Scott D. Holmberg

Declining Morbidity and Mortality among Patients with Advanced Human Immunodeficiency Virus Infection

BACKGROUND AND METHODS National surveillance data show recent, marked reductions in morbidity and mortality associated with the acquired immunodeficiency syndrome (AIDS). To evaluate these declines, we analyzed data on 1255 patients, each of whom had at least one CD4+ count below 100 cells per cubic millimeter, who were seen at nine clinics specializing in the treatment of human immunodeficiency virus (HIV) infection in eight U.S. cities from January 1994 through June 1997. RESULTS Mortality among the patients declined from 29.4 per 100 person-years in the first quarter of 1995 to 8.8 per 100 in the second quarter of 1997. There were reductions in mortality regardless of sex, race, age, and risk factors for transmission of HIV. The incidence of any of three major opportunistic infections (Pneumocystis carinii pneumonia, Mycobacterium avium complex disease, and cytomegalovirus retinitis) declined from 21.9 per 100 person-years in 1994 to 3.7 per 100 person-years by mid-1997. In a failure-rate model, increases in the intensity of antiretroviral therapy (classified as none, monotherapy, combination therapy without a protease inhibitor, and combination therapy with a protease inhibitor) were associated with stepwise reductions in morbidity and mortality. Combination antiretroviral therapy was associated with the most benefit; the inclusion of protease inhibitors in such regimens conferred additional benefit. Patients with private insurance were more often prescribed protease inhibitors and had lower mortality rates than those insured by Medicare or Medicaid. CONCLUSIONS The recent declines in morbidity and mortality due to AIDS are attributable to the use of more intensive antiretroviral therapies.

Mortality in the highly active antiretroviral therapy era: changing causes of death and disease in the HIV outpatient study.

Background: AIDS-related death and disease rates have declined in the highly active antiretroviral therapy (HAART) era and remain low; however, current causes of death in HAART-treated patients remain ill defined. Objective: To describe mortality trends and causes of death among HIV-infected patients in the HAART era. Design: Prospective, multicenter, observational cohort study of participants in the HIV Outpatient Study who were treated from January 1996 through December 2004. Measurements: Rates of death, opportunistic disease, and other non-AIDS-defining illnesses (NADIs) determined to be primary or secondary causes of death. Results: Among 6945 HIV-infected patients followed for a median of 39.2 months, death rates fell from 7.0 deaths/100 person-years of observation in 1996 to 1.3 deaths/100 person-years in 2004 (P = 0.008 for trend). Deaths that included AIDS-related causes decreased from 3.79/100 person-years in 1996 to 0.32/100 person-years in 2004 (P = 0.008). Proportional increases in deaths involving liver disease, bacteremia/sepsis, gastrointestinal disease, non-AIDS malignancies, and renal disease also occurred (P = <0.001, 0.017, 0.006, <0.001, and 0.037, respectively.) Hepatic disease was the only reported cause of death for which absolute rates increased over time, albeit not significantly, from 0.09/100 person-years in 1996 to 0.16/100 person-years in 2004 (P = 0.10). The percentage of deaths due exclusively to NADI rose from 13.1% in 1996 to 42.5% in 2004 (P < 0.001 for trend), the most frequent of which were cardiovascular, hepatic, and pulmonary disease, and non-AIDS malignancies in 2004. Mean CD4 cell counts closest to death (n = 486 deaths) increased from 59 cells/&mgr;L in 1996 to 287 cells/&mgr;L in 2004 (P < 0.001 for trend). Patients dying of NADI causes were more HAART experienced and initiated HAART at higher CD4 cell counts than those who died with AIDS (34.5% vs 16.8%, respectively, received HAART for 4 of more years, P < 0.0001; 22.4% vs 7.8%, respectively, initiated HAART with CD4 cell counts of more than 350 cells/&mgr;L, P < 0.001). Conclusions: Although overall death rates remained low through 2004, the proportion of deaths attributable to non-AIDS diseases increased and prominently included hepatic, cardiovascular, and pulmonary diseases, as well as non-AIDS malignancies. Longer time spent receiving HAART and higher CD4 cell counts at HAART initiation were associated with death from non-AIDS causes. CD4 cell count at time of death increased over time.

