Early severe morbidity and resource utilization in South African adults on antiretroviral therapy
© de Cherif et al; licensee BioMed Central Ltd. 2009
Received: 10 February 2009
Accepted: 15 December 2009
Published: 15 December 2009
High rates of mortality and morbidity have been described in sub-Saharan African patients within the first few months of starting highly active antiretroviral therapy (HAART). There is limited data on the causes of early morbidity on HAART and the associated resource utilization.
A cross-sectional study was conducted of medical admissions at a secondary-level hospital in Cape Town, South Africa. Patients on HAART were identified from a register and HIV-infected patients not on HAART were matched by gender, month of admission, and age group to correspond with the first admission of each case. Primary reasons for admission were determined by chart review. Direct health care costs were determined from the provider's perspective.
There were 53 in the HAART group with 70 admissions and 53 in the no-HAART group with 60 admissions. The median duration of HAART was 1 month (interquartile range 1-3 months). Median baseline CD4 count in the HAART group was 57 × 106 cells/L (IQR 15-115). The primary reasons for admission in the HAART group were more likely to be due to adverse drug reactions and less likely to be due to AIDS events than the no-HAART group (34% versus 7%; p < 0.001 and 39% versus 63%; p = 0.005 respectively). Immune reconstitution inflammatory syndrome was the primary reason for admission in 10% of the HAART group. Lengths of hospital stay per admission and inpatient survival were not significantly different between the two groups. Five of the 15 deaths in the HAART group were due to IRIS or adverse drug reactions. Median costs per admission of diagnostic and therapeutic services (laboratory investigations, radiology, intravenous fluids and blood, and non-ART medications) were higher in the HAART group compared with the no-HAART group (US$190 versus US$111; p = 0.001), but the more expensive non-curative costs (overhead, capital, and clinical staff) were not significantly different (US$1199 versus US$1128; p = 0.525).
Causes of early morbidity are different and more complex in HIV-infected patients on HAART. This results in greater resource utilization of diagnostic and therapeutic services.
Highly active antiretroviral therapy (HAART) has dramatically improved the prognosis of HIV infection in high-income countries by decreasing mortality, AIDS illnesses, and hospitalization [1–3]. In low-income countries patients start HAART with lower baseline CD4 counts than in high-income countries . Severely immune suppressed patients remain at increased risk of several complications within the first few months of starting HAART. First, although HAART reduces the incidence of AIDS-defining events, the incidence rates remain high in the first few months . Second, immune reconstitution inflammatory syndrome (IRIS) reactions occur more commonly with lower baseline CD4 counts . Third, many adverse drug reactions to antiretroviral agents occur more commonly with advanced disease . Therefore it is not surprising that there are high rates of early morbidity [8–10] and mortality [4, 8] in patients starting HAART in low-income countries.
Few studies have assessed the causes of early morbidity on HAART in sub-Saharan African ART programmes. Most studies report only HIV-related morbidity [8, 9] or on specific diseases like tuberculosis . The cohorts in the two West African studies [8, 9] had considerably higher baseline CD4 counts (71% >200 × 106 cells/L and median 290 × 106 cells/L) than is typical for sub-Saharan African programmes. A South African study included patients more typical of the region, but did not include IRIS as a diagnostic category . IRIS was also not reported in the West African studies [8, 9].
We conducted a cross-sectional study to determine whether the reasons for hospitalization (including all causes), outcomes, and resource utilization during hospital admissions were different in HIV-infected patients on HAART compared with HIV-infected patients not on HAART in South Africa, a country with a very high HIV prevalence.
Study Site Description
The study was conducted at G.F. Jooste Hospital, a secondary-level hospital providing care for indigent patients referred from primary health-care providers and clinics in the poor urban communities of metropolitan Cape Town.
