Article Text


Does palliative care improve outcomes for patients with HIV/AIDS? A systematic review of the evidence
  1. R Harding1,
  2. D Karus2,
  3. P Easterbrook3,
  4. V H Raveis2,
  5. I J Higginson1,
  6. K Marconi4
  1. 1Department of Palliative Care and Policy, Guy’s King’s and St Thomas’s School of Medicine, King’s College, London, UK
  2. 2Center for the Psychosocial Study of Health and Illness, Mailman School of Public Health, Columbia University
  3. 3Academic Department of HIV and GU Medicine, Guy’s King’s and St Thomas’s School of Medicine, King’s College, London, UK
  4. 4HIV/AIDS Bureau, Health Resources and Services Administration, US Department of Health and Human Services
  1. Correspondence to:
 Dr Richard Harding


Background: The need for palliative care in HIV management is underlined by the high prevalence of pain and symptoms, the toxicity, side effects, and virological failure associated with antiretroviral therapy, emergence of co-morbidities, continued high incidence of malignancies, late presentation of people with HIV disease, and the comparatively higher death rates among the infected individuals.

Methods: A systematic review was undertaken to appraise the effect of models of palliative care on patient outcomes. A detailed search strategy was devised and biomedical databases searched using specific terms relevant to models of palliative care. Data from papers that met the inclusion criteria were extracted into common tables, and evidence independently graded using well described hierarchy of evidence.

Results: 34 services met the inclusion criteria. Of these, 22 had been evaluated, and the evidence was graded as follows: grade 1 (n = 1); grade 2 (n = 2); grade 3 (n = 7); grade 4 (n = 1); qualitative (n = 6). Services were grouped as: home based care (n = 15); home palliative care/hospice at home (n = 7); hospice inpatient (n = 4); hospital inpatient palliative care (n = 4); specialist AIDS inpatient unit (n = 2); and hospital inpatient and outpatient care (n = 2). The evidence largely demonstrated that home palliative care and inpatient hospice care significantly improved patient outcomes in the domains of pain and symptom control, anxiety, insight, and spiritual wellbeing.

Conclusions: Although the appraisal of evidence found improvements across domains, the current body of evidence suffers from a lack of (quasi) experimental methods and standardised measures. The specialism of palliative care is responding to the clinical evidence that integration into earlier disease stages is necessary. Further studies are needed to both identify feasible methods and evaluate the apparent beneficial effect of palliative care on patient outcomes in the post-HAART era.

  • CCT, community care team
  • HAART, highly active antiretroviral therapy
  • HCA, home care assistance
  • HST, home support team
  • ITDU, infectious and tropical diseases unit
  • PCU, palliative care unit
  • POS, Palliative Outcome Scale
  • QoL, quality of life
  • RCT, randomised controlled trial
  • STAS, Support Team Assessment Schedule
  • HIV
  • palliative care
  • systematic review

Statistics from

The World Health Organization defines palliative care as1 “an approach that improves the quality of life of patients and their families facing the problems associated with life-threatening illness, through the prevention and relief of suffering by means of early identification and impeccable assessment and treatment of pain and other problems, physical, psychosocial and spiritual.”

UNAIDS estimates at least 40 million people living with HIV/AIDS in 2003, and during 2003 there were three million HIV/AIDS related deaths, and five million new infections.2 The high prevalence of HIV related manageable symptoms including fatigue, anorexia, weight loss, depression, agitation and anxiety, nausea and vomiting, diarrhoea, cough, dyspnoea, fever, sweats, and pruritus,3,4 as well as often inadequate pain control5,6 at all stages of HIV disease emphasise the need for palliative therapy.

Although widespread adoption of highly active antiretroviral therapy (HAART) in industrialised countries since 1996 has profoundly impacted on morbidity and mortality, the need for palliative care has continued. Firstly, HIV infected patients continue to die at a higher rate than the uninfected,7 and the virological failure rate with first and second line therapy remains high.8,9 Secondly, antiretroviral drugs are associated with a significant rate of toxicity and side effects10 such as peripheral neuropathy and gastrointestinal problems. Thirdly, as a result of survival prolongation, new co-morbidities have become apparent such as end stage liver disease secondary to hepatitis C co-infection, myocardial infarction,11 cerebrovascular disease,12 and diagnoses associated with significant neurological and cognitive impairment such as progressive multifocal leucoencephalopathy.13,14 Fourthly, some HIV related malignancies such as non-Hodgkin’s lymphoma, cervical carcinoma and colorectal and lung malignancies have not declined in incidence with HAART.15–17 Fifthly, a significant proportion of HIV infected individuals are unaware of their diagnosis, and present only with advanced disease.