Incidence of Types of Cancer among HIV-Infected Persons Compared with the General Population in the United States, 1992–2003

BACKGROUND Persons who are HIV-infected may be at higher risk for certain types of cancer than the general population. OBJECTIVE To compare cancer incidence among HIV-infected persons with incidence in the general population from 1992 to 2003. DESIGN Prospective observational cohort studies. SETTING United States. PATIENTS 54,780 HIV-infected persons in the Adult and Adolescent Spectrum of HIV Disease Project (47,832 patients) and the HIV Outpatient Study (6948 patients), who contributed 157,819 person-years of follow-up from 1992 to 2003, and 334,802,121 records from the Surveillance, Epidemiology, and End Results program of 13 geographically defined, population-based, central cancer registries. MEASUREMENTS Standardized rate ratios (SRRs) to compare cancer incidence in the HIV-infected population with standardized cancer incidence in the general population. RESULTS The incidence of the following types of non-AIDS-defining cancer was significantly higher in the HIV-infected population than in the general population: anal (SRR, 42.9 [95% CI, 34.1 to 53.3]), vaginal (21.0 [CI, 11.2 to 35.9]), Hodgkin lymphoma (14.7 [CI, 11.6 to 18.2]), liver (7.7 [CI, 5.7 to 10.1]), lung (3.3 [CI, 2.8 to 3.9]), melanoma (2.6 [CI, 1.9 to 3.6]), oropharyngeal (2.6 [CI, 1.9 to 3.4]), leukemia (2.5 [CI, 1.6 to 3.8]), colorectal (2.3 [CI, 1.8 to 2.9]), and renal (1.8 [CI, 1.1 to 2.7]). The incidence of prostate cancer was significantly lower among HIV-infected persons than the general population (SRR, 0.6 [CI, 0.4 to 0.8]). Only the relative incidence of anal cancer increased over time. LIMITATIONS Lower ascertainment of cancer in the HIV cohorts may result in a potential bias to underestimate rate disparities. Tobacco use as a risk factor and the effect of changes in cancer screening practices could not be evaluated. CONCLUSION The incidence of many types of non-AIDS-defining cancer was higher among HIV-infected persons than among the general population from 1992 to 2003.

Epidemiology of human immunodeficiency virus-associated opportunistic infections in the United States in the era of highly active antiretroviral therapy.

The incidence of nearly all AIDS-defining opportunistic infections (OIs) decreased significantly in the United States during 1992-1998; decreases in the most common OIs (Pneumocystis carinii pneumonia ¿PCP, esophageal candidiasis, and disseminated Mycobacterium avium complex ¿MAC disease) were more pronounced in 1996-1998, during which time highly active antiretroviral therapy (HAART) was introduced into medical care. Those OIs that continue to occur do so at low CD4+ T lymphocyte counts, and persons whose CD4+ counts have increased in response to HAART are at low risk for OIs, a circumstance that suggests a high degree of immune reconstitution associated with HAART. PCP, the most common serious OI, continues to occur primarily in persons not previously receiving medical care. The most profound effect on survival of patients with AIDS is conferred by HAART, but specific OI prevention measures (prophylaxis against PCP and MAC and vaccination against Streptococcus pneumoniae) are associated with a survival benefit, even when they coincide with the administration of HAART. Continued monitoring of incidence trends and detection of new syndromes associated with HAART are important priorities in the HAART era.

