Research Findings and Policy Implications of the Review of Highly Specialised Services in the Public Hospital Sector

Directorate: Health Financing & Economics

26^th June 2001

Department of Health

Republic of South Africa

Funding

This review was funded by the European Union’s Public Health Sectoral Support Programme. The conclusions and views expressed in this document are those of the authors, and do not reflect the official policy or position of the European Commission.

Acknowledgements

The successful completion of this review is due almost entirely to the extraordinary assistance and support of personnel in hospitals and provincial health departments across the country. Clinical and administrative staff gave generously and freely of their time to allow data collection an validation, and gave great priority to this exercise despite their numerous commitments. Staff from all nine provincial health departments rapidly and professionally collected and collated large quantities of information during the second phase of this exercise. Special mention must be made of the outstanding efforts made in this regard by the teams led by Dr. Krish Vallabhjee in the Western Cape and by Mr. Gerrie van der Merwe in KwaZulu Natal.

Many people have contributed significantly to the development of the concepts and analyses which underpin this review. Especially important intellectual contributions (above and beyond their efforts in facilitating the work of the review team) have been made by the following individuals: Drs. Peter Mitchell and Paul Ciapparelli of Groote Schuur Hospital; Prof. Ralph Kirsch and colleagues in the Faculty of Medicine at University of Cape Town, Dr. Revere Thomson (Tygerberg Hospital); Drs. Krish Vallabhjee and Mark Blecher, Provincial Administration of the Western Cape; Dr van der Walt, Acting Chief Executive at Pretoria Academic Hospital and Mr. Sagie Pillay, Chief Executive, Johannesburg Hospital. It is also essential to acknowledge the critical contribution of Drs. Tersia Mitchell and Nicolas Crisp, whose initial work on highly specialised services opened the door to the efforts which resulted in this document.

Project Team – Directorate: Health Financing & Economics

Introduction

Background

1.1 This paper presents the final results of the review of highly specialised services and conditional grants. It builds upon the following documents:

• Review of Conditional Grants for Health: Current Problems and Options for Change (May 2000)
• Review of Highly Specialised Services: Preliminary Results and Summary Report to PHRC (January 2001)
• Report to PHRC: Proposed basis for restructuring and allocating the central hospitals conditional grant from 2002/03 (January 2001)

1.2 The objectives of the review of Highly Specialised Services were as follows:

• To collect detailed data on HSS activity and resources in public hospitals
• To develop a costed model of HSS provision in the public sector to inform future planning and resource allocation

1.3 At its meeting in January 2001, PHRC accepted in principle the proposed approach for shifting from the current Central Hospitals Grant to a National Tertiary Services Grant. PHRC tasked the Directorate: Health Financing & Economics to extend the scope of the original review of Highly Specialised Services to incorporate all public hospitals providing some element of "tertiary" level services. This involved a substantial additional data collection effort beyond the 17 central and provincial tertiary hospitals originally surveyed, which required several months to complete. This report therefore presents the results of this extended review, and all findings and recommendations herein supersede those presented in the references noted in para 1.1 above.

Structure of Report

1.4 The report briefly describes the methods used in the review of highly specialised and tertiary services (with more detailed discussion of methods provided as annexes). Basic summary data are then presented on the distribution and costs of specialised and tertiary services, with annexes presenting more detailed provincial-level data. Please note that each province will be provided with a full set of data on their own hospitals and services in electronic format. Data on the distribution (or incidence) of expenditure on residents of each province are then provided, showing the extent of inequities in spending and service access under the current system.

1.5 The implications of this analysis for future policy are then discussed in more detail. Broad issues for developing a long-term policy on equitable access to tertiary care are presented, to initiate the policy debate and development process required to develop a long-range strategic direction.

1.6 A framework for the medium term restructuring of the conditional grant system is proposed in a companion paper ("Proposed revision of health conditional grant framework"; Directorate: Health Financing & Economics, June 2001), for phased implementation commencing in FY 2002/03.