The changing epidemiology of HIV disease in the era of HAART has resulted in new and evolving roles for palliative care, with a shift from the more conventional HIV palliative care of the 1980s and early 1990s to a greater focus on symptom control in patients who may continue to live for an extended period, or the need for active treatment for one HIV related condition and palliation for another simultaneously.18 However, in the developing world, where the majority of the HIV cases occur, and where antiretroviral drug availability remains limited, the principles of traditional palliative care still apply.

To date, there has been no comprehensive review of the different models of palliative care and their effectiveness for people affected by HIV/AIDS. Our objective was to systematically review the evidence base for the effectiveness of palliative care in improving patient outcomes in HIV/AIDS.


Search strategy

In December 2003, we searched the following databases: Medline (1980–2003), CINAHL (1982–2003), PsychINFO (1980–2003), Embase (1980–2003), Applied Social Sciences Index and Abstracts (1987–2003), and Social Sciences Citation Index (1981–2003). The search terms were (palliative/hospice*/terminal care/supportive care/terminally ill/palliat*/dying/end of life/home care/advanced/comfort care) and (HIV/AIDS). In order to obtain as comprehensive a search as possible,19 terms were selected to identify the range of models commonly used as proxies for palliative care, or care mixes that include palliation as a component.

The criteria for inclusion of studies were original research paper in English language reporting on human subjects; a minimum of 50% of study population infected with HIV (or HIV infected subsample analysed and reported separately); the provision of a basic service description and/or evaluation of patient outcomes or patient satisfaction with care. Studies that incorporated elements of palliative care (that is, pain and symptom control, psychosocial support, terminal care) were included regardless of whether they explicitly used the term.


Data items were extracted using a standard proforma: year of publication, setting/country, HIV disease stage and referral criteria to service, service description and palliative approach, sample size, study design, outcomes, and findings.

The studies were independently graded by two researchers (RH and DK) using a well described hierarchy of evidence20 (table 1). Studies were further grouped according to intervention model used. A formal meta-analysis with statistical pooling of results across studies was not possible because of the absence of both appropriate study designs and standardised outcome measures.

Table 1

 Grading criteria for studies


Characteristics of eligible studies

The 32 eligible studies originated from the following countries: United Kingdom (n = 10); United States (n = 9); sub-Saharan Africa (n = 5); Canada (n = 2); Australia (n = 1); France (n = 1); Italy (n = 2); Netherlands (n = 1); and Norway (n = 1) (table 2). Of those countries that adopted widespread use of HAART from 1996–7 (that is, excluding the five studies based in sub-Saharan Africa), 18 studies were published pre-HAART and nine post-HAART.

Table 2

 Service descriptions and evaluation evidence

Thirty four different palliative care services were described in 32 different studies, although only about half of these services (18/34) stated explicitly that their patient management was palliative. Of the 34 services, 22 had been evaluated in 17 studies. The different service models were home based care (n = 15); home palliative care/hospice at home (n = 7); hospice inpatient (n = 4); hospital inpatient palliative care (n = 4); specialist AIDS inpatient unit (n = 2); and hospital inpatient and outpatient care (n = 2) (table 2).

In general, these services offered various combinations of usual disciplines found in palliative care, including consultants in palliative medicine, palliative nurse specialists, pain and symptom control, on-call home care, education and advice, counselling, psychosocial care, pastoral/spiritual care, occupational therapy, social work, dietetics, a focus on family and patient support, and bereavement counselling.

Patient eligibility for the services was described for 30/34 services and ranged from any stage of HIV disease (n = 12) to only those with an AIDS diagnosis (n = 11), or advanced/terminal AIDS (n = 7). As would be expected, the services that offered care to all HIV infected patients were less likely to name palliation as part of the care package (2/12) than were services specifically for people with an AIDS diagnosis (7/11); or those for advanced disease services (6/7). The proportion of services that incorporated an evaluation of their service varied with the type of service model: home based care (n = 13/15); home palliative care and hospice at home (n = 5/7); hospital palliative care (n = 2/4); hospital inpatient and outpatient care (n = 1/2); hospice inpatient (n = 1/4); specialist AIDS inpatient unit = (0/2).

