Project Partners National Treasury of the Republic of South Africa (RSA) and Health Systems Trust (HST) Candy Day HST Emmanuelle Daviaud - SAMRC

Report (February 2014) For PROJECT Development and application of benchmarks for budgeting of nonnegotiable goods and services for South Africa s Provincial Departments of Health Project Partners National Treasury of the Republic of South Africa (RSA) and Health Systems Trust (HST) Candy Day HST Emmanuelle Daviaud SAMRC Report prepared for: National Treasury of the Republic of South Africa Report prepared by: Health Systems Trust 34 Essex Terrace, Westville 3630 Contact details: Tel.: +27 (0)31 266 9090 Fax: +27 (0)31 266 9199

TABLE OF CONTENTS Summary... 4 Methodology... 4 Results... 6 Conclusion... 8 Background... 9 Setup of the project... 9 Methodology... 9 Results... 11 1. and benchmarks... 11 Childhood vaccines... 11 HIV test kits... 12 HIV: female and male condoms... 12 ARV medicines and laboratory tests... 13 Blood... 15 Laboratory services... 15 Medical... 17 Medicines... 19 Benchmarks summary and comparison with bestperforming hospitals... 21 2. Suggested Budgets per province per NN... 22 Comparison of suggested budgets (1&2) with 2013/14 budgets... 24 3. Conclusion... 25 Annexure: Additional Tables... 26 2

TABLE OF TABLES Table 1. Childhood vaccines protocols and cost... 12 Table 2. ARV medicine and laboratory costs per patient year benchmark... 13 Table 3. Benchmarks summary... 21 Table 4. Suggested Budgets calculation per type of NN... 22 Table 5. Suggested Budgets by NN all levels of care... 23 Table 6. Suggested Budgets per capita... 24 Table 7. Comparison of Budget and Suggested Budgets... 24 Table 8. Vaccines expenditure and expenditure per population under 1 year... 26 Table 9. HIV testing kits expenditure and benchmarks... 26 Table 10. Condoms expenditure and benchmarks... 27 Table 11. ARV medicines expenditure per patient year... 27 Table 12. ARVrelated laboratory expenditure per patient year... 27 Table 13. Blood in acute hospitals expenditure and benchmark per PDE... 28 Table 14. Laboratory services in acute hospitals expenditure and benchmark per PDE... 28 Table 15. Medical in acute hospitals expenditure and benchmark per PDE... 28 Table 16. Medicines in acute hospitals expenditure and benchmark per PDE... 28 Table 17. Benchmarks: comparison with bestperforming hospitals... 29 Table 18. Blood : Suggested Budgets by province by level of care... 30 Table 19. Laboratory services: Suggested Budgets by province by level of care... 31 Table 20. Medical : Suggested Budgets by province by level of care... 32 Table 21. Medicines: Suggested Budgets by province by level of care... 33 Table 22. Difference between Suggested Budgets and Budget... 34 TABLE OF FIGURES Figure 1. on vaccines per child by province... 11 Figure 2. per capita total population on HIV testing kits by province... 12 Figure 3. per 15+ on condoms by province, 2012/13 only... 12 Figure 4. ARV medicines and laboratory expenditure per patient year by province, 2012/13... 14 Figure 5. Blood expenditure per PDE acute hospitals by province... 15 Figure 6. Laboratory services expenditure per PHC headcount by province... 16 Figure 7. Laboratory services expenditure per PDE acute hospitals by province... 17 Figure 8. Medical expenditure per PHC headcount by province... 17 Figure 9. Medical expenditure per PDE acute hospitals by province... 18 Figure 10. Medicines expenditure per PHC headcount... 19 Figure 11. Medicines expenditure per PDE acute hospitals by province... 20 3

Summary This project aims to benchmark budgeting for nonnegotiable goods and services, and is part of a range of projects funded by the Belgian Technical Cooperation Agency (BTC) to support National Treasury in costing and budgeting for National Health Insurance (NHI). The focus on nonnegotiable (NN) goods and services is led by the recognition that the share of total expenditure on goods and services is decreasing: over the past three years, Compensation of Employees increased by 12% whilst spending on goods and services increased by 4%. The Minister of Health seeks to ensure that funds for essential goods and services are protected. The objectives of the project are, therefore, for each selected set of nonnegotiable goods and services, to: quantify the last three years expenditure per province; define benchmarks for each of the nonnegotiables; quantify per province and per level of care the required budget, according to the benchmarks, for the next two years; compare existing budgets with suggested MediumTerm Framework (MTEF) budgets. Methodology Previous years expenditure per NN and suggested budgets are expressed in 2013 Rand. The average of the years 2010/11, 2011/12 and 2012/13 was calculated to represent expenditure per year at provincial level, using a combination of facility classification and budget subprogramme to classify the level of care. Benchmarks per NN are set as relative benchmarks, i.e. which may change over time to ensure affordability and sustainability. They were calculated as follows: barring vaccines and ARTrelated laboratory services and medicines which are costed as per protocol, two bottom benchmark levels were modelled: average (scenario 1) and 60 th centile (scenario 2). This approach to benchmarking implies that the benchmarks are likely to not reflect optimum treatment, but in the context of limited resources it aims at increasing resource allocation for the worstoff provinces. Suggested costs per headcount/pde by province were defined by the two benchmarks for provinces with unit cost below that level. For provinces above that level, their unit cost remains unchanged. Utilisation (number headcounts/pdes per level of care per province) is assumed to remain unchanged over the next two years. For childhood vaccines, increasing coverage towards the target 95% was modelled. For ARV medicines, utilisation projections include the National Department of Health (NDoH) projection of an annual additional 400 000 patients on treatment across the country. Suggested budgets (Budgets 1 & 2) were compared with the 2013/14 budgets for these nonnegotiables. Unit costs for protocolled benchmarks: childhood vaccines, ARTrelated laboratory services and medicines were provided by the NDoH. The selected NNs covered by the project are: Blood Childhood vaccines HIV/AIDS (condoms, HIV test kits, antiretroviral therapy (ART)related medicines and laboratory services) Laboratory services excluding ARTrelated Medical Medicines excluding antiretrovirals (ARVs) 4

