathophysiology appears to be influenced by high rates of traditional risk factors for CVD in HIV-positive populations (for example, smoking, alcohol and substance use) and the viral infection itself (for example, creating of a proinflammatory immune environment that may favour ©2014 International Medical Press 1359-6535 (print) 2040-2058 (online)
AVT-13-OA-2963_Gillis.indd 309
endothelial injury). The use of virologically suppressive antiretroviral therapy (ART) interrupts immunological decay and leads to normalization of the proinflammatory cytokine milieu, which is thought to ameliorate the risk of CVD, compared with untreated HIV infection [2]. However, these medications may also impact CVD risk through creation of abnormal lipid profiles, increased occurrence 309
06/06/2014 11:40:33
J Gillis et al.
of central obesity and metabolic syndromes, and potential effects on the endothelium [2–4]. Viral hepatitis coinfection is common in HIV with 30% HCV and 10% HBV coinfection rates globally [5]. In the context of HIV infection, HCV, and to a lesser extent HBV, have been shown to decrease hyperlipidaemia and hypercholesterolemia associated with ART [6–9]. In contrast to this positive impact on lipid profiles, HCV coinfection increases insulin resistance in HIV-positive individuals treated with ART [10,11] and likely increases the risk of type 2 diabetes [10]. Data on the association between HCV coinfection and subclinical CVD within HIV-positive populations are inconsistent. While the prevalence of carotid plaques has been shown to be higher among HCV-coinfected individuals than individuals infected with HIV alone [12], carotid intima-media thickness was not significantly increased in the presence of HCV coinfection [12–14]. Associations of HCV coinfection with endothelial function are mixed [11,14]. Clinical CVD outcomes in the context of HIV–viral hepatitis coinfection have been evaluated in few studies. Freiberg et al. [15] found a significant association between HCV coinfection and self-reported CVD in a cohort of HIV-positive individuals with past or current excess alcohol consumption. In a study of 19,424 HIV-infected individuals, HCV coinfection was found to increase the risk of cerebrovascular disease, but only a trend towards increased risk of acute myocardial infarction was identified [16]. The Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) cohort found no association between HCV or HBV coinfection and myocardial infarction or stroke [17]. Little data exist on the association of coinfection with HBV on the risk of CVD in the context of HIV infection. These studies demonstrate a lack of consensus on the association of viral hepatitis coinfection with the risk of CVD in HIV-positive individuals undergoing therapy. To address this uncertainty, we evaluated the impact of viral hepatitis coinfection on the incidence of new CVD in HIV-positive individuals treated with ART, adjusting for common CVD risk factors and competing causes of death.
lost to follow-up. The study follow-up period was from January 1995, the time of the first enrolment into the OCS, to January 2011, the time of the last follow-up. Data are collected from multiple sources including electronic medical charts and manual chart abstraction, linkages to laboratory databases and completion of questionnaires. Clinical and laboratory data are extracted every 6 months and include CD4+ T-cell counts, viral load measurements, hospitalizations and adverse events. HBV and HCV viral load, serology and genotype data were obtained from the Public Health Ontario Laboratories (PHOL), through which virtually all confirmatory HCV diagnostic testing in the province is conducted. Socio-behavioural and demographic data are collected from annually interviewer administered questionnaires.
Methods
Statistical methods
Study population and setting We studied participants enrolled in the Ontario HIV Treatment Network (OHTN) Cohort Study (OCS), a voluntary clinic-based cohort at 11 HIV care sites in Ontario, Canada [18]. The OCS was initiated in 2007 and includes new enrolees and participants from the HIV Ontario Observational Database (HOOD), the predecessor cohort study to the OCS, who consented to continue their enrolment or who had either died or were 310
AVT-13-OA-2963_Gillis.indd 310
Classification of hepatitis status Individuals were classified as having HBV if a positive laboratory test for HBV surface antigen or HBV viral load was ever present. Individuals were classified as HCV-infected if they were ever HCV antibody, viral load or genotype positive. All others were classified as HIV-monoinfected.
Outcome measures The outcome of interest was the time to first CVD event, which was defined as the number of years from ART initiation to the occurrence of any of coronary artery disease including atherosclerosis, chronic ischaemic heart disease and arteriosclerotic vascular disease; myocardial infarction; congestive heart failure; cerebrovascular accident or stroke; coronary bypass; angioplasty; and sudden cardiac death. Details about CVD events were abstracted from medical charts and classified according to ICD9 codes. For inclusion in the current analyses, participants had to have initiated ART and have not been diagnosed with CVD prior to ART initiation. Death from non-CVD causes was considered as a competing risk. Deaths for which cause was unknown were considered to be non-CVD-related since only 5% of deaths were CVD-related. Individuals who did not experience a CVD event or death from competing cause were censored on their last date of contact with the OCS.