Transmission of Human Immunodeficiency Virus Type 1 from a Seronegative Organ and Tissue Donor

BACKGROUND Since 1985, donors of organs or tissues for transplantation in the United States have been screened for human immunodeficiency virus type 1 (HIV-1), and more than 60,000 organs and 1 million tissues have been transplanted. We describe a case of transmission of HIV-1 by transplantation of organs and tissues procured between the time the donor became infected and the appearance of antibodies. The donor was a 22-year-old man who died 32 hours after a gunshot wound; he had no known risk factors for HIV-1 infection and was seronegative. METHODS We reviewed the processing and distribution of all the transplanted organs and tissues, reviewed the medical histories of the donor and HIV-1-infected recipients, tested stored donor lymphocytes for HIV-1 by viral culture and the polymerase chain reaction, and tested stored serum samples from four organ recipients for HIV-1 antigen and antibody. RESULTS HIV-1 was detected in cultured lymphocytes from the donor. Of 58 tissues and organs obtained from the donor, 52 could be accounted for by the hospitals that received them. Of the 48 identified recipients, 41 were tested for HIV-1 antibody. All four recipients of organs and all three recipients of unprocessed fresh-frozen bone were infected with HIV-1. However, 34 recipients of other tissues--2 receiving corneas, 3 receiving lyophilized soft tissue, 25 receiving ethanol-treated bone, 3 receiving dura mater treated with gamma radiation, and 1 receiving marrow-evacuated, fresh-frozen bone--tested negative for HIV-1 antibody. Despite immunosuppressive chemotherapy, HIV-1 antibody appeared between 26 and 54 days after transplantation in the three organ recipients who survived more than four weeks. CONCLUSIONS Although rare, transmission of HIV-1 by seronegative organ and tissue donors can occur. Improvements in the methods used to screen donors for HIV-1, advances in techniques of virus inactivation, prompt reporting of HIV infection in recipients, and accurate accounting of distributed allografts would help to reduce further this already exceedingly low risk.

The Increasing Burden of Mortality From Viral Hepatitis in the United States Between 1999 and 2007

BACKGROUND The increasing health burden and mortality from hepatitis B virus (HBV) and hepatitis C virus (HCV) in the United States are underappreciated. OBJECTIVE To examine mortality from HBV; HCV; and, for comparison, HIV. DESIGN Analysis of U.S. multiple-cause mortality data from 1999 to 2007 from the National Center for Health Statistics. SETTING All U.S. states and the District of Columbia. PARTICIPANTS Approximately 22 million decedents. MEASUREMENTS Age-adjusted mortality rates from HBV, HCV, and HIV. Logistic regression analyses of 2007 data generated 4 independent models per outcome (HCV- or HBV-related deaths) that each included 1 of 4 comorbid conditions and all sociodemographic characteristics. RESULTS Between 1999 and 2007, recorded deaths from HCV [corrected] increased significantly to 15,106, whereas deaths from HIV declined to 12,734 by 2007. Factors associated with HCV-related deaths included chronic liver disease, HBV co-infection, alcohol-related conditions, minority status, and HIV co-infection. Factors that increased odds of HBV-related death included chronic liver disease, HCV co-infection, Asian or Pacific Islander descent, HIV co-infection, and alcohol-related conditions. Most deaths from HBV and HCV occurred in middle-aged persons. LIMITATION A person other than the primary physician of the decedent frequently completed the death certificate, and HCV and HBV often were not detected and thus not reported as causes of death. CONCLUSION By 2007, HCV had superseded HIV as a cause of death in the United States, and deaths from HCV and HBV disproportionately occurred in middle-aged persons. To achieve decreases in mortality similar to those seen with HIV requires new policy initiatives to detect patients with chronic hepatitis and link them to care and treatment. PRIMARY FUNDING SOURCE Centers for Disease Control and Prevention.

