Vol. 44 No. 1 July 2012 Journal of Pain and Symptom Management 1

Original Article

Prevalence, Burden, and Correlatesof Physical and Psychological SymptomsAmong HIV Palliative Care Patientsin Sub-Saharan Africa: An InternationalMulticenter StudyRichard Harding, BSc, MSc, DipSW, PhD,Lucy Selman, BA, MPhil, PG Cert Pall Care, Godfrey Agupio, RN,Natalya Dinat, MD, FCOG, MPhil Pall Med, Julia Downing, RN, PhD,Liz Gwyther, MB, ChB, FCFP, Dip Pall Med, MSc, Thandi Mashao, RN,Keletso Mmoledi, CPN, MPH, BTech, Tony Moll, MB, ChB, BSc,Lydia Mpanga Sebuyira, BM, BCh, BA (Hons), MA, FRCP, Barbara Ikin, RN, BA,and Irene J. Higginson, BMedSci, BM BS, FFPHM, FRCP, PhDDepartment of Palliative Care, Policy, and Rehabilitation and The Cicely Saunders Institute of

Palliative Care (R.H., L.S., I.J.H.), King’s College London, London, United Kingdom; Hospice Africa

Uganda (G.A.), African Palliative Care Association (J.D.), and Infectious Diseases Institute (L.M.S.),

College of Health Sciences, Makerere University, Kampala, Uganda; Division of Palliative Care,

Department of Internal Medicine (N.D., K.M.), University of the Witwatersrand, Johannesburg, South

Africa; Hospice Palliative Care Association of South Africa (L.G., B.I.) and Palliative Medicine Unit

(L.G., T.M.), University of Cape Town, Cape Town, South Africa; and Philanjalo Hospice (T.M.),

KwaZulu Natal, South Africa

Abstract

Context. Despite HIV remaining life limiting and incurable, very little clinical

research focus has been given to the prevalence and related burden of physicaland psychological symptoms for those accessing palliative care. Despite evidenceof problems persisting throughout the trajectory and alongside treatment, scantattention has been paid to these manageable problems.

Objectives. This study aimed to measure the seven-day period prevalence andcorrelates of physical and psychological symptoms, and their associated burden, inHIV-infected individuals attending palliative care centers in sub-Saharan Africa.

Methods. Consecutive patients in five care centers across two countriescompleted the Memorial Symptom Assessment Scale-Short Form, with additionaldemographic and disease-oriented variables.

Results. Two hundred twenty-four patients participated. The most commonsymptoms were pain in the physical dimension (82.6%) and worry in the

Address correspondence to: Richard Harding, BSc, MSc,DipSW, PhD, Department of Palliative Care, Policy,and Rehabilitation, King’s College London, CicelySaunders Institute, Bessemer Road, Denmark Hill,

London SE5 9PJ, United Kingdom. E-mail:richard.harding@kcl.ac.uk

Accepted for publication: September 1, 2011.

� 2012 U.S. Cancer Pain Relief CommitteePublished by Elsevier Inc. All rights reserved.

0885-3924/$ - see front matterdoi:10.1016/j.jpainsymman.2011.08.008

2 Vol. 44 No. 1 July 2012Harding et al.

psychological dimension (75.4%). Interestingly, 71.4% reported hunger. Women,and those with worse physical function, were more likely to experience burden.However, being on antiretroviral therapy (ART) was not associated with global,physical, or psychological symptom burden.

Conclusion. This study is the first to report physical and psychological symptomburden in HIV-infected populations receiving palliative care in sub-SaharanAfrica. Despite increasing access to ART, these burdensome and manageableproblems persist. The assessment of these problems is essential alongsideassessment of ART virological outcomes. J Pain Symptom Manage 2012;44:1e9.� 2012 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved.

Key Words

HIV, palliative care, symptoms, Africa

IntroductionDuring 2008, there were an estimated 22.4

million people living with HIV infection insub-SaharanAfrica, and 1.4million deaths.1 Pal-liative care has been identified as an essentialcomponent of care services for people withHIV disease because of the variety of problemsthat may be experienced.2 HIV palliative carehas been shown to be effective in the manage-mentofpain, symptoms, andanxiety in a system-atic review of evidence,3 although the identifieddata were almost exclusively generated in high-income settings and prior to the advent of anti-retroviral therapy (ART). A dearth of evidenceof palliative care needs and outcomes hasbeen identified for sub-Saharan Africa4 andthe development of appropriate interventionsis hampered by a lack of data.

Although some data have been generated ontheneeds and symptoms of AfricanHIVpopula-tions5,6 and evidence suggests a high burden ofproblems in the palliative phase,7 populationswith advanced disease have been rarely investi-gated using well-validated outcome measures.Further, as a result of the challenges of opioidavailability and prescribing in Africa, the pallia-tive care research agenda has been dominatedby the study of pain and analgesia.8e12 This risksthe reduction of palliative care to pain relief, tothe detriment of a broader understanding ofthe pain, other symptoms, and psychological di-mensions that constitute HIV symptomatology.