Methods

Defining Specialised and Tertiary Services

2.1 Following the PHRC decision to extend the original Highly Specialised Services (HSS) analysis, a revised categorisation of "highly specialised and tertiary" services was developed. The original list of Highly Specialised Services defined in late 1999 proved on the ground to have included certain services that were not actually provided in the public sector, but also to have omitted a swathe of other services which required the inputs of "highly specialised" personnel and/or equipment, and which were provided in only a limited number of locations, hence clearly suggesting their inclusion in the analysis. At the same time, it rapidly became clear that any analysis which attempts to define "highly specialised", "tertiary" or "quaternary" qualities in terms of case-mix is currently impractical in the South African public sector due to lack of meaningful diagnostic data in al but a handful of hospitals. Use of alternative proxies, such as linking HSS status with the presence of personnel registered as members of a sub-specialty to run the risk of omitting the many important sub-disciplines which have not yet gained sub-specialty status, and places undue weight on the purely medical staffing aspects of service provision. As such, an expanded list (based upon the original HSS service list, but with key specialties and services added based upon on the ground experience from phase I of the review) based on service descriptions was felt to remain the most viable and practical approach to defining HSS given current data availability. The full schedule is presented as Annex A. The list is deliberately broad in order to allow a comprehensive analysis, and contains certain services which lie on the border between tertiary and regional hospital services. The impacts of excluding or including certain of these borderline services are explored in the analysis.

2.2 Considerable debate took place as to whether and how to incorporate tertiary mental health services into the analysis. The team originally felt strongly that psychiatric services should be included, but subsequent discussions with Prof. Freeman and practising psychiatrists in various provinces indicated significant problems in defining "tertiary" dimensions of mental health services in a consistent manner. Indeed, the only areas on which it was possible to reach a clear consensus on the subject of the appropriateness of national funding for a tertiary or supra-provincial service were a) the need for national funding of the Grahamstown maximum security facility, and b) the likely desirability of national funding for forensic psychiatric assessments, which currently are provided as a supra-provincial service (but for which no direct funding is attached). The former issue is not contentious, while the current review cannot address the latter service at the level of detail required to do it justice. A pragmatic decision was therefore taken to exclude mental health services from the scope of this review. This in no way implies that mental health services might not require some form of national funding arrangement in future, but recognises the inherent complexities of mental health care organisation and the fact that significant changes to the nature of mental health care provision are likely to occur in coming years.

Data Collection

2.3 In the 17 central and tertiary hospitals included in the original HSS analysis, a detailed validation exercise was conducted to quality assure the original data. Hospital management teams were asked to identify any highly specialised units from the revised list on which data were not collected during 2000, and these units then provided the required information.

2.4 In order to identify additional hospitals at which the defined highly specialised and tertiary services are provided, a rapid survey of provinces was undertaken during February 2001. Provinces were asked to coordinate the identification of each site at which any of the services defined in Annex A were provided. Having identified these locations, provinces then also coordinated data collection (using the standardised tools described in 2.5) for each service, and passed the resulting data back to Directorate: Health Financing & Economics (HFE). Most provinces completed this exercise by early May, but finalised data for all provinces were only fully received in early June.

2.5 Standardised data collection tools were used to collect data on each unit within a hospital providing a designated "highly specialised" service. Data were collected on the following areas:

• Staffing (staff directly employed in the provision of the service)
• Clinical equipment (type, quantity, age and replacement value)
• Activity (admissions [intensive care bed days], day cases, outpatient visits)
• Province of origin of patients treated

2.6 The data collection forms were completed by unit level staff (usually unit heads or their nominated representative). During the first phase, members of the HFE team provided direct support and advice to those completing the forms, and spent substantial time at each of the major hospitals. Given the harder timetable for completion of the extended data collection exercise, physical support was not feasible; however, the HFE team provided telephonic advice and support to back up very detailed instructions attached to the forms.

2.7 The data management structure used in the analysis is as follows:

• Unit data files on each individual service – the primary data unit
• Hospital summaries – aggregate unit data for each hospital
• Provincial summaries – aggregate all hospital summaries per province
• National summaries – aggregate all provincial summaries

All data files from hospital summaries upwards include dynamic links, allowing automatic updating of data as amendments are made.

Cost Model

2.8 Alongside the primary data collection exercise, a standardised average cost model was developed from the Price Waterhouse Coopers’ "Hospital Cost Analyses" dataset. When combined with routine specialty activity data (available for 8 of the 9 hospitals at which PWC undertook their analysis), this allowed the production of average specialty costs by hospital, and the calculation of national average specialty costs. A detailed description of the cost model, methods and data used is provided in Annex B.