Demographic and clinical characteristics were summarized using frequencies and percentages for categorical variables and medians and IQR for continuous variables. The demographic and clinical characteristics of HIV–HBV and HIV–HCV-coinfected individuals were compared with HIV-monoinfected individuals using c2 tests and Wilcoxon rank-sum tests, as appropriate. Incidence rates of CVD events and death were calculated for HIV-monoinfected, HIV–HBV and HIV– HCV-coinfected individuals. ©2014 International Medical Press
06/06/2014 11:40:33
HIV–viral hepatitis coinfection and risk of CVD
The time origin of interest, ART initiation, occurred prior to cohort enrolment for some individuals. While death is necessarily left truncated between the time origin and enrolment into the cohort, the availability of retrospective administrative data on comorbid conditions provided data on CVD events during this period. To our knowledge, no methods are currently available to handle differential left truncation in the context of time to event regression modelling. As such, data were left truncated between the time of ART initiation and enrolment in the OCS for those individuals who initiated ART prior to enrolment, and CVD events that occurred prior to enrolment were excluded from the analysis. Predictors of progression to CVD were explored using cumulative incidence function plots and univariable Fine and Gray models [19] accounting for the competing risk of non-CVD death. A multivariable Fine and Gray model was used to examine the impact of HBV and HCV coinfection after adjusting for factors known a priori to be associated with progression to CVD including age, sex, race, weight centered on average weight of males and females accordingly, baseline smoking status and year of ART initiation.
Imputation of missing smoking status at ART initiation Multiple imputation methods were used to impute missing smoking status at ART initiation [20]. Univariable cumulative logit models for smoking status and univariable logistic regression models were used to determine which covariates were significantly associated with baseline smoking status and missingness, respectively. Those covariates that were significant in the univariable left-truncated Fine and Gray models were also considered for inclusion in the imputation model. The final imputation model included HBV and HCV status, age, sex, race, risk factors for HIV acquisition, education, year of ART initiation, baseline regimen, previous CVD-related events (diabetes diagnosis, hypertension, dyslipidaemia, angina and peripheral vascular disease), CVD event or death and OCS site.
Results Of the 5,644 individuals who were enrolled in the OCS as of December 2010, 4,501 had initiated ART with no prior history of CVD and came from sites providing sufficient baseline demographic and clinical data. During the study period, 3,852/4,501 (86%) of participants were tested at least once for HCV. Of the 720 participants with at least one positive antibody result, 367 (51%) had confirmatory RNA testing. An additional 16 participants had only RNA-positive results, for a total of 736 HCV-positive individuals. 85% of the participants were tested at least once during the Antiviral Therapy 19.3
AVT-13-OA-2963_Gillis.indd 311
study period for HBV. In total, 349 participants were HIV–HBV-coinfected, 653 were HIV–HCV-coinfected and 83 were HIV–HBV–HCV tri-infected. Demographic and clinical characteristics at ART initiation of HIV–HBV and HIV–HCV-coinfected individuals were compared with HIV-monoinfected individuals (Table 1). In Table 1, tri-infected individuals were included in both of the HIV–HBV- and HIV– HCV-infected groups. HIV–HBV- and HIV–HCVcoinfected individuals were more likely to have been infected with HIV through injection drug use, had lower education and income, were smokers at initiation of ART, and had been diagnosed with HIV longer than HIV-monoinfected individuals. Smoking status at ART initiation was imputed for 692 (15%) of the study population. The proportion of individuals with missing smoking data was similar by coinfection status and CVD outcome (data not shown). A total of 167 CVD events and 613 non-CVD deaths were documented during the study period after enrolment into the OCS with a median (IQR) time of followup of 2.32 years (1.36–8.02). An additional 46 CVD events were excluded from the analysis because they occurred after ART initiation but prior to enrolment. The numbers of CVD events excluded from the analysis because they occurred prior to enrolment were 39, 4 and 4 for HIV, HIV–HCV and HIV–HBV, which was 25%, 10% and 18% of the CVD events for each group, respectively. Since the proportion of events excluded from the analysis due to left truncation was slightly higher among HIV-monoinfected individuals, exclusion of these events may have biased our findings slightly towards the null. Of the 46 patients with events that were excluded due to left-truncation, 4 had a subsequent CVD event during the study period. CVD event type is tabulated by coinfection status in Table 2. 25% and 16% of CVD events were attributed to myocardial infarction and cerebrovascular accident (stroke), respectively. 11% of the CVD events were cardiac procedures of coronary bypass and angioplasty. Coronary artery disease, congestive heart failure and sudden cardiac death respectively contributed 29%, 12% and 9% of events. The incidence (95% CI) of CVD (events per 1,000 person-years follow-up) was 8.70 (4.68, 12.7) for HIV– HBV, 9.62 (6.38, 12.9) for HIV–HCV-coinfected individuals and 7.59 (6.22, 8.96) for HIV-monoinfected individuals. The incidence (95% CI) of the competing risk, death (events per 1,000 person-years followup), was 40.3 (31.6, 48.9) for HIV–HBV, 34.9 (28.7, 41.1) for HIV–HCV-coinfected individuals and 27.3 (24.6, 29.9) for HIV-monoinfected individuals. Figure 1 depicts the cumulative incidence functions of CVD events and non-CVD death for HIV–HBV- and HIV– HCV-coinfected individuals. 311
06/06/2014 11:40:33
J Gillis et al.
Table 1. Demographic and clinical characteristics at initiation of antiretroviral therapy by viral hepatitis coinfection statusa Variables
HIV (n=3,416)
HIV–HBV (n=432)
P-value
HIV–HCV (n=736)
Demographics Age, years 36 (31–43) 36 (31–42) 0.61 36 (31–42) Sex Male 2,887 (85) 405 (94)
