Research on Additional Costs of Teaching in NHS Scotland Final

advertisement
SCRA (2003) 8
Research on Additional Costs of Teaching in NHS Scotland
Report for Standing Committee on Resource Allocation
Act Sub-Group
Final Report
October 2003
Martin Spollen
Paul Dixon
Giles Hindle
Alasdair Munro
Uzma Khan
Peter Wallace
Secta Consulting
Services
Analytical
Division
Scottish
Executive
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
CONTENTS
1) INTRODUCTION
2) METHODS OF ANALYSIS AND EVIDENCE OF COST
DIFFERENCES FROM THE LITERATURE
3) AN OVERVIEW OF THE PROJECT METHODOLOGY
4) MODELLING DIRECT COSTS
5) THE METHODS FOR ESTIMATING OVERALL COST
DIFFERENCES AND DIFFERENCES IN INDIRECT COSTS
6) STATISTICAL ANALYSIS OF ADDITIONAL AND INDIRECT
COSTS
7) CONCLUSIONS
REFERENCES
Annex 1
Annex 2
2
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
SECTION 1: INTRODUCTION AND CONTEXT
1.1 Introduction
This report sets out the findings of a research project commissioned to inform the
Standing Committee of Resource Allocation in NHSScotland on the quantum of
funding required by NHS Boards to support the costs of providing undergraduate
medical teaching (and the medical teaching of dental students) undertaken in
NHSScotland.
The research has involved two parallel research strands. Firstly, a detailed costing
exercise has been undertaken in conjunction with the four medical schools to capture
those costs directly identifiable with teaching delivery by NHSScotland. A parallel
research strand has tested the hypothesis that teaching hospitals in Scotland also incur
costs that are an indirect consequence of teaching.
Both studies are intended to inform a new mechanism to distribute revenue support
for teaching across Scotland’s teaching NHS Boards on a transparent and equitable
basis. If accepted, the new mechanism would replace the current ‘Additional Costs of
Teaching’ (ACT) revenue adjustment system currently in use.
1.2 Terms of Reference
The research has examined:
♦ the factors that influence the relative costs of teaching and non-teaching hospitals.
The aim of this aspect of the research has been to identify the range of factors that
effect these costs, and to establish an evidence base to support an assessment of:
(a)
the relative level of these costs;
(b)
the extent to which cost differences reflect teaching responsibilities as
distinct from other factors related to the specialist role of teaching
hospitals;
(c)
the impact of these costs on the overall cost structure of different Health
Boards; and
(d)
the key drivers that influence these costs (e.g. the number of medical
students and other factors).
♦ the influence of changing patterns of medical training on the additional cost of
teaching medical students.
♦ an analysis of the merits of alternative methods of distributing resources for the
additional costs of teaching.
3
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
1.3 Teaching context
1.3.1 Medical Schools
The research has focussed on NHSScotland’s input to teaching the undergraduates in
Scotland’s four Medical Schools – at the universities of Glasgow (252 students per
year), Edinburgh (244 students per year), Dundee (162 students per year) and
Aberdeen (184 students per year). There is a fifth Medical School in Scotland at St.
Andrew’s University, which has not historically been involved in ACT funding
because its students transfer to England for the practical NHS-based elements of their
training.
A medical degree is a five-year course, with an optional 6th year (normally squeezed
between the 3rd and 4t h years) which concentrates on scientific research training.
Students spend the early years of their programme based at the university, but
increasingly move into hospital, primary care and community settings as their
programme progresses.
Medical Schools use a variety of methods for the delivery of their programmes. These
include traditional lectures and seminars, as well as clinical skills sessions and
problem based learning (facilitated problem solving in small groups of around 10
students). Students also spend considerable amounts of time “on site” in hospitals, GP
practices and community settings. Special Study Modules (SSMs) allow students to
pursue particular areas of interest – some non-medical.
1.3.2 Tomorrow’s Doctors
The curricula at Scotland’s four major Medical Schools have been significantly
redesigned in the past 7 years based upon a report by the General Medical Council
entitled “Tomorrow’s Doctors”. First published in 1993, the report signalled a
significant change in the form of guidance from the GMC. The emphasis for medical
education moved from gaining knowledge to a learning process that includes the
ability to evaluate data as well as to develop skills to interact with patients and
colleagues.
Key aspects of this new approach to medical education include:
! Competence in key clinical skills and procedures;
! Competence in patient examination and decision making;
! Competence at managing a patient’s healthcare needs;
! Competence at communicating with patients and colleagues;
! Competence at retrieving and processing information;
! Appreciating the behavioural, ethical and legal aspects of healthcare;
! Appreciation of the role of the doctor within the health care system;
! A move away from memorising facts.
4
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
1.3.3 New Curricula
All four of the Medical Schools have responded to this report, although in different
ways. All curricula still concentrate on university based teaching in the early years of
the degree programme and move to clinical attachment approaches in the later years,
but a range of extra teaching styles and subjects have been introduced. Glasgow, for
example, have made a significant move to problem based learning and significantly
reduced the amount of lecture style delivery, thus freeing up a significant amount of
self-study time for students. All schools have developed a combination of clinical
skills, professional development, vocational studies and various GP-led activities. All
schools now have SSMs to give students the opportunity to specialise and OSCEs
(Objective Structured Clinical Exams) which examine the practice of being an
effective doctor through simulated patient examination.
1.3.4 NHS Input
The Medical Schools are closely linked with the NHS, especially the teaching NHS
trusts. NHS staff are involved in the delivery of medical undergraduate degrees in a
range of ways – from delivering classes, both on-site and at the university, to being
involved in student recruitment, course administration and development and student
examination. NHS trusts (both “teaching” and “non-teaching”) also provide the
facilities and staff for clinical attachment weeks, which form the dominant element of
the later years of the curricula. Students on clinical attachment will spend the whole
week on-site – at hospitals, GP practices and possibly community settings. During
these periods students will receive some formal education sessions, but will also
spend time observing and being shown the practice of being a doctor.
1.4 Curricula at the Four Medical Schools
This section will give a brief overview of the curricula at the four Medical Schools:
1.4.1 Glasgow University
The Glasgow curriculum is split into 3 sections. The first section (years 1 and 2) is
university based and consists of (a) university lectures (few compared to other
curricula) and problem based learning (PBL) sessions, (b) vocational study activities
and visits (GP led), (c) clinical skills training, (d) Fixed Resource Sessions (lab-based
activity), and 1 SSM.
The second section (year 3) is similar to years 1 and 2, but introduces NHS-based
delivery. PBLs are delivered on-site and Clinical Practice modules are delivered at
both hospital and GP practices. There are 2 SSMs. The third section (years 4 and 5)
concentrates on clinical attachment on-site at hospital, with 5 weeks at a GP practice.
There is also a period of intensive lectures in year 5 and 4 SSMs.
5
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
1.4.2 Edinburgh University
The Edinburgh curriculum is split into 2 sections. Section 1 covers years 1 and 2 and
is university-based. It involves lectures, various types of tutorials and practicals
(including case-based learning (CBL)) and Clinical Skills and Personal Professional
Development (CSPPD).
Section 2 covers years 3 to 5 and concentrates on clinical attachment on-site. There
are some lectures, tutorials and CSPPD sessions in years 3 and 4, and one of the 8
week modules in year 5 may be outside the UK. There is only 1 SSM.
1.4.3 Dundee University
The Dundee curriculum is split into 3 sections. Section 1 covers the first year and
centres around lectures and practical anatomy / physiology sessions delivered at the
university. There are also GP led sessions / visits, clinical demonstrations, emergency
care, behavioural sciences and 1 SSM.
The second section covers years 2 and 3 and delivery is based on-site. There are
lectures, PBLs, labs, clinical skills and ward teaching. There are also GP led sessions
and visits, and 2 SSMs. The final section covers years 4 and 5 and is predominantly
clinical attachment, especially year 4. Year 5 concentrates on preparation for a Junior
House Officer (JHO) post (shadowing and a short course) and 5 of the 6 SSMs for
this section occur in the 5th year.
1.4.4 Aberdeen University
The Aberdeen curriculum is split into 4 sections. The first section covers year 1 and is
university based. Lectures take most of the time, although there are also a GP led
Community Course, Practical Anatomy sessions, Communication Skills and 1 SSM.