Protease inhibitors and cardiovascular outcomes in patients with HIV-1

Protease inhibitors for treatment of HIV-1 have been linked with increased risk of hyperlipidaemia and hyperglycaemia. In a cohort of 5672 outpatients with HIV-1 seen at nine US HIV clinics between January, 1993, and January, 2002, the frequency of myocardial infarctions increased after the introduction of protease inhibitors in 1996 (test for trend, p=0.0125). We noted that 19 of 3247 patients taking, but only two of 2425 who did not take, protease inhibitors had a myocardial infarction (odds ratio 7.1, 95% CI 1.6-44.3; Cox proportional hazards model-adjusted for smoking, sex, age, diabetes, hyperlipidaemia, and hypertension-hazard ratio 6.5, 0.9-47.8). Our findings suggest that, although infrequent, use of protease inhibitors is associated with increased risk of myocardial infarction in patients with HIV-1.

Chronic Hepatitis C Virus Infection in the United States, National Health and Nutrition Examination Survey 2003 to 2010

Context Chronic hepatitis C virus (HCV) infection is an important public health issue. Using data from a U.S. household survey conducted between 2003 and 2010, the authors compared the estimated prevalence of chronic HCV infection and risk factors for infection with those from earlier periods. Contribution The estimated prevalence of chronic HCV infection in the United States has decreased. Risk factors are essentially unchanged from previous periods and were reported by only about one half of infected persons. Caution Homeless and incarcerated persons were not surveyed. Implication The burden of chronic HCV infection in the United States is substantial. National data on prevalence are useful for the design of programs for HCV screening, linkage to care, and treatment. The Editors Hepatitis C virus (HCV) infection is a treatable but underrecognized and underdiagnosed disease. An estimated 130 to 170 million persons, 2% to 3% of the worlds population, are living with HCV infection, and almost 500000 persons die of HCV-related conditions each year (primarily decompensated cirrhosis and liver cancer) (1). In the United States, previous estimates have consistently indicated that approximately 3 million or more persons have chronic HCV infection. An analysis of 21241 serum specimens from participants in NHANES (National Health and Nutrition Examination Survey), which provides nationally representative statistics on the health of the U.S. noninstitutionalized civilian population, indicated that 2.7 million persons (95% CI, 2.4 to 3.0 million persons) had chronic HCV infection between 1988 and 1994 (2). A similar analysis of 15079 NHANES specimens between 1999 and 2002 estimated that 3.2 million persons (95% CI, 2.7 to 3.9 million persons) had chronic HCV infection (3). These estimates do not include cases of chronic HCV not captured by NHANES, notably among homeless persons and persons who were incarcerated during the survey (4). The Institute of Medicine recently concluded that it is essential to know the dimensions and direction of this epidemic, which has major implications for health burden and costs for the United States (5). Current treatment can cure HCV in a substantial proportion of persons who complete therapy, thereby decreasing the risk for hepatocellular carcinoma and all-cause mortality. However, many persons infected with HCV remain untested and unaware of their infection, are unknown to the health care system, and are not captured in case-based surveillance because they are typically asymptomatic (6). Deaths among persons with HCV infection have superseded deaths in those with HIV infection (7). Surveillance for antibody to HCV (anti-HCV) and HCV RNA has been part of NHANES since the 1980s, although RNA testing for NHANES III was done retrospectively. Surveillance through such a large national survey presents the best measurement of the prevalence of anti-HCV and chronic HCV infection in the general U.S. population. Accordingly, using methods similar to analyses from 20 and 10 years ago (2, 3), we analyzed data from participants in NHANES between 2003 and 2010 to estimate the prevalence of HCV infection and to determine risk factors and exposures associated with chronic infection. Methods Survey Design The National Health and Nutrition Examination Survey, conducted by the Centers for Disease Control and Preventions National Center for Health Statistics, collects nationally representative data on the health and nutritional status of the U.S. noninstitutionalized civilian population. This survey uses a complex, stratified, multistage probability sampling design and collects information from approximately 5000 persons annually using standardized interviews, physical examinations, and tests of biological samples. Participants were interviewed in their homes using the interviewer-administered Computer-Assisted Personal Interviewing system to ascertain demographic characteristics and in the Mobile Examination Center to ascertain possible risks and exposures for HCV infection. Persons aged 16 years or older and emancipated minors were interviewed directly; an adult proxy provided information for participants younger than 16 years and for persons unable to answer the questions themselves. All participants provided written informed consent. More detailed information on survey design for NHANES, including approval from the National Center for Health Statistics Institutional Review Board (Hyattsville, Maryland), is