The clinical outcomes used in the evaluation of palliative care services varied. Four studies used the Support Team Assessment Schedule (STAS)21 which was designed to assess domains including pain and symptom control, patient and family anxiety, patient and family insight, and communication. Other studies assessed satisfaction with care, medical procedures and investigations, perceived quality of care, patient involvement, support, and quality of life (generally with non-standardised measures).

Of the 22 services that had been evaluated in 17 studies, the strength of the evidence in each study was graded as grade 1 (n = 1); grade 2 (n = 2); grade 3 (n = 7); grade 4 (n = 1). Six studies used only qualitative data.

Findings by evidence grade

Grade 1 (n = 1): strong evidence

In a randomised controlled trial (RCT), integrated multiprofessional case managed home care (n = 29) showed a possible advantage over standard home care (n = 28) in terms of both quality of wellbeing and survival (magnitude unreported, figurative representation of data only). However, this did not reach statistical significance.22

Grade 2 (n = 2): fairly strong evidence

In an observational study 10 patients accessing home palliative care in addition to hospital care were compared to 32 who were attending hospital care only. Small sample size, death and attrition disallowed statistical analysis, and no difference in quality of wellbeing was reported although number of hospital admissions and mean number of days for each admission and overall costs were lower for those receiving home care. In a further observational study, relative to 77 patients receiving standard hospital inpatient care, 10 patients attending a hospital inpatient palliative care unit (PCU) were less likely to have been prescribed antiretrovirals (34% v 10%) and antimicrobial agents (60% v 0%), more likely to have ceased treatment for curable infections (23% v 80%), and more likely to have been prescribed step 3 analgesics (18% v 80%).23 Depression was also far less common in the PCU (55% v 10%), although a similar frequency of pain was assessed (46% v 50%). However, owing to the small sample size and group size imbalance, statistical analysis was not possible.

Grade 3 (n = 7): weaker evidence

The following studies used single group longitudinal methods. In a home AIDS palliative team, care was rated as good/excellent by 100% of respondents, and qualitative comments reported better communication with palliative care staff than with staff outside the team, and identified helpful elements of care as talking and support.24 Among patients receiving home hospice/palliative specialist AIDS care, statistically significant improvements over time were identified in the following domains: symptom control, from 6% to 33% reporting no problem from entry to last week of care, p = 0.00009),25 and from 4% to 16% p = 0.05 for the first 2 weeks and from 4% to 24% p = 0.01 from referral to the point of death26 and significant (magnitude unreported) improvements for inpatients of a generic hospice27; pain control (from 51% to 66% after 2 weeks reporting no problem, p = 0.01, and from 51% to 66% from referral to death p = 0.0001)26; family insight from entry to death from 33% to 67% reporting as no problem p = 0.0006)25; patient anxiety (from 7% to 16% from referral to 2 weeks reporting no problem p = 0.05, and from 7% to 47% from referral to death p = 0.00005)26; patient insight p = 0.04 (magnitude of change not reported) and spiritual wellbeing p = 0.005 (magnitude of change not reported).28 Conversely, increased pain and discomfort were reported during attendance from 2 weeks to the final 48 hours for patients of another home AIDS hospice team (from 54% to 68%).29 Although reductions in pain (from severe pain for 31% at admission to 3% in last days) and significant improvement for symptom control (unquantified) were found among patients of an inpatient generic hospice providing AIDS care, no improvement was found for most severe symptoms reported at admission,27 and for a community palliative care team, symptom control worsened from referral to death, with the proportion of those reporting severe symptoms rising from 22% to 46% (p = 0.04).28 The study reporting significant symptom improvements from referral to end of life also found 33% of patients having severe symptoms in the last week before death.26 In a further study, specialist AIDS palliative pharmacy input was found to improve complex drug management and patient care (although the effect was not quantified).30

Grade 4 (n = 1): weak evidence

Using a Likert rating scale of 1 = excellent to 5 = poor, a cross sectional evaluation of 52 patients accessing HIV home care found high patient satisfaction with care (mean score 1.6) and personal control over care (mean score 2.2), and with both communication (84.6% reporting a professional available to talk) and regularity of contact with staff (80.6% satisfied), though perceived health status remained unchanged.31

Qualitative (n = 6)

Qualitative data from patients receiving home based care reported patients particularly valuing information giving, support, welfare assistance, and specialist clinical intervention and referrals (focus group data, number unspecified)32 as well as enhanced human dignity33 (patient interviews, number unspecified). Patient interviews in other studies reported less disturbance of daily routine34 (nine interviews)35 (five interviews) fewer hospital visits and high satisfaction and quality of care34 (nine interviews)36 (265 interviews). However, quality and satisfaction with care were tempered by the anxiety and fear induced by being cared for at home for a small unspecified number of patients35 (of n = 5 interviewed patients) and poor weekend/out of hours and primary care communication34 (of nine interviews).