This study aimed to determine the seven-dayprevalence and associated burden of physicaland psychological symptoms among HIV pa-tients receiving palliative care in sub-Saharan

Africa, and to identify correlates of symptomburden.

MethodsThis study used an international, multicen-

ter, cross-sectional design, and used a well-validated symptom scale.

Participating SitesThe study was undertaken in five palliative

care facilities, four in South Africa, and onein Uganda. The facilities serve periurban andurban areas with a range of home care, daycare, hospital consulting, and palliative inpa-tient services.

RecruitmentInclusion criteria were adult patients (at least

18 years old) with a confirmed HIV diagnosisknown to the patient, receiving palliative care,with sufficient physical and cognitive ability toparticipate in interviews. Patients with a knownHIV diagnosis were recruited consecutively. Allinformation and consent forms and tools weretranslated from English (forward and back)into the principle languages of Luganda,Runyankole, SeSotho, Runyoro, SeTswana,isiXhosa, and two isiZulu dialects (Gautengand KwaZulu Natal). Informed consent wasobtained from all the participants. The studywas reviewed and approved by the EthicalReview Boards of the Universities of CapeTown, KwaZulu Natal, and Witwatersrand; theUgandan National Council for Science andTechnology; Hospice Africa Uganda; and the

Vol. 44 No. 1 July 2012 3HIV Symptoms in Sub-Saharan African Palliative Care

Hospice Palliative Care Association of SouthAfrica.

Translation and Data CollectionThe following patient demographic and clin-

ical data were collected: age, gender, householdsize, number of children responsible for, loca-tion of home (urban, periurban, rural), primaryplace of palliative care (home, inpatient/outpa-tient unit, day care facility), physical functionalstatus (using the Eastern CooperativeOncologyGroup [ECOG] functional status score13), timeunder care in weeks, current ARTuse, whetherthe patient had received a prior AIDS diagnosis,and most recent CD4 count. We elected to col-lect data on thenumber of children that respon-dents were responsible for, rather than numberof biological children. This was because adultsmay often care for children other than theirown, for example, grandchildren, nephews,and nieces, a situation that has been exacer-bated by AIDS-related deaths. The MemorialSymptom Assessment Scale-Short Form(MSAS-SF) was used to measure the seven-dayperiod prevalence and associated burden ofmultidimensional symptoms. The MSAS-SF of-fers three subscale indices: Physical SymptomDistress Index (MSAS-Phys), PsychologicalSymptom Distress Index (MSAS-Psych), andGlobal Distress Index (MSAS-GDI).14 Each ofthese three subscales has a possible score rangeof 0e4. This well-validated, multidimensionalinstrument captures the presence and distressof 26 physical and four psychological symptoms.It has good psychometric properties, with sub-scale Cronbach’s alpha coefficients of0.76e0.87, andone-day test-retest reliability cor-relation coefficients of 0.86e0.94.14 Its use hasbeen well reported among HIV and Africanpopulations.5,15e19 Time to complete theMSAS was recorded.

Following a study in Uganda to determinewhether additional items were necessary formeasurement among cancer and HIV palliativecare populations (involving patient and staff fo-cus group studies and item testing, data submit-ted), the following items were included in thepool of physical symptom items: bad smell/odor; sores/lumps on genitals; discharge fromgenitals; difficulty moving; difficulty walking;poor vision and poor hearing; hunger (KarenFrame, MSc dissertation, unpublished).

The MSAS-SF, demographic record, and in-formation and consent sheets were translatedfrom English into the main local languages re-ported above. Translation was carried out atthe participating sites and cross-checked by stafffluent in bothEnglish and the relevant local lan-guage. The University of KwaZulu Natal carriedout the Natal Zulu translation and the Univer-sity of Cape Town, the isiXhosa translation.

Research nurses read aloud the questionnaireitems and entered the patient’s self-report re-sponse on the patient’s behalf. Self-completewas not used because of limited literacy, and allquestionnaires were completed using researchnurses to enter responses to reduceanypotentialbias through using a mixture of self-completeand researcher completion. Research nursesthen entered the data into purpose-designedExcel spreadsheets, subsequently imported intoSPSS (SPSS Inc., Chicago, IL) for analysis.

AnalysisDescriptive analysis was undertaken for pa-

tient characteristics and MSAS-SF scores. Foreach item within MSAS-SF, the prevalenceand associated burden were calculated. Sub-scale scores of global, physical, and psycholog-ical distress were calculated using the originalsubscales and calculation methods of theMSAS-SF (i.e., not including the additional Af-rican items). The total number of seven-dayperiod prevalent symptoms also was calculatedfor each respondent, and the mean and stan-dard deviation (SD) for the sample. Physicaland psychological symptoms are reported sep-arately and in descending order of prevalence.