A cross-sectional study was conducted to assess the spectrum of disease and resource utilization of medical admissions to G.F. Jooste Hospital among HIV-infected patients on HAART (HAART group) compared with those not on HAART (no-HAART group) during a nine-month period from July 2003 through March 2004. The patients on HAART accessed antiretroviral therapy predominantly from donor-funded programs in G.F. Jooste Hospital's catchment area, because the national public sector HAART roll-out program only commenced in 2004. The donor-funded programs included Médecins sans Frontières (MSF) at Khayelitsha ; Hannan Crusaid at Gugulethu  (both operated at public sector primary care clinics); and the joint Nelson Mandela Foundation-South African Medical Association's Tshepang project, GF Jooste Hospital clinic. These programs followed the 2002 WHO guidelines for HAART in resource-limited settings, initiating HAART at CD4+ counts < 200 × 106 cells/L or with clinical AIDS .
Patients in the HAART group were identified from a registry of HIV-infected patients on HAART who were admitted to the medical wards. The registry was maintained by two resident infectious disease physicians. Patients in the no-HAART group were matched by gender, month of admission, and age group (18-29; 30-49; and >49 years) to correspond with the first admission of each patient in the HAART group. Subsequent admissions of patients in both groups during the study period were also captured.
The study was approved by the University of Cape Town's Research Ethics Committee.
Data capture and analysis
The following data were extracted by retrospective chart review: demographic information, WHO clinical stage prior to admission, tuberculosis treatment or prophylaxis, cotrimoxazole prophylaxis, HAART regimen and date of commencement, treatment administered in hospital (medication, fluids, or blood products), investigations during hospitalization (cross-checked with the laboratory's electronic database), CD4+ cell count within six months of admission (if this was unavailable, the total lymphocyte count was recorded), length of stay, discharge diagnoses, discharge disposition (home/step-down facility/tertiary hospital), and survival. The data were recorded in a spreadsheet (Excel, version 97) and checked for illogical or missing entries by the database analyst. Data were analyzed with EpiInfo, version 6. Barlett's test was used for homogeneity of variance, which was shown to differ between the two groups, therefore the Kruskal-Wallis test was used to analyse differences between the outcomes and costs of the two groups.
The study physician classified the primary reason for admission (identified from the discharge diagnoses and a chart review) into the following diagnostic categories: AIDS-defining events, adverse drug reactions, HIV-related conditions, and other medical conditions not necessarily related to HIV. In the WHO clinical staging classification, pulmonary tuberculosis is defined as stage 3, whilst extrapulmonary tuberculosis is stage 4 (AIDS). Because it was not always clear from discharge diagnoses which form of tuberculosis patients had, we did not separate tuberculosis by site and included all forms of tuberculosis as AIDS illnesses. HIV-related conditions were recorded primarily according to WHO clinical stages 1 to 4 . IRIS was defined as paradoxical deterioration of an opportunistic disease that was diagnosed prior to HAART commencement.
Costing took the perspective of the health-care provider. The median cost per inpatient day was calculated separately for non-curative care costs and for patient-specific diagnostic and therapeutic costs. Patient-specific laboratory and therapeutic costs were calculated using a standard ingredients approach, which involves linking the usage of resources to the cost of each resource . As the study did not use timesheets to calculate the time spent by clinical staff with each patient, costs for these clinical staff were included in non-curative care costs.
Non-curative care costs, including recurrent overhead and hotel costs, capital costs, and clinical staff costs, were calculated by the study economist using a standard step-down approach . Hotel costs include rent on buildings, utilities, ward staff (laundry, kitchen, building maintenance), administration, and other nonspecific services and stores that cannot be directly linked to patient utilization. Hotel costs at G.F. Jooste medical wards and its step-down ward, Carnation, were assumed to be the same. Capital costs include assets that are not used up in one year, such as buildings, furniture, and medical equipment. The replacement costs of buildings and equipment were based on data from the National Health Facilities Audit undertaken by the National Department of Health (Rod Bennet, personal communication). These data were annualized using an estimated working life of 30 years for buildings and 5 years for equipment at a rate of 8 percent. This real interest rate is the return on long-term government bonds in South Africa and is similar to the rate of 10 percent in a comparable secondary facility in East Africa .
Clinical staff costs included salaries of doctors (attending physicians, medical registrars, and interns) and nurses (chief nurses, registered nurses, nurses-in-training, and nurse assistants). The clinical staff cost per inpatient day was calculated by establishing the full cost of employment (including benefits) of doctors and nurses together with the number of full time equivalent staff working in the medicine wards. Multiplying these two allowed us to estimate the annual cost of these clinical inputs. The clinical staff cost per inpatient day was then calculated by the formula:
Annual clinical staff cost/(number of beds × average bed occupancy × 365 days per year).
