Secondary Logo

Journal Logo

Brief Report: Clinical Science

Predictors of Mortality for Patients With Advanced Disease in an HIV Palliative Care Program

Shen, Jennifer M MD, MPH; Blank, Arthur PhD; Selwyn, Peter A MD, MPH

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: December 1st, 2005 - Volume 40 - Issue 4 - p 445-447
doi: 10.1097/01.qai.0000185139.68848.97


Although mortality from AIDS in the United States has dramatically declined since 1996, the rate of decline has diminished and AIDS and associated comorbidities remain a significant cause of death in many young adult populations.1 In the era of highly active antiretroviral therapy (HAART), AIDS has been transformed from a rapidly and uniformly fatal disease into a more prolonged illness with exacerbations and remissions, a growing cumulative disease burden, significant therapy-related toxic effects, and increasing medical and psychiatric comorbidities.2 As patients survive longer in the latter stages of progressive HIV disease, disease-specific “curative” therapies and symptom-specific “palliative” care must be integrated to meet the new challenges of advanced HIV disease.2-5 Moreover, little is known about the prognostic factors that may predict mortality in the late-stage population of long-surviving chronic patients in the HAART era. In this follow-up study, we describe the causes and potential predictors of death in a cohort of late-stage HIV patients in a previously described HIV palliative care program in a large urban teaching hospital.3


Montefiore Medical Center is a large urban medical center located in the Bronx, New York. In 1999, the Montefiore Medical Center was awarded a grant to develop an HIV palliative care program by the Health Resources and Services Administration (HRSA) through the Ryan White Care Act as 1 of 6 US sites in a Special Projects of National Significance (SPNS) initiative on HIV and palliative care.4 The HIV palliative care program was developed within Montefiore Medical Center's extensive delivery network, focusing on the integration of palliative care and HIV services for traditionally underserved AIDS populations in inpatient and outpatient settings.3

As previously described, patients were referred to the HIV palliative care team by care providers throughout the medical center, primarily from inpatient medical and AIDS services.3 Enrollment started on July 1, 2000; patients were referred for clinical consultation, most commonly to address issues of pain and symptom management, goals of care, advance care planning, end-of-life care, and psychosocial problems for patients and/or caregivers. Patients were evaluated by the palliative care consultation team through a standardized intake assessment, including history and physical examination as well as assessment of symptoms using the Memorial Symptom Assessment Scale (MSAS)6 and other instruments, including the Karnofsky Performance Status Scale7 and the Rapid Disability Rating Scale (RDRS).8 Activities of daily living (ADL), as categorized by the RDRS, included eating, walking, mobility, bathing, dressing, self-toileting, continence, and a category representing time spent in bed. All patients' medical records were also reviewed for HIV-related history, laboratory results, and other clinically relevant information. After the patient's death or completion of follow-up, all relevant medical record data were reviewed using a standardized chart review instrument to assess disposition, outcome, and cause of death. Follow-up occurred until patients died, were discharged to a nursing home or other long-term care facility, or otherwise left the program, or until March 1, 2003, when the project ended. Quality of the ascertainment of cause of death was addressed by double review by 2 reviewers of a 20% sample of cases, with 95% concordance.

Data were entered and cleaned in Access 98 and analyzed using SAS System Software, version 9.0 (SAS Institute, Cary, NC). Analyses of predictors of mortality were conducted using the χ2 test and logistic regression analysis. Traditional markers of HIV progression (eg, CD4+ count, HIV viral load) as well as demographic variables and measures of functionality (eg, Karnofsky score, number of ADL impairments) were included in the univariate analysis. Those found to have significance on univariate analysis were included in the multivariate analysis. The Cox proportional hazard model was used for multivariable analysis, estimating the relative risk of death and 95% confidence intervals (CIs), adjusting for other independent risk factors. For analysis of mortality, only baseline data were used to determine predictors of death during follow-up. The output of the multivariable model was assessed using the Hosmer and Lemeshow goodness-of-fit test.9 This project and its evaluation components were approved by the Institutional Review Board of Montefiore Medical Center.


In 230 patients followed by the palliative care program, the median age at enrollment was 43 years, with a range of 21 to 79 years. More than half of the patients were male (56%); 54% of patients identified themselves as Hispanic and 39% as black. The most common type of exposure to HIV was injection drug use (41%), followed by heterosexual contact (32%) and male same-sex contact (11%). Most patients (89%) had been diagnosed with an AIDS-defining illness and/or a prior CD4+ T-lymphocyte count of <200 cells/mm3; 75% were referred from inpatient hospital services. Clinical features demonstrate the advanced disease stage of most patients enrolled in the program. Median baseline values included CD4+ count of 39 cells/mm3 (range: 0-987 cells/mm3; 25th-75th percentile, 10-148 cells/mm3), HIV viral load of 65,202 copies/mL (range: <50 to >750,000 copies/mL; 25th-75th percentile, 3063-339,000 copies/mL), Karnofsky score of 30 (range: 10-100), and 5 impairments in ADL (range: 0-8 impairments).

