South Africa is home to both the largest number of people infected with HIV and people on antiretroviral treatment (ART) worldwide.1 To plan for a program that currently covers more than 1.5 million patients, or 20% of the world's population on ART in low- and middle-income countries,1,2 information on the cost of ART provision is needed. Since 2004, a number of studies have been undertaken that have resulted in cost estimates of ART provision for a variety of different settings and models of care in South Africa.3–7 These studies almost exclusively focus on large clinics in urban settings, while recent policy changes have led to a refocusing of the national ART programme on increasing coverage of currently underserved semiurban and rural communities.
To reach the target of 80% ART coverage of the eligible population by the end of financial year 2010/2011 set by the 2007 National Strategic Plan,8 the 2010 National ART Guidelines introduced nurse initiation and management of ART.9 Involving the nurse cadre, together with strengthened political leadership, has lead to a rapid increase in the number of clinics accredited for ART provision from 497 in January 2010 to 1668 by January 2011.10 Most of the additional clinics are primary health care clinics (PHCs) outside large urban centers. To help with planning and budgeting to sustain this effort, information on the cost of ART provision in more remote settings is urgently needed.
This study compares resource utilization and the cost of providing ART to large numbers of patients in 2 different settings in the South African public sector: an urban clinic in central Johannesburg and a semiurban township-based clinic in the North West province, using identical methods. To correct for differences in case mix and disease severity, the development of per-patient costs is then compared to patients' CD4 cell-count development over time.
Data on the cost of ART provision were collected in 2 South African cohorts of patients initiating ART at a clinic attached to Charlotte Maxeke Johannesburg Academic Hospital (CMH), an urban tertiary academic hospital in the inner city of Johannesburg, Gauteng, between April 2006 and August 2008 (n = 181), and at a clinic attached to Tshepong Hospital (TWC), a secondary hospital in the township of Jouberton close to Klerksdorp, North West Province, between January 2007 and December 2008 (n = 184). Both clinics started providing routine ART care as part of the national ART rollout in the public sector commencing in 2004. Although TWC is situated in a densely populated area with formal and informal semiurban housing, it is further removed from central urban infrastructure than CMH.
A consecutive sample of patients was enrolled into the study during the period of treatment preparation, before ART initiation. Patients' consent was obtained for 3-monthly interviews and review of their clinical information.
To enter each clinic, patients had to fulfill eligibility criteria as detailed in the 2004 national ART guidelines: they had to have tested HIV-positive and have at least one recorded CD4 cell count of <200 cells/μL and/or World Health Organization stage 4 HIV disease.11 Before treatment initiation, patients underwent a series of preparatory visits, including medical examination, laboratory tests, and other diagnostic tests, as indicated, and up to 3 individual or group counseling sessions with an adherence counselor. Only those patients initiated on ART during their participation in the study were included in the analysis.
Cost and Resource Use Included
We collected information on the economic costs incurred for each patient in a cohort of patients initiated on ART in each of the 2 clinics from the health care provider perspective, using a microcosting approach to resource use and costs at the outpatient and hospital level.12 We included all health care resources used by patients from up to 3 months before ART initiation until either 12 months after ART initiation or, in case of patients dying, defaulting from care, or being down-referred to another clinic during the first year on ART, until the last visit to the clinic. Down-referral is a process by which patients who are stable on therapy and have achieved an undetectable viral load get referred to a clinic at primary health care level closer to their home. The cost analysis includes the cost of drugs (antiretroviral and nonantiretroviral drugs), diagnostic and monitoring tests (including laboratory tests and radiological examinations), labor, overheads, staff training in HIV management, infrastructure, medical equipment, and furniture.