The second section runs up to Easter in year 3 and is similar to the first section, but
includes more site-based activities – for example ward based teaching (WBT) and an
Introduction to Clinical Skills course. There are 2 SSMs.
The third section covers the end of year 3 and year 4 and is site-based – 9 five-week
rotations including 5 weeks in a GP practice. There is 1 SSM. The final section covers
year 5 and takes place in a range of locations including clinical, GP or Mental Health,
medical and non-medical electives. [NOTE: All curricula include at least 3 OSCEs]
6
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
1.5 Costs associated with undergraduate teaching in NHS settings
1.5.1 What are the main cost drivers
Undergraduate medical students are placed in a variety of NHS contexts including the
community and general hospitals as well as the more traditional setting of teaching
hospitals. Supervising and accommodating students will incur costs in all these
settings and an increasing body of research is examining both the scale of these costs
and the types of resources needed to support student placements. In the most widely
researched of settings, the major teaching hospital, the following three sets of factors
(not all associated with teaching) are thought to contribute to increased costs of care.
There are some aspects of teaching hospitals, such as the ability to benefit from
economies of scale, that may enable some forms of care to be delivered more cheaply.
(1) Cost drivers directly associated with teaching, including
•
•
•
•
•
•
•
•
•
staff time with students (contributing to increased staff to patient ratios)
staff time spent on preparation and curriculum development
a greater level of laboratory tests and facilities associated with demonstrating
tests
more medical illustrations and teaching aids
larger offices and dedicated teaching space
wider corridors and generally larger public spaces
additional library facilities
increased portering and security
additional hotel facilities for students, catering, accommodation etc.
The focus in much of the literature, and this project, is on revenue funding, so capital
costs and depreciation tend to be excluded. However, they will include maintenance
and utilities costs associated with the extra space used for students and teaching.
Most of these costs will be related to student numbers, but the relation may not be
simple. The OR modelling that forms part of this project can be used to explore the
nature of the relationship between these direct costs and student numbers.
(2) Factors that may be closely linked to undergraduate medical training, including
•
•
•
•
•
•
funded research;
personal research;
tendency to develop and use innovative treatments
higher staff skill mix, from nursing grades to prestigious consultant
appointments
training of nurses; and
training of other staff - such as technical occupations and paramedics.
7
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
(3) Factors that are less obviously linked to teaching, but which frequently, if not
inevitably characterise teaching hospitals.
•
•
•
•
•
•
•
higher staff, buildings and maintenance costs due to metropolitan locations
more complex case mix due to their role as centres for tertiary referrals
higher than average morbidity in their catchment areas (due to inner city
deprivation) resulting in greater and more complex demand for general
hospital services and A&E
maintaining expensive specialties that are not often found outside teaching
centres
a wider range of patient support services
above average quality of care
superior grade accommodation
A literature review circulated earlier in this project summarises estimates of the
relative costs of these different factors, though the evidence is very incomplete and
difficult to synthesise. However it is possible to say that some of the non-teaching
influences can have a major impact of costs and that it is important to control for their
effects if one is to establish what additional costs are due to teaching.
1.5.2 Direct versus indirect costs
For the purposes of this project we have drawn a distinction between those costs of
placements that can be directly observed and recorded, such as staff time and obvious
use of physical resources and those that may be integral to a teaching environment but
not so easily or directly measurable, such as personal research. We have labelled the
costs associated with the former set of factors the direct costs of teaching and those
associated with the latter the indirect costs. Separate strands of the project address
each of these costs. It is worth noting that in our analysis of the indirect costs we try
to exclude other types of costs such as those due to case-mix or metropolitan location
that are often present in teaching hospitals but are arguably not an inevitable
consequence of their teaching status.
8
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
SECTION 2: METHODS OF ANALYSIS AND EVIDENCE OF COST
DIFFERENCES FROM THE LITERATURE
2.1 Cost differences reported in the literature
There is a limited literature on the costs of both undergraduate and postgraduate
training of doctors in hospitals and how these contribute to the revenue costs of
teaching hospitals. As a review of this material was presented earlier in the project
only key points that bear directly on the analyses are presented here.
Most empirical studies of the cost of medical training provide some estimate of the
overall difference in gross cost per case between teaching and non-teaching hospitals
- and in a few cases (summarised in Linna et al - 1998) also compare these differences
with the funds made available for teaching.
Table 1 presents a representative selection of estimates of overall cost differences.
Most of these report excess costs per case of between 5 and 25% in teaching and
compared to non-teaching hospitals. When interpreting these results it is important to
remember that the systems for funding medical training can very greatly between the
countries studied, and that variations in methods of measurement and analysis may
also influence the results.
Table 1 Estimates of additional costs in teaching hospitals
(Adapted and expanded from Linna et al - 1998)
Country
USA
USA
USA
USA
Spain
Additional cost
10-25%
8-15%
0-15%
1.4%
3-11
Spain
3.1%
Spain
11.1%
Finland
England
England
15%
15%
4-43%
Source
Sloan et al 1983
Zuckerman et al 1994
Granneman et al 1986
Gaynor M and Anderson GF (1995)
Lopez-Casanovas and Wagstaff
(1996)
Gonzalez-Lopez Valcarcel B and
Barber P (1996)
Wagstaff A and Lopez-Casasnovas
(1996
Linnakko and Linna (1995)
Culyer et al
Foote et al (1988)
Relatively few studies provide similar estimates for the additional costs of individual
departments or services. The lack of available data and the difficulty of controlling for
confounders are largely to blame for the limited evidence at this level. Results from a
small numbers of studies are shown in Table 2, covering many of the services and
departments that are generally reckoned to cost more in teaching settings. The range
amongst even this small set of results suggests there is little consensus on how much
extra each of these costs.
9
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Table 2 Estimates of additional costs in
teaching hospitals
Staff costs
Medical pay 7% higher in teaching
institutions
Cost of nursing higher
Nursing costs 9.5%
specific departments and services in
(Culyer et al 1978)
- Busby et al 1972
(Culyer et al 1978)
Pathology/Lab tests
8% increase in dept cost
Culyer et al 1978
Twice as high
Busby et al 1972
Increased tests account for 56% of Scroeder & O’Leray
differences in costs
Operating theatre
11% (unit cost)
25%?
(Culyer et al 1978)
AUC (1965)
X-ray/Radiology
Twice as high
5% (dept cost)
5%
Busby et al 1972
(Culyer et al 1978)
AUC (1965)
Medical records
(dept cost- proportion unspecified)
25%
(Culyer et al 1978)
AUC (1965)
Library
50-70%
AUC (1965)
Catering
12% (dept cost)
(Culyer et al 1978)
Cleaning
4.5% increase in “domestic” service costs
(Culyer et al 1978)
10
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
2.3. Approaches to estimating costs and cost differences
Studies to estimate cost differences and their causes fall into two broad groups.
Bottom-up approaches that study the processes of teaching and the resources required
and top-down studies that focus on the differences in the costs of patient care between
teaching and non-teaching institutions.
2.3.1 Bottom-up approaches
Bottom-up approaches are so described because they aim to build-up a picture of the
activities and costs associated with teaching. This can be done in two rather different
ways, by observing the processes and collecting data on activities by other means, or
by simulation.
Observational methods for estimating the inputs to teaching
In this group of studies the main aim is to study the actual processes of teaching and
record factors such as the amount of staff time involved and the use of facilities.
Research may be based on observation, but also use other methods such as surveys
and activity diaries. Examples of the latter include surveying students on the their
contact time with clinicians in various teaching settings (e.g. Sheldon, 1990, 1991a
and 1991b; and Weinberg et al 1994). Examples of studies that observe the activity
patterns of both medical and non-medical staff in teaching institutions to establish
proportions of time spent on tasks directly or indirectly related to teaching include
Snijders et al. (1987). Other examples of these approaches include Institute of
Medicine (1974); Perrin (1987); Perrin and Magee (1982), and Rayner (1985).
Although there are relatively few examples of this type of work, its potential strength
is in supplying everyday detail on the process of teaching in hospitals; hence it should
help measure the relative effort spent on teaching and other activities.
It has three main weaknesses. First, detailed observation in hospitals is very costly.