In this systematic review of 17 studies, we found that most of the evidence supporting the benefits of palliative care for HIV infected patients is based on seven studies categorised as offering grade 3 evidence. These largely demonstrated that home palliative care and inpatient hospice care significantly improved patient outcomes in the domains of pain and symptom control, anxiety, insight and spiritual wellbeing. However, the need for improvements in pain and symptom management was identified even among specialist services.

Only one study employed the most rigorous evaluation methodology of an RCT (that is, grade 1), and although this study suggested that patient outcomes might be improved in terms of quality of wellbeing and survival, this did not reach statistical significance. The lack of grade 1 evaluations is unsurprising given the extensive literature documenting the challenges and logistical problems with palliative care trials.37,38 The present review identified a recent RCT of AIDS homecare that failed to adequately recruit and retain patients.39 Similar methodological challenges were present in a recent review of palliative and hospice care teams’ effect on cancer patient outcomes which found only studies of mainly grade 2 or 3 quality.40 The accompanying meta-analysis documented benefits in the domains of pain, symptom control, and trends towards satisfaction and therapeutic intervention, with evidence generally strongest for home care. The qualitative studies identified in this current review were of generally poor methodology, with sample sizes, data collection, and analysis methods seldom reported.

There are several important limitations and caveats to the interpretation of the findings from this systematic review. The lack of (quasi) experimental evaluations means that firm conclusions cannot be drawn regarding the comparative benefit of palliative care. A number of evaluations also did not adequately identify services as palliative although they aimed to provide services such as home based, specialist, and/or terminal care. Optimal terminal HIV/AIDS care requires a primary focus on palliation, and palliation requires clear and precise terminology to avoid the inconsistency found in clinical instruction for HIV/AIDS terminal care.41 The heterogeneous clinical studies included in this review reflect the diverse settings, clinical management strategies, and disease stages relevant to palliative care in HIV disease. A lack of detailed service description for many of the studies meant it was not possible to identify the effective elements of the service that might be replicated, although exceptions were found. For example, the study by Selwyn et al clearly defined the range of patient clinical needs at baseline that were related to the goals of palliative care, and subsequently reported on their full or partial resolution.42 A further limitation in evaluating palliative care is the set of methodological challenges. These include small sample sizes and attrition,38 selection bias recruitment and gate-keeping particularly among those with advanced disease,43 aggregation of data from heterogeneous populations, the selection of appropriate outcomes in the measurement of broad psychosocial as well as clinical outcomes,44 and the resulting lack of successful RCTs.45 In the absence of successful RCTs in areas where compassionate but effective care is needed, over-reliance on RCTs may not provide the evidence required.46 Other methods such as quasi-experimental and longitudinal cohort designs may be more appropriate for robust but feasible protocols.

There is doubtless greater activity in HIV/AIDS patient management in the field of specialist palliative care than was identified by this review, and the tendency for publication of studies with positive findings may mean an under-representation of studies reporting null findings. We attempted to minimise the bias resulting from terminology by including a broad range of common terms in the search strategy—for example, palliative care approaches being termed comfort care.41 This review also was guided by the findings from a recent evaluation of systematic reviews of palliative care services.47 Similarly, the grading criteria selected were purposefully broad and inclusive, allowing a full body of evidence rather than being narrow (and RCT focused).

