Since the advent of HAART in 1996 , the number of deaths from AIDS has declined dramatically [2,3]. Furthermore, the number of US death certificates that mention HIV disease has decreased from 47 977 in 1995 to 16 061 in 1999 . This decline has largely been attributed to HAART and improved treatment for AIDS-defining opportunistic illnesses [2,5]. The continued approval of antiretroviral drugs in new classes  and updated guidelines for the use of antiretroviral therapy (ART) for HIV-infected patients, such as those developed by a panel convened by the US Department of Health and Human Services , may contribute to further reducing HIV-related mortality.
Among HIV-infected persons, the increased longevity conferred by HAART and improved prophylaxis and treatment for opportunistic illnesses [2,5] has led to increases in the proportions of deaths caused by diseases not directly attributed to HIV infection [4,8,9] and to increases in morbidity secondary to the adverse effects of HAART, including renal disease, liver disease, heart disease, and diabetes [10–15]. Furthermore, longer survival may result in an increasing number of deaths from age-related diseases in HIV-infected patients (e.g., heart disease, cancer, renal disease) [16,17]. Studies regarding changes in causes of death for HIV-infected patients as the availability of HAART increases are limited in that they have relatively small sample sizes, collected data prior to 2000, were conducted at a single site, or reflect one type of medical care system [4,8,9,17–20]. To address these limitations, the present study has used data from the Adult/Adolescent Spectrum of HIV Disease (ASD) project. This large, diverse cohort enabled examination of perimortal conditions, including less frequent ones, during a period spanning the changing availability of HAART.
The primary goals were to describe the perimortal conditions (i.e., diagnosed conditions that caused death, as documented on the death certificate, or were present at death, as documented in medical records) that occurred among HIV-infected patients in a multicenter clinical cohort in the United States and to compare trends in these conditions during periods that reflected the increasing availability of HAART: pre-HAART (1992–1995), early HAART (1996–1999), and contemporary HAART (2000–2003). In addition, because the use of ART affects the prevalence of perimortal conditions and HIV/AIDS–related mortality , annual mortality rates and annual ART prevalence for ASD participants were analyzed and standardized according to the characteristics of decedents to provide the context for the observed trends in perimortal conditions.
In collaboration with state and local health departments, the Centers for Disease Control and Prevention (CDC) established ASD to monitor the spectrum and frequency of diseases in HIV-infected persons aged 13 years or older who received HIV care at any of the more than 100 inpatient and outpatient facilities in 11 US metropolitan areas . ASD was a longitudinal surveillance project conducted from 1990 through 2004. Trained data abstractors reviewed medical records for the 12-month period preceding a patient's first observation in ASD and every 6 months thereafter until December 2003 or the patient died, relocated, or was lost to follow-up. Loss to follow-up was defined as no medical record documentation of an encounter between patient and physician for at least 18 months; ASD could capture a death after loss to follow-up. Data for each patient included information on demographics, HIV transmission category, prescriptions of ART, CD4 T lymphocyte counts, AIDS-defining opportunistic illnesses, and as many as 10 diagnosed conditions present at death, which were abstracted from inpatient or outpatient medical records, death certificates, autopsy reports, and HIV/AIDS surveillance data. Thus, perimortal conditions included but were not limited to causes of death recorded on death certificates. Until January 2003, perimortal conditions were recorded in ASD software by using the codes from the ninth revision of the International Classification of Diseases (ICD); thereafter, codes from the tenth revision were also used.
Data were taken from the seven ASD metropolitan areas that utilized death certificates and medical records for the ascertainment of perimortal conditions: Seattle (Washington), Los Angeles (California), Denver (Colorado), Dallas (Texas), New Orleans (Louisiana), Atlanta (Georgia), and Bayamon (Puerto Rico).
The proportions of deaths were estimated for the various perimortal conditions that were reported, and trends in those conditions were calculated among patients who died during 1992–2003. Decedents were excluded if only HIV disease or a nonspecific diagnosis, such as cardiac arrest, was reported or if diagnosis at the time of death was missing. Among decedents who had at least one reported perimortal condition, the analysis covered the perimortal conditions reported among at least 1% of decedents during one of the three HAART periods. Proportionate mortality for a given perimortal condition was defined as the proportion of deaths for which that condition was reported among all deaths for which at least one perimortal condition was reported. The sum of the proportions of all perimortal conditions exceeded 100% because multiple perimortal conditions could be reported per death. To control for changes in patients' demographic characteristics and access to care during the study period, proportionate mortalities for each period were standardized according to the overall distribution of patients by sex, race/ethnicity, age at death, HIV transmission category, and lowest observed CD4 cell count for all decedents during 1992–2003. It was assumed that the lowest CD4 cell count, compared with the first CD4 cell count in ASD or the CD4 cell count before death, would most accurately reflect the most advanced stage of disease before ART initiation and that it would thus be a marker for accessing care. Chi-square statistics were used to test associations of sex, race/ethnicity, age at death, HIV transmission category, and lowest CD4 cell count with calendar period and for differences between patients included and excluded from the analyses.
For each perimortal condition, logistic regression was used to estimate the likelihood that decedents were diagnosed with the condition during the early HAART period compared with the pre-HAART period, during the contemporary HAART period compared with the early HAART period, and as a linear trend through all three periods. Multivariable models controlled for potential confounding by sex, race/ethnicity, age at death, HIV transmission category, and lowest CD4 cell count.
Annual mortality rates for this cohort study population were calculated as the total number of deaths, with or without specified perimortal conditions, divided by the total number of person-years of prospective observation per year. The calculation of ART prevalence was not limited to patients clinically eligible for ART because eligibility changed during the observation period, and this restriction would have excluded decedents who were never clinically eligible for ART. In addition to considering use of any ART, annual ART prevalence was also estimated by type of HAART-based therapy: either protease inhibitor (PI) or nonnucleoside reverse transcriptase inhibitor (NNRTI). To control for changes in patient characteristics over time, the mortality rates and ART prevalence were standardized (as with the proportionate mortality) to the overall distribution of sex, race/ethnicity, age at death, HIV transmission category, and lowest CD4 cell count of ASD patients who died from 1992 to 2003. Annual trends in mortality rates and ART prevalence were estimated by using multivariable Poisson and logistic regression, respectively, controlling for sex, race/ethnicity, age at death, HIV transmission category, and lowest CD4 cell count.
Analyses used SAS statistical software (version 9.1; SAS Institute, Cary, North Carolina, USA). Statistical significance was assessed at the 0.05 level, and all hypothesis tests were two sided.
From 1992 through 2003, a total of 13 895 deaths occurred; for 9225 (66.4%), specific perimortal conditions were documented. Of these, 5407 (58.6%) deaths occurred during the pre-HAART period, 2722 (29.5%) in the early HAART period, and 1096 (11.9%) during the contemporary HAART period. Of the total 9225 decedents, 16% were female, 47% were men who had sex with men but did not inject drugs, 39% were black, 93% were 25–54 years of age at death, and the lowest CD4 cell count for 73% was < 100 cells/μl (Table 1). The distributions of the decedents by sex, race/ethnicity, age at death, HIV transmission category, and lowest CD4 cell count differed significantly for the three periods (P < 0.001). Compared with decedents included in the analysis, decedents excluded on the basis of nonspecific perimortal diagnoses (17.9%) differed significantly by sex, HIV transmission category, race/ethnicity, and lowest CD4 cell count. Compared with decedents included in the analysis, decedents excluded owing to missing perimortal diagnoses (15.8%) differed significantly by HIV transmission category, race/ethnicity, and lowest CD4 cell count (Table 1).
Thirty-four perimortal conditions were diagnosed for 1% or more of the decedents. Of the 9225 decedents, 92.7% had one or more of these 34 perimortal conditions. Thirteen perimortal conditions were infectious diseases (Table 2), which were reported for 64.5% of decedents; the other 21 conditions were noninfectious diseases, reported for 54.0%. The most commonly diagnosed conditions were pneumonia (23.3%; excludes pneumonias owing to pathogens grouped in one of the other 33 categories of perimortal conditions), nontuberculous mycobacterial disease (12.0%), and septicemia (11.9%).
The trend in proportionate mortality during 1992–2003 reflected a significant increase for perimortal conditions such as liver disease (excluding viral hepatitis), viral hepatitis, hypertension, and alcohol abuse and a significant decrease for perimortal conditions such as pneumocystosis, nontuberculous mycobacterial disease, and cytomegalovirus disease (Table 2). When the multiyear periods were compared, significant changes in proportionate mortality were observed for several perimortal conditions. For example, from the early to contemporary HAART period, the proportionate mortality for viral hepatitis, tuberculosis, and diabetes mellitus increased significantly; proportionate mortality for nontuberculous mycobacterial disease, Kaposi sarcoma, cytomegalovirus disease, anemia, and chorioretinitis decreased significantly. These same trends occurred when the early HAART period was compared with the pre-HAART period, except for the proportionate mortalities associated with anemia and diabetes mellitus, which did not change significantly, and tuberculosis, which decreased significantly (Table 2).
From 1992 through 2003, 133.6 deaths per 1000 person-years occurred during 98 313 person-years of observation. Standardized to the overall distribution of demographic and other characteristics of the population of ASD decedents, the overall mortality rate for 1992–2003 was 260.6/1000 person-years. The standardized annual mortality rate dropped from a high of 487.5 per 1000 person-years in 1995 to a low of 100.6/1000 person-years in 2002 (Fig. 1). The annual rate, modeled by Poisson regression, underwent the largest year-to-year decline during the early HAART period [rate ratio (RR), 0.71; 95% confidence interval (CI), 0.69–0.73], a smaller decrease during the contemporary HAART period (RR, 0.93; 95% CI, 0.89–0.97), and an increase during the pre-HAART period (RR, 1.06; 95% CI, 1.04–1.08).
Of 36 256 persons prospectively monitored during 1992–2003, an average of 72.5% (75.7% when standardized to our study population of decedents) were prescribed ART annually. Standardized ART prescriptions decreased during the pre-HAART period to 58.8% in 1995 and then increased (during the early and contemporary HAART periods) to 83.5% in 2002. During the early HAART period, prescriptions for PIs and NNRTIs increased; during the contemporary HAART period, prescriptions for PIs decreased and prescriptions for NNRTIs remained constant (Fig. 1).
This analysis of data from a large cohort of HIV-infected patients in care during 1992–2003 demonstrated significant changes in the proportionate mortality for several perimortal conditions and a decline in annual mortality rates. Specifically, overall trends in proportionate mortality for many infectious AIDS-defining opportunistic illnesses decreased significantly. Although the proportionate mortality for infectious diseases decreased, infectious diseases continued to contribute most perimortal conditions, even during the most recent period. This finding may be attributable to the relative youth of the cohort members, since their deaths are less likely to be associated with chronic noninfectious diseases, which increase in prevalence among older populations . The greatest decrease in annual mortality rates occurred during the early HAART period. During the contemporary HAART period, the trends in overall ART and NNRTI prevalence leveled off and the trends in PI prevalence declined among this cohort, in which ART was prescribed for three quarters annually. Concurrently, the dramatic decline in mortality slowed. Identifying the reasons for this is beyond the scope of this study but could include the consequences of delaying HIV care  and barriers to HAART effectiveness, such as ART-associated toxicity, nonadherence, and resistance [7,24].
In addition to the improved immune status of patients receiving care during the HAART era, the changes in the proportionate mortalities for perimortal conditions likely reflected multiple factors, including changes in screening practices and evolving endeavors (e.g., antimicrobial prophylaxis guidelines, effective opportunistic illness-specific therapy, and public health control efforts [4,8,19]). For example, the increasing trend in proportionate mortality for viral hepatitis should be interpreted with caution because recommendations to screen for hepatitis C in HIV-infected patients were not issued until late 1998 ; hence, this increase could reflect increases in the number of diagnoses rather than an increase in prevalence or incidence. Other analyses have also shown statistically significant increases in the number of viral hepatitis diagnoses reported at death during the HAART era [4,8,19]. As for evolving endeavors, recommendations were amended on when to discontinue prophylaxis for primary and secondary opportunistic illnesses, new anti-cytomegalovirus therapies were approved, and concerted public health programs to control tuberculosis were established [26–28]. However, the increasing number of cases of multidrug-resistant tuberculosis nationally reported  and the increase in proportionate mortality for tuberculosis in ASD during the contemporary HAART period, compared with the early HAART period, could be early indicators that the trend is reversing.