Apart from populationrelated benchmarks, benchmarks relate to PHC facilities (Clinics and Community Health Centres) and acute hospitals (District Hospitals DH, Regional Hospitals RH, Tertiary Hospitals TH, Central Hospitals CH). This report includes revised protocol costs for ART medicines and ARTrelated laboratory services which have increased significantly. The worksheet is also included for reference; however, it cannot be used directly with the new figures as manual adjustments have had to be made to deal with data issues in some provinces, overwriting formulae in some areas. 5

Results Benchmarks: A summary of benchmarks per NN, expressed in 2013 Rand, is presented below. These benchmarks are applied to provinces whose expenditure is below the benchmark; in other provinces current expenditure is maintained. Benchmarks per Headcount/PDE NN: Blood NN: Laboratory services excl. ART related NN: Medical NN: Medicines excl ART related PHC Headcount Average 0.07 5.75 3.06 18.90 60th centile 0.07 6.50 3.53 21.17 DH Average 18.33 50.37 60.59 80.50 60th centile 20.92 52.13 66.37 81.02 RH Average 43.00 77.33 107.48 79.58 60th centile 46.78 82.04 116.04 85.57 TH Average 75.19 112.35 210.90 172.75 60th centile 78.68 122.83 258.96 191.64 CH Average 120.26 155.13 380.44 233.20 60th centile 114.07 191.01 427.86 257.59 Vaccines per child <1 HIV (Test kits) Per child <1 1 811.46 Per pop not on ARV HIV (Male and female condoms) Per pop 15+ Average 1.67 60th centile 1.76 Average 4.38 60th centile 5.03 HIV ARV Drugs Per patient year 1 668.49 HIV ARV Labs Per patient year 306.09 In order to view the benchmarks per PDE in perspective, the 2012/13 expenditure per PDE in bestperforming hospitals was extracted. Bestperforming hospitals were identified through the National Facilities Audit carried out by Health Systems Trust for the NDoH in 2011/12; as such, their selection reflects the criteria used in that study which focuses on inputs and processes. More work is required to include outcome data in selection of bestperforming hospitals. Data from bestperforming hospitals show surprisingly wide ranges within each level of care. Comparison between benchmarks and average/60 th centiles of the displayed hospitals shows that the benchmarks are lower than the bestperforming hospitals in many instances, but are also higher in many other instances. Suggested Budgets: From the combined expected level of utilisation and benchmark per province per level of care, the following budgets are suggested for all levels of care combined (breakdown per level of care is shown in the Annexure). The Suggested Budget1 for the country would amount to R1.6 billion for childhood vaccines, R1.4 billion for blood, R4.2 billion for laboratory services (R759 million for ARVrelated laboratory services and R3.5 billion for other laboratory services), R4.6 billion for medical, R10 billion for medicines (R4.1 billion for ARVs and R6 billion for other medicines), and R274 million for HIV test kits and condoms. If the nonprotocolbased benchmark is set at the 60 th increase by 3% (+ R737.5 million). centile, the total of Suggested Budget1 would 6

Suggested Budget1: Benchmark according to Average or Protocol Suggested Budgets in 2013 Rands ('000) Childhood Vaccines NN: Blood NN: Laboratory services excl. ARV related NN: Medical NN: Medicines excl ARV HIV Tests and Condoms ART medicines ART related Labs EC R 203 325 R 157 079 R 379 638 R 507 891 R 757 995 R 31 025 R 435 338 R 79 864 FS R 88 042 R 77 800 R 171 075 R 263 374 R 323 661 R 14 428 R 226 625 R 41 575 GP R 358 509 R 485 890 R 1 149 190 R 1 489 430 R 1 469 091 R 64 156 R 955 232 R 175 240 KZN R 367 962 R 298 834 R 656 781 R 860 002 R 1 371 423 R 54 207 R 1 306 913 R 239 757 LP R 187 266 R 83 416 R 245 494 R 274 382 R 506 908 R 24 448 R 286 159 R 52 497 MP R 113 933 R 71 845 R 178 661 R 192 762 R 439 088 R 22 234 R 380 504 R 69 805 NC R 36 961 R 23 358 R 73 280 R 87 715 R 125 378 R 6 213 R 45 347 R 8 319 NW R 100 448 R 44 297 R 144 348 R 152 743 R 295 761 R 18 794 R 254 734 R 46 732 WC R 175 717 R 185 908 R 487 698 R 766 848 R 761 452 R 38 254 R 247 674 R 45 437 Total R 1 632 163 R 1 428 426 R 3 486 164 R 4 595 146 R 6 050 757 R 273 759 R 4 138 526 R 759 226 To consider the Suggested Budgets in perspective, these projections are expressed below in expenditure per capita. Demographic structure, HIV prevalence and levels of utilisation largely explain differences between provinces. Comparison with budgets 2013/14: Comparison of NDoH budgets (October 2013) for the NNs studied with the Suggested Budget1 shows an underfunding of all NNs except for laboratory services, but only if this category in the budget excludes ARVrelated laboratory expenditure. Overall, the budgets are lower than Suggested Budget1 (benchmark: Average) by a total amount of R3.5 billion, representing an additional 2.5% on the total budget. The large shortfall in the HIV/AIDS Conditional Grant (R1.5 billion) can be partly explained by a reassessment of the cost of protocols for ART which saw a significant increase. This shortfall should in fact be higher, as some elements of the HIV/AIDS Conditional Grant (staff, equipment, others) are not costed in the Suggested Budget1. Medicines and childhood vaccines are also projected to have large shortfalls. ('000) Childhood Vaccines Difference: 201314 Budget with Suggested Budget1 (Benchmark: Average) NN: Blood NN: Laboratory services excl ARV related NN: Medical NN: Medicines excl ARV related HIV & AIDS : Conditional Grant Total 879 405 69 728 417 455 467 342 1 188 138 1 325 380 3 512 537 Total Positive values Negative values Budget higher than Suggested Budget Budget lower than Suggested Budget 7