2.9 The version of the cost model finally incorporated into this analysis provides a step-down costing approach which allocates all hospital overhead costs to end-user specialties. The cost model has been set up to exclude from these specialty costs the direct staff costs of the units involved; data gathered directly from the units concerned are used instead. In regional hospitals, model outputs are scaled down by a factor of 0.6, reflecting the overall relationship between regional and tertiary hospital costs per PDE as identified in the recent National Health Accounts analysis. This scaling factor is likely to be more than sufficient to account for higher overhead and teaching costs at the academic central hospitals.

2.10 The unit cost estimates are then combined to generate standardised cost estimates for each service on the basis of:

Unit direct staff costs

plus
No. Admissions x standard specialty cost per admission

plus
No. Day cases x standard specialty cost per bed day

Plus
No. Outpatient visits x standard specialty cost per visit

2.11 Hence, the model generates service level cost estimates, driven by activity levels, which can be aggregated by hospital, province or service, and which provide for an equitable basis for funding equivalent services. At the same time, the incorporation of local staff cost data ensures that the key determinant of fixed costs allows for the actual scale and organisation of units to be reflected in the model.

Results – Current Patterns of Provision and Expenditure

Distribution of Highly Specialised Services

3.1 Final data returns from provinces provided information on 531 highly specialised service units at 62 hospitals nationally, excluding psychiatric services (see Annex D for a listing of all hospitals included). Table 1 summarises the location of these services by province, in ascending order of frequency. Sub-provincial breakdowns are available in spreadsheet form.

3.2 Table 1 illustrates clearly the range of availability of services encountered – from truly unique national specialist services (e.g. heart transplantation) to services provided in several locations in most provinces. However, it also clearly shows the disparities which mark the provision of tertiary services between provinces. Thus, renal dialysis might well be considered to be a regional service in Gauteng, where it is provided at six sites – yet four provinces provide this service at only one site each. Similarly, nine separate sites possess MRI and/or CT scanners in Gauteng, but none in Mpumalanga.

Table 1: Distribution of Highly Specialised Services

3.2 Cumulatively, the services included in this review generate a significant workload – over 436,000 admissions per year, nearly 95,000 day cases, and over 1.6 million outpatient visits. Annex C summarises activity per service. A detailed sub-provincial breakdown is available in spreadsheet form, but is not reproduced in this report due to space constraints.

3.3 Table 2 summarises overall highly specialised service workload by province. It clearly indicates the importance of specialised service provision and activity in Gauteng and the Western Cape, which account for more than 50% of all tertiary activity. However, it also clearly indicates the very significant levels of activity occurring in Eastern Cape, Free State and KwaZulu Natal. Finally, it shows strikingly just how little tertiary service activity is undertaken in the "northern" provinces of Northern Cape, North West, Mpumalanga and Northern Province.

Table 2: Highly Specialised Services Workload by Province

Equipment

3.4 The review also collected detailed data on the availability, age and estimated replacement value of major clinical equipment (defined generally as clinical equipment with a replacement value of more than R10,000 per item). Whilst it is not pretended that the data on equipment are 100% complete, they are presented here as they provide a useful insight into the importance of understanding equipment for capital planning, rolling replacement and maintenance. It is also important to note that data on equipment were more readily and comprehensively available in the larger tertiary hospitals, and that the more capital-intensive a service is, the better able they are to provide information their equipment inventory. All equipment data reflect items actually in place – they are not "wish lists".

3.5 A number of tertiary services are inherently highly capital-intensive. This is shown clearly in Table 3, which ranks services nationally by equipment value.

Table 3 – Equipment Value and Age by Service

3.6 The table clearly shows that a relatively small number of highly capital-intensive specialties contain the majority of the equipment stock by value. Thus, the top five services (out of 43 services in all) possess over 50% of the equipment stock by value; the top third of services contain 80% of the equipment stock by value. It is probably no coincidence that the most capital-intensive service, radiation oncology, also possesses the oldest equipment. The average equipment ages reported in Table 3 obviously conceal a significantly wider range of ages – both within individual units and between different units providing the same service. Equipment age is of significance mainly as an indicator of relative obsolescence and for life-cycle planning for equipment replacement. While there are no hard and fast rules regarding the useful lifespan of medical equipment, the pace of technical change and physical depreciation suggest, as a rule of thumb, that a ten year life-cycle is probably realistic. If a ten-year life-cycle were considered reasonable, the averages presented above should clearly alert us to the fact that a significant portion of the clinical equipment in use in tertiary services is either due for replacement now, or shortly will be (the average age should, ideally, stabilise around five years if all equipment were being replaced after ten years).