Secondly, there is the problem of deciding what activities, or more often, what part of
an activity, should be associated with undergraduate teaching. This is a particular
problem in teaching hospitals. When one considers the range of functions that are
typically carried out in such hospitals it is clear that any single task could be
contributing to several if not all of the following:
•
•
•
•
•
•
•
•
providing health care directly
supporting other institutions providing health care
undergraduate medical education;
postgraduate studies and training;
funded research;
personal research;
training of nurses; and
training of other staff - such as technical occupations and paramedics.
11
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Although it is theoretically possible to apportion parts of each activity to different
functions, the practical difficulties are obvious and results would be open to question.
For these reasons, some authors (e.g. Perrin 1987) have doubted whether an
observational approach can ever arrive at accurate estimates of teaching costs.
A third and related problem, is that observation may overlook some activities because
they do not appear to be directly contributing to teaching when, in fact, they are
indirectly supporting teaching or more generally contributing to the culture of a
teaching hospital.
Simulation modelling
Simulation modelling provides a cheaper and quicker alternative to observation for
estimating the costs directly involved in teaching. As adopted in this project, the
method considers what activities and resource are necessary to deliver that part of
undergraduate medical training that takes place in the NHS. It uses a combination of
syllabus information, student and medical sub-dean reports, and accounts from others
involved in undergraduate teaching to construct an operational model of the inputs to
the training. Both national and local sources are then used to attach costs to these
inputs.
The resulting model can estimate the costs of providing the existing syllabi to current
numbers of students, or it can be used to explore the cost consequences of varying,
syllabus, training methods and student numbers. The model can be used to predict
costs at different levels, such as medical school, trust and hospital, provided there is
sufficient detail to populate the model at each of these levels.
The main advantages of this approach, as for much simulation, is that it relatively
cheap and can be used to explore hypothetical scenarios. However, there are several
disadvantages. There will be costs associated with data collection, as the model needs
to be grounded in the details of how teaching is carried-out and the costs of its various
components. It is potentially open to the main criticism of observational methods, that
it may not adequately address the problem of multi-functional activities. It will also
be very limited in its coverage indirect cost drivers.
2.3.2 Top-down approaches
A second class of methods for investigating teaching costs has a very different
starting point. It examines the differences in costs between teaching and non-teaching
institutions using data on the costs of care and hospital services. Again, such studies
can be crudely grouped into two types:
those that simply compare the costs of teaching and other hospitals, albeit with so
controls for confounders such as case mix and those that try to model costs at all types
of hospitals with teaching status or teaching load as one of the independent variables.
12
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Cost comparisons
The principle of comparing the costs per case in a teaching hospital with a nonteaching hospital is straightforward., but the practice is more complicated as studies
need to control for confounding factors such as case mix. The methodological issues
associated with these comparisons include those listed below. They are discussed in
more detail in relation to the present exercise in Section .
What are teaching hospitals - can they be easily distinguished from other hospitals by,
say, teaching load
•
•
•
•
•
How to control for case-mix - both at the level of specialty and HRG within
specialty
How to control for economies of scale
How to control for market forces factors.
How to control for other external factors that may influence costs
How to find suitable data and deal with high levels of variation between small
numbers of hospitals
Regression modelling of hospital costs
Simple (OLS) multiple regression techniques are still used to identify the excess costs
of teaching. Typically, such modelling attempts to predict the overall cost per case by
using variables associated with teaching and those associated with case-mix and
environmental factors. There is an increasing interest in replacing simple OLS with
more sophisticated methodologies, but all these approaches share the common
problem of how to obtain adequate data on costs and more especially on potential cost
drivers. The problem is compounded by the relatively small number of institutions
available for analyses and over-time variations in costs and organisation of teaching.
That most existing studies are cross-sectional (i.e. based on a single point in time) has
prompted some criticism (e.g. . Foster, 1987). Other well-known difficulties are how
to control for supply side effects and deal with extensive collinearity amongst the cost
drivers.
The problem of the potential endogeneity of supply factors has led to the use of two
stage least squares methods - as for example by Milne et al (1989).
The problem of having relatively few teaching institutions to include in the analysis is
compounded by the tendency of some authors to include large numbers of variables in
their model - drastically reducing the degrees of freedom. The resulting models,
especially when packed with supply and activity variables, have very high values for
r-squared (typically in excess of 97% as in Milne et al), but it is arguable whether they
help us understand the sources of variation between hospitals.
In the English context this has been of particular interest because critics such as
Bevan (1999) and Sheldon (1999) have argued that, in particular, the facilities
component of SIFT may be supporting historical inefficiencies in teaching hospitals.
Hence the related concern that regression analyses will not be able to identify
13
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
inefficiency and that other methods should be used. Two current contenders include
data envelope analysis (recommended by Bevan 1999 and applied by, amongst others,
Morey et al 1995) and a modification of OLS - frontier cost functions as used again,
amongst others, by Linna et al (1998).
14
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
SECTION 3 AN OVERVIEW OF THE PROJECT METHODOLOGY
3.1 Overview of main methods used
The approach finally adopted by this study has two main components
Operational research models of the costs of delivering the syllabi of the four Scottish
Medical Schools.
These models provide an estimate of what we describe as the "Direct Costs of
teaching undergraduate in NHS settings". They are designed to compute estimates for
each of the medical schools, but with suitable data can provide cost estimates at health
board, trust of even hospital levels.
Comparisons of the costs of care at teaching and non-teaching hospitals in Scotland
The first stage of these comparisons involves estimating the overall difference in cost
between the two types of hospitals having controlled for potential confounders such
as case-mix and economies of scale. The direct costs of teaching (as estimated by the
OR models) are then subtracted from the cost differences to provide estimates of what
we describe as the "indirect costs of teaching". These estimates are both computed
for Scotland as a whole and for each of the major teaching hospitals.
The next stage of the work tries to account for these indirect costs in two ways.
Firstly, by comparing them with the sources of income that are largely limited to
teaching hospitals, such as research funding. Secondly, we examine whether the costs
ratios between teaching and other hospitals in respect of staff, nursing, theatre and
other identifiable budget heads are similar to the overall cost ratios between the two
types of institution, or whether certain of these activities seem to account for a
disproportionate amount of the cost differences.
Finally, we use the results from the above and information on relative teaching loads
to investigate the differences between the cost bases of the major teaching hospitals.
15
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Schematically the main components of the methods are as follows
♦ Construct OR models of the DIRECT costs of delivering the medical schools
syllabi
♦ Compare the overall costs of care at Major teaching Hospitals (MTH) with those
at other hospitals
♦ Control for potential confounders (incl case-nix, economies of scale and MFF)
♦ Produce corrected estimates of the gross cost differences
♦ Subtract the estimates from the direct cost modelling to give an estimate on
indirect costs
♦ Try to identify the main contributors to these indirect costs (e.g. staff costs,
research etc)
♦ Try to explain the differences between Teaching Hospitals in these indirect costs
16
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
SECTION 4 THE DIRECT COST MODELS
4.1 Introduction
This chapter describes the direct cost modelling of input by NHSScotland to the
curricula at the four main Medical Schools in Scotland – Glasgow, Edinburgh,
Dundee and Aberdeen. An Excel-based model has been developed to capture direct
costs of teaching support for each Medical School, which gives a cost quantum for
each academic year (including the optional science year and the medical education of
dental students).
The models have been populated using a Delphi approach involving documentary
evidence, interviews with stakeholders, a workshop and direct input by the Medical
Schools over a three-month period.
The term ‘direct costs’ refers to those costs incurred by NHSScotland that can be
directly attributed to the curriculum of a particular medical school. This has generally
been reflected in the direct cost models by activities undertaken by students – for
example, lectures, seminars, clinical / vocational skills sessions, clinical attachments
(both in primary and secondary care), evaluation, recruitment, etc. However, a
number of overheads have been included – for example, administration, hospital subdeans.