Conclusion With pressure on budgets, and the large size of fixed expenditure (e.g. on staff), the share of the total budget spent on goods and services has decreased over the years, which constitutes a serious threat to quality of care. Analysis of the past expenditure per NN illustrated significant data quality issues, but overall showed large variations between provinces, with some provinces at worryingly low levels of expenditure. Some variations in expenditure may reflect data quality issues. In some cases, low/high expenditure may be partly due to miscoding of expenditure by item, or inaccurate allocation of expenditure to individual facilities and level of care. For example, the total expenditure on vaccines per province stands at R825 million in 2012/13, whilst according to the NDoH, the figure is R1.2 billion, which indicates coding issues. The same coding issues may affect the breakdown of the budgets. Data from some provinces were much more consistent and comprehensive than in others. Variation in hospital expenditure also partly reflects the issue of mixed levels of care in hospitals, as in combos (KZN) or hospital complexes (EC). Large outliers need to be investigated to assess whether these are justified by different scopes of services due to the local situation or due to other issues, including inefficiencies. Childhood vaccine costs as per protocol appear very high, with the combined impact of wastage factors, largely due the packaging of vaccines in multidose vials, adding 50% to the cost per dose. Cost levels for ARVrelated drugs and laboratory services were revised and increased substantially in the reference model, resulting in high suggested budgets for these NNs. The setting of benchmarks for expenditure per unit of activity per level of care (for nonprotocolled NNs) was aimed at lifting the more underresourced provinces to the average level between provinces and the 60th centile of the provinces, whilst maintaining the level of expenditure in betterresourced provinces. This approach assists in mapping out an incremental frameworkfor benchmarking, with higher benchmark levels being introduced over time whilst reducing disparities between provinces. This approach to benchmarking, despite data limitations, could assist provinces and national departments in moving towards better and more equitable funding as well as better resource allocation within budgets. 8

Background South Africa has embarked on reengineering its health system based on a primary health care approach which focuses clear emphasis on promotive and preventive health care, while concomitantly implementing National Health Insurance (NHI) which aims to ensure that everyone has access to appropriate, efficient and goodquality health services. Strengthening the PHC approach and improving service provision will entail major changes in health service delivery structures, administrative processes and management systems, and these efforts are underpinned by the commitment inherent in the National Department of Health (NDoH) Negotiated Service Delivery Agreement (NSDA). This project aims to benchmark budgeting for nonnegotiable health goods and services (NNs) and is part of a range of projects funded by the Belgian Technical Cooperation Agency (BTC) to support National Treasury in costing and budgeting for NHI. The focus on nonnegotiable goods and services is led by the recognition that the share of total expenditure on goods and services has decreased over the past three years. Compensation of employees increased by 12%, whilst spending on goods and services increased by 4%. The Minister of Health seeks to ensure that funds for essential goods and services are protected. The project objectives are, therefore, for each selected set of nonnegotiable goods and services, to: quantify the last three years expenditure per province; define benchmarks for each of the nonnegotiables; quantify per province and per level of care the required budget, according to the benchmarks, for the next two years compare existing Mediumterm Framework (MTEF) budgets with suggested MTEF budgets. Setup of the project A wideranging series of meetings and consultations was convened in order to define the scope, priorities and methodology of the investigation. Methodology This study has been carried out in consultation with Treasury and the NDoH, and the approach and methodology used were agreed with both Ministries. Previous years expenditure per NN is expressed in 2013 Rand. The average of the years 2010/11, 2011/12 and 2012/13 was calculated to represent expenditure per year at provincial level, using DHIS facility classification, followed by the budget subprogramme to classify the level of care. is expressed by headcount for clinics and community health centres, and by PDE for hospitals. Vaccines expenditure is expressed as vaccine expenditure per child under 1: although some vaccination takes place at 18 months, 6 years and 12 years, 88% of the target cost of vaccines is for children under 1 year of age. For HIV and AIDS NNs, expenditure on test kits was calculated as expenditure per capita total population, and expenditure on male and female condoms was calculated as expenditure per capita 15 and over. on antiretrovirals (ARVs) is expressed as ARV medicine cost per patient year. Benchmarks per NonNegotiable. Benchmarks are set as relative benchmarks, i.e. which may change over time to ensure affordability and sustainability. Barring vaccines and ARV medicines, two benchmark levels were modelled: average (scenario 1) and 60 th centile (scenario 2). This approach to benchmarking implies that the benchmarks are likely to not reflect optimum treatment, but in the context of limited resources, the calculation aims at increasing resource allocation for the worstoff provinces. Suggested costs per headcount/pde by province were defined by the two bottom benchmarks for provinces with unit costs below that level. For provinces above that level, their unit costs remain unchanged. A comparison of unit costs was made between modelled unit costs and unit costs of topperforming facilities according to the National Facilities Audit performed in 2011/12. 9

Utilisation (number headcounts/pdes per level of care per province) is assumed to remain unchanged over the next two years. For childhood vaccines, the provincial coverage was compared against the target 95%, and an annual increase in the provincial coverage of the projected under 1 population at 2% a year was modelled for provinces under the 95% target. For ARV medicines and laboratory tests, utilisation projections include the NDoH projection of an annual additional 400 000 patients on treatment across the country, spread between provinces using the current distribution of ARV patients across provinces. Each nonnegotiable thus has an average yearly expenditure per province per unit of analysis as relevant (headcount, PDE, population category). Cost per headcount and per PDE is broken down per level of care, whilst expenditure for HIV and AIDS and vaccines is presented for all levels of care combined, although these costs are essentially carried at PHC level. Suggested provincial budgets (Scenario 1 & 2) per NN per level of care were modelled by applying modelled unit costs to the projected utilisation level. Budget per capita per NN was calculated. Comparison of suggested and existing budgets for the next two years was made and NNs respective share of total budget was calculated. For large outliers, with current unit costs largely over the province below the average level possibly pointing to inefficiency, the suggested budget is modelled using the province below unit cost, and possible savings are quantified. Information was extracted from the following sources: DHIS was the source of data for the categorisation of facilities, number of headcounts, PDEs, number of patient years on ARV treatment, total population, population 15+ and number of children under 1. data were extracted from the national Basic Accounting System (BAS) database provided by Treasury. Existing MTEF budgets for each province were extracted from NDoH data dated 23 August 2013. Benchmarks as per protocols: vaccines and ARV medicine and laboratory costs were provided by the NDoH. Existing budgets were provided by the NDoH and National Treasury. Some data problems were encountered in the analysis of provincial expenditure and were dealt with in the following way: data from financial subprogrammes showed that 20% of the expenditure by NN could not be allocated by level of care. Data were thus extracted by facility (referencing financial subprogramme under which these expenditures were coded). not linked to facilities was then allocated to level according to the budget subprogramme. This process drastically reduced the expenditure which could not be allocated by level of care. Some provinces had data missing for some years, and in these instances, the average of the data available for the other years was used. Some provinces have very low or no values for some items; for example, e.g. medicines in Limpopo, are covered under a different subprogramme in the provincial accounts, by all levels of care combined. This Province s data were excluded from the calculation of the benchmark. Given the time limitations of the project, it was agreed with Treasury and the NDoH that a subset of NNs would be covered in the study. 10