Correlates of symptom burden were identi-fied using regression analyses. Four modelswere constructed, each model with a depen-dent variable as follows: global distress (model1), physical distress (model 2), psychologicaldistress (model 3), and total number of symp-toms (model 4). Univariate linear regressionanalyses were conducted to test the associationof various factors with the dependent variable.The independent variables entered stepwisewere age (continuous), gender (two levels ofmale/female), prior AIDS diagnosis (two levelsof yes/no), current ARTuse (two levels of yes/no), functional status (five levels of ECOG sta-tus), household size (continuous), andwhether they were responsible for children(two levels of yes/no). Following each

4 Vol. 44 No. 1 July 2012Harding et al.

univariate regression, multivariate regressionmodels were constructed. Independent vari-ables from the univariate analyses above wereentered stepwise into the multivariate modelif significant at the conservative 25% level.20

Cases with missing data were excluded fromthe multivariate models. For each model, the95% confidence interval (95% CI) was calcu-lated for the unstandardized b coefficient,and r2 presented to determine the amount ofvariance explained by the multivariate model.

ResultsSample Characteristics

Two hundred twenty-four patients were re-cruited in 2009, n¼ 192 in South Africa andn¼ 32 in Uganda. The mean age was 36.5 years(median¼ 35, SD¼ 9, min¼ 20, max¼ 70),and 164 (73.2%) were female. The mainplace of palliative care and recruitment washome care n¼ 149 (66.5%), inpatient n¼ 44(19.6%), outpatient n ¼ 18 (8.0%), and daycare n¼ 13 (5.8%). Patients had been receivingpalliative care for a median of 12 weeks. Respon-dents’ home locationwas: urbann¼ 50 (23.3%);periurban n¼ 35 (15.6%); and rural n¼ 139(62.1%). Of the 224 patients, 180 (80.4%) wereresponsible for children (mean number of chil-dren for which responsible was 2.7, range 1e11).

Of the 224 recruited HIV patients, 198(88.5%) had a prior AIDS diagnosis, and 110(49.1%) were currently receiving ART. A CD4count was available for only 122 patients(54.5%). Their ECOG functional status scoreswere fully active n¼ 49 (21.9%); restrictedn¼ 59 (26.3%); ambulatory n¼ 49 (21.3%);limited self-care n¼ 61 (27.2%); and com-pletely disabled n¼ 6 (2.7%).

Symptom Prevalence and BurdenThe seven-day period symptom prevalence

and associated burden are reported in Table 1.The mean number of symptoms was 18.1(SD¼ 6.9, median¼ 19). The five most preva-lent symptoms were pain (82.6%), feeling sad(75.4%), feeling drowsy (74.1%), worrying(73.2%), and lack of energy (71.9%). The symp-toms reported as having themost severe burden(i.e., scored ‘‘very much’’) are particularly im-portant clinically. Thefivemost prevalent severesymptoms were hunger n¼ 81 (36.2%), pain

n¼ 79 (35.3%), weight loss n¼ 62 (27.7%),numbness n¼ 59 (26.3%), and lack of energyn¼ 56 (25.0%).The mean Global Distress Index was 1.74

(SD¼ 0.81), the Physical Distress Index was1.48 (SD¼ 0.82), and the Psychological DistressIndex was 1.56 (SD¼ 0.88). The MSAS-SF tookamean of 25.0 minutes to complete (SD¼ 7.9).

Correlates of Symptom BurdenThe univariate and multivariate models to de-

termine associations with symptom burden arepresented in Table 2. It is notable that genderand functional status were the sole and consis-tent correlates for the multivariate models. Inthe multivariable analyses, gender and func-tional status were correlated to global distress(b¼ 0.315, P¼ 0.006 and b¼ 0.280, P¼ 0.001,respectively), physical burden (b¼ 0.331,P¼ 0.004, respectively), number of symptoms(b¼ 2.793, P¼ 0.004 and b¼ 2.334, P¼ 0.001,respectively) and psychological burden (ECOGonly, b¼ 0.157,P¼ 0.001). In eachcase, being fe-male andhaving worse physical functionwere as-sociated with higher burden. It is notable thatART use was not associated with burden, andneither was a prior AIDS diagnosis. Further, theresponsibility of children did not affect psycho-logical burden, and family household size wasnot associated with any index of burden, that is,having a family at home did not significantlyaffect symptom burden.