4.2. The Direct Cost Model
4.2.1 Overview of the Modelling Process
The direct costing uses a cost model for each medical school, which has been
populated using a Delphi type approach, vis:
1. Examine the curriculum at each of the four Medical Schools in order to identify
the types and volume of teaching activity that comprises each academic year. For
example, “1 hour university lecture” would be one such activity type. Information
required would be the number of lectures delivered per year and the size of the
lecture delivery group. An important part of this process is to apportion costs
between the NHS and the Medical School, as the model will ultimately only count
NHS incurred costs. For example, a university lecture delivered at the Medical
School by an NHS member of staff would incur costs to both the Medical School
and the NHS;
2 . For each activity type, quantify the amount of resources used to deliver an
instance of each activity. Types of resources might be staffing, facilities, travel,
etc. Resources are costed at Scotland-wide unit costs (Blue Book). Initially, this
information was generated from documentary evidence of curricula and meetings
with ACT officers and Medical School administrators;
3. In order to validate user input at (2), above, engage in a Delphi process by direct
liaison and interviews with key informants, and a workshop for stakeholders (9th
17
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
of May 2003). Final population of the models has been co-ordinated by ACT
Officers and Medical Schools over the summer of 2003. All four medical schools
have confirmed significant progress has been made populating the models,
however, it is clear that work will continue to increase the accuracy and
transparency of the models;
4. Finally, record the output of each Board’s model and thus determine an overall
Scotland direct cost quantum, which could, in principle, be distributed between
the four Boards.
4.2.2 Structure of the Direct Cost Model
There are four cost models, one for each of the Medical Schools at Glasgow,
Edinburgh, Dundee and Aberdeen. The models essentially link the curricula at a
particular Medical School to Scotland-wide unit costs (see figure below).
DCM - shape of the model
* Model feeds explicitly from resources to activities to curriculum
Level 3
Level 2
Level 1
Year 1 Curriculum
Activity 1
Resource 1
Activity 2
Resource 2
Activity 3
Resource 3
Resource 4
The models are based in Microsoft Excel and each have 48 worksheets. The structure
of the worksheets is as follows:
!
!
!
!
!
!
Sheet 1 – Title Page
Sheet 2 – Cost Summary Page – total direct cost of teaching for NHS
Sheets 3-8 – Summary Pages for each academic year (the curricula)
Sheet 9 – Administration Costs (overhead costs)
Sheets 10-43 – Definition of Activity Types (A1 to A33)
Sheets 44-48 – Resource Unit Costs (staffing, facilities, etc.)
Each academic year has a well-defined curriculum. The sheets 3-8 record the activity
types and quantum for each of the five compulsory years of the degree, plus the
optional science year (labelled Year 6 in the model). These sheets are linked to the
sheets A1 to A33 which give a cost for each instance of a particular activity type (by
18
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
year). For example, the cost of a lecture delivered at the Medical School by an NHS
consultant might be £300. The worksheet relating to university lectures will outline
how this £300 is made up – for example, 3 hours time (preparation, delivery, travel)
and 30 miles travelling. The activity sheets are linked to Scotland-wide unit costs for
staffing, facilities, consumables, etc.
Because the model is linked, any input can be changed at any time and the model will
automatically update. For example, if the amount of time needed to prepare a lecture
changes from 1hr to 2hrs, the activity sheet can be updated and the cost totals will be
adjusted automatically. Similarly, if the hourly rate for an NHS consultant changes,
this can be altered on the staffing resource sheet and the cost totals will be adjusted
automatically.
4.2.3 Populating the Model – Delphi Approach
The curricula for each academic year at each Medical School is complex, but
generally well understood, and therefore can be regarded as an objective input to the
model. Similarly, the Scotland-wide resource unit costs are taken from the Scottish
Blue Book, although consensus estimations of has also been used.
However, definitions of each activity type in terms of the resources used in a
particular instance of the activity – i.e. the resources used to deliver a university
lecture, or the resources used on a typical clinical attachment week – required careful
estimations by key stakeholders. In order to populate the models with these activity
definitions, a Delphi process was implemented. A Delphi process enables a document
– in this case an Excel Model – to be developed by a group of stakeholders until it has
reached an agree final state.
The first stage in this process involved a series of meetings (with ACT Officers,
Medical School staff, General Practitioners, trusts, finance staff, the BMA, NES) to
generate an initial population for each of the models. The second stage involved a
mail shot and a workshop (9th May 2003) where key stakeholders were allowed to
evaluate the initial assumptions of the models, suggest alternative (or additional)
activity definitions and interact “live” with the four models. This enabled the initial
estimations to be developed and the costings fine-tuned. The final stage of the Delphi
process involved the Medical Schools taking ownership of the models – co-ordinated
by the ACT officers – to reach an agree final costing (August 2003). To date the
medical schools have confirmed significant progress has been made towards a final
costing, although it is clearly an ongoing process.
19
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
4.3 Results of the Direct Cost Model
There is a direct cost model for each of the Medical Schools for Glasgow, Aberdeen,
Edinburgh and Dundee. Each model gives the total direct costs of the NHS input to
the Medical School curricula.
The results reveal that the aggregate direct cost quantum for the four medical schools
is £57.6 million, based on work to date.
The breakdown of this total between the schools is shown in the table below.
Total Cost of NHS Input to Medical School Curricula by year
Year
Year 1
Year 2
Year 3
Year 4
Year 5
Year 6
Total
Glasgow
(£000s)
564
763
3,802
6,953
7,346
276
19,704
Edinburgh
(£000s)
578
757
5,196
4,802
6,136
61
17,530
Dundee
(£000s)
258
669
4,557
2,505
2,096
0
10,084
Aberdeen
(£000s)
216
423
2,823
3,811
3,005
6
10,285
4.4 Current ACT Allocations
The table below gives the allocation of ACT funding between the four NHS Boards
for 2001/2002. The total figure was £86.2 million. It can be seen that the funding split
between the four NHS Boards is roughly proportional to student numbers, although
two Boards receive proportionally more money (Glasgow and Tayside) and two
receive proportionally less (Lothian and Grampian).
Comparison of ACT shares in 2001-02 to proportion of student body in each
School
Board
Allocation (01/02)
% of ACT total
% of total student
(£ millions)
body
Glasgow
27.2
31.6
29.9
Lothian
24.3
28.2
29.0
Tayside
17.4
20.2
19.2
Grampian
17.3
20.0
21.9
Total
86.2
100.0
100
20
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
4.5 Options for using results for resource allocation
There are four basic methods available, which are discussed below.
Option 1: Make allocations to Boards on the basis of the number of students
supported in the local NHS at a single Scotland-wide average cost per student across
all years
1. Derive a Scotland-wide average cost per student (for all activities together) based
on the aggregated DC Model returns from each of the four Schools.
2. Collect information for the medical Schools on the number of student weeks
placed per Board (or allocate to the Medical School’s host Board for onward
distribution).
3. Total up each Board’s allocation from the no. of student supported.
4. Top-slice the DC quantum (sum of 3 across the four Schools) – and allocate per
relative shares determined at 3, above.
Option 2: Similar to option 1 – except that a different Scotland-wide cost is
applicable to students in each year of the curriculum.
1. As per Option 1, except the calculation would be undertaken for each year
individually to reflect differential cost of support by year of study.
Option 3: Delivery Unit costing
1. Derive a Scotland-wide average cost per student for each activity type
2. Collect information from the Medical Schools on the volume of teaching activity
delivered by each institution (teaching and non-teaching hospital) including that
delivered in primary care settings
3. Apply the average unit from 1, to determine a ‘budget’ for each institution
4. Aggregate across institutions in each Board to arrive at a Board level total.
Either teaching Boards or the Medical Schools could run the models to make such
cost estimates.
Option 4: Allocations by student week
1. Use the direct cost models to derive the cost of a student week – differentiating
settings where difference teaching activities are typically carried out (e.g. Primary
Care, Major Teaching Hospitals, Non-Teaching Hospitals);
2. Allocate funding to all NHS Boards where teaching is being delivered – based on
the number of weeks supported in each setting;
3. Funds to be allocated through a central body – such as NHS Education for
Scotland.
All options require top-slicing of the aggregate quantum of funding identified.
21
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
All options envisage a separate compensation mechanism for the indirect cost
consequences of having a teaching status.
4.6 Example of Allocation using Direct Cost Model Results
This section shows the outworking of Option 1, and compares funding shares between
the four Schools with that to host NHS Boards under ACT in 2001/02.
It can be seen that, using this approach to allocation, two Boards would receive
proportionally more money (Grampian and Lothian) and two boards would receive
proportionally less (Glasgow and Tayside).