The selected NNs are as follows: Results Blood Childhood vaccines HIV (condoms, tests, ARVs) Laboratory services Medical Medicines The results are presented in two groups: Past expenditure and benchmarks: yearly expenditure per province per unit of analysis as relevant (headcount, PDE, population category) for the past three financial years expressed in 2013 Rand, the average of the yearly values, and the benchmarks as average and 60 th centile of the provincial average values Comparison of suggested and existing budgets 1. and benchmarks per province and benchmarks are presented in the following sections via graphs. Detailed tables are presented in the Annexure. Childhood vaccines Recorded expenditure for South Africa increased over the three years under investigation and reached R825.5 million in 2012/13, or an average of R854 in 2012/13 per child under 1. For the purpose of the study and ease of reference given that the breakdown of actual vaccine expenditure per age is not available, all childhood vaccines are expressed per child under 1. However, R1.2 billion is budgeted for this NN at a national level; this discrepancy may point to the fact that vaccine expenditure may be recorded under other codes. As a consequence, expenditure on vaccines per child under 1 ranges from R108 in Gauteng to R1 441 in KZN. Figure 1. on vaccines per child by province The national benchmark has been set according to protocol regarding the number of doses required for each component. Cost by dose includes the wastage factor. The wastage factor varies with the packaging of the vaccine: for the BCG packaged by 20 doses, the wastage factor is 80%, whilst Rotavirus, packaged by single dose, has a wastage factor of 5%. Overall, wastage factors increase the cost by 50%. The cost of vaccines stands at a total of R1 811.5 per child, 80% of which is for children under 1. The cost per child is high and the NDoH may need to reassess its reference costs and wastage factors. 11

Table 1. Childhood vaccines protocols and cost Number Doses per Child Cost per Child Name Price Units Unit cost Wastage Cost <1 18 mths 6 Yrs 12 Yrs <1 18 mths 6 Yrs 12 Yrs BCG 33.21 20 1.7 80% 3.0 1 3.0 Oral Polio 27.05 10 27.1 40% 37.9 2 75.7 DaPTIPVHib 127.60 1 127.6 5% 134.0 3 1 401.9 134.0 HepB 50.59 10 50.6 25% 63.2 3 189.7 Rotavirus 73.90 1 73.9 5% 77.6 2 155.2 PCV 166.35 1 166.4 5% 174.7 3 524.0 Measles 61.57 10 61.6 50% 92.4 1 1 92.4 92.4 Td 57.28 10 57.3 25% 71.6 1 1 0.0 71.6 71.6 Total 2013 Rds 1441.9 226.3 71.6 71.6 1811.5 HIV test kits on HIV testing kits decreased by 53% between 2010/11 and 2011/12, and increased again by 23% in 2012/13. Higher expenditure in 2010/11 was linked to the testing campaign, which may be restarted. per capita across provinces ranges from R1.11 in the Free State to R2.28 in the Northern Cape, higher than in KZN (2.13). Figures for Limpopo Province were not available. The average and 60 th centile, which will be used for benchmarks for the provinces with lowest expenditures, stand at R1.67 and R1.76 respectively. Figure 2. per capita total population on HIV testing kits by province HIV: female and male condoms for 2010/11 was available for only a few provinces. The 2011/12 expenditure increased threefold in 2012/13, and is likely to represent increased use of condoms; the benchmark will thus be calculated not from the average of 2011/12 and 2012/13, but from 2012/13 only. The average and 60 th centile stand at R4.38 and R5.03 respectively. Figure 3. per 15+ on condoms by province, 2012/13 only 12

ARV medicines and laboratory tests The benchmarks according to protocol and expenditure are presented by patient years on treatment. The number of patient years was calculated from the monthly numbers of patients on treatment (District Health Information System DHIS), in order to reflect use of ARVs by patients who dropped out before the end of the year and by those who started later during the year. In the protocol, the cost of ARV medicine by patient years stands at R1 668 and for ARTrelated laboratory services at R306. Table 2. ARV medicine and laboratory costs per patient year benchmark ARV Drugs Costs 2013/14 2012/13 ARVPatients Year Adults 1 680 94% Children 1 491 6% Average per patient year 1 668 100% ARV Labs/diagnostics Costs 2013/14 2012/13 ARVPatients Year Adults 304 94% Children 333 6% Average per patient year 306 100% MeyerRath, G. 2013. National ART Cost Model, South Africa. Health Economics and Epidemiology Research Office, Boston University/ University of the Witwatersrand, Johannesburg. The year 2012/13 is used as reference to reflect recent protocols and costs of ARV medicines and laboratory services. Costs presented here reflect expenditure for these items coded under Conditional Grant (subprogramme 2.6). For ARV medicines, the country average cost per patient year stands at R1 669 in 2012/13, although variations can be observed between provinces, ranging from R1 202 in KZN to R1 744 in the Free State. The Northern Cape Province is a large outlier at R3 038. Surprisingly, the country average is identical to the protocol. For ARTrelated laboratory tests, the country average expenditure stood at R700, considerably higher than the protocol level of R306. This may point to other laboratory expenditures being coded under subprogramme 6, affecting in turn the analysis of nonartrelated laboratory expenditure and benchmarks under other programmes. Under subprogramme 2.6, laboratory services expenditure ranged from R215 in the Western Cape to R1 176 in the Northern Cape. 13

Figure 4. ARV medicines and laboratory expenditure per patient year by province, 2012/13 14