DiscussionThis study is the first to identify the preva-

lence, burden, and correlates of pain and otherphysical and psychological symptoms in an HIVpalliative care population in sub-Saharan Africa.Our sample reflects the demographics ofHIV-infected persons accessing care, that is,they are relatively young (mean age 36.5 years),and largely female (73.2%),21 which is notablydifferent from palliative care populations inhigh-income countries, where HIV is lessprevalent.The data were sampled from sites where

palliative care is being delivered alongsideART, that is, in line with current guidancefrom the World Health Organization (WHO).Although palliative care is advocated alongsidetreatment options, it is arguably rarely achieved

Table 1Seven-Day Period Symptom Prevalence (n¼ 224) in Descending Order of Prevalence

Symptom Prevalence Missing

Burden (Total¼ 100%)

Not Present Not at All A Little Bit Somewhat Quite a Bit Very Much Missing

Physical problemsPain 82.6% (n¼ 185) 0 17.4% (n¼ 39) 2.2% (n¼ 5) 15.2% (n¼ 34) 9.8% (n¼ 22) 19.6% (n¼ 44) 35.3% (n¼ 79) 1 (0.4%)Feeling drowsy/tired 74.1% (n¼ 166) 0 25.9% (n¼ 58) 1.3% (n¼ 3) 16.5% (n¼ 37) 13.8% (n¼ 31) 18.8% (n¼ 42) 23.7% (n¼ 53) 0Lack of energy 71.9% (n¼ 161) 0 28.1% (n¼ 63) 0 14.3% (n¼ 32) 9.4% (n¼ 21) 23.2% (n¼ 52) 25.0% (n¼ 56) 0Hungera 71.4% (n¼ 160) 0 28.6% (n¼ 64) 2.2% (n¼ 5) 10.7% (n¼ 24) 10.7% (n¼ 24) 11.6% (n¼ 26) 36.2% (n¼ 81) 0Numbness/tingling

hands or feet66.5% (n¼ 149) 0 33.5% (n¼ 75) 0.9% (n¼ 2) 10.7% (n¼ 24) 11.6% (n¼ 26) 17.0% (n¼ 38) 26.3% (n¼ 59) 0

Dry mouth 61.6% (n¼ 138) 0 38.4% (n¼ 86) 3.6% (n¼ 8) 13.4% (n¼ 30) 11.6% (n¼ 26) 15.6% (n¼ 35) 16.5% (n¼ 37) 2 (0.9%)Weight loss 60.3% (n¼ 135) 0 39.7% (n¼ 89) 0.9% (n¼ 2) 10.3% (n¼ 23) 8.9% (n¼ 20) 12.5% (n¼ 28) 27.7% (n¼ 62) 0Itching 58.9% (n¼ 132) 0 41.1% (n¼ 92) 0.4% (n¼ 1) 13.4% (n¼ 30) 10.3% (n¼ 23) 10.3% (n¼ 23) 24.6% (n¼ 55) 0‘‘I do not look like

myself’’58.0% (n¼ 130) 0 42.0% (n¼ 94) 0.9% (n¼ 2) 10.7% (n¼ 24) 9.4% (n¼ 21) 14.7% (n¼ 33) 22.3% (n¼ 50) 0