Data on patients' resource use were obtained via a retrospective review of the clinic files of study patients. The number of laboratory tests was confirmed by a review of the patient records of the public sector National Health Laboratory Service at each hospital; the number of radiological examinations by a review of the electronic database of the radiology department where available. With regards to non-ARV costs, we made a number of assumptions and exclusions. All patients with CD4 cell counts less than 200 cells/μL were assumed to be prescribed cotrimoxazole for the prevention of opportunistic infections. Every patient was assumed to have been prescribed multivitamins throughout their stay in the clinic. We excluded all tuberculosis treatment costs because, although the first dose of such treatment might be prescribed at the ART clinic, tuberculosis is managed entirely off-site at PHCs. To focus the analysis on the cost of ART provision alone, we excluded this resource use.
Data on unit costs came from government drug depots (medication), the public sector National Health Laboratory Service (laboratory costs), the hospitals' radiology departments (x-ray examinations and ultrasound), clinic management (staff numbers and levels, numbers of patients, and visits per year), and the finance, human resources, and asset and store departments of both hospitals involved (staff salaries, equipment, supplies, and overheads). Data were collected in an electronic format where available and otherwise abstracted from printouts and paper-based price lists.
All fixed cost data for one clinic were entered into an Excel spreadsheet to impute totals and calculate the average fixed cost per patient-month, taking into account the number of patients in both ART and in pre-ART care in the clinic, the average number of visits for both of these patient subgroups by year, and the number of months spent in the clinic by each of the study patients. Space and utilities costs for the entire hospital were allocated by using the ratio of clinic-to-hospital space as calculated from hospital floor plans and measurements of the clinic space.
Patient files were reviewed for information on the dates of admission and discharge, discharge diagnoses, and names of hospitals that patients were admitted to. To capture the cost of inpatient resource use under ART only, we excluded any hospitalizations that occurred before ART initiation. To complement the information available from the files, we asked patients during 3-monthly interviews after ART initiation about any hospital stays since the last interview and recorded length of stay (LOS), hospital, and diagnosis where available. If there was incongruence between interviews and files with regards to hospitalization events, we deferred to the information from the interviews. We assumed that the LOS of those hospitalizations with missing admission and/or discharge dates was the same as the average LOS of those patients with complete dates.
For the unit cost of inpatient days, we used the cost per patient-day equivalent (PDE) of the hospitals that the patients were admitted to. This information, which divides total hospital expenditure during a financial year by the total number of visits to the hospital, is collected by the management of all public sector hospitals in South Africa. For the denominator, the number of visits to day wards or outpatient clinics is weighted by their average duration in relation to a full day of inpatient stay so that the result represents the average cost per inpatient day. Cost per PDE is calculated annually, allowing us to use the cost specific to the year that a patient was hospitalized in.
CD4 Cell-Count Data
In both clinics, patients' responsiveness to ART is measured by regular CD4 cell-count measurements, which are undertaken once before and every 3–6 months after treatment initiation. The clinic files of participants and the electronic laboratory database were reviewed for laboratory data on participants' CD4 cell counts for the duration of the study.
Data Collection and Analysis
Data on resource use, clinical status, and CD4 cell count was collected using paper templates from which data were entered into Excel spreadsheets, with 10% of the sample double entered for quality control. Mistakes were found in 5% of the double-entered data and most often pertained to the dosages of nonantiretroviral drugs, a component that does not contribute much to total cost (see below).
Capital costs were annualized using the government depreciation rate of 10% for telephones, 20% for computers and other electronic equipment, 6.7% for other office equipment, and 1% for buildings.13 For patients who were in care for less than a year because of defaulting, death, or down-referral, mean total costs were annualized. Unit costs were taken from the year in which the corresponding resources were used (2006–2008) and were adjusted for inflation to 2011 USD, using the 2009 average conversion rate of 1 USD = 8.442 ZAR and the country's average Consumer Price Inflation Index,14 with one exception: Because the prices of ARV drugs have halved on average since the data collection period, we used 2011 unit costs for ARV drugs rather than upward-adjust their cost for inflation. All costs are presented in 2011 USD.