Blood Blood for PHC were recorded in few provinces, and in these provinces the average cost per PHC headcount stood at R0.07 and at R0.07 for the 60 th centile. In hospitals, expenditure per PDE ranged from an average of R18.33 in district hospitals, to R43.00 in regional hospitals, R75.19 in tertiary hospitals and R120.06 in central hospitals. Low expenditure in district hospitals is likely to reflect the fact that some hospitals do not perform surgery due to a shortage of doctors. The 60 th centile stands at R20.92 in district hospitals, R46.78 in regional hospitals and R78.68 in tertiary hospitals, and at R114.07 in central hospitals. Figure 5. Blood expenditure per PDE acute hospitals by province District Hosps WC Average 60th Centile 23 Regional Hosps WC Average 60th Centile 44 NW 13 NW 47 NC 14 NC 33 MP 25 MP 49 LP 21 LP 34 KZN 15 KZN 49 GP 29 GP 57 FS 17 FS 47 EC 9 EC 28 0 5 10 15 20 25 30 0 10 20 30 40 50 60 Tertiary Hosps NC Average 67 60th Centile Central Hosps 60th Centile Average WC 114 MP 96 KZN 207 LP KZN 66 86 GP 101 EC 61 FS 59 0 20 40 60 80 100 120 0 50 100 150 200 250 Laboratory services Due to the issue of a backlog in payment for services to the National Health Laboratory Services (NHLS) in KwaZuluNatal Province (KZN), the artificially low cost of this NN in KZN was excluded from calculations at PHC and Acute Hospital level in order to avoid distortions. NonARVrelated laboratory expenditure ranged from R2.04 per PHC headcount in the FreeState to R11.15 in Gauteng. The average of the provinces stands at R5.75 and the 60 th centile at R6.50. However, when all PHC level laboratory expenditure, including that from the HIV/AIDS Conditional Grant, is taken into account, the picture is very different, with Free State, North West, Northern Cape and Mpumalanga having the highest level of expenditure. 15

Figure 6. Laboratory services expenditure per PHC headcount by province In Acute Hospitals, laboratory expenditure per PDE ranged from an average of R50.37 in district hospitals, to R76.47 in regional hospitals, R112.35 in tertiary hospitals and R155.13 in central hospitals. The 60th centile stands at R52.13 in district hospitals, R82.04 in regional hospitals and R122.83 in tertiary hospitals, and at R191.01 in central hospitals. Gauteng Province has the highest level of expenditure for all levels of hospitals. 16

Figure 7. Laboratory services expenditure per PDE acute hospitals by province District Hosps WC Average 60th Centile 70 Regional Hosps WC Average 60th Centile 89 NW 59 NW 99 NC 37 NC 64 MP 48 MP 77 LP 31 LP 56 KZN KZN GP 106 GP 119 FS 22 FS 61 EC 29 EC 53 0 20 40 60 80 100 120 0 20 40 60 80 100 120 140 Tertiary Hosps NC Average 60th Centile 123 Central Hosps WC Average 60th Centile 200 MP 150 KZN LP KZN 66 GP 185 EC 110 FS 80 0 50 100 150 200 0 50 100 150 200 250 Medical Due to data problems, expenditure for Limpopo Province could not be included. Medical expenditure ranged from R2.9 per PHC headcount in the Eastern Cape to R6.3 in the Western Cape. The average of the provinces and the 60 th centile were similar at R4.4. Figure 8. Medical expenditure per PHC headcount by province In Acute Hospitals, expenditure per PDE on medical ranged from an average of R60.6 in district hospitals, to R107.5 in regional hospitals, R210.9 in tertiary hospitals and R380.4 in central hospitals. The 60 centile stands at R66.4 in district hospitals, R116 in regional hospitals and R259 in tertiary hospitals, and at R427.9 in central hospitals. The Eastern Cape Province had a low level of expenditure in all hospitals. 17

Figure 9. Medical expenditure per PDE acute hospitals by province District Hosps WC Average 60th Centile 89 Regional Hosps WC Average 60th Centile 160 NW 53 NW 115 NC 82 NC 104 MP 56 MP 76 LP LP KZN 53 KZN 108 GP 82 GP 118 FS 37 FS 130 EC 32 EC 49 0 20 40 60 80 100 0 50 100 150 200 Tertiary Hosps NC Average 60th Centile 259 Central Hosps WC Average 60th Centile 428 MP 193 KZN 510 LP KZN EC 128 263 GP FS 261 322 0 50 100 150 200 250 300 0 100 200 300 400 500 600 18

Medicines Due to data problems, expenditure for Limpopo Province could not be included. Medicines expenditure (excluding ARVs) ranged from R12 per PHC headcount in KZN to R25.3 in Mpumalanga. The average of the provinces stood at R18.9 and the 60 th centile was similar at R21.2. If expenditure from the HIV/AIDS Conditional Grant is added in All PHC Combined, KZN expenditure per headcount remains surprisingly the lowest at R13, with Gauteng being the highest at R32 per PHC headcount. The Western Cape shows almost no variation, indicating that the Conditional Grant had been allocated to PHC facilities. Figure 10. Medicines expenditure per PHC headcount In acute hospitals, expenditure per PDE on medicines showed wide variations between provinces, as well as lower cost in regional hospitals than in district hospitals in North West, Mpumalanga and Gauteng Provinces. For district hospitals, spending ranged from a threeyear average of R41.01 in the Eastern Cape to R109 in Mpumalanga. The North West is an outlier at R182 per PDE. These high costs in district hospitals may indicate that some medicines expenditure for PHC is in fact reflected in district hospitals. The average across provinces stood at R80.50 and the 60 th centile at R81.02. In regional hospitals, the lowest cost is observed in Western Cape at R51.05, rising to R111.81 in KZN. The average across provinces stood at R79.58 and the 60 th centile at R85.57. In tertiary hospitals, expenditure per PDE ranged from R80.04 in the Eastern Cape to R260.98 in KZN. The average across provinces stood at R172.75 and the 60 th centile at R191.64. In central hospitals, the threeyear average ranged from R150.05 in Gauteng to R338.01 in KZN. The average across the four provinces stood at R233.20 and the 60 th centile at R257.59. 19

Figure 11. Medicines expenditure per PDE acute hospitals by province District Hosps WC Average 43 60th Centile Regional Hosps WC Average 51 60th Centile NW 182 NW 86 NC 44 NC 68 MP 109 MP 103 LP LP KZN 59 KZN 112 GP 92 GP 73 FS 74 FS 85 EC 41 EC 58 0 50 100 150 200 0 20 40 60 80 100 120 Tertiary Hosps NC Average 60th Centile 192 Central Hosps WC Average 187 60th Centile MP 158 KZN 338 LP KZN EC 80 261 GP FS 150 258 0 50 100 150 200 250 300 0 100 200 300 400 20