Given the improved immune status associated with HAART, the significant increase in proportionate mortality for septicemia and the nonsignificant findings for pneumonia and meningitis were unexpected. Studies of death [4,8,17,19] have demonstrated both increasing and decreasing trends in septicemia and no change in the proportion of deaths caused by unspecified pneumonia during 1992–1999 .
Changes in HAART could also have been associated with decreases in proportionate mortality for specific noninfectious conditions. For example, the availability of tenofovir as an alternative to zidovudine , where zidovudine is associated with the side-effect of anemia, could have contributed to the decline in anemia among our cohort, whose use of prescribed zidovudine declined during the study period (data not shown). Also, the decrease in chorioretinitis may be attributed to decreases in other perimortal conditions that cause chorioretinitis and respond favorably to HAART and opportunistic illness prophylaxis (e.g., cytomegalovirus and toxoplasmosis).
In tandem with interventions involving HAART, chemotherapy advancements for the treatment of Kaposi sarcoma have continued . Our observed decline in perimortal Kaposi sarcoma has also been seen in other studies of deaths of HIV-infected persons [4,8,19]. However, we found no changes in proportionate mortality of cancers other than Kaposi sarcoma. The trends in these cancers among HIV-infected patients are inconsistent , and that inconsistency is reflected in other death analyses [4,8,17,19,20].
For less common perimortal conditions, our analysis did not detect any significant trends. For example, some studies have reported declining trends in primary multifocal leukoencephalopathy and cryptococcosis among decedents; others have found no trend [4,8,19]. Limited data on the occurrence of candidiasis before versus during the HAART era make a similar comparison difficult. Although we found a significantly decreasing trend in proportionate mortality for dementia, our definition of dementia was not limited to dementia secondary to HIV infection. Other death analyses have found conflicting results for HIV-related dementia [4,8,17,19].
Although this study could not measure the side effects of HAART , they may have contributed to our observed increasing trend in proportionate mortality for diabetes mellitus, ischemic heart disease, and hypertension, and they could explain the increases in proportionate mortality for kidney disease and liver disease in other analyses of causes of death [4,17,19,20]. Another consideration is that even though the prevalence of a condition (e.g., gastrointestinal hemorrhage ) is low, its association with mortality may be high. We also found an increasing trend in proportionate mortality for alcohol abuse. The speculated alcohol-induced adverse immunomodulation  and the decreased HAART adherence associated with active alcohol abuse and illicit drug use  may contribute to the presence of these perimortal conditions.
Although we controlled for potential confounding factors, residual confounding and differences in the design of our analysis compared with that of others (especially in terms of patients' access to care, type of care facility, study calendar period, and limitation to causes of death rather than perimortal conditions) could explain differences in the magnitude of our trends in proportionate mortality compared with the trends identified by other researchers. Furthermore, we could not compare our trends for non-AIDS related illnesses in ASD with those in a comparable HIV-seronegative cohort. Therefore, we were unable to determine whether the use of HAART and opportunistic illness prophylaxes has modified these trends among HIV-infected patients to be more similar to the trends among comparable HIV-seronegative persons.
Our findings must be interpreted in the context of several limitations. First, the generalizability of our results may be limited because the decedents with known perimortal conditions in the seven ASD metropolitan areas may not have been representative of all deaths among the HIV-infected population in the United States. However, the ASD catchment areas included large geographic areas that were the source of care for a substantial proportion of HIV-infected patients. Second, the data may have been affected by incomplete ascertainment of vital status; ASD patients who were lost to follow-up during 1992–2003 (27.7%) were more likely to be younger, white, and have higher lowest CD4 cell count. Third, data on perimortal conditions were missing or were not specific for 34% of decedents, precluding our ability to estimate disease-specific perimortal rates. Fourth, although changes in ICD coding could have led to discontinuity in trends, these changes had the potential to affect approximately only 1 of 12 study years. Finally, although ASD used a standardized data collection tool, the medical information available for abstraction was dependent on the individual clinician's clinical management style.
