|Home | About | Journals | Submit | Contact Us | Français|
We sought to determine the incidence and clinical characteristics of sudden cardiac death (SCD) in patients with HIV.
As the HIV-infected population ages, cardiovascular disease prevalence and mortality are increasing; however, the incidence and features of SCD have not yet been described.
Records of 2860 consecutive patients in a public HIV clinic in San Francisco, CA between April 2000 and August 2009 were examined. Identification of deaths, causes of death, and clinical characteristics were obtained by search of the National Death Index and/or clinic records. SCDs were determined using published retrospective criteria: (1) ICD10 code for all cardiac causes of death and (2) circumstances of death meeting WHO criteria.
Of 230 deaths over 3.7 median years’ follow-up, 30 (13%) met SCD criteria, 131 (57%) were due to AIDS, 25 (11%) other (natural) diseases, and 44 (19%) overdose/suicides/unknown. SCDs accounted for 86% (30/35) of all cardiac deaths. The mean SCD rate was 2.6 per 1,000 person-years (95%CI 1.8-3.8), 4.5-fold higher than expected. SCDs occurred in older patients than AIDS deaths (mean 49.0 vs. 44.9 years, p=0.02). Compared to AIDS and natural deaths combined, SCDs had higher prevalence of prior MI (17% vs. 1%, p<0.0005), cardiomyopathy (23% vs. 3%, p<0.0005), heart failure (30% vs. 9%, p=0.004), and arrhythmias (20% vs. 3%, p=0.003).
SCDs account for most cardiac and many non-AIDS natural deaths in HIV-infected patients. Further investigation is needed to ascertain underlying mechanisms, which may include inflammation, antiretroviral therapy interruption, and concomitant medications.
As HIV-infected individuals experience longer life expectancy with antiretroviral therapy, their rates of cardiovascular disease (CVD) are increasing. One of the most feared manifestations of CVD is sudden cardiac death (SCD), responsible for 5-15% of total deaths in the United States.1, 2 Many SCDs occur in patients previously undiagnosed with CVD,3 making identification of high-risk populations important for screening and prevention.
Individuals with HIV have higher rates of CVD than uninfected controls, likely due to a combination of traditional risk factors, HIV-related inflammation, and antiretroviral therapy.4, 5 Cardiovascular abnormalities strongly associated with SCD are prevalent in HIV-infected individuals, including cardiomyopathy,6 pulmonary hypertension7 and prolonged QTc.8 While most persons with HIV still die of AIDS,9 given the increased prevalence of CVD in this population SCD is likely an important contributor to overall mortality. Thus, we sought to determine the incidence, clinical characteristics, and predictors of SCD over the past decade in a large cohort of individuals receiving care at an urban, public HIV clinic.
We conducted a single-center, retrospective cohort study in a public HIV specialty clinic in San Francisco, California, USA. We included 2860 consecutive patients, all ≥18 years old with documented HIV infection, enrolled between April 1, 2000 and August 31, 2009. The study was approved by the Institutional Review Board of the University of California, San Francisco.
Deaths were identified using the Social Security Death Index and/or clinic records. Cause of death was obtained through clinical chart review and search of either the National Death Index Plus database (through 2008) or direct evaluation of death certificates from the San Francisco Department of Public Health (in 2009). Charts were available for all patients who died and were independently evaluated by two reviewers for cause of death, circumstances of death, and traditional cardiovascular risk factors. Disagreements were resolved by consensus with a third reviewer.
SCDs were defined as deaths meeting two published criteria: (1) primary ICD-10 code for all cardiac causes1, 2 and (2) circumstances of death meeting World Health Organization (WHO) criteria for SCD (death within 1 hour of symptom onset if witnessed or within 24 hours of being observed alive and symptom-free if unwitnessed)10 or unexpected out-of-hospital death.11 Cardiac ICD-10 codes were: disease of heart (I00 to I09, I11, I20 to I51); congenital heart disease (Q20 to Q24); and ill-defined cause of death (R95 to R99).2 Deaths in hospice or due to overdose, violence, suicide, cancer, or opportunistic infections were excluded. All unexpected out-of-hospital deaths classified as SCD were confirmed as not meeting criteria for AIDS death (below).
For all SCDs, we recorded the following: symptoms reported during clinic visits (chest pain, dyspnea, syncope, palpitations), cardiac medications at last clinic visit, and results of all cardiac studies.
AIDS deaths were defined as deaths meeting at least two of three published criteria: (1) primary ICD-10 code for HIV-disease related illness (B20 to 24); (2) circumstances of death involving HIV-related infection or illness; or (3) most recent CD4 count <50 cells/mm3.12 Remaining deaths were classified as due to trauma, suicide, substance overdose, and other natural causes (which included non-AIDS-related cancer, liver disease, sepsis, seizure, renal disease, pulmonary disease, and diabetes).