Allocation of Direct Cost Model Total between Boards by Number of Students
Board
Allocation
% of total
DCM
% of
change
(01/02)
Allocation
total
(%)
(£ millions)
(£ millions)
Glasgow
27.2
31.6
17.2
29.9
-1.7
Lothian
24.3
28.2
16.7
29.0
+0.8
Tayside
17.4
20.2
11.1
19.2
-1.0
Grampian
17.3
20.0
12.6
21.9
+1.9
Total
86.2
100
57.6
100
Although the quantum of funding has significantly reduced, the share of resources
under the proposed mechanism is, for each teaching Board, within 2% points of the
ACT shares.
Final comment
The direct costing exercise has collated together – for the first time – a vast database
of comparable information on activities and costs across all four Medical Schools.
This is a significant achievement given the complexity of the curricula and the variety
in delivery between Schools.
This information now makes possible much more detailed analysis of individual cost
structures, NHS inputs and overhead recharges that has hereto been possible.
However, further work should aim to refine the final estimate of direct costs – noting
that this was part of the agreed process with Trusts, Boards and the Medical Schools
in order to engage them in the process in the available timescale. In particular, such
work should focus on the comparability of Costing Models completed and returned by
each School - issues of comprehensiveness, costing bases in relation to capital
charges, and recharging non-activity related ‘administration’ costs.
Analytical Services Division (SEHD) are now taking forward the final development
of proposals for a suitable formula based on the above options and information from
the direct costing exercise completed in this work.
22
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
SECTION 5 THE METHODS FOR ESTIMATING OVERALL COST
DIFFERENCES AND DIFFERENCES IN INDIRECT COSTS
5.1 Alternative methods and sets of analyses
This section of the report describes the methods used to estimate overall cost
differences and indirect costs, but before describing the main approach in more detail,
it is worth noting that it was chosen after other methods had been tried. Three
alternatives were explored. Previous project reports have described some of this work
and further details will be given in appendices to the final report.
•
Predicting the indirect costs of Scottish teaching hospitals using cost data on
English specialties (3.2.1).
•
Predicting gross Scottish TH costs using Scottish specialty level costs. (3.2.2)
•
Investigating the impact of teaching load on cost (3.3.3)
5.1.1. Predicting the indirect costs of Scottish teaching hospitals using cost data
on English specialties
The first set of analyses conducted for the project used cost data on English
specialties to predict the additional (indirect) cost of Scottish teaching hospitals. For
our purposes, the English Reference Costs Database has three advantages over the
Scottish Cost Book:
1) It is based on a larger sample of teaching hospitals (28) and should therefore
provide more robust estimates of specialty costs
2) Case-mix controls can be more rigorous because the English Reference Cost
data are directly compatible with the English specialties
3) The English Reference Costs specifically exclude any direct costs of teaching
that can be specifically identified set off against SIFT income. These are
netted out by trust and hospital accountants prior to computing the HRG costs.
So cost differences based on these figures should only reflect indirect costs;
and using these figures eliminates the need for separate estimates of direct
costs.
The main steps in these analyses were as follows:
•
•
The costs are obtained by trust, HRG and specialty.
These are aggregated to produce mean values for each specialty across all acute
teaching trusts and across all other acute trusts - a group of trusts that are
designated as intermediate teaching are excluded.
23
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
•
•
•
•
Figures are computed for three sets of episodes: elective and non-elective inpatients and day-cases.
Two sets of cost ratios are computed. One for unstandardised costs per specialty,
the other is case-mix standardised, by HRG within specialty. Standardisation is
quite rigorous: the average costs per HRG are applied to the same national
average bundle of HRGs per specialty in each institution.
Two further sets are computed, both standardised and unstandardised, to include
the cost of I/P episodes that exceed the HRG trim points.
Figures are computed that both do and do not take account of market forces
factors – predominately based on labour market variations.
This approach proved to have both practical and theoretical limitations, several of
which are described here. It included some unjustifiable assumptions, such as settings
specialty cost weights to 1.0 for specialties that were on average cheaper in THs.
(Though this might have been corrected by an economy of scale adjustment)
Applying English cost ratios to Scottish hospitals was complicated by definitional and
organisational differences between English and Scottish specialties. Moreover, the
organisation of teaching and the infrastructure of Scottish THs may be different from
their English counterparts.
In all, though this work was of considerable interest and produced estimates for the
costs of teaching were broadly similar to present levels of ACT, it would be hard to
justify using these results for resource allocation in Scotland.
5.1.2 Predicting gross Scottish TH costs using Scottish specialty level costs.
Quite extensive analyses were carried out on the differences in specialty case costs
between major teaching hospitals (MTHs) and other hospitals in Scotland. This
approach has several merits, not least that it provides a starting point for examining
the differential impact of teaching on specialties, a potentially important issue as
undergraduates tend to be placed in a limited number of specialties.
The difficulties with this approach almost all arise from the small number of cases
available for specialty level comparisons. That there are only eight MTHs in Scotland
some specialties only occur in these eight limits the potential (and robustness) of
direct comparisons and is the principal reason why we chose to adopt an alternative
method using hospital wide comparisons combined with indirect standardisation.
24
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
5.1.3 The impact of teaching load on cost
A substantial proportion of the literature on the costs of medical teaching attempts to
establish a relation between some measure of teaching load and cost. We have found
statistically significant positive correlations between teaching load (measured as
student weeks per in-patient admission) and cost across both the hospitals associated
with three English medical schools and the hospitals (other than MTHS in Scotland)
Further work on these correlations was abandoned. The English analyses could not be
extended due to difficulties of getting information on student placements. The
Scottish analyses were discontinued as the project changed emphasis and focussed on
the MTHs rather than the teaching loads in other hospitals.
By grouping hospitals into MTHS and others, the comparisons presented in the body
of this report treat teaching status as a binary rather than a continuous variable. The
argument for this approach is that MTHs have infrastructual features (giving rise to
indirect costs) that make them qualitatively different from other hospitals. However it
reduces the opportunities for exploring the impact of different levels of teaching.
5.2 The methods of the cost comparison in more detail.
The cost comparisons that form the basis of much of the rest of this report are quite
straightforward and the methodology should be obvious from the presentation of
results in Sections. However, it may be worth making a few preliminary points on
some definitional questions and on the methods used to control for confounders in the
comparisons.
5.2.1 What is a major teaching hospital (MTH)?
A basic difficulty in trying to estimate the indirect costs of teaching is the lack of a
clear distinction between teaching and non-teaching hospitals. The data in Annex 3 of
the earlier draft report on indirect costs indicate that almost every district general
hospital in Scotland has a significant levels of teaching responsibility as measured by
the number of student weeks. Indeed, relatively small hospitals such as Borders
General and Roodlands have a ratio of student weeks to patient activity, which is
higher than is found in some of the ‘major’ teaching hospitals. Non-teaching hospitals
are now confined to very small hospitals that, for various reasons, may not provide a
reasonable baseline for assessing the costs of providing treatment at major hospitals
Edinburgh Royal Infirmary in the absence of teaching.
Given the limitations of the available data, it is not easy to identify an alternative
approach to classifying hospitals. One possibility might be to focus the analysis on the
additional indirect costs of providing services in those hospitals that are generally
recognised as major teaching hospitals. Teaching responsibilities now appear to be
widely spread across different hospitals. According to the data in Annex 3 every NHS
25
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Board in Scotland with the exception of the 3 island Boards now has hospitals that
provide a significant number of student placements. Since our interest is in the
relative effect of teaching status on costs, it may be reasonable to focus on those
hospitals where teaching responsibilities are more heavily concentrated and where the
effects on indirect costs are likely to be more significant. These include:
•
•
•
•
•
•
Edinburgh Royal Infirmary
Western General
Glasgow Royal Infirmary
Western/Gartnavel
Ninewells
Aberdeen Royal Infirmary
Two hospitals which are sometimes regarded as major teaching hospitals because of
the scale of teaching carried out are:
•
•
Southern General
Raigmore
Specialist hospitals which are regarded as having major teaching status include the
following maternity and children’s hospitals:
•
•
•
•
•
•
•
Aberdeen Maternity
Queen Mother’s, Yorkhill
Princess Royal Maternity Hospital
Simpsons
RHSC, Yorkhill
RHSC, Edinburgh
Royal Aberdeen Children’s
For the moment we will assume that the eight hospitals listed above can be used as
the teaching wing of the comparison (these are the MTHs in this report), but a case
could easily be made for reducing or increasing this sub-set. The number of student
placements and the ratio of placements to caseload at these institutions are shown in
Table XX.