Benchmarks summary and comparison with bestperforming hospitals A summary of benchmarks per NN is presented below. These benchmarks are applied to provinces whose expenditure is below the benchmark; in other provinces current expenditure is maintained. Table 3. Benchmarks summary Benchmarks per Headcount/PDE NN: Blood NN: Laboratory services excl. ART related NN: Medical NN: Medicines excl ART related PHC Headcount Average 0.07 5.75 3.06 18.90 60th centile 0.07 6.50 3.53 21.17 DH Average 18.33 50.37 60.59 80.50 60th centile 20.92 52.13 66.37 81.02 RH Average 43.00 77.33 107.48 79.58 60th centile 46.78 82.04 116.04 85.57 TH Average 75.19 112.35 210.90 172.75 60th centile 78.68 122.83 258.96 191.64 CH Average 120.26 155.13 380.44 233.20 60th centile 114.07 191.01 427.86 257.59 Vaccines per child <1 HIV (Test kits) Per child <1 1 811.46 Per pop not on ARV HIV (Male and female condoms) Per pop 15+ Average 1.67 60th centile 1.76 Average 4.38 60th centile 5.03 HIV ARV Drugs Per patient year 1 668.49 In order to viewthe benchmarks per PDE in perspective, 2012/13 expenditure per PDE in bestperforming hospitals was extracted. Bestperforming hospitals were identified through the National Facilities Audit carried out by Health Systems Trust (HST) for the NDoH in 2011/12. Data from bestperforming hospitals show surprisingly wide ranges within each level of care. Comparison between benchmarks and average/60 th centiles of the displayed hospitals shows that the benchmarks are lower than the bestperforming hospitals in many instances, but are also higher in many other instances: District hospitals: the benchmarkaverage expenditure per PDE is lower than the bestperforming hospitals average by 7% for blood, lower by 17% for laboratory services, higher by 1% for medical, and higher by 3% for medicines. The benchmark60 th centile is higher than the bestperforming hospitals 60 th centile by 2% for blood, lower by 34% for laboratory services, lower by 24% for medical, and higher by 6% for medicines. Regional hospitals: the benchmarkaverage expenditure per PDE is lower than the bestperforming hospitals average by 18% for blood, lower by 32% for laboratory services, lower by 25% for medical, and lower by 12% for medicines. The benchmark60 th centile is lower than the bestperforming hospitals 60 th centile by 21% for blood, lower by 9% for laboratory services, lower by 4% for medical, and lower by 6% for medicines. Tertiary hospitals: the benchmarkaverage expenditure per PDE is lower than the bestperforming hospitals average by 17% for blood, lower by 89% for laboratory services, lower by 10% for medical, and higher by 3% for medicines. The benchmark60 th centile is lower than the bestperforming hospitals 60 th centile by 32% for blood, lower by 84% for laboratory services, higher by 11% for medical, and higher by 17% for medicines. Central hospitals: the benchmarkaverage expenditure per PDE is higher than the bestperforming hospitals average by 2% for blood, lower by 15% for laboratory services, higher by 11% for medical, and higher by 31% for medicines. The benchmark60 th 21

centile is higher than the bestperforming hospitals 60 th centile by 5% for blood, similar for laboratory services, higher by 18% for medical, and higher by 43% for medicines. 2. Suggested Budgets per province per NN Suggested budgets in 2013 Rand were calculated as shown in the following examples, and tables are displayed in the Annexure. As explained in the methodology section, different methods are used for different NNs: Table 4. Vaccines: increasing coverage for provinces under the 95% target, maintaining coverage for those at this level, and the benchmark is as per protocol. Coverage figures are based on DHIS population, currently used by the NDoH. Data from Census 2011 give lower coverage figures. ARVs: the benchmark is as per protocol, and increasing number of patient years to reflect additional 400 000 a year for South Africa. HIV test kits: the target population excludes those already on ARVs, and the benchmark is based on average cost per capita across provinces and 60 th centile. Blood, laboratory services, medical and medicines: utilisation (headcounts/pdes) remains the same over the next two years. The benchmark is based on average cost per headcount/pde across provinces and 60 th centile. Where actual expenditure is higher than the benchmark, actual expenditure is used as the benchmark. Suggested Budgets calculation per type of NN For NNs with benchmarks as per protocol Actual Suggested Coverage Population <1 (DHIS) Benchmark Vaccines Coverage 2013 2014 2015 2014 2015 per <1 201314 201415 (DHIS) EC 82.6 85 88 132 051 134 759 R 203 325 R 214 818 R 1 811.46 FS 95.1 95.1 95.1 51 106 49 191 R 88 042 R 84 743 ARV Drugs Patients Years 2013 % Patients Years 2013 Distribution Additional 400 000 per year for SA Patients Years 2014 Patients Years 2015 Benchmark per Patient Year 201314 201415 EC 218 840 11% 42 077 260 917 302 994 R 1 668.49 R 435 338 R 505 543 FS 113 922 5% 21 904 135 826 157 730 R 226 625 R 263 172 For NNs with budgets calculated from benchmark per unit of activity Laboratory services District Hospitals PDEs per PDE Benchmark per PDE Scenario 1: Average Scenario 2: 60th Centile EC 1 870 399 1 870 399 1 870 399 29.35 R 50.37 R 52.13 R 94 209 R 94 209 R 97 497 R 97 497 WC 1 266 675 1 266 675 1 266 675 70.42 R 70.42 R 70.42 R 89 195 R 89 195 R 89 195 R 89 195 Combining the information per NN per level of care, the suggested budget in 2013 Rand is presented by province, using the two benchmark approaches, all levels of care combined to match the format used by the NDoH. Note that these budgets do not reflect expenditure on specialised hospitals. The suggested Budget1 for the country would amount to R1.6 billion for childhood vaccines, R1.4 billion for blood, R4.2 billion for laboratory services (R 759 million for ARVrelated laboratory services and R3.5 billion for other laboratory services), R4.6 billion for medical, R10 billion for medicines (R4.1 billion for ARVs and R6 billion for other medicines) and R274 million for HIV test kits and condoms. If the nonprotocolbased benchmark is set at the 60 th increase by 3% (+ R737.5 million). centile, the total of Suggested Budget1 would 22