For the comparison between cost and CD4 cell count by 3-month intervals since ART initiation, the results of CD4 cell-count tests were attributed to the same 3-month time period since ART initiation as cost, depending on the date on which the test was undertaken. If CD4 cell counts were missing for 1 or 2 consecutive 3-month periods, their values were interpolated linearly from CD4 cell counts in the adjacent periods. Baseline CD4 cell counts were defined as the lowest CD4 cell count during the 3 months before ART initiation. We used 2-sided t tests and Wilcoxon rank sum tests to compare mean cohort CD4 cell count, mean and median cost in each 3-month period on ART, and mean and median cost associated with each CD4 cell-count stratum, respectively, in-between clinics. All statistical analysis was done in SAS version 9.1.
The study was approved by the Human Research Ethics Committee of the University of the Witwatersrand and the Ethics Committee of the London School of Hygiene and Tropical Medicine.
Patient Status and Length of Follow-Up
Cohort length of follow-up differed between the 2 settings, largely as the result of higher rates of down-referral at TWC during the second half of the study period. Twenty-eight percent of patients at TWC had been down-referred by month 9 and 39% by month 12, as opposed to 2% and 3% at CMH. This resulted in TWC contributing 1878 patient-months to the study, compared to 2040 at CMH. Default and death rates were comparable in both clinics, with 1 patient (1%) having died and 16 patients (9%) having defaulted by 12 months at CMH, as compared with 4 (2%) and 11 patients (6%) at TWC, respectively.
Development of CD4 Cell Count Over Time
In both clinics, cohort CD4 cell counts increased steadily under ART (Table 1). Differences in mean CD4 cell count by 3-month period on treatment between the 2 clinics were not significant at the <0.05 level, except for months 1–3 after initiation, where the average CD4 cell count at TWC was much lower, and for month 7–9 after initiation, where it was higher.
Cost and Resource Use
Outpatient Cost and Resource Use
Table 2 summarizes outpatient resource use and unit costs for both clinics. At CMH, the average total cost per patient per year on ART was $1040 (95% confidence interval [CI]: $800 to $1280). The average outpatient cost was $692 (67% of total cost). The cost of antiretroviral drugs contributed 16% to this cost, with laboratory tests and radiology contributing 50%, fixed costs including staff and overhead costs 31%, and non-ARV drugs 2% (Table 3). Of fixed cost, 95% were due to staff cost, 4% due to building and utility costs, and 1% due to equipment and supplies (data not shown).
At TWC, the average cost of ART provision per patient per year on ART was $1115 (95% CI: $776 to $1453) (Table 2). The average outpatient cost was almost the same as at CMH, $697, or 68% of total cost. The largest contributor to outpatient cost was fixed cost with 63%, followed by ARV drug cost (13%), diagnostics cost (21%), and non-ARV drug costs (3%). Of fixed cost, 85% was due to staff cost, 1% due to building cost, and 14% due to equipment and supplies.
Inpatient Cost and Resource Use
At CMH, 26 (14%) of the 181 study patients required inpatient care after the initiation of ART, at a mean LOS of 9 days (range, 1–30 days). At a cost per day equivalent to the cost per PDE in 2007/2008 of $158 ($176 in 2011 values), the resulting mean cost per hospitalized patient was $1663 (95% CI: $1103 to $2041) and mean cost of inpatient care among all patients on ART, whether hospitalized or not, was $348 (28% of total cost) (Table 3). The average time on ART at hospitalization was 132 days (range, 13–363 days). Thirteen of the hospitalizations were due to opportunistic infections, 7 of which occurred in the first 3 months after initiation, the typical time window for the development of immune reconstitution syndrome. At least 8 other admissions were related to side effects of ART (pancytopenia, lactic acidosis, peripheral neuropathy, and gastric ulcers).
At TWC, only 12 (7%) of the 184 study patients required inpatient care after the initiation of ART, at an average LOS of 28 days (range, 1–63 days). At a cost per PDE of $226 in 2007/2008 ($252 in 2011 values), this led to a mean cost per hospitalized patient of $3824 (95% CI: $1143 to $6505) and a mean cost of inpatient care among all patients on ART of $418 (32% of total cost). The distribution of hospitalizations over time was rather different than at CMH, with only 23% of hospitalizations and cost within months 1–3 and 54% of hospitalizations and total inpatient cost falling into the time period of months 7–9 on ART. The average time on ART at hospitalization was 107 days (range, 2–232 days). Four of the hospitalizations were due to opportunistic infections, with only one occurring in the typical time window for the development of immune reconstitution syndrome. Three of the other admissions were related to side effects of ART (progressive neuropathy and lactic acidosis).