available from the survey documentation at www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm. Laboratory Testing Qualitative determination of anti-HCV in blood serum or plasma was measured using direct solid-phase enzyme immunoassay with an anti-HCV screening chemiluminescence immunoassay (VITROS Anti-HCV Immunodiagnostic System, Ortho Clinical Diagnostics, Rochester, New York). Screening reactive specimens were then tested using a confirmatory recombinant immunoblot assay (RIBA) (RIBA HCV 3.0 Strip Immunoblot Assay, Chiron, Emeryville, California), an in vitro qualitative immunoassay for the detection of anti-HCV in human serum or plasma. Samples with positive results on RIBA testing were reported as confirmed positive for anti-HCV, those with results that were negative were reported as negative for anti-HCV, and those with indeterminate results were reported as indeterminate. In clinical practice, it is most important to identify persons who are currently infected; however, for surveillance purposes, we are interested in having a reliable measure of both those who are currently infected and those who were ever infected. Although the sensitivity and specificity of anti-HCV tests have improved over time for at-risk populations, estimating the true prevalence in a low-risk, low-prevalence population, such as that sampled in NHANES, requires a confirmatory test, such as RIBA, to eliminate false-positive results from our estimates of persons ever infected. Serum samples that were confirmed positive or indeterminate for anti-HCV were further tested for HCV RNA using an in vitro nucleic acid amplification test for the quantitation of HCV RNA in human serum or plasma. We used the COBAS AMPLICOR HCV Test, version 2.0 (Roche Diagnostics, Indianapolis, Indiana), on the COBAS AMPLICOR Analyzer (Roche Diagnostics) for samples from 2005 to 2010 and the COBAS AmpliPrep/TaqMan HCV Test, version 2 (Roche Diagnostics), on the COBAS TaqMan 48 Analyzer (Roche Diagnostics) for samples from 2003 to 2004. We considered persons to have chronic HCV infection if results of their test for anti-HCV were confirmed positive or indeterminate and results of their test for HCV RNA were positive. Our comparison group comprised persons who tested negative for anti-HCV; these participants were considered to be never infected with HCV. Those who tested positive for anti-HCV but negative for HCV RNA (resolved infections; n= 90) and those who had no serum available for RNA testing (n= 51) were not included in our analyses, except for estimation of overall anti-HCV prevalence, because we wanted to focus on chronic HCV infection. Statistical Analysis SAS-Callable SUDAAN, Release 10.0 (Research Triangle Institute, Research Triangle Park, North Carolina) (8), a statistical package designed to analyze complex survey data, was used for analysis. Estimates were weighted to represent the total U.S. noninstitutionalized civilian population and to account for oversampling and nonresponse to the household interview and physical examination. Two-year sample weights (WTMEC2yr) were further adjusted to account for the fact that not all examination participants were tested for anti-HCV, not all participants who tested positive or indeterminate for anti-HCV had samples available for HCV RNA testing, and multiple years of data were used. A P value less than 0.05 was considered statistically significant. We analyzed demographic characteristics (age at interview, sex, race/ethnicity, birthplace, education, and income), potential risk factors or exposures (receipt of blood or a blood product before 1992, any past injection drug use, and number of lifetime sexual partners), and a proxy for sexual risk (antibodies to herpes simplex virus type 2) (9, 10). Although sex is not a usual method of transmission of HCV infection (11), number of sexual partners and presence of herpes simplex virus type 2 antibodies are included because these are indices of increased likelihood of being infected with HCV and have been used in previous analyses. We restricted most analyses to persons aged 20 years or older because only 2 persons aged 6 to 19 years had evidence of chronic HCV infection (that is, tested positive for HCV RNA) and because data on drug use and sexual behaviors among those younger than 20 years are not available from NHANES public-use data files. Ages included in our reporting of risks and exposures reflect age eligibility for a particular question or laboratory test rather than a focus on particular age groups. We used bivariate analyses to estimate demographic characteristics of persons with chronic HCV infection and those who were never infected and to estimate the prevalence of potential risk factors or exposures among population subgroups. Chi-square tests were used for statistical comparisons between subgroups. Simple (unadjusted) and multivariate logistic regression analyses were used to identify factors associated with chronic HCV infection. We performed separate logistic analyses for persons aged 20 to 59 years and those aged 60 years or older because NHANES does not query those older than 59 years about sexual behaviors or drug use. We retained 3 potential confoundersage at interview, sex, and race/ethnicityin all multivariate models. Variables that were associated with HCV infection in previous NHANES analyses (2, 3) were included in multivariate models to identify factors independently associated with chronic HCV infection. Variables not inclu