Baseline characteristics were abstracted from the clinic’s medical record upon enrollment. We recorded the following variables: age, gender, race, HIV risk factor(s), CD4 count, and viral load. For all deaths, we searched medical records for evidence of CVD [ICD-9 codes for ischemic heart disease (410-414), disease of the pericardium, endocardium and/or myocardium (420-424), cardiomyopathy (425), conduction disorders and arrhythmias (426-427), heart failure (428), complications of heart disease (429)] or CVD risk factors [hypertensive disease (401-405), disorders of lipid metabolism (272), chronic kidney disease (585), diabetes mellitus (250)].
Differences in prevalent CVD and risk factors between SCDs vs. AIDS and other natural deaths were assessed with Fisher’s exact tests and Wilcoxon rank sum or Kruskal-Wallis tests. We also used Poisson regression to compare year-specific crude mortality rates, and to estimate expected background SCD rates by age, race, and sex, using all previously identified SCDs (n=252) and the city-wide SCD rate (0.373 per 1,000 person-years) in San Francisco in 200713 and population sizes by age, race, and sex.14 The resulting expected background rates were then used in combination with follow-up time in the entire cohort to calculate the expected numbers of SCDs in the HIV cohort and the standardized mortality ratio.
Of 2860 cohort patients, 2478 (87%) were men and 1515 (53%) were Caucasian; among those with a known HIV risk factor (n=2482), 1778 (72%) were men who have sex with men, 165 (7%) were injection drug users, and 519 (21%) had heterosexual risk. At entry, median age was 39 years (IQR 33-45), median CD4 count was 353 cells/mm3 (IQR 175-551, mean 390), median log viral load was 4.1 copies/mL (IQR 2.9-4.9, mean 3.9) and 578 (21%) had undetectable HIV RNA (< 200 copies/mL), indicative of successful antiretroviral therapy.
Of 230 deaths occurring over a median 3.7 years of follow-up, 30 (13%; 95% CI 9-18%) met criteria for SCDs, 131 (57%; 95% CI 50-63%) were due to AIDS, 25 (11%; 95% CI 7-16%) due to other natural diseases, and 44 (19%; 95% CI 14-25%) due to overdose, suicides, or unknown causes [Figure 1]. SCDs accounted for 30/35 (86%; 95% CI 70-95%) of all cardiac deaths. Although SCD did not increase as a proportion of total deaths, by 2003 SCD was often the leading cause of non-AIDS natural deaths. During the 10-year period, the mean SCD rate was 2.6 per 1,000 person-years (95% CI 1.8-3.8) versus 11.4 per 1,000 person-years (95% CI 9.6-13.6) for AIDS death [Figure 2].
Three SCDs underwent autopsy; causes of death were myocardial infarction (MI) (2), and severe cardiomyopathy. More than half of SCDs had a history of tobacco, alcohol, or drug use. Five patients (17%) had family history of CVD and 80% had either known CVD or CVD risk factors. At their final clinic visit, 33% reported chest pain, palpitations, syncope, and/or dyspnea; 83% were prescribed cardiac medication. Thirteen (43%) received echocardiograms: 8 showed moderately-to-severely reduced ejection fraction, 7 diastolic dysfunction, and 3 pulmonary hypertension. Three of six patients who underwent stress testing demonstrated ischemia; coronary angiography in 2 patients demonstrated no significant stenoses. Of 23 patients with electrocardiograms, 1 had atrial fibrillation, 8 met criteria for left ventricular hypertrophy, 1 had prolonged QTc, and 4 showed evidence of prior MI.
Compared to AIDS deaths, SCDs victims were similar with respect to ethnicity and gender but a mean 4.1 years older, with higher CD4 count (median 312 cells/mm3 vs. 87, p=0.0001) and lower viral load (median 3.8 copies/mL vs. 4.8, p=0.009) [Table 1]. Of patients with laboratory studies within 90 days of death, 12/15 (80%; 95% CI 52-96%) SCDs had CD4 counts >200 cells/mm3 and 8/15 (53%; 95% CI 27-79%) had HIV RNA <200 copies/mL, compared to 36 (55%; 95% CI 43-68%) and 18 (28%; 95% CI 17-40%) of 65 AIDS deaths.
Compared to AIDS and other natural deaths combined, SCDs had a higher prevalence of prior MI (17% vs. 1%, p<0.0005), cardiomyopathy (23% vs. 3%, p<0.0005), heart failure (30% vs. 9%, p=0.004), arrhythmias (20% vs. 3%, p=0.003), hypertension (67% vs. 27%, p<0.0005), and hyperlipidemia (20% vs. 6%, p=0.03), but a similar prevalence of diabetes mellitus, chronic renal disease, and chronic pulmonary disease [Table 2].
Based on an expected 6.73 SCD deaths (given 2007 San Francisco SCD rate and population), the standardized morality ratio for SCD in this population was 30/6.73 = 4.46.