We also analyzed whether patients requiring inpatient care also had higher outpatient resource use. At CMH, patients with inpatient resource use had a mean outpatient cost of $848 (95% CI: $553 to $1143) compared to a mean outpatient cost of patients without inpatient resource use of $662 (95% CI: $612 to $712) (P > 0.2). At TWC, patients with inpatient resource use had a mean outpatient cost of $734 (95% CI: $677 to $792), compared to a mean outpatient cost of patients without inpatient resource use of $693 (95% CI: $668 to $718) (P > 0.1). In summary, although in both clinics mean outpatient cost was somewhat higher for those patients who required inpatient care, the differences were not significant at the <0.05 level.
Development of Cost With Time on Treatment and With CD4 Cell Count
Total cost, including both outpatient and inpatient cost, varied with time on treatment in both clinics, in part as a result of the treatment protocol asking for specific monitoring during the first visits, in part due to the higher incidence of opportunistic infections during the first months on treatment when CD4 cell counts were still low, some of which required hospitalization. As can be seen in Table 4, the highest cost per 3-month period on ART was found in the first 3 months after ART initiation in both clinics, after which cost decreased with every 3 months on ART except for the 7- to –9-month period at TWC which had high inpatient cost. Additionally, in both clinics, cost was highest in the lowest CD4 cell count stratum (<50 cells/μL). Differences in mean cost between clinics were not significant at the <0.05 level in any time period or CD4 cell-count stratum except for the 8–12 months on ART time period, where care at TWC was cheaper. Differences in median cost (data not shown), however, were significant at the <0.05 level for the 1- to 3-month to 7- to 9-month periods and the <50 and 50–199 cells/μL CD4 cell-count strata.
Our study shows that the cost of ART provision is similar between an urban and a semiurban site, and that in both sites cost is highest at low CD4 cell counts. The average outpatient cost of ART provision during the first 12 months after initiation was, at $692 and $697, respectively, almost identical in both clinics. Nonetheless, average total cost per patient-year at TWC was 6% higher than average total cost at CMH, with the entire difference borne by the difference in inpatient cost, which was a result of a number of patients with very long lengths of stay at TWC. The median outpatient cost per patient-year in 2011 USD of 5 previously published analyses of the cost of ART provision in the South African public sector is $1377 (range, $1204–$1438); median inpatient cost in the public sector is $369 (range, $258–$660).3–7 The annual per-patient cost found in both clinics examined in this paper fall within the spectrum found in these past analyses. In our analysis, 2 factors are associated with higher cost in either clinic: low CD4 cell counts and the higher risk of hospitalizations in the first 3 months on treatment. Both factors have been described as drivers of the cost of ART provision in previous studies, for both the private sector5 and the public sector of South Africa.3,4,15 At both clinics in our study, patients with CD4 counts below 50 cells/μL drive costs significantly in the first 3 months of treatment, costing between double and triple that of those initiated at CD4 cell counts above that threshold.
Inpatient care periods varied significantly between the 2 sites, although the small number of patients requiring inpatient care in both cohorts limit the generalizability of this difference. From subsequent discussions with staff at the sites, we believe this is driven by a higher illness threshold for admission in the semiurban site, radiology investigations and specialist referrals during inpatient care taking longer to arrange, as well as the fact that patients live further away from the hospital and are poorer, leading local clinicians to retain patients in inpatient care for longer.