Table 5. Suggested Budgets by NN all levels of care Suggested Budget1: Benchmark according to Average or Protocol Suggested Budgets in 2013 Rands ('000) Childhood Vaccines NN: Blood NN: Laboratory services excl. ARV related NN: Medical NN: Medicines excl ARV HIV Tests and Condoms ART medicines ART related Labs EC R 203 325 R 157 079 R 379 638 R 507 891 R 757 995 R 31 025 R 435 338 R 79 864 FS R 88 042 R 78 018 R 171 075 R 263 374 R 323 661 R 14 428 R 226 625 R 41 575 GP R 358 509 R 485 854 R 1 149 190 R 1 489 430 R 1 469 091 R 64 156 R 955 232 R 175 240 KZN R 367 962 R 298 812 R 656 781 R 860 002 R 1 371 423 R 54 207 R 1 306 913 R 239 757 LP R 187 266 R 83 416 R 245 494 R 274 382 R 506 908 R 24 448 R 286 159 R 52 497 MP R 113 933 R 71 845 R 178 661 R 192 762 R 439 088 R 22 234 R 380 504 R 69 805 NC R 36 961 R 26 601 R 73 280 R 87 715 R 125 378 R 6 213 R 45 347 R 8 319 NW R 100 448 R 44 297 R 144 348 R 152 743 R 295 761 R 18 794 R 254 734 R 46 732 WC R 175 717 R 186 116 R 487 698 R 766 848 R 761 452 R 38 254 R 247 674 R 45 437 Total R 1 632 163 R 1 432 036 R 3 486 164 R 4 595 146 R 6 050 757 R 273 759 R 4 138 526 R 759 226 Suggested Budget2: Benchmark according to 60th Centile or Protocol Suggested Budgets in 2013 Rands ('000) NN: Childhood Vaccines NN: Blood NN: Laboratory services excl. ARV related NN: Medical NN: Medicines excl ARV HIV Tests and Condoms ART medicines ART related Labs EC R 203 325 R 167 942 R 412 920 R 598 810 R 828 314 R 34 591 R 435 338 R 79 864 FS R 88 042 R 77 933 R 191 564 R 280 966 R 341 224 R 15 458 R 226 625 R 41 575 GP R 358 509 R 469 701 R 1 149 190 R 1 613 013 R 1 548 223 R 65 134 R 955 232 R 175 240 KZN R 367 962 R 300 877 R 714 374 R 917 458 R 1 442 889 R 59 030 R 1 306 913 R 239 757 LP R 187 266 R 86 565 R 265 393 R 312 111 R 481 815 R 27 247 R 286 159 R 52 497 MP R 113 933 R 71 845 R 189 062 R 209 257 R 395 042 R 22 550 R 380 504 R 69 805 NC R 36 961 R 28 109 R 73 996 R 90 316 R 133 712 R 6 745 R 45 347 R 8 319 NW R 100 448 R 45 433 R 151 098 R 156 286 R 313 636 R 19 477 R 254 734 R 46 732 WC R 175 717 R 180 640 R 487 698 R 766 848 R 792 096 R 38 764 R 247 674 R 45 437 Total R 1 632 163 R 1 429 046 R 3 635 295 R 4 945 065 R 6 276 954 R 288 995 R 4 138 526 R 759 226 SA Childhood Vaccines NN: Blood Difference Suggested Budget2 and Budget 1 ('000) NN: Laboratory services excl. ARV related NN: Medical NN: Medicines excl ARV HIV Tests and Condoms ART medicines ART related Labs Budget2 Budget1 R R 2 991 R 149 131 R 349 919 R 226 197 R 15 236 R R R 737 492 % Variation 0% 0.2% 4% 8% 4% 6% 0% 0% 3% Total In order to understand the variation in suggested budgets per province, utilisation level and budget per capita are presented in the following table. Suggested budgets are a function of the demographic structure and size of population (childhood vaccines), HIV prevalence, level of utilisation and previous expenditure by headcount or PDE. 23

Table 6. Suggested Budgets per capita Suggested Budgets in 2013 Rands Provincial PHC Headcounts/ Capita Total Population Acute Hospitals PDEs/ Capita Childhood Vaccines NN: Blood NN: Laboratory services excl. ARV related Suggested Budget1 per Capita Total Population NN: Medical NN: Medicines excl ARV HIV Tests and Condoms ART medicines ART related Labs EC 2.62 0.53 30.3 23.4 56.6 75.7 113.0 4.6 64.9 11.9 FS 2.54 0.53 29.8 26.3 57.9 89.1 109.5 4.9 76.7 14.1 GP 2.04 0.53 31.4 42.6 100.7 130.5 128.8 5.6 83.7 15.4 KZN 2.86 0.62 33.9 27.5 60.5 79.2 126.3 5.0 120.4 22.1 LP 2.70 0.48 35.2 15.7 46.1 51.5 95.2 4.6 53.7 9.9 MP 2.44 0.48 30.5 19.2 47.9 51.6 117.6 6.0 101.9 18.7 NC 2.92 0.40 31.6 20.0 62.7 75.0 107.3 5.3 38.8 7.1 NW 2.23 0.34 28.3 12.5 40.6 43.0 83.2 5.3 71.7 13.1 WC 2.56 0.50 30.2 32.0 83.9 131.9 131.0 6.6 42.6 7.8 Total 2.51 0.52 31.7 27.7 67.7 89.2 117.4 5.3 80.3 14.7 Suggested Budgets in 2013 Rands Childhood Vaccines NN: Blood NN: Laboratory services excl. ARV related Suggested Budget2 per Capita Total Population NN: Medical NN: Medicines excl ARV HIV Tests and Condoms ART medicines ART related Labs EC 30.3 25.0 61.6 89.3 123.5 5.2 64.9 11.9 FS 29.8 26.3 64.8 95.1 115.4 5.2 76.7 14.1 GP 31.4 41.2 100.7 141.4 135.7 5.7 83.7 15.4 KZN 33.9 27.7 65.8 84.5 132.9 5.4 120.4 22.1 LP 35.2 16.3 49.8 58.6 90.5 5.1 53.7 9.9 MP 30.5 19.2 50.6 56.1 105.8 6.0 101.9 18.7 NC 31.6 21.3 63.3 77.3 114.4 5.8 38.8 7.1 NW 28.3 12.8 42.5 44.0 88.2 5.5 71.7 13.1 WC 30.2 31.0 83.9 131.9 136.3 6.7 42.6 7.8 Total 31.7 27.7 70.6 96.0 121.8 5.6 80.3 14.7 Comparison of suggested budgets (1&2) with 2013/14 budgets Comparison of suggested budgets with 2013/14 budgets is made difficult by the fact that some categories do not have similar boundaries between the budget from the NDoH (October 2013) and the categories of NNs studied. Some categories are similar: childhood vaccines, blood and medical. Laboratory services in the budget amounts to R3.9 billion; in the Suggested Budget1, laboratory services excluding ARVrelated stands at R3.5 billion, but ARVrelated laboratory services is R759 million, putting laboratory services at R4.2 billion. The same issue arises for medicines. The category Medicines in the budget amounts to R4.8 billion; in the Suggested Budget1, medicines excluding ARVs stands at R6 billion, but ARV drugs is R4.1 billion, putting medicines at R10.2 billion. On the other hand, the budget displays the amount for HIV/AIDS Conditional Grant. The HIV/AIDS Conditional Grant covers the NNs: HIV testing kits, condoms, ARVrelated laboratory services and ARV drugs, but it also covers some staff expenditure, equipment and other spending. So the sum of the NNs, studied in this report, in the suggested Budget1 (R5.1 billion) would be expected to be smaller than the Conditional Grant budget, which in fact stands at R3.85 billion. Overall, the budgets are lower than Suggested Budgets1 by a total amount of R3.5 billion, representing an additional 2.5% on the total budget. This excludes the additional amount for the areas in the Conditional Grant not covered in the NNs analysed in this report. Table 7. Comparison of Budget and Suggested Budgets ('000) Childhood Vaccines Difference: 201314 Budget with Suggested Budget1 (Benchmark: Average) NN: Blood NN: Laboratory services excl ARV related NN: Medical NN: Medicines excl ARV related HIV & AIDS : Conditional Grant Total 879 405 69 728 417 455 467 342 1 188 138 1 325 380 3 512 537 Total Positive values Negative values Budget higher than Suggested Budget Budget lower than Suggested Budget Detailed tables with provincial breakdown are presented in the Annex. The 2013/14 budgets include specialised hospitals, which are not included in the suggested budgets. However, these hospitals represent a very small component of the NNs expenditure. Based on 2012/13 expenditure, PHC and acute hospital care represent 99.7% of blood, 99% of laboratory services, 24