While AIDS remained the leading cause of mortality, SCD was disproportionately represented in this urban HIV-infected cohort, accounting for 13% of all deaths and 86% of cardiac deaths, at a rate 4.5-fold higher than expected. Compared to AIDS deaths, SCDs occurred in older patients with better control of their HIV disease, as measured by CD4 count and viral load.
In the general population, most SCD is due to coronary artery disease.15 SCDs in this cohort reflect the age (mean 49 years) and gender distribution (93% male) of HIV patients presenting with acute coronary syndromes,5 and prior MI was strongly associated with SCD. This study also replicates in the HIV population other risk factors associated with SCD: cardiomyopathy,16 heart failure,17 arrhythmias,18 hypertension,19 and hyperlipidemia.19
SCD victims had modest immunodeficiency, with similar CD4 counts (median 312 cells/mm3) and viral loads (median 3.8 log copies/mL) as the full cohort (353 and 4.1, respectively), suggesting that individuals are susceptible to SCD even in the setting of mild HIV disease. This finding is consistent with the Strategies for Management of Antiretroviral Therapy (SMART) study, in which treatment reduced non-AIDS mortality primarily among individuals with CD4 counts greater than 350 cells/μL.20 However, of SCD victims with recent laboratory studies, over half had undetectable viral loads, suggesting that even patients on effective therapy remain at risk.
While our study does not address the mechanism(s) underlying SCD in the setting of HIV, other large cohorts may provide insight. In the Data collection on Adverse events of anti-HIV Drug (DAD) study, 16 of 36 fatal MIs were reported as secondary to “unclassifiable coronary event (such as sudden death).”4 In the SMART study, levels of IL-6 and D-dimers at entry were strongly associated with CVD and unwitnessed deaths, although few such events occurred.21 Chronic HIV-associated inflammation is thought to underlie many non-AIDS conditions including CVD, consistent with studies demonstrating increased all-cause mortality and acute MI risk among patients with higher levels of serum hs-CRP.22, 23 Future investigations may help clarify the role of inflammation and antiretroviral therapy in SCD.
HIV-infected patients also have higher prevalence of prolonged QTc,8 a risk factor for malignant arrhythmia and cardiovascular mortality in otherwise healthy subjects.24 A transgenic murine model of HIV infection has demonstrated acquired sodium and potassium channelopathy,25, 26 suggesting that HIV may also directly affect cardiac depolarization and repolarization, thereby predisposing infected individuals to malignant ventricular arrhythmias.
This study is limited by its retrospective data collection. We employed commonly used criteria for retrospective ascertainment of SCD.1, 11, 13 including requirement for a primary cardiovascular cause on death certificate plus circumstances of death meeting WHO criteria. Furthermore, all unexpected out-of-hospital deaths meeting criteria for AIDS death were excluded as SCD. However, the sensitivity and specificity of WHO classification and death record ascertainment for SCD are limited for actual cardiac and arrhythmic causes.1, 13 Therefore, although we cross-checked multiple records, some SCDs may still have been misclassified. Notably, we classified sudden deaths in patients with low CD4 counts as SCD that other studies classified as AIDS-related9 or unknown,12 thus results across methods are not fully comparable. Therefore, while previous studies may underestimate SCD rates, in the absence of definitive data such as rhythm strips or autopsy, we may overestimate them. Unfortunately, autopsy confirmation was available for only a few SCDs, consistent with low autopsy rates in other population studies of SCD.1, 11
Our assessment of prevalent conditions is limited by coding practices. Furthermore, clinical charts were insufficient to accurately ascertain duration of antiretroviral therapy; thus we are unable to evaluate potential associations between treatment duration and SCD. Finally, while we adjusted for age, race, and sex, our estimated standardized mortality ratio for SCD may be biased high, as patients with prevalent CVD are over-represented in this cohort. However, this may also be an important causal link between HIV and SCD.
In summary, we found that SCDs comprise an unexpectedly high proportion of overall deaths in this urban HIV cohort, with most cardiac deaths presenting suddenly. SCD occurred at a rate more than four times expected in the general population, with similar risk factors. As cardiac symptoms were common in victims of SCD, aggressive primary prevention of CVD should be considered in HIV-infected individuals, especially those with traditional risk factors. While implantable cardioverter-defibrillators have been shown to be life saving in certain clinical settings,16 no studies have evaluated their utility specifically for individuals with HIV. As we seek to reduce mortality in an aging HIV-infected population, greater attention must be directed towards the mechanisms underlying SCD, with the goal of identifying at-risk individuals and ultimately preventing sudden death.
We thank My Linh Van, B.S. and Ben Colburn, B.S. for assistance with data collection.
This work was supported by NIH/NHLBI grants 5R01 HL102090 (ZHT), 5R01 HL095130 (PYH), and 5R01 HL091526 (PYH), Veterans Affairs Merit Review grant R01 NS501132 (JKW), and NIH grant K24 AI51982 (DVH).
None of the authors have any relationships to industry
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.