Clinical assessment of Hiv-associated lipodystrophy in an ambulatory population

ObjectiveTo identify clinical factors associated with prevalence of fat atrophy (lipoatrophy) and fat accumulation (lipoaccumulation) in HIV-1 infected patients. DesignEvaluation of HIV-1 infected patients seen for routine care between 1 October and 31 December 1998 in the eight HIV Outpatient Study (HOPS) clinics. SettingEight clinics specializing in the care of HIV-1 infected patients. PatientsA total of 1077 patients were evaluated for signs of fat maldistribution. InterventionsA standardized set of questions and specific clinical signs were assessed. Demographic, clinical and pharmacological data for each patient were also included in the analysis. Main outcome measuresDemographic, immunologic, virologic, clinical, laboratory, and drug treatment factors were assessed in stratified and multivariate analyses for their relationship to the presence and severity of fat accumulation and atrophy. ResultsIndependent factors for moderate/severe lipoatrophy for 171 patients were increasing age, any use of stavudine, use of indinavir for longer than 2 years, body mass index (BMI) loss, and measures of duration and severity of HIV disease. Independent risk factors for moderate/severe fat accumulation for 104 patients were increasing age, BMI gain, measures of amount and duration of immune recovery, and duration of antiretroviral therapy (ART). The number of non-drug risk factors substantially increased the likelihood of lipoatrophy. If non-drug risk factors were absent, lipoatrophy was unusual regardless of the duration of drug use. ConclusionsHIV-associated lipodystrophy is associated with several host, disease, and drug factors. While prevalence of lipoatrophy increased with the use of stavudine and indinavir, and lipoaccumulation was associated with duration of ART, other non-drug factors were strongly associated with both fat atrophy and accumulation.

Survival Benefit of Initiating Antiretroviral Therapy in HIV-Infected Persons in Different CD4+ Cell Strata

Context When to start antiretroviral therapy (ART) for HIV infection is controversial. Starting too early exposes patients to side effects and uncertain benefits. Starting too late deprives patients of benefits. Contribution In this cohort study, HIV-infected patients with baseline CD4+ cell counts between 0.201 and 0.350 109 cells/L who began ART immediately had lower mortality rates than those who started therapy after their CD4+ cell count decreased to less than 0.201 109 cells/L. Optimal timing of therapy is unclear when the CD4+ cell count is greater than 0.350 109 cells/L. Cautions A randomized, controlled trial is the best way to identify the optimal timing of ART. The Editors Optimal timing of antiretroviral therapy (ART) initiation for persons with HIV infection is of great clinical and public health importance. Therapy reduces HIV-related mortality and morbidity for patients with substantial CD4+ cell depletion (<0.100 109 cells/L) who initiate treatment (1). Although data demonstrate the viral suppressive and immunologic (CD4+ cell count) benefits of therapy in persons with higher CD4+ cell counts (2-8), long-term improvements in disease-associated morbidity and mortality with earlier therapy are less clear (9, 10). In such patients, the potential benefits of ART and highly active ART (HAART) will probably be weighed against possible untoward sequelae of earlier treatment, including the development of metabolic abnormalities; emergence of drug-resistant virus, with resultant exhaustion of effective remaining therapies; cost; and access (9, 11-13). Current treatment guidelines allowing for the delay of ART until a lower CD4+ thresholdusually 0.350 109 cells/L or, for some patients, 0.200 109 cells/Lreflect a lack of consensus on the benefits of earlier initiation of therapy (13, 14). Sparse data exist on which to base specific recommendations for the initiation of ART relative to CD4+ cell count. Longitudinal data comparing ART recipients to appropriate comparison groups not receiving ART (especially patients with CD4+ cell counts > 0.200 109 cells/L) are limited (10, 15). Analyses that include extended follow-up data on such patients are critical because these patients are unlikely to develop or die of an HIV-related condition over the short term, in contrast to those who start therapy with lower CD4+ cell counts. Another challenge is related to the relatively brief time that HAART has been available (since early 1996), making comparative longitudinal studies of sufficient duration difficult. We compare mortality rates among ambulatory HIV-infected patients who initiated ART and those who delayed ART in various CD4+ strata. Patients were enrolled in the HIV Outpatient Study (HOPS), a dynamic cohort of ambulatory HIV-infected patients demographically representative of treated HIV-infected patients in the United States. Methods HOPS HOPS is an ongoing prospective observational cohort study into which patients have been continuously recruited and followed since 1993 (1, 16). Study sites are 10 clinics (8 private, 2 public) in 8 U.S. cities that provide care for more than 2400 HIV-infected patients per year. Participating physicians have extensive experience treating HIV-infected patients. Information is abstracted from outpatient charts at each visit and entered electronically by trained staff; it is then compiled centrally and reviewed and edited before being analyzed. Information abstracted includes demographic characteristics and risk factors for HIV infection; symptoms; diagnosed diseases (both definitive and presumptive); medications prescribed, including dose and duration; and laboratory values, including CD4+ cell counts and measurements of plasma HIV-1 RNA (viral load). Selection of Patients for Analysis We identified HOPS participants who had at least two CD4+ measurements and reliable data on ART initiation and use for at least 30 consecutive days from January 1994 through March 2001. We defined HAART as the use of at least three drugs simultaneously, including one protease inhibitor or non-nucleoside reverse-transcriptase inhibitor, or any regimen with at least two full-dose protease inhibitors. Three patient subgroups were analyzed: those observed to have a pre-ART CD4+ cell count of 0.501 to 0.750 109 cells/L, those with a pre-ART CD4+ cell count of 0.351 to 0.500 109 cells/L, and those with a pre-ART CD4+ cell count of 0.201 to 0.350 109 cells/L. Patients could be in more than one subgroup if they had a pre-ART CD4+ cell count in more than one of the defined ranges. Thus, analyses within a subgroup are distinct from analyses in other subgroups. We then stratified patients in each subgroup into one of three treatment groups: those who began ART while still in the same CD4+ subgroup range (subsequently called patients who initiated ART), those who began ART after their CD4+ cell count decreased to less than the CD4+ subgroup range (subsequently called patients who delayed ART), and those