98.9% of medical and 95.8% of medicines. Suggested budgets can then be compared with total budgets without creating major distortions. 3. Conclusion With pressures on budgets, and the large size of fixed expenditure (e.g. on staff), the share of the total budget spent on goods and services has decreased over the years, leading to the call by Minister Motsoaledi to protect selected goods and services to improve the quality of health care: the nonnegotiables. This study has defined two sets of bottom benchmarks for selected NNs: one benchmark based on the average of the provincial expenditure over the last three years, and the second one on the 60 th centile of these provinces. Analysis of the past expenditure per NN showed large variations between provinces, with some provinces at a worryingly low level of expenditure. Some variations in expenditure may reflect data quality issues. In some cases, low/high expenditure may be partly due to miscoding of expenditure by item, or inaccurate allocation of expenditure to individual facilities and level of care. For example, the total expenditure on vaccines per province stands at R825 million in 2012/13, while according to the NDoH figures, the amount is R1.2 billion, which suggests that there might be coding issues. The same coding issues may affect the breakdown of the budgets. Data from some provinces were much more consistent and comprehensive than in others. Variation in hospital expenditure also partly reflects the issue of mixed levels of care in hospitals, as in combos (KZN) or hospital complexes (EC). Large outliers need to be investigated to assess whether these are justified by different scopes of services due to the local situation or due to other issues, including inefficiencies. Suggested budgets were calculated for each selected nonnegotiable. Benchmarks were set by protocols for childhood vaccines, ARV drugs and ARVrelated laboratory services. For blood Supplies, laboratory services excluding ARVrelated, medical and medicines excluding ARVs, benchmark levels were set in a way which would allow lifting the leastresourced provinces to the level of the average expenditure per headcount/pde across the past three years and across provinces. Provinces with higher expenditure would keep their previous level of expenditure per unit of activity per level of care. These benchmarks do not include additional expenditures linked to campaigns. A second set of benchmarks was based on protocols for some NNs, and on the 60 th centile of the average three years expenditure per province. The Suggested Budget based on this second set of benchmarks would be 3% higher than that based on the first set of benchmarks. Comparison of budgets for the NNs studied with the Suggested Budget1 shows an underfunding for all NNs except for laboratory services, but only if this category in the budget excludes ARVrelated laboratory expenditure. Overall, the budgets are lower than Suggested Budgets1 (benchmark: Average) by a total amount of R3.5 billion, representing an additional 2.5% on the total budget. The shortfall should in fact be higher, as it does not include the additional budget for the areas in the HIV/AIDS Conditional Grant not covered in the NNs analysed in this report. The large shortfall in the HIV/AIDS Conditional Grant (R1.5 billion) can be partly explained by a reassessment of the cost of protocols for ART, which saw a significant increase. Medicines and childhood vaccines are also projected to have large shortfalls. Improved funding for the selected nonnegotiables may need to use an incremental approach focusing first on lifting poorer provinces to the average level of expenditure for each NN per level of care, with the aim of moving towards a higherlevel set of benchmarks (60 th centile). 25

Annexure: Additional Tables Table 8. Vaccines expenditure and expenditure per population under 1 year Table 9. HIV testing kits expenditure and benchmarks 26

Table 10. Condoms expenditure and benchmarks Table 11. ARV medicines expenditure per patient year Table 12. ARVrelated laboratory expenditure per patient year Total expenditure on ART labs 2013 Patient Years 2013 Cost / Patient Year Benchmark Prov FY 2011 FY 2012 FY 2013 Total EC 239 053 623 157 058 392 190 870 097 776 975 039 218 840 872 FS 149 498 657 151 766 723 87 453 281 387 121 154 113 922 768 GP 5 359 305 52 170 947 159 066 827 2 063 422 358 480 186 331 KZN LP 89 094 703 123 498 207 133 453 964 548 166 217 143 849 928 MP 117 576 072 101 275 668 95 888 926 816 941 294 191 276 501 NC 26 726 958 32 313 255 26 813 670 188 116 992 22 796 1 176 NW 131 338 667 103 292 543 103 920 380 570 568 490 128 052 812 WC 18 841 374 23 479 463 26 779 983 454 024 847 124 503 215 777 489 359 640 513 303 824 247 129 7 626 635 881 2 080 397 Average 700 306 27

Table 13. Blood in acute hospitals expenditure and benchmark per PDE Table 14. Laboratory services in acute hospitals expenditure and benchmark per PDE Table 15. Medical in acute hospitals expenditure and benchmark per PDE Table 16. Medicines in acute hospitals expenditure and benchmark per PDE 28