Since HAART became available to HIV-infected patients obtaining care in the United States, the proportions and numbers of many infectious perimortal diseases and annual mortality rates have declined. Further investigation is needed to determine the causes of the increase in proportionate mortality for several noninfectious perimortal diseases in a population with many prevalent comorbidities (e.g., psychiatric disorders, substance use, and alcohol abuse [34,35]), which could interfere with the effectiveness of HAART.
We would like to acknowledge the Adult/Adolescent Spectrum of HIV Disease Working Group. The findings and conclusions in this report are those of the authors and do not necessarily represent the view of the Centers for Disease Control and Prevention.
Sponsorship: Metropolitan areas were funded through cooperative agreement with the Centers for Disease Control and Prevention.
1. Bozzette SA, Joyce G, McCaffrey DF, Leibowitz AA, Morton SC, Berry SH, et al
. Expenditures for the care of HIV-infected patients in the era of highly active antiretroviral therapy. N Engl J Med 2001; 344:817–823.
2. Fleming PL, Ward JW, Karon JM, Hanson DL, De Cock KM. Declines in AIDS incidence and deaths in the USA: a signal change in the epidemic. AIDS 1998; 12(Suppl A):S55–S61.
3. Murphy EL, Collier AC, Kalish LA, Assmann SF, Para MF, Flanigan TP, et al
. Highly active antiretroviral therapy decreases mortality and morbidity in patients with advanced HIV disease. Ann Intern Med 2001; 135:17–26.
4. Selik RM, Byers RH Jr, Dworkin MS. Trends in diseases reported on US death certificates that mentioned HIV infection, 1987–1999. J Acquir Immune Defic Syndr 2002; 29:378–387.
5. Hogg RS, Yip B, Chan KJ, Wood E, Craib KJ, O'Shaughnessy MV, et al
. Rates of disease progression by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA 2001; 286:2568–2577.
6. Yeni P. Update on HAART in HIV. J Hepatol 2006; 44(Suppl 1):S100–S103.
7. Deeks SG. Antiretroviral treatment of HIV infected adults. BMJ 2006; 332:1489.
8. Crum NF, Riffenburgh RH, Wegner S, Agan BK, Tasker SA, Spooner KM, et al
. Comparisons of causes of death and mortality rates among HIV-infected persons: analysis of the pre, early, and late HAART (highly active antiretroviral therapy) eras. J Acquir Immune Defic Syndr 2006; 41:194–200.
9. Smith DK, Gardner LI, Phelps R, Hamburger ME, Carpenter C, Klein RS, et al
. Mortality rates and causes of death in a cohort of HIV-infected and uninfected women, 1993–1999. J Urban Health 2003; 80:676–688.
10. Barbaro G, Di LG, Cirelli A, Grisorio B, Lucchini A, Hazra C, et al
. An open-label, prospective, observational study of the incidence of coronary artery disease in patients with HIV infection receiving highly active antiretroviral therapy. Clin Ther 2003; 25:2405–2418.
11. Barrios A, Garcia-Benayas T, Gonzalez-Lahoz J, Soriano V. Tenofovir-related nephrotoxicity in HIV-infected patients. AIDS 2004; 18:960–963.
12. Brown TT, Cole SR, Li X, Kingsley LA, Palella FJ, Riddler SA, et al
. Antiretroviral therapy and the prevalence and incidence of diabetes mellitus in the Multicenter AIDS Cohort Study. Arch Intern Med 2005; 165:1179–1184.
13. Puoti M, Torti C, Ripamonti D, Castelli F, Zaltron S, Zanini B, et al
. Severe hepatotoxicity during combination antiretroviral treatment: incidence, liver histology, and outcome. J Acquir Immune Defic Syndr 2003; 32:259–267.
14. Reilly RF, Tray K, Perazella MA. Indinavir nephropathy revisited: a pattern of insidious renal failure with identifiable risk factors. Am J Kidney Dis 2001; 38:E23.
15. Tanuma J, Ishizaki A, Gatanaga H, Kikuchi Y, Kimura S, Hiroe M, et al
. Dilated cardiomyopathy in an adult human immunodeficiency virus type 1-positive patient treated with a zidovudine-containing antiretroviral regimen. Clin Infect Dis 2003; 37:e109–e111.