Although total cost at the outpatient level is very similar, the contribution of single cost items is different between the settings, reflecting differences in clinic operations: At TWC, fixed cost is almost twice as expensive per patient than at CMH, while average ARV drug cost is only 33% of total outpatient cost, and average diagnostic cost is less than half that at CMH. In all past estimates of ART provision in South Africa, ARV cost contribute at least 50% to total outpatient cost.3–7,16–18 The higher ART and diagnostics cost at CMH can be explained by more patients having been switched to second line therapy, its academic status and better access to specialist care in the clinic. In contrast, the higher fixed cost at TWC is a result of the higher staff numbers, with the nurse contingent being almost double that of CMH due in part to nurses doubling as translators between doctors and patients. The higher equipment cost for TWC is a result of the clinic operating out of the hospital's previous nurses' quarters, thus requiring higher initial refurbishment and equipment expenditures than CMH where the clinic operates out of a previously established outpatient clinic.
Patients' status after 12 months on ART and time in the cohort differed significantly between the 2 clinics, mostly as a result of much higher down-referral at TWC (39% at TWC vs. 3% at CMH). This did not, however, lead to a bias towards sicker patients remaining in the cohort, as from 7 months on ART onwards the mean cohort CD4 cell counts were higher at TWC than at CMH, though not significantly. The increase in down-referral to PHCs in the patients' immediate communities at TWC during the study period resulted in part from an analysis of the association between patients' travel cost and loss to follow-up which showed a strong linear relationship between distance travelled and risk of defaulting from care in the study cohort.19,20 Acting on these findings in 2007 the provincial government of the North-West province accredited a number of additional ART clinics in the immediate vicinity of TWC and the clinic management of TWC accelerated down-referral to these sites as a means of preventing further patients defaulting from ART care. Although this resulted in lower numbers of patients defaulting from care during the remainder of the study, it did lead to a higher overall loss from the cohort through down-referral.
The generalizability of the results of this study is limited by a number of factors. Average costs at both facilities were collected for the first year on treatment and in 2 clinics only and are at least in part a result of specific operational circumstances (such as the high rate of down-referral during the study period at TWC); however, costs were similar between the 2 clinics, as well as compared to other studies, suggesting they are useful to guide policy makers. Differences in clinic populations in terms of poverty level and educational background could have impacted follow-up rates, rendering direct comparison of cost between the clinics difficult–although the distribution of baseline CD4 cell counts, as a proxy for disease severity and case-mix, was very similar at both clinics. The analysis is limited to the resources used by the cohort while in the care of the clinics under study, which means that not all cost of care is included, particularly for patients that were down-referred before the end of the study period. Since the cost of care at those downreferral sites is likely to be lower, this would mean that we over-estimate the cost of care for the semiurban site somewhat. Finally, both clinics are attached to a hospital complex, and most South African patients will in the future be initiated at primary care clinics, limiting generalizability of staff and fixed costs.
The cost of providing ART in the public sector of South Africa has been well researched but analysis has so far focused on urban centers. The results presented in this paper show that provision in semiurban settings costs almost the same at the outpatient level as in urban settings, while inpatient cost tends to be higher as a result of longer lengths of stay. Significant cost reduction within the program is possible, through identifying, retaining, and initiating patients on ART at a higher CD4 cell count, hence preventing hospitalization, which is the major preventable driver of cost. Task shifting away from professional staff may also offer cost reductions.
Cost at both clinics analyzed here fall within the spectrum of cost estimates that are already available for South Africa and that have been used as the average cost per adult patient in government budgeting and planning exercises,21,22 suggesting that provision at this cost level is expected by government, planned for, and, as a result, potentially sustainable. Recent reductions in staff cost due to task shifting and in the cost of antiretroviral drugs due to the opening of the South African market to international bidders have brought the cost of ART provision further down.22,23
The authors thank Kgomotso Thloaele, Albertina Dambuza, Motlalepule Letsapa, and Belinda Dambuza for recruiting and interviewing the patients of the 2 longitudinal study cohorts, the many patients who agreed to be part of them and sacrificed time and offered information toward this study, and staff at Tshepong and Charlotte Maxeke Hospitals and the National Health Laboratory Services for assistance with providing cost and resource use data, in particular, Prof Jeffrey Wing, Dr Sagie Pillay, and Ms Wiebe-Randerjee. The authors are grateful to Lilani Kumaranayake, Rosa Legood, Richard Grieve, Charlotte Watts, and Peter Vickerman for help in conceptualizing the analytical framework and methods used in this article.