Over a median follow-up of 126 days (range: 1-823 days) 120 patients (54%) died, 27% were discharged to a nursing home or other long-term care facility, 1% self-discharged (against medical advice) from the palliative care program, and 7% were medically discharged because of resolution of all palliative care-related issues requiring ongoing follow-up.

Of the 120 deaths, analysis of cause-specific mortality indicated that 43 (36%) patients died of end-stage AIDS, 23 (19%) of non-AIDS-defining cancers, 22 (18%) of bacterial pneumonia or sepsis, 15 (13%) of liver failure and/or cirrhosis, 9 (8%) of cardiac or pulmonary disease (eg, congestive heart failure, coronary artery disease, chronic obstructive pulmonary disease), 4 (3%) of end-stage renal disease, 2 (2%) of amyotrophic lateral sclerosis, and 2 (2%) of unknown causes.

Comparison of patients who died with those who did not die during follow-up indicated that death was not predicted by gender, baseline symptoms on the MSAS, HIV risk behavior, HIV disease stage, baseline CD4+ count, or baseline HIV viral load. Cox proportional hazards analysis of mortality revealed that age >65 years (risk ratio = 18.37, 95% CI: 3.92 to 86.06; P = 0.0002) and total ADL impairments (risk ratio per impairment = 1.14, 95% CI: 1.09 to 1.19; P < 0.0001) were the only significant predictors of mortality. Because the Karnofsky score and ADL were closely correlated, we ran the Cox model with each of these terms separately; the model using the Karnofsky score yielded a risk ratio for death of 1.03 (95% CI: 1.02 to 1.04; P < 0.0001) for each 1-point decrease in score (ie, a decrease of 10 points conferred an increased risk of dying of approximately 30%).


In this study of 230 late-stage HIV patients with AIDS enrolled in an HIV palliative care program in the Bronx, New York, we found that markers of impaired functional status, such as Karnofsky score and impairments in ADL, were the strongest predictors of mortality, in addition to age older than 65 years. Although a significant body of data exists validating CD4+ cell count and HIV viral load as powerful predictors of survival and disease progression among those with early and moderately advanced disease, limited data exist relating to those with extremely advanced disease.10 Previous studies have shown that assessment of functional status as indicated by ADL may be more predictive of inpatient mortality with AIDS than CD4+ counts.11 Traditional clinical markers of HIV disease progression may be even less useful in predicting mortality for patients with extremely late-stage HIV disease, especially in the HAART era. Instead, as shown by our study, functionality, as measured through the Karnofsky score or ADL impairments, may be a stronger indicator of prognosis in this chronically ill population near the end of life.

The importance of functional status in predicting mortality has been shown in other diseases and populations as well. For instance, difficulty with several instrumental ADL has been associated with increased cardiovascular mortality in men and women aged 65 years or older.12 Other studies have shown that functional status is an important predictor of hospital outcomes in older patients,13 including mortality in patients with pneumonia and congestive heart failure.14 Outcomes have also been predicted by the Karnofsky score in patients with cancer15 and acute myocardial infarction.16 Patterns of functional decline can also be observed in different types of illness trajectories, such as sudden death, cancer death, death from organ failure, and frailty.17 Assessing ADL function is imperative for advising patients about long-term care needs, assessing the need for home care and other supportive services, or evaluating the needs of a patient's caregiver.18 An awareness of the importance of functional status can help to guide care planning as well as being a potentially important predictor of mortality risk.

An interesting finding in our patient population was the significant contribution of mortality from non-AIDS-defining illnesses, including cancer and end-stage liver disease. These results corroborate previous findings that patients with AIDS are now living long enough to experience growing morbidity and mortality from coexisting conditions that are not intrinsically related to HIV.19-23 As AIDS becomes more of a manageable chronic disease, some of the common comorbidities that occur in patients with AIDS could pose more immediate risk of short-term mortality than HIV infection itself. Unlike our preliminary analysis of this cohort,3 in which the existence of serious non-AIDS-related comorbidities (eg, advanced liver, cardiac, pulmonary, or renal disease) had been marginally significant in predicting short-term mortality on univariate analysis, the current analysis did not find these conditions to be significant. This suggests the need for more stratified analysis of larger samples based on established prognostic and staging criteria for these other illnesses, which should be evaluated prospectively in patients with coexisting HIV disease.

Although the generalizability of these findings may be limited because of the small sample size and unique care setting of the patients enrolled in the study, these data suggest that there is value in recognizing functional status as an important predictor of mortality in patients with extremely advanced disease. Markers of functional status can easily be taken into account in gauging disease prognosis, in coordinating therapeutic and palliative care interventions, and in helping to determine goals of care. Discussions about the use of life-prolonging interventions or the implementation of treatment-limiting advance directives could be informed by data about patients' functional status, the estimated reversibility of functional impairments, and the impact of these impairments on quality of life and goals of care. This study also demonstrates the need for additional prospective cohort studies to investigate predictors of mortality in patients with advanced-stage HIV disease.5

Identifying predictors of death and prognostic variables for patients with advanced disease in the HAART era can help to inform the process of planning and coordinating care in this evolving clinical environment. The challenge is for us to integrate disease-specific and palliative care interventions approaching the end of life for patients and their families effectively as AIDS becomes a less stereotypic and more complex chronic disease.