Patient-specific diagnostic and therapeutic costs included laboratory investigations, radiology, intravenous fluids and blood, and non-ART medications. Laboratory costs were provided by the National Health Laboratory Service. Drug costs were obtained from the hospital pharmacy price list, which is derived from the 2003 Provincial Tender Price List. Radiology and special investigations were costed from the central government's Uniform Fee Schedule (UPFS) 2002, inflated to the 2003 level using the consumer price index, less mortgage bonds [19, 20].
All costs were converted from South African Rands to U.S. dollars at the average bank rate at the midpoint of the study (November 2003): U.S. $1 = 6.73231 Rand . The total cost for each admission was calculated by the length of stay multiplied by non-curative costs per day, plus total diagnostic and therapeutic costs.
Between July 2003 and March 2004, 72 admissions of HIV-infected individuals on HAART met the inclusion criteria, but 2 patients' folders were lost, leaving 70 admissions analyzed (53 patients, 17 with 2 admissions). Two-hundred thirty-seven HIV-infected individuals not on HAART who had been admitted to G.F. Jooste during the study period were identified, and 204 patient folders from those admissions were found. Fifty three patients in the no-HAART group were matched with the first admission of the 53 patients in the HAART group representing 60 admissions (6 were admitted twice, and one was admitted thrice). Five patients in the no-HAART group were matched by season of admission as no exact month match could be made.
Ninety-seven percent of patients admitted on HAART were on a non-nucleoside reverse transcriptase inhibitor (NNRTI) regimen with dual NRTI. One patient was on self-funded dual NRTI antiretroviral therapy, but was included in the study in the HAART group. The median duration of antiretroviral therapy was 1 month (interquartile range 1-3 months; range 1-7 months).
Demographics and characteristics of HIV-infected patient admissions at G.F. Jooste Hospital.
Number of Admissions
32 (IQR 29-35)
32 (IQR 27-40)
p = 0.871
p = 0.647
p = 0.000
public primary-care clinic
secondary hospital clinic
tertiary hospital clinic
private medical doctor
Prior WHO Stage
p = 0.040
CD4 Count × 106 cells/L
Median CD4 count
57 (IQR 15-115)
16 (IQR 10-316)
p = 0.96
p = < 0.001
Primary Reason for Admission
(n = 70) n (%)
(n = 60) n (%)
p = 0.005
p < 0.001
Lactic acidosis (stavudine)
Pancreatitis (stavudine or lamivudine)
p = 1
Severe bacterial infection
p = 0.676
Deep venous thrombosis
Length of Stay and Outcome
(n = 70)
(n = 60)
Median Length of Stay, days (IQR)
p = 0.63
p = 0.139
p = 0.076
(n = 56)
(n = 54)
Median Cost in US$ of Hospitalization
Non curative costs G.F. Jooste plus step-down ward*
p = 0.595
Diagnostic and therapeutic cost **
p = 0.001
Total hospitalization cost per admission
p = 0.525
We analysed diagnoses, outcomes and resource utilization in 130 secondary hospital HIV admissions comparing patients who recently started HAART with those not on HAART in a high HIV prevalence area. We found that the primary reasons for admission were more likely to be due to adverse drug reactions and less likely to be due to AIDS events in patients on HAART. IRIS emerged as an important cause of early morbidity, accounting for 10% of the primary causes of admission in the HAART group. IRIS together with adverse drug reactions accounted for a third of all deaths in patients on HAART.