Cough 57.1% (n¼ 128) 0 42.9% (n¼ 96) 3.6% (n¼ 8) 10.7% (n¼ 24) 12.5% (n¼ 28) 12.5% (n¼ 28) 17.9% (n¼ 40) 0Sweats 56.3% (n¼ 126) 0 43.8% (n¼ 98) 3.1% (n¼ 7) 8.9% (n¼ 20) 10.3% (n¼ 23) 16.1% (n¼ 36) 17.9% (n¼ 40) 0Difficulty walkinga 55.4% (n¼ 124) 0 44.6% (n¼ 100) 0.4% (n¼ 1) 12.1% (n¼ 27) 9.8% (n¼ 22) 11.2% (n¼ 25) 21.9% (n¼ 49) 0Changes in skin 53.6% (n¼ 120) 0 46.4% (n¼ 104) 0.9% (n¼ 2) 11.6% (n¼ 26) 7.6% (n¼ 17) 13.4% (n¼ 30) 19.6% (n¼ 44) 0Dizziness 50.5% (n¼ 113) 0 49.6% (n¼ 111) 0 12.5% (n¼ 28) 11.6% (n¼ 26) 11.6% (n¼ 26) 14.7% (n¼ 33) 0Difficulty sleeping 49.1% (n¼ 110) 0 50.9% (n¼ 114) 0 6.3% (n¼ 14) 10.7% (n¼ 24) 10.3% (n¼ 23) 21.9% (n¼ 49) 0Difficulty seeinga 44.6% (n¼ 100) 0 55.4% (n¼ 124) 0 12.1% (n¼ 27) 6.3% (n¼ 14) 10.3% (n¼ 23) 16.1% (n¼ 36) 0Difficulty movinga 44.2% (n¼ 99) 0 55.8% (n¼ 125) 0.4% (n¼ 1) 9.8% (n¼ 22) 8.9% (n¼ 20) 9.8% (n¼ 22) 14.7%(n¼ 33) 1 (0.4%)Lack of appetite 41.5% (n¼ 93) 0 58.5% (n¼ 131) 0.4% (n¼ 1) 9.8% (n¼ 22) 4.9% (n¼ 11) 11.6% (n¼ 26) 14.3% (n¼ 32) 1 (0.4%)Muscle achesa 40.2% (n¼ 90) 0 59.8% (n¼ 134) 0 9.4% (n¼ 21) 11.2% (n¼ 25) 6.7% (n¼ 15) 12.9% (n¼ 29) 0Difficulty concentrating 39.7% (n¼ 89) 0 60.3% (n¼ 135) 0.9% (n¼ 2) 12.9% (n¼ 29) 7.1% (n¼ 16) 8.9% (n¼ 20) 9.8% (n¼ 22) 0Nausea 38.8% (n¼ 87) 0 61.2% (n¼ 137) 1.3% (n¼ 3) 7.1% (n¼ 16) 7.6% (n¼ 17) 12.9% (n¼ 29) 9.8% (n¼ 22) 0Shortness of breath 37.1% (n¼ 83) 0 62.9% (n¼ 141) 0.4% (n¼ 1) 8.5% (n¼ 19) 5.4% (n¼ 12) 10.7% (n¼ 24) 12.1% (n¼ 27) 0Feeling bloated 35.7% (n¼ 80) 0 64.3% (n¼ 144) 0 8.0% (n¼ 18) 8.0% (n¼ 18) 9.8% (n¼ 22) 9.8% (n¼ 22) 0Problems urinating 33.0% (n¼ 74) 0 67.0% (n¼ 150) 0.4% (n¼ 1) 6.3% (n¼ 14) 4.5% (n¼ 10) 8.0% (n¼ 18) 13.8% (n¼ 31) 0Constipation 32.1% (n¼ 72) 0 67.9% (n¼ 152) 1.3% (n¼ 3) 10.3% (n¼ 23) 4.0% (n¼ 9) 9.4% (n¼ 21) 7.1% (n¼ 16) 0Swelling arms/legs 29.9% (n¼ 67) 0 70.1% (n¼ 157) 0.4% (n¼ 1) 4.5% (n¼ 10) 5.4% (n¼ 12) 8.5% (n¼ 19) 10.7% (n¼ 24) 0Difficulty hearinga 30.8% (n¼ 69) 0 69.2% (n¼ 155) 0.4% (n¼ 1) 9.4% (n¼ 21) 7.1% (n¼ 16) 5.8% (n¼ 13) 8.0% (n¼ 18) 0Changes in food taste 29.9% (n¼ 67) 0 70.1% (n¼ 157) 0.4% (n¼ 1) 11.6% (n¼ 26) 8.0% (n¼ 18) 3.6% (n¼ 8) 5.4% (n¼ 12) 0Sexual problems 26.8% (n¼ 60) 0 73.2% (n¼ 164) 0.4% (n¼ 1) 4.9% (n¼ 11) 2.7% (n¼ 6) 8.0% (n¼ 18) 10.7% (n¼ 24) 0Discharge from genitalsa 26.8% (n¼ 60) 0 73.2% (n¼ 164) 0 5.8% (n¼ 13) 5.8% (n¼ 13) 4.0% (n¼ 9) 10.7% (n¼ 24) 0Sores/lumps on genitalsa 26.8% (n¼ 60) 0 73.2% (n¼ 164) 0.4% (n¼ 1) 2.2% (n¼ 5) 5.4% (n¼ 12) 8.5% (n¼ 19) 9.8% (n¼ 22) 1 (0.4%)Hair loss 25.4% (n¼ 57) 0 74.6% (n¼ 167) 0.9% (n¼ 2) 8.0% (n¼ 18) 4.5% (n¼ 10) 4.5% (n¼ 10) 7.1% (n¼ 16) 1 (0.4%)Diarrhea 24.6% (n¼ 55) 0 75.4% (n¼ 169) 0 4.5% (n¼ 10) 6.3% (n¼ 14) 4.5% (n¼ 10) 9.4% (n¼ 21) 0Vomiting 21.9% (n¼ 49) 0 78.1% (n¼ 175) 0 6.3% (n¼ 14) 5.4% (n¼ 12) 4.5% (n¼ 10) 5.8% (n¼ 13) 0Bad smell/odora 20.5% (n¼ 46) 0 79.5% (n¼ 178) 0.4% (n¼ 1) 5.8% (n¼ 13) 2.7% (n¼ 6) 3.6% (n¼ 8) 8.0% (n¼ 18) 0

(Continued)

Vol.

44No.

1July2012

5HIV

Symptom

sin

Sub-Saharan

African

Palliative

Care

Table1

Continued

Symptom

Prevalence

Missing

Burden

(Total¼

100%

)

NotPresent

Notat

All

ALittleBit

Somew

hat

QuiteaBit

VeryMuch

Missing

Difficu

ltyswallowing

19.2%

(n¼43

)0

80.8%

(n¼18

1)0

5.4%

(n¼12

)4.0%

(n¼9)

5.4%

(n¼12

)4.5%

(n¼10

)0

Mouth

sores

18.3%

(n¼41

)0

81.7%

(n¼18

3)0

4.5%

(n¼10

)2.2%

(n¼5)

3.1%

(n¼7)

8.5%

(n¼19

)0

Symptom

Prevalence

Missing

Burden

(Total¼

100%

)

NotPresent

Rarely

Occasionally

Frequen

tly

Alm

ost

Constan

tly

Missing

Psych

ologicalproblems

Fee

lingsad

75.4%

(n¼16

9)0

24.6%

(n¼55

)11

.2%

(n¼25

)28

.6%

(n¼64

)19

.6%

(n¼44

)16

.1%

(n¼36

)0

Worrying

73.2%

(n¼16

4)0

26.8%

(n¼60

)10

.7%

(n¼24

)24

.1%

(n¼54

)19

.2%

(n¼43

)18

.8(n

¼42

)1(0.4%)

Feelingirritable

70.1%

(n¼15

7)0

29.9%

(n¼67

)16

.5%

(n¼37

)17

.9%

(n¼40

)17

.9%

(n¼40

)17

.4%

(n¼39

)1(0.4%)

Fee

lingnervous

48.2%

(n¼10

8)0

51.8%

(n¼11

6)12

.1%

(n¼27

)18

.3%

(n¼41

)9.4%

(n¼21

)8.5%

(n¼19

)0

MSA

S-SF

¼Mem

orial

Symptom

Assessm

entScale-Sh

ort

Form

.aItem

notin

original

MSA

S-SF.

6 Vol. 44 No. 1 July 2012Harding et al.

in high-income settings.22 Our data representAfrican successes in delivering such a model ofintegrated care, as half (49.1%) were takingART while receiving palliative care. However,it is also of concern that CD4 counts werenot available in almost half of participants(45.5%), because an indicator of immune func-tion (and disease progression) is importantfor palliative care clinicians to be able to provideappropriate care. The absence of CD4counts disallowed us from including this as anindependent variable in the multivariateanalyses.With respect to symptom prevalence, it is

interesting to note that both physical andpsychological problems were among the fivemost prevalent symptoms. As the samplewas receiving palliative care, and coverageand access to palliative care services are verylimited in sub-Saharan Africa, we may hy-pothesize that prevalence would be muchhigher in the general HIV-infected popula-tion. Further, the high prevalence of thesesymptoms (between 71.9% and 82.6%) sug-gests that patients who are able to receivepalliative care require more effective symp-tom control. The inclusion of the additionalitems reveals that one of the most severeproblems is that of hunger (36.2% reportingthis as burdening them ‘‘very much’’), whichposes a significant challenge to health careservices in the context of low- and middle-income countries.The associations with burden reveal that

those on ART do not have significantly differ-ent symptom burden, and this is an importantclinical message: assessment and palliative careare equally important when ART is present.This is in line with WHO policy that, for pa-tients with HIV disease, palliative care is indi-cated alongside treatment. Previous studieshave determined a similar finding for outpa-tients, that is, that those on treatment do nothave a lower symptom burden.16,19 Further re-search is needed to determine whether thoseon ART experience different symptom clustersthat constitute their burden compared withthose not on ART. Although it is less surprisingthat physical function is associated with bur-den in the analytic models, it is of concernthat women experience higher global andphysical burden, and a greater number ofsymptoms.

Table 2Associations With Symptom Burden

Independent Variables

Univariate Analysis Multivariate Analysis

b P 95% CI for b b P 95% CI for b

Model 1: Global Distress Subscale, r2¼ 19.4%Age 0.005 0.425 �0.007, 0.017 d d dGender 0.279 0.024 0.037, 0.521 0.315 0.006 0.094, 0.537AIDS 0.059 0.728 �0.277, 0.395 d d dART �0.010 0.930 �0.227, 0.208 d d dECOG 0.278 0.001 0.194, 0.363 0.280 0.001 0.196, 0.364Household size 0.020 0.258 �0.015, 0.055 0.013 0.405 �0.018, 0.045Children 0.096 0.492 �0.372, 0.179 d d d

Model 2: Physical Distress Subscale, r2¼ 16.9%Age 0.006 0.346 �0.006, 0.018 d d dGender 0.311 0.012 0.068, 0.554 0.331 0.004 0.106, 0.556AIDS 0.176 0.308 �0.163, 0.515 d d dART �0.020 0.859 �0.239, 0.200 d d dECOG 0.258 0.001 0.172, 0.345 0.262 0.001 0.177, 0.347Household size 0.006 0.732 �0.029, 0.041 d d dChildren �0.047 0.737 �0.324, 0.229 d d d

Model 3: Psychological Distress Subscale, r2¼ 13.7%Age 0.004 0.533 �0.009, 0.017 d d dGender 0.114 0.386 �0.145, 0.374 d d dAIDS 0.051 0.777 �0.306, 0.409 d d dART �0.034 0.769 �0.264, 0.196 d d dECOG 0.157 0.001 0.061, 0.253 0.157 0.001 0.061, 0.253Household size 0.007 0.700 �0.030, 0.044 d d dChildren �0.108 0.464 �0.399, 0.182 d d d

Number of Symptoms, r2¼ 19.1%Age 0.059 0.259 �0.044, 0.161 0.075 0.118 �0.019, 0.169Gender 2.339 0.027 0.273, 4.405 2.793 0.004 0.881, 4.705AIDS 0.816 0.581 �2.091, 3.723 d d dART �0.650 0.492 �2.513, 1.213 d d dECOG 2.304 0.001 1.575, 3.033 2.334 0.001 1.617, 3.051Household size 0.044 0.770 �0.252, 0.340 d d dChildren 0.505 0.672 �1.846, 2.857 d d d

CI¼ confidence interval; ART¼antiretroviral therapy; ECOG¼ Eastern Cooperative Oncology Group.

Vol. 44 No. 1 July 2012 7HIV Symptoms in Sub-Saharan African Palliative Care

When compared with previous data fromWakeham et al.,5 using the same tool inHIV outpatients in Uganda, our sample hadhigher global distress (1.74 vs. 1.28), physicaldistress (1.48 vs. 1.1), and psychological dis-tress (1.56 vs. 0.91). Our sample reportedthe same three of the five most prevalentsymptoms as the Wakeham et al. study, al-though prevalence was higher for each symp-tom in our sample: pain (82.6% vs. 76.0%),feeling drowsy (74.1% vs. 61.0%), and lackof energy (71.9% vs. 61.0%). Interestingly,two psychological problems (sadness andworry) were among the five most prevalentsymptoms in our sample but not in the Wake-ham et al. study. This may reflect the poorerpsychological morbidity of our sample, whichwas receiving palliative care and, therefore,may have had greater disease progression,

complex pain and other symptoms, or ARTside effects. Comparing our sample with can-cer patients recruited during the same studypresented here,23 the same five most preva-lent symptoms with very similar prevalenceand the same mean number of symptomswere reported. However, the HIV patients inthe present analysis reported higher burdencompared with the cancer patients (GlobalDistress Index 1.74 vs. 1.61, Physical DistressIndex 1.56 vs. 1.41, and Psychological DistressIndex 1.48 vs. 1.33). Therefore, the preva-lence of the most common symptoms in palli-ative care is the same across cancer and HIV,but the associated burden is higher in HIVpatients. This may reflect the complex trajec-tory of HIV disease, its treatment, and thepsychological impact of living with an HIVdiagnosis.

8 Vol. 44 No. 1 July 2012Harding et al.

There are a number of limitations to ourstudy and data. First, the cross-sectional designcannot establish causality. Second, the abilityto participate in data collection may excludethose with poorest function; therefore, the es-timates of prevalence may be lower for oursample. Third, the absence of biologicalmarkers potentially weakened the power ofthe model, although around one-fifth of thevariance was explained. Fourth, although theMSAS is a tool with proven psychometric prop-erties, it has not undergone full validation inAfrica. However, previous published data inHIV and African populations have demon-strated its utility and lack of floor/ceiling ef-fects, and the addition of items generatedfrom African palliative care settings enhancedits ability to measure problems of concern toour population. Last, although the study toolswere translated by our university partners andcross-checked by clinical researchers fluent inboth languages, we did not use a forward-backward formal process.

Our study further develops the previous lit-erature24 by increasing the sample size andnumber of centers and identifying associationswith burden, and confirms previous studiesthat show that HIV outpatients on ART inhigh-income settings do not have lower symp-tom burden3,16 compared with those not cur-rently on treatment.

We recommend that HIV patients under pal-liative care receive multidimensional care thatreflects the nature of their problems, that is,the social (e.g., hunger), the psychological(e.g., worry and sadness), and the physical(e.g., pain). Patients also may experience spir-itual distress and have associated needs, andfurther research in this area is needed. The ef-fective provision of multidimensional care onlycan be achieved by taking into account thecommunication and information needs of pa-tients and families,25 and family-based care isessential to reflect the family-wide potentialimpact of advanced disease (as demonstratedby our data on household size). Further clini-cal research studies are required to determinethe symptom burden in nonpalliative carepopulations in Africa to ensure that the highprevalence and burden of symptoms are man-aged in all settings where HIV-infected personspresent for care, and that this is deliveredalongside ART.

Disclosures and AcknowledgmentsThis study was supported by the BIG Lottery

Fund UK (grant number IG/1/010141040).The authors have no conflicts of interest todeclare.The authors would like to acknowledge the

support of the BIG Lottery Fund, and thefive clinical research centers. They are alsograteful to the patients and families who par-ticipated and to Lucy Bradley for articlemanagement.

References1. UNAIDS. AIDS epidemic update 2009. Availablefrom http://data.unaids.org/pub/Report/2009/JC1700_Epi_Update_2009_en.pdf. Accessed January20, 2011.

2. WorldHealthOrganization. Palliative care. 2005.Available from http://www.who.int/hiv/topics/palliative/PalliativeCare/en/. Accessed December19, 2010.

3. Harding R, Karus D, Easterbrook P, et al. Doespalliative care improve outcomes for patients withHIV/AIDS? A systematic review of the evidence.Sex Transm Infect 2005;81:5e14.

4. Harding R, Higginson IJ. Palliative care in sub-Saharan Africa. Lancet 2005;365:1971e1977.

5. Wakeham K, Harding R, Bamukama D, et al.Symptom burden in HIV infected adults at time ofHIV diagnosis in rural Uganda. J Palliat Med 2010;13:375e380.

6. Peltzer K, Phaswana-Mafuya N. The symptomexperience of people living with HIV and AIDS inthe Eastern Cape, South Africa. BMC Health ServRes 2008;8:271.

7. Kikule E. A good death in Uganda: survey ofneeds for palliative care for terminally ill peoplein urban areas. BMJ 2003;327:192e194.

8. Gwyther L, Rawlinson F. Symptom control inpalliative care: essential for quality of life. S AfrMed J 2004;94:437.

9. Harding R, Powell RA, Kiyange F, Downing J,Mwangi-Powell F. Provision of pain- and symptom-relieving drugs for HIV/AIDS in sub-Saharan Africa.J Pain Symptom Manage 2010;40:405e415.

10. Logie DE, Harding R. An evaluation of a mor-phine public health programme for cancer andAIDS pain relief in Sub-Saharan Africa. BMC PublicHealth 2005;5:82.

11. Maritz J, Benatar M, Dave JA, et al. HIV neurop-athy in South Africans: frequency, characteristics,and risk factors. Muscle Nerve 2010;41:599e606.

12. Mphahlele N, Mitchell D, Kamerman P. Valida-tion of the Wisconsin Brief Pain Questionnaire in

Vol. 44 No. 1 July 2012 9HIV Symptoms in Sub-Saharan African Palliative Care

a multilingual South African population. J PainSymptom Manage 2008;36:396e412.

13. Oken M, Creech R, Tormey D, et al. Toxicityand response criteria of the Eastern CooperativeOncology Group. Am J Clin Oncol 1982;5:649e655.

14. Chang VT, Hwang SS, Feuerman M, Kasimis BS,Thaler HT. The Memorial Symptom AssessmentScale Short Form (MSAS-SF). Cancer 2000;89:1162e1171.

15. Brechtl JR, Breitbart W, Galietta M, Krivo S,Rosenfeld B. The use of highly active antiretroviraltherapy (HAART) in patients with advanced HIV in-fection: impact on medical, palliative care, and qual-ity of life outcomes. J Pain Symptom Manage 2001;21:41e51.

16. Harding R, Molloy T, Easterbrook P, Frame K,Higginson IJ. Is antiretroviral therapy associatedwith symptom prevalence and burden? Int J STDAIDS 2006;17:400e405.

17. Karus D, Raveis VH, Alexander C, et al. Patientreports of symptoms and their treatment at threepalliative care projects servicing individuals withHIV/AIDS. J Pain Symptom Manage 2005;30:408e417.

18. Selwyn PA, Rivard M, Kappell D, et al. Palliativecare for AIDS at a large urban teaching hospital:program description and preliminary outcomes.J Palliat Med 2003;6:461e474.

19. Harding R, Lampe FC, Norwood S, et al. Symp-toms are highly prevalent among HIV outpatientsand associated with poor adherence and unpro-tected sexual intercourse. Sex Transm Infect 2010;86:520e524.

20. Altman DG. Practical statistics for medical re-search. London, UK: Chapman and Hall/CRC,1990.

21. Snow RC, Madalane M, Poulsen M. Are mentesting? Sex differentials in HIV testing in Mpuma-langa Province, South Africa. AIDS Care 2010;22:1060e1065.

22. Selwyn PA, Forstein M. Overcoming the false di-chotomy of curative vs. palliative care for late-stageHIV/AIDS: ‘‘Let me live the way I want to live, untilI can’t’’. JAMA 2003;290:806e814.

23. Harding R, Selman L, Agupio G, et al. The preva-lence andburdenof symptoms among cancer patientsattending palliative care in 2 African countries. Eur JCancer 2011;47:51e56.

24. Shawn ER, Campbell L, Mnguni MB,Defilippi KM, Williams AB. The spectrum of symp-toms among rural South Africans with HIV infec-tion. J Assoc Nurses AIDS Care 2005;16:12e23.

25. Selman L, Higginson IJ, Agupio G, et al. Meet-ing information needs of patients with incurableprogressive disease and their families in South Afri-ca and Uganda: multicentre qualitative study. BMJ2009;338:b1326.