1. UNAIDS. AIDS Epidemic Update. Geneva, Switzerland: UNAIDS; 2009.
2. Health budget vote policy speech. Presented at: the National Assembly by Minister of Health, Dr A Motsoaledi, May 31, 2011; Cape Town, South Africa. Available at: www.info.gov.za/speech
Accessed July 8, 2011.
3. Cleary S, McIntyre D, Boulle A. The cost-effectiveness of antiretroviral treatment in Khayelitsha, South Africa—a primary data analysis. Cost Eff Resour Alloc. 2006;4:20.
4. Harling G, Wood R. The evolving cost of HIV in South Africa. J Acquir Immune Defic Syndr. 2007;45:348–354.
5. 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(12):e1000189.
6. Long L, Fox M, Sanne I, et al.. The high cost of second-line antiretroviral therapy for HIV/AIDS in South Africa. AIDS. 2010;24:915–919.
7. 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.
8. National Department of Health, Republic of South Africa. National Strategic Plan for HIV and AIDS & STIs 2007-2011. Pretoria, South Africa: National Department of Health; 2007.
9. National Department of Health, Republic of South Africa. Clinical Guidelines for the Management of HIV & AIDS in Adults and Adolescents. Pretoria, South Africa: National Department of Health; 2010.
10. South African National AIDS Council and National Department of Health. Presented at: Programme Implementation Committee Meeting; January 26, 2011; Pretoria, South Africa.
11. National Department of Health, South Africa. National Antiretroviral Treatment Guidelines. Pretoria, South Africa: National Department of Health; 2004.
12. Drummond M, O'Brien B, Stoddart G, Torrance G, eds. Methods for the Economic Evaluation of Health Care Programmes. 3rd ed. New York, NY: Oxford University Press; 2005.
13. Accounting Standards Board South Africa. Standard of Generally Recognised Accounting Practice: Property Plant and Equipment (GRAP 176). Lynwood Ridge, South Africa: Accounting Standards Board South Africa. 2004.
14. Kumaranayake L. The real and the nominal? Making inflationary adjustments to cost and other economic data. Health Policy Plan. 2000;15:230–234.
15. Long L, Rosen S, Sanne I. Stable outcomes and costs in South African patients' second year on antiretroviral treatment. Presented at: International AIDS Economics Network Symposium; August 1–2 2008; Cuernavaca, Mexico.
16. 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.
17. 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]. Presented at: XVIIth International AIDS Conference; August 3–8, 2008; Mexico City, Mexico.
18. Kevany S, Meintjes G, Rebe K, et al.. Clinical and financial burdens of secondary level care in a public sector antiretroviral roll-out setting (G F Jooste Hospital). S Afr Med J. 2009;99:320–325.
19. Meyer-Rath G, Kumaranayake L, Variava E, et al.. Transport costs of patients accessing antiretroviral treatment (ART) in Gauteng and the North-West province. Presented at: South African AIDS Conference; 2007; Durban, South Africa.
20. Meyer-Rath G, Kumaranayake L, Variava E, et al.. Getting people to the pills: transport costs, socio-economic status and reasons for defaulting from antiretroviral treatment in public sector clinics in South Africa. Presented at: International AIDS Economics Network pre-conference meeting; August 1–2, 2008; Cuernavaca, Mexico.
21. Cleary S. The Costs of the National Strategic Plan on HIV and AIDS & STIs 2007-2011. Pretoria, South Africa: National Department of Health, South Africa; 2007.
22. 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. Presented at: XVIII International AIDS Conference; August 18–23, 2010; Vienna, Austria.
23. 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. Presented at: South African AIDS Conference; June 7–10, 2011; Durban, South Africa.
Keywords:© 2012 Lippincott Williams & Wilkins, Inc.
HIV; economics; middle-income; antiretroviral treatment; cost analysis; rural