26
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Table 1 : Hospitals regarded as "teaching hospitals" for the purposes of
preliminary cost comparisons
Trust
South Glasgow University
Hospitals NHS Trus
Highland Acute Hospitals NHS
Trust
North Glasgow University
Hospitals NHS Trus
North Glasgow University
Hospitals NHS Trus
Lothian University Hospitals NHS
Trust
Grampian University Hospitals
NHS Trust
Tayside University Hospitals NHS
Trust
Lothian University Hospitals NHS
Trust
Hospital
Southern General
(SGH)
Raigmore, Inverness
Western / Gartnavel
Glasgow Royal
Infirmary
Western General,
Edinburgh
Aberdeen Royal
Infirmary
Ninewells
Edinburgh Royal
Infirmary
A
B C D
E
3711 183 2.2
3 9 6 4.96
815
3071 227 3.9
5 0 1 7.39
580
4902 285 3.2
1 4 4 5.82
882
4114 403 5.5
1 4 3 9.81
729
2640 484 7.6 18.3
1 6 0
6
637
5641 537 5.7
7 2 0 9.52
942
5131 690 8.3 13.4
4 7 1
6
832
4391 702 9.3 15.9
749
0 0 8
9
A: nos. of beds B:nos. of I/P discharges C: nos. of student week placements;
D:nos. of students weeks per bed;
e:100 times the no. of students week per patient discharged
5.2.2 Controlling for case-mix
There are two principal reasons why major teaching hospitals are likely to have a
more complicated and costly case-mix than other hospitals. Firstly, they function as
tertiary referral centres and, secondly, their metropolitan and urban locations may
result in their acting as general hospitals to a relatively deprived population with
above average levels of ill-health. Moreover, some of the most costly specialties
involving high staffing and equipment levels are more likely to be found in THs that
elsewhere. For these reasons, any cost comparison needs to control for both the
combination of specialties and the case mix within specialties.
The Scottish Hospital Cost Book provides figures for average cost per case by
specialty (or rather, by budget line). The figures are separated into inpatients
outpatients and day cases. They can be used to compute the relative costs of each
specialty that can be used as controls in cost comparisons.
Controlling for case-mix within specialties is rather more difficult as the Scottish
Hospital Cost data does not provide figures for the costs of HRGs. These have to be
obtained from the English Reference Costs database and combined with Scottish
hospital activity data to compute a case-mix weighting for each specialty in each of
the Scottish hospitals. Weightings were computed for inpatients and day cases in most
27
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
specialties, but the English Reference Cost data is too limited in its coverage of
outpatients to include them in the analyses.
Although this method of case-mix control is widely accepted (the English weights are
also used in compiling performance summaries for Scottish hospitals) it is sometimes
argued that HRG based weights inadequately represent the additional cost and
complexity of cases seen by tertiary centres. There is little direct evidence to confirm
or dispute this assertion, but the English cost data shows that the proportion of cases
out with the standard HRG trim points is, on average, higher in non-teaching than
teaching hospitals. This suggests that teaching hospitals do not have to admit people
for longer because they may be at the more complex or severe end of an HRG.
5.2.3 Controlling for economies of scale
Of the eight Scottish hospitals with the highest caseload, only one (The Royal
Alexandria) is not a MTH. If economy of scale effects can be demonstrated amongst
the Scottish hospitals, then the observed cost differences between the MTHs and other
hospitals may understate the full cost differences due to teaching and related effects.
The method used to control for economies of scale are widely reported in later
Sections of this report so only the key features are described here.
•
•
•
•
•
•
•
MTHS are excluded from the analysis as their cost /volume relationship is
confounded by the costs associated with teaching
(Hence) the analyses are based on data from 22 Medium to large hospitals
with wide case-mix
The Scottish average cost per specialty is combined with the specialty case
mix of these 22 hospitals to compute an expected average cost per case.
The ratio of actual to expected cost id plotted against caseload.
A logarithmic curve is fitted to the plot and extrapolated to the higher
caseloads found in the MTHs.
The ratios in the extrapolated part of the plot are used to estimate the
economies of scale in the MTHs and adjust the cost differences between
MTHs and other hospitals.
The analyses are repeated with a control for HRG case-mix as well as
specialty.
28
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
5.2.4 Controlling for market forces factors.
The metropolitan locations of many MTHS may have implications for wage costs.
This is certainly an issue in England and especially in London. The research tests
whether wage costs have a significant influence on the costs of Scottish MTHs.
The wage cost figures used by the study are those computed by The University of
Warwick Institute for Employment Research for the DoH (NHS Labour Market
Forces for Great Britain, 2002 Update) They are used in English resource allocation.
The figures are smoothed estimates from the New Earnings Survey Panel Data Set
and for the first time in 2002 there are estimates for each of the 21 NES areas in
Scotland. They report a 19% difference between the highest are lowest wage areas of
Scotland.
5.3 Data availability and reliability
The cost comparisons are mostly based on information from the Scottish Hospitals
Cost Book, Scottish Hospital Activity Data and The English NHS Reference Costs
Database. As has been previously mentioned in the discussion of case-mix
adjustment, the English Reference Costs are not just used for the purposes of
comparison, but also because Scottish hospital cost data does not provide costings for
HRGs.
Several problems arise is merging these data sets. Principal of which are the
differences between the English and Scottish specialties and the differences between
both sets of specialties and the budget lines used in the Scottish cost analyses. ISD has
provided conversion tables to assist in the merging, but some inaccuracies and some
loss of detail will result from the definitional differences.
The reliability of hospital cost data is an issue, especially for small specialities with
high variation in case-cost. The project has carried out several sensitivity tests
including a repeat analysis on a second year's data and noting the effect of removing
individual and groups of hospitals from the analyses.
The main problem of data availability (apart from the lack of costed HRGs in
Scotland) lies in the difficulties of obtaining indicators or good proxies for the various
factors that are cited as contributing to the extra cost of teaching hospitals, such as
research and innovation. The difficulties have been widely noted elsewhere and they
pose severe limits on the practicality of multivariate modelling.
Even where measures can be found, there is the problem that they may refer to
characteristics that are often barely present outside MTHs and, in effect, function as a
dummy variable that simply distinguishes MTHs from other hospitals.
The small number of MTHs in Scotland means that it is difficult to do any modelling
of the differences between teaching hospitals, especially as there is such a wide
variation in their cost bases. The small numbers also impinge on the comparisons as
29
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
there are wide confidence intervals around any MTH averages and models are very
sensitive to the exclusion of individual cases.
We have already noted that between hospital variations become even more extreme
when analyses are conducted at the specialty level and the sample size problem is
exaggerated because only a relatively small subset of specialties are common to most
Scottish hospitals and a considerable minority of specialities are largely confined to
the MTHs. The larger number of hospitals in England meant that the original SIFT
analyses could be based on specialty costs, but that work has been criticised (see e.g.
Bevan) for its attempts to use average or other summary measures for cost ratios that
exhibit very high levels of variation.
30
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
SECTION 6 STATISTICAL ANALYSIS:
COSTS
MEASURING ADDITIONAL
This section of the report details the findings from the statistical analysis to estimate
the extent of additional costs in major teaching hospitals. The analysis has been
carried out at hospital level, but concentrates mainly on the results from the ‘major
teaching hospitals’ (MTHs). While it is recognised that most hospitals undertake
significant amounts of teaching activity, the purpose of this study is primarily to
determine whether the traditional teaching centres have cost structures that make them
inherently more expensive than others.
The definition of major teaching hospitals, therefore, includes only the 8 main
teaching centres as defined in Section 5.2.1. Note that, the Sick Childrens Hospitals
have been excluded here primarily because the nature of the activity in these hospitals
makes it difficult to compare their costs structures against those of other hospitals.
The analysis begins with measuring overall cost differences between the MTHs and
the District General Hospitals (DGHs). Given that indirect costs are not tangible, the
method used is a ‘top-down’ approach. Once overall cost differences have been
calculated, the measurable elements can then be deducted – this includes an
adjustment for direct costs, and research costs. Any remaining unexplained cost can
then be attributed to the indirect costs of the major teaching hospitals.
6.1
Finding a suitable control for measuring overall costs
The Economies of Scale Effect: The starting point for calculating additional costs for
the major teaching hospitals is to find a suitable basis against which comparisons can
be made. The size of the hospital is a critical factor in assessing costs – it is expected
that as hospital size increases, the costs per unit will tend to fall (the economies of
scale effect). The major teaching hospitals on the whole, have significantly higher
patient volumes. This would suggest, therefore, that the cost ratios of these hospitals
need to compared with the cost ratio of a non-teaching hospital of a similar size in
order to make an accurate estimate of any additional costs that might incur as a result
of teaching activities.
Accounting for complexity of case-mix: A key feature of teaching hospitals is the
complexity of case-mix that arises from the greater range of activities carried out in
the larger hospitals. The effect of this will be to raise the relative costs of the
teaching hospital. In determining additional costs, the impact of case-mix has to be
netted out because it artificially increases the costs of a major teaching hospital when
comparing it to a non-teaching hospital of a similar size. All the results in this
analysis have been adjusted for case-mix. Chapter 6 highlight the value of the casemix adjustment.
Speciality Standardisation: In order to carry out a fair comparison, the analysis
requires adopting a method that allows hospital-wide comparisons to be made, by
standardising for specialty. The analysis, therefore, excludes some of the more costly
specialties that are specific to MTHs. In doing so, the cost comparisons are
31
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
effectively controlling for the combination of specialties, as well as the case-mix
within them.
Thus, in calculating overall additional costs for teaching hospitals, certain adjustments
are made that take into account factors that might distort the figures. The results
produced will show the extent to which costs are greater in MTHs compared to those
expected from a hospital of a similar size, and with no significant teaching
responsibilities.
6.2
Methodology used for measuring overall costs
The primary data source for the analysis is the ‘Scottish Health Service Costs’ book,
for the year 2001/02. The analysis was also carried out using data for the previous
year to check for consistency in results.
Step 1: Estimating cost ratios for District General Hospitals
The sample size consisted of 19 DGHs. All hospitals with a caseload of patients
(inpatients plus daycases) around 10,000 or below are excluded from the analysis due
to concerns regarding the robustness of data for the smaller hospitals. The economies
of scale effect which shows the extent of additional costs for MTHs is also very
sensitive to the trend-line that is based on the DGHs. Smaller hospital outliers can
significantly influence this trend-line and it is therefore, more appropriate to exclude
these hospitals.
Costs are computed across 22 specialties that are found to be common to all DGHs
and MTHs (see Annex A for a detailed list of hospitals and specialties).
Cost ratios are then obtained for each DGH, , where the cost ratio is defined by:
Actual Hospital Cost
Expected Hospital Cost
Actual Cost = sum of total costs for each speciality
Expected Cost* = sum of the total expected cost for each speciality
*the expected costs are derived by calculating a weighted average cost for
each activity across of all hospitals in the sample. In effect, this gives the
‘national average’ cost for each component
The advantage of using the cost ratio method is that it allows comparisons to be made
between hospitals by highlighting their relative differences in costs.
32
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Step 2: Plotting the results, and establishing the economies of scale effect
The cost ratios for the DGHs were plotted against hospital size (measured by
inpatients plus daycases). A trend-line, based only on the data points for the 19
DGHs, was added to the chart. Chart 1 below indicates that there is a negative
relationship between cost ratios and caseload: that is, cost ratios are falling as
hospital size increases.
By projecting forward the trend-line to account for increasing hospital size beyond
55,000 patients, the result shows a further expected decline in cost ratios (albeit, at a
decreasing rate). This is the economies of scale effect – theoretically, cost ratios for
larger DGHs with no significant teaching commitments can be estimated using the
trend-line. It is against these hospitals, that cost ratios for the Major Teaching
Hospitals can be compared.
Chart 1: Gross Cost Ratio of DGHs
Actual:Expected Gross Cost Ratios, 2001/02
1.60
1.40
1.20
Perth RI
Dr Grays
A:E
1.00
Borders General
Hairmyres The Ayr
Victoria Kirkcaldy
0.80
Crosshouse
Stobhill
Falki8rk RI
Wishaw
Queen Margaret
Stirling RI
St John's
D & G Royal
Inverclyde Royal
Monklands
Royal Alexandra
Victoria Infirmary
y = -0.0234Ln(x) + 1.231
0.60
0.40
0.20
0.00
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
Caseload
DGHs
Log. (DGHs)
Step 3: Calculating cost ratios for Major Teaching Hospitals
MTHs were excluded from the above calculation of expected costs. Cost ratios are
now calculated in a similar fashion, by using the national average expected costs and
applying these to activity in the Major Teaching Hospitals. Chart 2 shows the plot
of cost ratios for MTHs.
33
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Chart 2: Gross Cost Ratio for All Hospitals
Actual:Expected Gross Cost Ratios, 2001/02
1.60
Western General
1.40
Edinburgh RI
Ninewells
1.20
Glasgow RI
Perth RI
Dr Grays
A:E
1.00
Borders General
Hairmyres The Ayr
Aberdeen RI
Stobhill
Falki8rk RI
Raigmore Wishaw
Queen Margaret
Stirling RI
St John's
D & G Royal
Inverclyde Royal
Monklands
Victoria Kirkcaldy
0.80
Crosshouse
Western/Gartnavel
Southern General
Royal Alexandra
Victoria Infirmary
y = -0.0234Ln(x) + 1.231
0.60
0.40
0.20
0.00
0
10000
20000
30000
40000
50000
60000
70000
80000
Caseload
DGHs
Major Teaching Hospitals
Log. (DGHs)
The results show that the 8 Major Teaching Hospitals have higher cost ratios than all
the DGHs. In addition, there is a significant gap (measured vertically) between cost
ratio for each hospital and the point at which it touches the trend-line. This gap
measures the additional cost of a MTH, compared to what would be expected for a
DGH of a similar size.
There appears to be some evidence of economies of scale within the group of major
teaching hospitals - i.e. the ratio of actual to expected costs in major teaching
hospitals falls as the size of the teaching hospital increases. The small sample size,
however, makes it impossible to offer any solid conclusions without further
investigation into what factors might determine relative costs.
The tables below shows the additional costs incurred each of the Major Teaching
Hospitals , and also the results by Health Board. At this stage, note that there has
been no adjustment for case-mix.
34
90000
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Table 1: Additional Gross costs, by Major Teaching Hospital
Major Teaching
Hospital
Aberdeen RI
Edinburgh RI
Glasgow RI
Ninewells
Raigmore
Southern General
Western/Gartnave
l
Western General
TOTAL
6.3
mix
Additional
Cost
£m
11.7
30.7
16.2
28.2
0.1
4.4
14.1
19.3
%
difference
124.8
-
12%
36%
19%
31%
0%
6%
15%
46%
The Additional Costs of MTHs: Results of the Gross Costs Analysis (caseadjusted)
Chart 3 shows the plot of cost ratios for MTHs, after case-mix adjustments have been
applied. The main points to note are:
(i) the general relationship between cost ratios and hospital size stays the same
(ii) overall cost ratios for the Major Teaching Hospitals are reduced when the effect
of more complex cases are taken out. The difference between these ratios and
those for unweighted gross costs indicate the extent to which MTH costs are
higher due to these hospitals having a more complicated case-mix.
35
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Chart 3: Cost Ratios for Weighted Gross Costs, by Hospital
Ratio of Actual:Expected Weighted Hospital Costs (2001-02)
1.40
Western General
Forecast
Edinburgh RI
1.20
Falkirk RI
Perth RI
Borders General
Dr Gray's, Elgin
1.00
Stirling RI
D & G RI
Queen Margaret
0.80
Hairmyres
St John's
Stobhill
The
Inverclyde
Ayr
Victoria
Kirkcaldy
Ninewells
Glasgow RI
Crosshouse
Wishaw
Southern General
Monklands
Raigmor
e
Victoria
Aberdeen RI
Western/Gartnavel
Royal Alexandra
Infirmary
y = -0.0141Ln(x) + 1.1284
0.60
0.40
0.20
0.00
0
10000
20000
30000
40000
50000
60000
70000
80000
Caseload
DGH
Major Teaching Hospitals
Log. (DGH)
Table 2 shows the change in additional costs after adjusting for case-mix. It indicates
that case-mix adjustment reduces the additional costs of Major Teaching Hospitals by
around £34.4 million.