who never received ART (untreated patients). The closest (in time) CD4+ cell count available within 6 months before or 2 weeks after ART initiation was used to define the CD4+ cell count at the start of therapy. By definition, because patients who delayed ART had to have at least 1 additional CD4+ measurement during follow-up, patients who initiated ART and those who were untreated were also required to have at least 1 additional CD4+ measurement during follow-up to reduce potential bias in the analysis as a result of differential time under observation. For all treatment groups, time under observation began with the date of the earliest CD4+ cell count within the CD4+ stratum in which the patient was analyzed. Patients included in the analyses of the CD4+ subgroups of 0.201 to 0.350 109 cells/L and 0.351 to 0.500 109 cells/L were those whose earliest CD4+ cell count within the subgroup-defined range was observed after 1 January 1994. The analysis of the CD4+ subgroup of 0.501 to 0.750 109 cells/L was limited to those whose earliest CD4+ cell count within this range was observed between January 1994 and December 1995. This allowed longer elapsed time to observe clinical events. For analysis, the observation period for each patient ended at 6 months after the last contact with a HOPS clinic or at death. We analyzed all deaths, including those not directly due to AIDS or indirectly from conditions exacerbated by HIV infection (such as hepatic, renal, or cardiac disease). Causes of death were ascertained through review of clinic and hospital charts, death certificates, and national AIDS surveillance data. Deaths from suicide (one patient in the CD4+ subgroup of 0.201 to 0.350 109 cells/L who delayed ART and one patient in the CD4+ subgroup of 0.501 to 0.750 109 cells/L who initiated treatment) were treated as censored. Statistical Analysis We used SAS software, version 8.0 (SAS Institute, Inc., Cary, North Carolina), for all analyses. Patient characteristics were compared by chi-square test or the Fisher exact test for categorical variables and the Wilcoxon rank-sum test or t-test for continuous variables. We analyzed mortality rates per 1000 person-years and calculated the relative risk for death, 95% CIs, and approximate two-sided P values for each subgroup (17). Cox proportional-hazards regression was used to estimate hazard ratios, adjusted for age, sex, race, insurance status, viral load (log scale) at time of first ART (a dummy variable was used to include patients missing viral load data), receipt of HAART, and CD4+ cell count at the time of first observation within each stratum. Role of the Funding Source The funding source participated in the design, conduct, analysis, and reporting of the study and in the decision to submit the manuscript for publication. Results We evaluated data from 1464 HIV-infected HOPS participants. Of these patients, 596 who initiated ART had at least one additional CD4+ measurement after ART initiation, and 175 who delayed ART had at least one additional recorded CD4+ cell count in a higher stratum before ART initiation. We compared the demographic and baseline characteristics of patients described in this report to those of the larger overall group of HOPS participants and found no meaningful differences (data not shown). We analyzed data from 399 patients (340 who initiated and 59 who delayed ART) with pre-ART CD4+ cell counts between 0.201 and 0.350 109 cells/L, 327 patients (240 who initiated and 87 who delayed ART) with pre-ART CD4+ cell counts between 0.351 and 0.500 109 cells/L, and 122 patients (55 who initiated and 67 who delayed ART) with pre-ART CD4+ cell counts between 0.501 and 0.750 109 cells/L. Median years of follow-up for patients who initiated and those who delayed ART, by CD4+ subgroup, were as follows: 3.8 and 3.9 years for the subgroup of 0.201 to 0.350 109 cells/L, 4.1 and 4.2 years for the subgroup of 0.351 to 0.500 109 cells/L, and 5.4 and 5.3 years for the subgroup of 0.501 to 0.750 109 cells/L, respectively. Table 1 shows the demographic, immunologic, virologic, and care characteristics of patients who initiated ART and those who delayed ART, by CD4+ subgroup. Across subgroups, at least 69% of patients were men, 64% were younger than 40 years of age, 62% were white, and 35% had private health care insurance. Patients who initiated ART and those who delayed ART did not differ significantly except for the following: Patients in the CD4+ subgroup of 0.351 to 0.500 109 cells/L with private insurance tended to initiate rather than delay ART, and men in the CD4+ subgroup of 0.501 to 0.750 109 cells/L tended to delay therapy. Table 1. Characteristics of the HIV Outpatient Study Patients Who Initiated or Delayed Antiretroviral Therapy, by Preantiretroviral CD4+ Cell Count Stratum In general, most patients in a CD4+ subgroup who delayed ART initiated therapy in the next lowest CD4+ subgroup, that is, those who did not start in one subgroup started approximat