Table 4: Average Equipment Value per Site

3.7 It is also important to identify capital intensity and likely equipping costs for future service planning purposes. A crude measure of capital equipment cost per unit can be constructed by dividing total equipment value by number of sites, to give an average equipment cost per service delivery site. This measure is relatively crude, as it makes no allowance for different scales of operation, but it is highly informative nonetheless. Table 4 presents this measure. It provides a similar ranking of capital intensity to Table 3, but is significant as it sheds new light on a number of services which might otherwise be considered to lie on the "borderline" between tertiary and regional level services. Thus, Neonatal Intensive Care, Cardiology, Renal Dialysis and Intensive Care have, in certain fora, all been discussed as potential candidates for inclusion in a future regional hospitals care package. Given the high cost of equipping (and subsequently maintaining) such facilities at regional hospitals, this analysis strongly suggests that a more critical approach be used in specifying which services should be rolled out at the regional level of provincial health systems.

3.8 Table 5 below further emphasises the very low levels of tertiary care provision within the provinces of North West, Northern, Mpumalanga and Northern Cape. Their lower average equipment age reflects the fact that such clinical equipment as does exist in North West, Northern and Mpumalanga provinces has been installed in recent years through the policy of strengthening tertiary service provision in under-served areas. It is equally important to note that, in terms of capital infrastructure, KwaZulu Natal and Eastern Cape do not fall far below the equipment levels seen in Gauteng and Western Cape.

Table 5: HSS Clinical Equipment by Province

Costs and Expenditure

3.9 The cost model generates estimates of total expenditure on highly specialised services by hospital, province and nationally. Space constraints make it difficult to present more than a summary of the model outputs in this report; once again provinces will be provided with all the data and modelling files used to allow them to undertake a detailed inspection of data and methods.

3.10 If all the designated highly specialised services (except psychiatry) from the list in Annex A are included, the model generates an overall national estimate of specialised service expenditure of some R4.9 billion (in 2001/02 prices). Table 6 summarises the breakdown of expenditure by province for this "full" definition. This total expenditure exceeds all the potential funds available by combining the current central hospitals, health professionals and redistribution grants (R4.69 billion).

Table 6: Estimated Expenditure on "Full" Highly Specialised Services

3.11 This expenditure clearly takes place both in designated central hospitals (which are in receipt of national funding via the central hospitals grant), and other hospitals which do not receive such national funding. The distribution of expenditure between central and non-central hospitals is revealing, as shown in Table 7. Key points to note are that very substantial levels of expenditure (R450 million and R500 million respectively) are incurred on specialised services outside central hospitals in the Eastern Cape and KwaZulu Natal. While significant expenditure on specialised services does occur outside the central hospitals of Gauteng and Western Cape, the great majority of such expenditure (89% and 82% respectively) in both provinces does, in fact, occur in the designated central hospitals. Furthermore, even using this widest possible definition of specialised services, the current central hospitals grant allocation for Gauteng and Western Cape appears to slightly over-fund these provinces for services rendered in central hospitals, while slightly under-funding Free State and KwaZulu Natal.

Table 7: "Full" Expenditure – Central & Non-Central Hospitals

3.12 As discussed earlier, the "full" definition of highly specialised services is a very broad one, potentially incorporating various services which ultimately may not warrant inclusion as "tertiary" services. To illustrate the potential impacts of more conservative definitions of specialised services, a sensitivity analysis was undertaken of several options, each of which excludes progressively more "borderline" regional services. This analysis is summarised in Tables 8 and 9.

Sensitivity Analysis – Service Exclusion Options

Table 8: Sensitivity Analysis of Service Exclusions on Total Costs

Table 9: Sensitivity Analysis – Share of Expenditure by Province

3.13 Obviously, the progressive exclusion of borderline services reduces the total level of expenditure on those services that remain defined as " highly specialised", both at the aggregate national level and at the provincial level. One obvious implication of this analysis is that the precise definition of "tertiary" services can be flexed somewhat, to allow consistency with the available grant funding envelope. Equally important, though, are the data from Table 9. These show very clearly that changing the scope of defined services actually has only a very limited impact on the inter-provincial distribution of expenditure. Thus, excluding borderline "regional" services has an almost uniform impact on all provinces, and would not lead to any significant proportionate change in potential grant allocations.

Current Levels of Benefit from Expenditure on Highly Specialised Services