1. Centers for Disease Control and Prevention. Summary of notifiable disease-United States, 2001. MMWR Morb Mortal Wkly Rep. 2003;50:1-22.
2. Selwyn PA, Forstein M. Overcoming the false dichotomy of curative vs. palliative care for late-stage HIV/AIDS: “Let me live the way I want to live, until I can't.” JAMA. 2003;290:806-814.
3. Selwyn PA, Rivard M, Kapell D, et al. Palliative care for AIDS at a large urban teaching hospital: program description and preliminary outcomes. J Palliat Med. 2003;6:461-474.
4. O'Neill JF, Marconi K, Surapruik A, et al. Improving HIV/AIDS services through palliative care: an HRSA perspective. J Urban Health. 2000;77:244-254.
5. Alexander CS, Back A, eds, for the Workgroup on Palliative and End-of-Life Care in HIV/AIDS. Integrating Palliative Care into the Continuum of HIV Care. Missoula, MT: Robert Wood Johnson Foundation; 2004. Available at:
6. Portenoy RK, Thaler HT, Kornblith AB, et al. The Memorial Symptom Assessment Scale: an instrument for the evaluation of symptom prevalence, characteristics, and distress. Eur J Cancer. 1994;30:1326-1336.
7. Mor V, Laliberte L, Morris JN, et al. The Karnofsky Performance Status Scale. An examination of its reliability and validity in a research setting. Eur J Cancer. 1984;9:2002-2007.
8. Linn MW, Linn BS. The Rapid Disability Rating Scale. J Am Geriatr Soc. 1982;30:378-382.
9. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: John Wiley & Sons; 2000.
10. Coakley E, Samore M, Gillis J, et al. The values of quantitative serum HIV-1 RNA levels and CD4 cell counts for predicting survival time among HIV-positive individuals with CD4 counts of ≤50 × 106 cells/L. AIDS. 2000;14:1147-1153.
11. Justice AC, Aiken LH, Smith HL, et al. The role of functional status in predicting inpatient mortality with AIDS: a comparison with current predictors. J Clin Epidemiol. 1996;49:193-201.
12. Fried LP, Kronmal RA, Newman AB, et al. Risk factors for 5-year mortality in older adults. JAMA. 1998;279:585-592.
13. Covinsky KE, Justice AC, Rosenthal GE, et al. Measuring prognosis and case mix in hospitalized elders. J Gen Intern Med. 1997;12:203-208.
14. Davis RB, Iezzoni LI, Phillips RS, et al. Predicting in-hospital mortality: the importance of functional status information. Med Care. 1995;33:906-921.
15. Hwang SS, Scott CB, Chang VT, et al. Prediction of survival for advanced cancer patients by recursive partitioning analysis: role of Karnofsky performance status, quality of life, and symptom distress. Cancer Invest. 2004;22:278-287.
16. Brezinski D, Stone P, Muller J, et al. Prognostic significance of the Karnofsky Performance Status score in patients with acute myocardial infarction: comparison with the left ventricular ejection fraction and the exercise treadmill test performance. Am Heart J. 1991;121:1374-1381.
17. Lunney JR, Lynn J, Foley DJ, et al. Patterns of functional decline at the end of life. JAMA. 2003;289:2387-2392.
18. Walter LC, Brand RJ, Counsell SR, et al. Development and validation of a prognostic index for 1-year mortality in older adults after hospitalization. JAMA. 2001;285:2987-2994.
19. Sansone RG, Frengley JD. Impact of HAART on causes of death of persons with late-stage AIDS. J Urban Health. 2000;77:165-175.
20. Valdez H, Chowdhry TK, Asaad R, et al. Changing spectrum of mortality due to HIV: analysis of 260 deaths during 1995-1999. Clin Infect Dis. 2001;32:1487-1493.
21. Selwyn PA, Goulet JL, Molde S, et al. HIV as a chronic disease: long-term care for patients with HIV at a dedicated skilled nursing facility. J Urban Health. 2000;77:187-203.
22. Puoti M, Spinetti A, Ghezzi A, et al. Mortality from liver disease in patients with HIV infection: a cohort study. J Acquir Immune Defic Syndr. 2001;24:211-217.
23. Welch K, Morse A. The clinical profile of end-state AIDS in the era of highly active antiretroviral therapy. AIDS Patient Care STDs. 2002;16:75-81.

mortality; prognostic markers; palliative care

© 2005 Lippincott Williams & Wilkins, Inc.