In addition, it may be that current activity in less developed regions is not fully represented, and to address this, end of life HIV care in sub-Saharan Africa has been independently surveyed.48 The identification of evidence is also hampered by studies combining diagnosis groups. For example, one study49 reported palliative care consultations in an acute care hospital where only 8.3% of patients had HIV disease, a second study of the benefits of symptom control by a hospital based palliative care consultant and pharmacist also had only 8% of patients with HIV,50 and a third study of parenteral antibiotics in an inpatient palliative care unit reported 15% of patients having an underlying HIV diagnosis.51

The potential for a selection bias in HIV palliative care evaluation was demonstrated in a study that found patients reported lower levels of pain, better insight, less anxiety, and fewer spiritual problems than in providers’ assessments.24 Patient self reports are also subject to bias because of more unwell patients being unable or unwilling to participate.23 There is currently a modest amount of clinical research into HIV palliative care, and future efforts should include both qualitative and quantitative approaches to inform feasibility, acceptability, processes, and preferences as well as outcomes. In addition, well standardised quantitative palliative specific outcome measures such as the Support Team Assessment Schedule (STAS)21 and the Palliative Outcome Scale (POS)52 should be considered, as well as a range of quality of care53 and quality of life measures54 in HIV management.

Our findings have several implications for clinical practice. A review of HIV care found that service delivery models are changing: home hospital care is provided for proportionately fewer patients, and more frequently for those with less advanced disease.55 Palliative care should be part of this shift towards earlier intervention, particularly as our review found palliation to be provided towards advanced and terminal disease stages. Most of the studies were based on patients with end stage disease, and few studies have explored the benefits of palliative care through the disease trajectory. Such longitudinal studies could usefully evaluate the benefits of integrated palliative care, and should be a priority for future research.

Key messages

  • The clinical evidence demonstrates that patients with HIV infection require palliative care throughout the disease trajectory in order to control pain and symptoms, to promote adherence through reduction of side effects and toxicity associated with antiretroviral therapy, to manage life limiting co-morbidities such as cancers and end stage liver disease, and to provide quality end of life care for those whom antiretroviral therapy fails or who are unable to access it

  • This systematic review of patient outcomes found home palliative care and inpatient hospice care to significantly improve patient outcomes in the domains of pain and symptom control, anxiety, insight, and spiritual wellbeing

  • Quality HIV clinical care should offer palliation as a flexible, integrated approach when needed from diagnosis to the end of life across the range of institutional and home care settings alongside new therapeutics

Service objectives and the evaluation evidence would benefit from an explicit inclusion of palliative care in service aims and multiprofessional mix.56 The historical distinction between early/late stage disease that placed palliative care as solely terminal is inappropriate in the light of the clinical evidence that patients need pain and symptom management (and psychosocial care) throughout the disease trajectory.5,6

Nursing homes,57,58 brain impairment units,59 hospices,60 hospital palliative care teams,61,62 prison hospice,63 and hospital paediatric wards41 face challenges of providing both curative and palliative care for people with HIV/AIDS. Therefore, clinical need throughout the disease trajectory and the range of settings in which patient care is delivered speaks of the need for specialist education and the inclusion of palliative care expertise into the multiprofessional care team. HIV/AIDS palliative care may also be seen as a specialism within palliative care; provider data from a generic hospice highlighted staff concern that the low number of HIV/AIDS referrals prevented them from maintaining skills, fearing poor care as a likely outcome.64

Further challenges are presented by resource poor countries and some poorer communities in wealthier nations where the epidemic is most pressing. Psychosocial/“total” care is likely to include additional issues such as orphan care and food security, and care programmes need to consider coverage and not simply strive to provide high quality service to a few patients.65 Palliative care remains a novel concept in less developed regions. It is largely absent from the clinical training syllabus for nurses and doctors, and restricted availability of opioids necessitate strategic approaches to ensure the expansion of palliative care.66

Patients with HIV/AIDS report needing emotional support, communication, and pain control from their physicians,67 and those patients with advanced AIDS describe a good death as having pain/symptom control, optimum quality of life, choosing where to die, having control of their treatment, and having their spiritual needs addressed.68 Palliative care remains an essential component of clinical care, embracing pain and symptom management, advance care planning, and psychosocial support. The changing epidemiology of HIV, and the widely divergent management of patients in the developed world compared to those in developing countries, present clear and necessary roles for the integration of palliative approaches to patient management throughout the disease trajectory. The integration of palliative care throughout all stages of patient care, incorporating flexibility and specialist advice needed, shows the curative versus palliative approach to be false.69,70 The need for palliative care continues alongside new therapeutics and should not be associated solely with terminal care.


This study was conceived by IJH, VR, and KM; the study protocol was developed by RH and reviewed and agreed by all authors; RH searched the databases, all authors reviewed the search outputs, and data graded by RH and DK; the manuscript was drafted by RH and PE, and reviewed and commented on by all authors.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles