Worldwide, the introduction of antiretroviral therapy (ART) has resulted in a large decrease in hospital admissions amongst HIV-positive patients. Analyses from 7 North American and European countries showed a decrease in frequency, average length of hospital stay, and cost per stay of between 32% and 77% in patients on ART compared with those not on ART.1–9 The reduction in cost because of decreased need for inpatient care has been used to make the economic case for public-sector provision of ART in many high-income countries.10–18 However, initiating ART at low CD4 cell counts has been strongly associated with high inpatient costs.3,4 Data from low- and middle-income countries show that a significant number of HIV-infected patients on ART still require hospitalization, especially those initiating at low CD4 counts.19–23
Few studies have measured rates of hospitalization in a single cohort both before and after ART initiation to evaluate the effect of treatment on hospital admissions.10,17,22,24,25 South Africa, a middle-income country, started its public-sector ART program in 2004. Although several studies have examined the program’s cost and cost-effectiveness,19,20,22–30 only a handful have included a description of inpatient cost in patients on and off ART in the public sector.20,24,26,27 None of these studies controlled for patients’ CD4 count, making it hard to compare studies across cohorts with different levels of disease severity.
To establish whether ART reduces hospitalizations while controlling for the patients’ CD4 counts, we compared hospitalization rates and costs in a South African cohort of HIV-positive patients before and after ART initiation stratified by patients’ current CD4 count.
We analyzed data from an adult HIV cohort study31–33 conducted from July 2003 to October 2010 at Chris Hani Baragwanath Hospital, a large, urban, tertiary hospital in Soweto in Gauteng Province, and Tintswalo Hospital, a rural hospital in Mpumalanga Province. Patients were recruited after testing HIV positive in the same hospital and were provided with pre-ART HIV care (regular clinic visits for CD4 count monitoring and nurse-led care for opportunistic infections) and initiated on ART once diagnosed with WHO stage 4 disease or a CD4 count <200 cells/mm3.34,35 Recruiting at HIV testing allowed us to include a sizable group of patients who were followed up until they became ART eligible and then were started on ART and continued to be followed up. Eligible patients for this analysis were ≥ 18 years old with at least 1 follow-up visit and 1 CD4 count after enrollment into the study. Participants were interviewed about their demographic and socioeconomic characteristics and medical history at baseline and about admission and discharge dates of hospitalizations and reasons for admission both at baseline and at follow-up visits 4 to 7 months apart. CD4 counts were collected at enrollment and up to 6 monthly thereafter.
Patient baseline characteristics were summarized using descriptive statistics. Hospital admissions occurring during pre-ART and ART periods were stratified by most recent CD4 count in the same 6-month time period as the admission. For this, we divided person-time for each subject into 6-month periods, starting at enrollment into the study for pre-ART person-time and the date of treatment initiation for person-time on ART. For each 6-month period, a patient contributed 1 observation indicating whether hospitalization occurred in this period, and a current CD4 cell count, which was the first CD4 cell count within that period of observation. For missing CD4 count data (30.2%), we created 25 randomly imputed datasets each for the pre-ART and the ART populations, with missing values modeled on existing data (hospitalization, site, square root of available CD4 counts, and time from either enrollment or ART initiation) and took the mean of the imputed CD4 counts for each missing observation.36 We estimated incident rate ratios (IRR) of hospitalization stratified by CD4 counts in the same 6-month period.
We estimated the cost of hospitalizations from the healthcare provider perspective using cohort data on length of stay (LOS) and the 2008 cost per patient-day equivalent (PDE) of the hospitals’ districts.37 The cost per PDE is a proxy of cost per inpatient day and is collected by all public-sector hospitals in South Africa, dividing total hospital expenditure during a financial year by the total number of hospital visits.38 The cost per PDE for Baragwanath Hospital was USD 164.19, and the cost per PDE for Tintswalo Hospital was USD 176.29. All costs are presented in 2009 USD using the 2009 average currency conversion rate of 1 USD = 7.11 ZAR.
Ethical approval was granted by review boards of the University of the Witwatersrand and Boston University.
Of 3906 patients in our analysis, 140 (3.6%) initiated ART before the enrollment into the study and 913 (23.4%) initiated ART after being enrolled. Overall, patients were predominately women (76.6%) with a median age of 33 years [interquartile range (IQR), 28–39]. Pre-ART patients had a median CD4 count of 269 cells/mm3 (IQR, 136–442) at study enrollment, whereas ART patients had a median CD4 count of 154 cells/mm3 (IQR, 91–239) at study enrollment, which declined to a median CD4 count of 117 cells/mm3 (IQR, 57–183) at treatment initiation. Median time in pre-ART care before the initiation onto ART was 7.0 months (IQR, 1.5–15.9). Patients on treatment were predominately (71%) treated with stavudine, lamivudine, and efavirenz, the most common first-line regimen in South Africa until 2010.
Frequency of Hospitalizations
Among the 3906 patients, 534 hospitalizations occurred during a median follow-up of 13.1 months (IQR, 6.3–28.2); 344 (64%) hospitalizations were in pre-ART patients, whereas 190 (36%) occurred after ART initiation (Table 1). Most patients had a single admission; however, 28 patients in the pre-ART period and 19 patients in the ART period had more than 1 admission, with a maximum of 5 and 4 admissions per patient in the pre-ART and ART period, respectively. The leading reasons for admission in patients not on ART were pulmonary tuberculosis (TB) (15.1%), Pneumocystis jirovecii pneumonia (6.4%), and trauma (5.5%); in patients on ART, there were cases of pulmonary TB (15.2%), P. jirovecii pneumonia (7.9%), and headache of any kind (6.8%). The incidence of hospitalizations related to pulmonary TB was 0.74 and 1.0 per 100 patient years in the pre-ART and ART period, respectively; the incidence of admissions for extrapulmonary TB was 0.13 and 0.3 per 100 patient years, respectively. During the first 6 months on ART, the incidence of admissions for pulmonary TB was, at 2.8 per 100 patient years, almost 3 times as high, pointing at the possibility of immune constitution syndrome.
As current CD4 count increased, the rate of hospitalization decreased. Hospitalization rates were highest for patients with CD4 counts ≤ 100 cells/mm3. Patients with a CD4 count > 350 cells/mm3 had a reduction in the rate of hospitalization compared with patients with a CD4 count < 100 cells/mm3 of 70% pre-ART and of 80% under ART [pre-ART IRR, 0.3; 95% confidence interval (CI): 0.2 to 0.5; ART IRR, 0.2; 95% CI: 0.1 to 0.3]. Hospitalization rates were higher for ART patients than pre-ART patients in all CD4 strata, with most of this difference being driven by the rural cohort, a much smaller population. When removing events unlikely to be HIV related (trauma and accidents; 43 events in the pre-ART period and 6 in the ART period), events in patients initiating ART with a CD4 cell count above 200 cells/mm3, and all events in the first 3 months after ART initiation, the average rates of hospitalization in the pre-ART and ART cohorts did not change, and the effect was still significant (Tables 2–4).
Cost of Hospitalizations
Mean LOS per hospitalization was 8.7 days (95% CI: 7.5 to 9.9) for pre-ART patients and 10.1 days (8.4–11.8) for ART patients and decreased with increasing CD4 count in both populations (Table 1). Mean LOS was slightly higher amongst ART vs. pre-ART patients in all CD4 strata except at >350 cells/mm3 and was higher in the rural clinic, regardless of ART status. As a result, the inpatient cost per patient year was higher for ART patients in every stratum and higher in the rural than in the urban site in almost all strata, partly because of the higher rural cost per PDE. The resulting mean inpatient cost per patient year for ART patients was 63% higher than for pre-ART patients (USD, 117 vs. 72). Regardless of treatment status, hospital stays were longest and most costly in patients with a CD4 count < 100 cells/mm3, with mean inpatient cost per patient year being 4 times higher at <100 cells/mm3 than at >350 cells/mm3 in the pre-ART period and 9 times higher in the ART period. Figure 1 shows inpatient cost as a function of CD4 count over the lifetime of a representative patient, extrapolated from the mean inpatient cost per CD4 cell count stratum found in our analysis.
Impact on the Average Inpatient Cost Per Patient in the National Treatment Program
We parameterized a previously published model of the cost of the South African national ART program40,41 with the results of this analysis. Between financial years 2012/2013 and 2016/2017, the total inpatient cost of patients on ART is projected to increase from USD 85 million per year to USD 121 million (5% of total program cost) as a result of a planned increase in patient numbers from the current 1.7 million to 3.6 million in 2017. However, the mean inpatient cost per patient year on ART will decrease by 9% from USD 37 to USD 34 as a result of a maturation of the cohort on ART and redistribution into higher CD4 counts. From 2010/2011 onward, the average annual inpatient cost of patients on ART is lower than that of patients not on ART (Fig. 2).
Our study shows that, as in high-income countries,3,4 hospitalizations in HIV-infected adults in South Africa are more frequent, longer, and more costly at lower CD4 counts. We found this to be true regardless of ART status. Patients on ART were hospitalized more often and for longer durations than pre-ART patients. This difference can be explained partly by a higher risk of immune constitution syndrome in patients initiating ART at lower CD4 counts, especially in a population with high TB coinfection rates.21,42 The incidence of hospitalizations related to pulmonary and extrapulmonary TB, the opportunistic infections most frequently associated with immune constitution syndrome in South Africa,21,42 was higher in the ART than in the pre-ART cohort and highest in the 6 months immediately after ART initiation. Similarly, the difference in hospitalization frequency and cost between the pre-ART and ART populations was driven by the rural population and could at least partly because of a bias of physicians toward patients on ART who they have already invested in and whose prognosis is far better.
Similar to our analysis, 3 of 4 studies of the cost of inpatient care for public-sector patients on ART in South Africa showed an increase in inpatient care cost for patients on ART, with a median inpatient cost per stay in 2009 USD of USD 1769 (range, 1319–2080).19,20,27 Our mean cost per stay of patients on ART of USD 1642 is comparable. A study of a private South African medical aid program showed a dramatically increased inpatient cost in the 6 months around ART initiation, when CD4 cell counts are at their lowest.22 When comparing our mean annual per patient inpatient cost of USD 110 to the median cost of outpatient care in 2009 USD for patients on ART in South Africa from a number of published studies, USD 1233 (range, 1078–1287),23,28–30 inpatient care adds about 10% to the total annual per patient cost of a patient on ART. Thus, it accounts for a small but not trivial share of the total cost of caring for HIV/AIDS patients in South Africa.
A potential limitation of our study is the assumption that the published cost per PDE is a good proxy for the inpatient costs of HIV-positive patients, which could lead to an over- or underestimation of real inpatient cost. However, a recent in-depth study of the inpatient cost of patients on ART in a different hospital in Johannesburg has shown that cost per PDE is very similar to total per day cost as evaluated in a bottom-up cost analysis using the detailed review of inpatient files.19 Second, since our study cohort had a higher median CD4 count at ART initiation than most public-sector clinics in South Africa, the cost of inpatient care for patients on ART was lower than is likely in routine care. Finally, while diagnoses were available for all admissions included in this study, their accuracy was somewhat limited by the experience and expertise of the attending healthcare workers and their access to diagnostic modalities, especially in the rural cohort.
Our findings provide evidence to support earlier initiation of ART in low- and middle-income countries. We saw a decrease in hospital admission rates by 50%, and of cost by 250%, when comparing CD4 200–350 to ≤100 in the pre-ART period. Currently, allowing patients’ CD4 counts to drop to very low levels before initiating them on ART burdens the health system 3-fold: first, through the high cost of inpatient care immediately before and after ART initiation, then with the cost of life-long ART, and finally with the high cost of end-of-life care once limited treatment options are exhausted. One of the benefits of initiating patients on ART at higher CD4 counts could be avoiding the first of these costs. In the absence of sufficient drug options to avoid the third, terminal cost, and in a situation of decreasing international funding for ART programs in low- and middle-income countries, avoiding the depletion of patients’ CD4 cells and the associated high likelihood of expensive inpatient care is one of the few options available to national ART programs to reduce the costs of HIV care.
The authors thank the patients attending the study clinics for the readiness to share their data.
1. Beck EJ, Mandalia S, Griffith R, et al.. Use and cost of hospital and community service provision for children with HIV infection at an English HIV referral centre. Pharmacoeconomics 2000;17:53–69.
2. Baum SE, Morris JT, Gibbons RV, et al.. Reduction in human immunodeficiency virus patient hospitalizations and nontraumatic mortality after adoption of highly active antiretroviral therapy. Mil Med. 1999;164:609–612.
3. Mouton Y, Alfandari S, Valette M, et al.. Impact of protease inhibitors on AIDS-defining events and hospitalizations in 10 French AIDS reference centres. Federation National des Centres de Lutte contre le SIDA. AIDS. 1997;11:F101–F105.
4. Krentz HB, Auld MC, Gill MJ. The high cost of medical care for patients who present late (CD4 <200 cells/microL) with HIV infection. HIV Med. 2004;5:93–98.
5. Nykamp D, Barnett CW, Lago M, et al.. Cost of medication therapy in ambulatory HIV-infected patients. Ann Pharmacother. 1997;31:303–307.
6. Krentz HB, Auld MC, Gill MJ. The changing direct costs of medical care for patients with HIV/AIDS, 1995-2001. CMAJ. 2003;169:106–110.
7. Garattini L, Tediosi F, Di Cintio E, et al.. Resource utilization and hospital cost of HIV/AIDS care in Italy in the era of highly active antiretroviral therapy. AIDS Care. 2001;13:733–741.
8. Stoll M, Claes C, Schulte E, et al.. Direct costs for the treatment of HIV-infection in a German cohort after the introduction of HAART. Eur J Med Res. 2002;7:463–471.
9. Kyriopoulos JE, Geitona MA, Paparizos VA, et al.. The impact of new antiretroic therapeutic schemes on the cost for AIDS treatment in Greece. J Med Syst. 2001;25:73–80.
10. Sendi PP, Craig BA, Meier G, et al.. Cost effectiveness of highly active antiretroviral therapy in HIV-infected patients. AIDS. 1999;13:1115–1122.
11. Freedberg KA, Losina E, Weinstein MC, et al.. The cost effectiveness of combination antiretroviral therapy for HIV disease. N Engl J Med. 2001;344:824–831.
12. Schackman BR, Goldie SJ, Weinstein MC, et al.. Cost-effectiveness of earlier initiation of antiretroviral therapy for uninsured HIV-infected adults. Am J Public Health. 2001;91:1456–1463.
13. Beck E, Mandalia S, Gaudreault M, et al.. The cost-effectiveness of highly active antiretroviral therapy, Canada 1991-2001. AIDS. 2004;18:2411–2418.
14. Miners A, Sabin C, Trueman P, et al.. Assessing the cost-effectiveness of highly active antiretroviral therapy for adults with HIV in England. HIV Med. 2001;2:52–58.
15. Le Pen C, Rozenbaum W, Downs A, et al.. Une analyse cout-efficacité du changement des schémas thérapeutiques dans le VIH depuis 1996 [in French]. Therapie. 2002;57:27–33.
16. Kowalik E, Jakubczyk M, Niewada M, et al.. The cost-effectiveness of antiretroviral regimens containing protease inhibitors (PIS) or nonnucleoside reverse transcriptase inhibitors (NNRTI) in the treatment of HIV-infected individuals in Poland. Value Health. 2002;5:569. Abstract.
17. Lacey L, Youle M, Trueman P, et al.. A prospective evaluation of the cost effectiveness of adding lamivudine to zidovudine-containing antiretroviral treatment regimens in HIV infection. European perspective. Pharmacoeconomics. 1999;15(suppl 1):39–53.
18. Pinto JL, Lopez Lavid C, Badia X, et al.. Análisis coste-efectividad del tratamiento antirretroviral de gran actividad en pacientes infectados por el VIH asintomáticos [in Spanish]. Med Clin (Barc). 2000;114(suppl 3):62–67.
19. Long L, Fox M, Rosen S. Cost of hospitalization
for those presenting at an HIV treatment center in South Africa [THPE0859]. Paper presented at: XVIII International AIDS Conference, July 18-23, 2010; Vienna, Austria.
20. Smith de Cherif TK, Schoeman JH, Cleary S, et al.. Early severe morbidity and resource utilization in South African adults on antiretroviral therapy. BMC Infect Dis. 2009;9:205.
21. Murdoch DM, Venter WD, Feldman C, et al.. Incidence and risk factors for the immune reconstitution inflammatory syndrome in HIV patients in South Africa: a prospective study. AIDS. 2008;22:601–610.
22. Leisegang R, Cleary S, Hislop M, et al.. Early and late direct costs in a Southern African antiretroviral treatment programme: a retrospective cohort analysis. PLoS Med. 2009;6:e1000189.
23. Stearns BK, Evans DK, Lutung P, et al.. Primary estimates of the costs of ART care at 5 AHF clinics in sub-Saharan Africa [MOPE0706]. Paper presented at: XVIIth International AIDS Conference, August 3-8, 2008; Mexico City, Mexico.
24. Harling G, Wood R. The evolving cost of HIV in South Africa--Changes in Health Care Cost With Duration on Antiretroviral Therapy for Public Sector Patients. J Acquir Immune Defic Syndr. 2007;45:348–354.
25. Marseille E, Kahn JG, Pietter C, et al.. The cost-effectiveness of home-based provision of antiretroviral therapy in rural Uganda. Appl Health Econ Health Policy. 2009;7:229–243.
26. Cleary S, McIntyre D, Boulle A. The cost-effectiveness of antiretroviral treatment. Cost Eff Resour Alloc. 2006;4:20.
27. Thomas LS, Manning A, Holmes CB, et al.. Comparative costs of inpatient
care for HIV-infected and uninfected children and adults in Soweto, South Africa. J Acquir Immune Defic Syndr. 2007;46:410–416.
28. Martinson N, Mohapi L, Bakos D, et al.. Costs of providing care for HIV-infected adults in an urban HIV clinic in Soweto, South Africa. J Acquir Immune Defic Syndr. 2009;50:327–330.
29. Long L, Rosen S, Sanne I. Stable outcomes and costs in South African patients’ second year on antiretroviral treatment [abstract]. Paper presented at: International AIDS Economics Network Symposium, August 1-2, 2008; Cuernavaca, Mexico.
30. Rosen S, Long L, Sanne I. The outcomes and outpatient costs of different models of antiretroviral treatment delivery in South Africa. Trop Med Int Health. 2008;13:1005–1015.
31. Chhagan V, Luiz J, Mohapi L, et al.. The socioeconomic impact of antiretroviral treatment on individuals in Soweto, South Africa. Health Sociol Rev. 2008;17:95–105.
32. Golub J, Pronyk P, Mohapi L, et al.. Isoniazide preventive therapy, HAART and tuberculosis risk in HIV-infected adults in South Africa: a prospective cohort. AIDS. 2009;23:631–636.
33. Hanrahan CF, Golub JE, Mohapi L, et al.. Body mass index and risk of tuberculosis and death. AIDS. 2010;24:1501–1508.
34. National Department of Health, Republic of South Africa. National Antiretroviral Treatment Guidelines. Pretoria, South Africa: Jacana Publishers; 2004.
35. World Health Organization. Towards Universal Access: Scaling Up Priority HIV/AIDS Interventions in the Health Sector. Geneva, Switzerland; 2008.
36. Schafer JL. Analysis of Incomplete Multivariate Data. Boca Raton, FL: Chapman & Hall/CRC; 1997.
38. National Department of Health. National Indicator dataset for South Africa. Pretoria, South Africa:National Department of Health; 2005.
39. Havlir DV, Getahun H, Sanne I, et al.. Opportunities and challenges for HIV care in overlapping HIV and TB epidemics. JAMA. 2008;300:423–430.
40. Meyer-Rath G, Pillay Y, Blecher M, et al.. Total cost and potential cost savings of the national antiretroviral treatment (ART) programme in South Africa 2010 to 2017 [WEAE0201]. Paper presented at: XVIII International AIDS Conference, July 18-23, 2010; Vienna, Austria.
41. Meyer-Rath G, Pillay Y, Blecher M, et al.. The impact of a new reference price list mechanism for drugs on the total cost of the national antiretroviral treatment programme in South Africa 2011 to 2017 . Paper presented at: South African AIDS Conference, June 7-10, 2011; Durban, South Africa.
42. Meintjes G, Rabie H, Wilkinson RJ, et al.. Tuberculosis-associated immune reconstitution inflammatory syndrome and unmasking of tuberculosis by antiretroviral therapy. Clin Chest Med. 2009;30:797–810.