16. Cooley TP. Non-AIDS-defining cancer in HIV-infected people. Hematol Oncol Clin North Am 2003; 17:889–899.
17. Palella FJ, Baker RK, Moorman AC, Chmiel JS, Wood KC, Brooks JT, et al
. Mortality in the highly active antiretroviral therapy era: changing causes of death and disease in the HIV Outpatient Study. J Acquir Immune Defic Syndr 2006; 43:27–34.
18. Burckhardt B, Sendi P, Pfluger D, Zimmerli W, Nuesch R, Bucher HC, et al
. Rare AIDS-defining diseases in the Swiss HIV Cohort Study. Eur J Clin Microbiol Infect Dis 1999; 18:399–402.
19. Louie JK, Hsu LC, Osmond DH, Katz MH, Schwarcz SK. Trends in causes of death among persons with acquired immunodeficiency syndrome in the era of highly active antiretroviral therapy, San Francisco, 1994–1998. J Infect Dis 2002; 186:1023–1027.
20. Sansone GR, Frengley JD. Impact of HAART on causes of death of persons with late-stage AIDS. J Urban Health 2000; 77:166–175.
21. Farizo KM, Buehler JW, Chamberland ME, Whyte BM, Froelicher ES, Hopkins SG, et al
. Spectrum of disease in persons with human immunodeficiency virus infection in the United States. JAMA 1992; 267:1798–1805.
22. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System
. Atlanta, GA: Centers for Disease Control and Prevention; http://www.cdc.gov/brfss/
(accessed 30 January 2007).
23. Samet JH, Freedberg KA, Stein MD, Lewis R, Savetsky J, Sullivan L, et al
. Trillion virion delay: time from testing positive for HIV to presentation for primary care. Arch Intern Med 1998; 158:734–740.
24. d'Arminio Monforte A, Cozzi Lepri A, Rezza G, Pezzotti P, Antinori A, Phillips AN, et al
. Insights into the reasons for discontinuation of the first highly active antiretroviral therapy (HAART) regimen in a cohort of antiretroviral naive patients. AIDS 2000; 14:499–507.
25. Centers for Disease Control and Prevention. Recommendations for prevention and control of hepatitis C virus (HCV) infection and HCV-related chronic disease.MMWR Morb Mortal Wkly Rep
26. Centers for Disease Control and Prevention. Guidelines for preventing opportunistic infections among HIV-infected persons: 2002. Recommendations of the US Public Health Service and the Infectious Diseases Society of America.MMWR Recomm Rep
27. Centers for Disease Control and Prevention. Trends in tuberculosis: United States, 2005
. MMWR Morb Mortal Wkly Rep
28. Martin DF, Sierra-Madero J, Walmsley S, Wolitz RA, Macey K, Georgiou P, et al
. A controlled trial of valganciclovir as induction therapy for cytomegalovirus retinitis. N Engl J Med 2002; 346:1119–1126.
29. Tulpule A, Groopman J, Saville MW, Harrington W, Friedman-Kien A, Espina BM, et al
. Multicenter trial of low-dose paclitaxel in patients with advanced AIDS-related Kaposi sarcoma. Cancer 2002; 95:147–154.
30. Mbulaiteye SM, Parkin DM, Rabkin CS. Epidemiology of AIDS-related malignancies an international perspective. Hematol Oncol Clin North Am 2003; 17:673–696, v.
31. Chalasani N, Wilcox CM. Etiology and outcome of lower gastrointestinal bleeding in patients with AIDS. Am J Gastroenterol 1998; 93:175–178.
32. Isaki L, Kresina TF. Directions for biomedical research in alcohol and HIV: where are we now and where can we go? AIDS Res Hum Retroviruses 2000; 16:1197–1207.
33. Cargill VA, Stone VE. HIV/AIDS: a minority health issue. Med Clin North Am 2005; 89:895–912.
34. Bing EG, Burnam MA, Longshore D, Fleishman JA, Sherbourne CD, London AS, et al
. Psychiatric disorders and drug use among human immunodeficiency virus-infected adults in the United States. Arch Gen Psychiatry 2001; 58:721–728.
35. Galvan FH, Bing EG, Fleishman JA, London AS, Caetano R, Burnam MA, et al
. The prevalence of alcohol consumption and heavy drinking among people with HIV in the United States: results from the HIV Cost and Services Utilization Study. J Stud Alcohol 2002; 63:179–186.