The patients in our study were severely immune suppressed, as assessed by WHO clinical staging and CD4 cell counts, which is typical of patients starting HAART in Africa. We found that tuberculosis was by far the commonest diagnosis in both the HAART and no-HAART admissions. In the West African studies tuberculosis was the second commonest diagnosis [8, 9], but it was the commonest in another South African study , which reflects the higher TB incidence in South Africa. However, we found that adverse drug reactions were more commonly the primary reason for admission among our HAART group than the other South African study . Adverse drug reactions were not assessed in the West African studies. In a recent study of severe adverse drug reactions in South African medical inpatients, among HIV-infected patients those on HAART were 10 times more likely than those not on HAART to be admitted to hospital with an adverse drug reaction , which is in keeping with our finding that adverse drug reactions were far more commonly the primary reasons for admission in the HAART group. IRIS accounted for 10% of our admissions and was not assessed in either of the West African studies or the other South African study.
Hospitalization costs are a major component of costs for HIV care in developed countries, particularly in patients with more advanced disease . The hospital sector is where HIV currently has the greatest impact on the public health budgets in South Africa, where HIV admissions account for 25% of all hospital admissions and consume one eighth of the public-health budget . Costs of delivering HAART are in part offset by savings on hospitalization. A South African study of the cost-effectiveness of HAART demonstrated dramatic reductions in the costs of hospitalization and a shift of care delivery to the outpatient setting for patients accessing HAART . We found that there were significantly more costs for diagnostic and therapeutic interventions of patients on HAART compared with controls. The higher resource utilization for diagnostic and therapeutic services in the HAART group likely reflects the greater complexity of diagnosis in these patients. The similar median total costs among both arms of the cohort arise from the fact that the hospitalization costs are being driven by per diem non-curative costs. Higher per admission costs for patients on HAART compared with those not on HAART were also reported in another South African study . In our study, the recording of comorbidities and inclusion of associated diagnostic and therapeutic costs carries forward the work by Guiness et al in Kenya, which was limited to costs related to the primary reason for admission .
Our findings have several implications. First, starting HAART at higher baseline CD4 counts would reduce the incidence of IRIS, many adverse drug reactions, and HIV-related morbidity. This would result in a shift from inpatient to outpatient HIV care, which would free substantial resources for the many other competing health care priorities in Africa. Second, the use of safer antiretroviral drugs (e.g. tenofovir or efavirenz instead of stavudine or nevirapine) will reduce morbidity as adverse drug reactions were the primary reason for admission in 34% of our HAART group. The World Health Organization has recommended this strategy in resource-limited settings, but recognizes that agents with better safety profiles are currently more costly . Third, the large burden of tuberculosis we found, which is typical of other African studies, emphasises that prevention, diagnosis and treatment of tuberculosis should be integrated into antiretroviral programmes. Fourth, differential diagnosis and management in patients presenting with early morbidity on HAART is complex and difficult in resource-limited settings with limited diagnostic facilities. The differentiation of IRIS from new opportunistic disease or drug toxicity is particularly complex. Training manuals for health care workers in Africa should include the causes of early morbidity on HAART, as well as the importance of their early recognition and appropriate management.
Our study has a number of limitations. We collected data retrospectively, which is always less reliable and complete than in prospective studies. We could only detect large difference between patients in the HAART and no-HAART groups given our relatively small sample size and comparisons between the two groups should be interpreted with caution. Our study was also only limited to inpatients and should not be extrapolated to outpatients. However, detailed costing was done on all 130 admissions, which involved a considerable amount of data collection. Limitations of our cost analysis include lack of a detailed assessment of clinical staffing costs, and exclusion of the indirect and direct non-health care costs borne by patients and their families. Importantly, future cost savings from HAART, notably reduction in rates of hospitalisation, were not assessed. Our study was conducted early on in the roll out of antiretroviral therapy when the numbers of patients on HAART in our catchment area was considerably lower than it is currently, but CD4 counts at HAART initiation were lower than they are currently, which should result in lower hospitalization rates and possibly different reasons for hospitalisation. CD4 counts were discouraged on inpatients as a cost-saving measure, but were done as outpatients. Because only a minority of patients in the no-HAART group was in HIV care while all those on HAART were, we had access to CD4 counts in most of the HAART group and few of the no-HAART group. Finally, although access to diagnostic and therapeutic interventions was similar for both groups, we could not exclude the possibility that clinicians looking after patients on HAART might have been more active in investigating and treating them due to a perceived better prognosis.
In conclusion, our study provides disease and economic information that may assist with the planning and delivery of comprehensive HIV/AIDS care in Africa. A high proportion of admissions for patients on HAART is due to adverse drug reactions or IRIS. Further large prospective cohort studies should be conducted to identify the nature of the early morbidity on HAART in Africa more accurately, and to identify strategies to prevent, diagnose and manage these morbidities.
The authors would like to thank Mrs. Natalie Christians for her research assistance and Mr. Horatious Duma for his assistance with file retrieval.
Sponsorship: This study was partially funded by Mount Holyoke College (South Hadley, Massachusetts, USA) through a Mary P. Dole Medical Fellowship awarded to Dr. Smith de Cherif and supported by funding from the University of New Mexico and the University of Miami. Graeme Meintjes was partly funded by a Guy Elliott fellowship.
- Ledergerber B, Egger M, Opravil M, Telenti A, et al: Clinical progression and virologic failure on highly active antiretroviral therapy in HIV-1 patients: a prospective cohort study. Lancet. 1999, 353: 863-868. 10.1016/S0140-6736(99)01122-8.View ArticlePubMedGoogle Scholar
- Palella F, Delaney KI, Moorman A, Loveless M, et al: Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med. 1998, 338: 853-860. 10.1056/NEJM199803263381301.View ArticlePubMedGoogle Scholar
- Paul S, Gilbert H, Lande L, Vaamonde C, et al: Impact of antiretroviral therapy on decreasing hospitalization rates of HIV-infected patients in 2001. AIDS Res Hum Retroviruses. 2002, 18: 501-506. 10.1089/088922202317406646.View ArticlePubMedGoogle Scholar
- Braitstein P, Brinkhof MW, Dabis F, et al: Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries. Lancet. 2006, 367: 817-24. 10.1016/S0140-6736(06)68337-2.View ArticlePubMedGoogle Scholar
- Michelet C, Arvieux C, Francois C, Besnier J, et al: Opportunistic infections during highly active antiretroviral treatment. AIDS. 1998, 12: 1815-1822. 10.1097/00002030-199814000-00013.View ArticlePubMedGoogle Scholar
- French MA, Price P, Stone S: Immune restoration disease after antiretroviral therapy. AIDS. 2004, 18: 1615-1627. 10.1097/01.aids.0000131375.21070.06.View ArticlePubMedGoogle Scholar
- Carr A, Cooper D: Adverse effects of antiretroviraltherapy. Lancet. 2000, 356: 1423-1430. 10.1016/S0140-6736(00)02854-3.View ArticlePubMedGoogle Scholar
- Moh R, Danel C, Messou E, Ouassa T, Gabillard D, Anzian A, et al: Incidence and determinants of mortality and morbidity following early antiretroviral therapy initiation in HIV-infected adults in West Africa. AIDS. 2007, 21: 2483-91. 10.1097/QAD.0b013e3282f09876.View ArticlePubMedGoogle Scholar
- Seyler C, Messou E, Gabillard D, Inwoley A, Alioumn A, Anglaret X: Morbidity before and after HAART initiation in Sub-Saharan African HIV-infected adults: a recurrent event analysis. AIDS Res Hum Retroviruses. 2007 Nov;23(11):1338-47. 2007, 23: 1338-1347. 10.1089/aid.2006.0308.Google Scholar
- Harling G, Orrell C, Wood R: Healthcare utilization of patients accessing an African national treatment program. BMC HealthServ Res. 2007, 7: 80-10.1186/1472-6963-7-80.View ArticleGoogle Scholar
- Antiretroviral Therapy in Low-Income Countries Collaboration of the International epidemiological Databases to Evaluate AIDS (IeDEA); ART Cohort Collaboration, Brinkhof MW, Egger M, Boulle A, May M, Hosseinipour M, Sprinz E, et al: Tuberculosis after Initiation of Antiretroviral Therapy in Low-Income and High-Income Countries. Clin Infect Dis. 2007, 45: 1518-1521. 10.1086/522986.View ArticleGoogle Scholar
- Coetzee D, Hildebrand K, Boulle A, Maartens G, et al: Outcomes after two years of providing antiretroviral treatment in Khayeltisha, South Africa. AIDS. 2004, 18: 887-895. 10.1097/00002030-200404090-00006.View ArticlePubMedGoogle Scholar
- Bekker L, Orrell C, Reader L, Matoti K, et al: Antiretroviral therapy in a community clinic: early lessons from a pilot project. S Afr Med J. 2003, 93: 458-462.PubMedGoogle Scholar
- World Health Organization: Scaling up Antiretroviral Therapy in Resource-limited Settings: Guidelines for a Public Health Approach; Executive Summary. 2002, Geneva: World Health OrganizationGoogle Scholar
- World Health Organization: Acquired Immunodeficiency Syndrome (AIDS). Interim proposal for a WHO staging for HIV infection and disease. Wkly Epidemiol Rec. 1990, 65: 221-4.Google Scholar
- Cleary S, McIntyre D, Boulle A: The cost-effectiveness of antiretroviral treatment in Khayelitsha, South Africa: a primary data analysis. Cost Eff Resourc Alloc. 2006, 4: 20-10.1186/1478-7547-4-20.View ArticleGoogle Scholar
- Barnum H, Kutzin J: Public hospitals in developing countries: resource use, cost, financing. 1993, Baltimore, Maryland: Johns Hopkins University Press for the World BankGoogle Scholar
- Guinness L, Arthur G, Bhatt S, et al: Costs of hospital care for HIV-positive and HIV-negative patients at Kenyatta National Hospital, Nairobi, Kenya. AIDS. 2002, 16: 901-8. 10.1097/00002030-200204120-00010.View ArticlePubMedGoogle Scholar
- National Department of Health: Uniform Fee Schedule for Patients Attending Public Hospitals. 2002, accessed December 7, 2005, [http://www.doh.gov.za/programmes/index.html]Google Scholar
- Drummond M, Jefferson T: Guidelines for authors and peer reviewers of economic submissions to the BMJ. BMJ. 1996, 313: 275-283.View ArticlePubMedPubMed CentralGoogle Scholar
- Exchange rates. accessed April 1, 2005, [http://www.x-rates.com/d/ZAR/USD/hist2003.html]
- Post F, Wood R, Maartens G: CD4+ and total lymphocyte counts as predictors of HIV disease progression. QJM. 1996, 89: 505-8.View ArticlePubMedGoogle Scholar
- Mehta U, Durrheim D, Blockman M, Kredo T, Gounden R, Barnes K: Adverse drug reactions in adult medical inpatients in a South African hospital serving a community with a high HIV/AIDS prevalence: prospective observational study. Br J Clin Pharmacol. 2008, 65: 396-406. 10.1111/j.1365-2125.2007.03034.x.View ArticlePubMedGoogle Scholar
- Levy A, James D, Johnston K, Hogg R, Harrigan P, Harrigan B, Sobolev B, Montaner J: Direct costs of HIV/AIDS Care. Lancet Infect Dis. 2006, 6: 171-177. 10.1016/S1473-3099(06)70413-3.View ArticlePubMedGoogle Scholar
- Badri M, Cleary S, G Maartens, Pitt J, Bekker L-G, Orrell C, Wood R: When to initiate highly active antiretroviral therapy in sub-Saharan Africa? A South African cost-effectiveness study. Antivir Ther. 2006, 11: 1-10.Google Scholar
- Badri M, Maartens G, Mandalia S, Bekker LG, Perod J, Platt R, et al: Cost-Effectiveness of Highly Active Antiretroviral Therapy in South Africa. PLOS Med. 2006, 3: 1-9. 10.1371/journal.pmed.0030004.Google Scholar
- Thomas LS, Manning A, Holmes CB, Naidoo S, Linde van der F, Gray GE, Martinson NA: Comparative costs of inpatient care for HIV-infected and uninfected children and adults in Soweto, South Africa. J Acquir Immune Defic Syndr. 2007, 46: 410-6. 10.1097/QAI.0b013e318156ec90.View ArticlePubMedGoogle Scholar
- World Health Organization: Antiretroviral therapy for HIV infection in adults and adolescents in resource-limited settings: towards universal access. Recommendations for a public health approach. 2006, Geneva: World Health OrganizationGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2334/9/205/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.