Table 2: Additional Costs of MTHs, adjusted for case-mix
Major Teaching
Hospital
Aberdeen RI
Edinburgh RI
Glasgow RI
Ninewells
Raigmore
Southern General
Western/Gartnavel
Western General
TOTAL
Additional
Cost
£m
6.4
21.8
10.8
22.6
-1.9
8.2
8.0
14.7
90.4
%
difference
6%
25%
12%
24%
-3%
12%
8%
34%
-
The results can also be aggregated to Health Board level as shown in Table 3.
36
90000
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Table 3: Summary of Gross Costs by NHS Health Board
Health Board
Grampian
Greater
Glasgow
Lothian
Tayside
Total ACT
Estimates of
Additional
Costs (£m)
Current
Allocation
(£m)
6.4
26.8
36.6
22.6
6%
11%
29%
25%
92.3*
-
Does not sum to £90m as Raigmore has been excluded from the Health Board
analysis.
Give that it has overall negative additional costs, this would reduced the total figure
here.
6.4
Limitations to the Analysis:
Some caution should be exercised when interpreting the calculated ACT figure of
£90m. The statistical analysis carried out only includes those specialties which are
also present among the DGHs. Other specialties that are limited in DGHs, or only
exclusive to MTHs have been excluded. For example, neurosurgery is excluded
because it is almost exclusively performed in the MTHs. The study recognises,
however, that there are excess teaching costs associated with such specialties, which
are not accounted for by this particular analysis.
6.5
Further Analysis: Additional Costs by Cost Category
The differences in cost ratios for the major teaching hospitals can be explored further,
by looking at the various direct cost components which make up the Gross costs
figure that was used to compute the results in the above section. There are seven
main categories: Dental & Medical, Nursing, Group Allocated, Theatre, PAMS,
Laboratory and Other. Annex B details the contents of each of these groupings.
The pie-chart below shows the relative proportions of each of these categories within
total costs for MTHs.
37
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Chart 4
Direct Costs as a proportion of the Total Actual Costs for MTHs
Other
PAM
5%
Medical & Dental
13%
6%
Pharmacy
9%
Laboratory
5%
Group Allocated
30%
Theatre
10%
Nursing
22%
Source: Scottish Health Service Costs, 2001/02
Each of the cost components were analysed separately. The purpose of this was to
determine whether MTHs have uniformly higher costs that cannot be explained, or
whether certain features of MTHs, such as the costs of medical and dental staff can
account for the higher cost ratios found in the earlier results. The main findings for
the individual cost components show that:
Medical and Dental, Group Allocated and Nursing costs all display strong evidence of
economies of scale. For each of these cost categories, cost ratios for the Major
Teaching Hospitals lie at points that are significantly higher than expected for similar
sized hospitals.
By way of example, the chart below shows the plot for Medical and Dental cost
ratios. These are adjusted for case-mix complexity. Tables 4 & 5 highlight the
additional costs in monetary terms, by individual hospital and by Health Board.
38
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Chart 5: Cost Ratios for Weighted Medical Costs
Ratio of Actual:Expected (Weighted) Medical Costs (2001-02)
1.60
Borders
1.40
Western General
Perth RI
Ninewells
Glasgow RI
Southern General
Monklands
1.20
Aberdeen RI
Dr Gray's Elgin
Queen Margaret
Cost Ratio
1.00
D&G RI
Falkirk RI
Edinburgh RI
St John's
Hairmyres
Stobhill
Victoria Kirkcaldy
Inverclyde
Stirling RI
Western/Gartnaval
Royal Alexandra
Wishaw
The Ayr
Crosshouse
Raigmore
Victoria Infirmary
0.80
0.60
y = -0.3189Ln(x) + 4.3382
0.40
0.20
0.00
0
10000
20000
30000
40000
50000
60000
Caseload
DGHs
Major Teaching Hospitals
Table 4: Additional Medical Cost by Hospital
Major Teaching
Hospital
Aberdeen RI
Edinburgh RI
Glasgow RI
Ninewells
Raigmore
Southern General
Western/Gartnave
l
Western General
TOTAL
Additional
Cost
£m
5.4
1.9
4.0
7.1
-1.3
2.7
3.5
2.3
%
difference
25.6
-
41%
18%
41%
61%
-17%
35%
29%
48%
39
Log. (DGHs)
70000
80000
90000
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Table 5: Summary of Additional Medical Costs by Health Board
Health Board
Grampian
Greater
Glasgow
Lothian
Tayside
Total ACT
Estimates of
Additional
Costs (£m)
Current
Allocation
(£m)
5.4
10.2
4.2
7.1
41%
42%
31%
61%
26.9*
(*Raigmore
exl)
-
There appears to be little evidence of any relationship between hospital size and cost
ratios for laboratory costs, theatre, pharmacy and PAM costs.
However, with
laboratory costs the cost ratios are still persistently greater than those classed as nonteaching hospitals. Also, within the Major Teaching Hospitals, the chart shows that
cost ratios tend to fall as voume increases (Ninewells is an outlier, although it does
not affect the downward trend apparent among MTHs).
Chart 6: Cost Ratio
Ratio of Actual: Expected (Weighted) Laboratory Costs (2001-02)
2.50
2.00
Edinburgh RI
Raigmore
Cost Ratio
Western General
Ninewells
Southern General
1.50
D&G RI
Borders
Falkirk RI
1.00
Glasgow RI
Stobhill
Aberdeen RI
Monklands
Victoria Infirmary
Gilbert Bain
Victoria Kirkcaldy
Wishaw
Stirling RI
Perth RI Queen Margaret
St John's
Inverclyde
Dr Gray's Elgin
Crosshouse
The Ayr
Western/Gartnaval
Royal Alexandra
Hairmyres
0.50
0.00
0
10000
20000
30000
40000
50000
60000
70000
80000
Caseload
DGHs
Major Teaching Hospitals
In calculating the extent of additional costs, it is impossible to estimate what the cost
ratio would be for a non-teaching hospital of a similar size, that can be compared to
MTHs, given that no relationship exists on which to base the assumptions.
40
90000
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
The additional costs have therefore, been calculated on the difference in cost ratios
compared to the national average. These are given for both hospital level and Health
Board.
Table 6 : Additional Laboratory Costs, by Hospital
MTH
Additional
Costs (£m)
Aberdeen RI
1.3
Edinburgh RI
2.6
Glasgow RI
1.2
Ninewells
2.3
Raigmore
1.6
Southern General
1.0
Western/Gartna
-0.1
vel
1.4
Western General
TOTAL
11.2
Table 7: Summary of Additional Lab Costs by Health Board
Health Board
Grampian
Greater
Glasgow
Lothian
Tayside
Total ACT
Estimates of
Additional
Costs (£m)
Current
Allocation
(£m)
1.3
2.1
4
2.3
32%
23%
88%
76%
10*
(excl.
Raigmore)
-
41
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Analysis Summary and Implications
This section of the report has looked at estimating the extent of overall additional
costs in Major Teaching Hospitals (MTHs). The purpose of this is to provide a
starting point from which indirect costs can be calculated.
The main findings indicate that:
♦ Overall, major teaching hospitals are shown to have costs that are greater than
would be expected for hospitals of a similar size. Across the 8 MTHs, this is
estimated to be around £124.8 million.
♦ Complexity of case-mix accounts for around £34 million of the additional costs.
Adjusting for this results in an estimated additional cost of teaching hospitals at
£90.4m.
♦ The additional costs are explained to some extent by medical and dental, group
allocated and nursing costs where cost ratios are persistently higher than those
found for District General Hospitals. Laboratory costs are also significantly
higher than the national average.
♦ The general pattern of high cost ratios within MTHs does vary between cost
category. For example, while Edinburgh Royal Infirmary has relatively lower
cost ratios for Medical and Dental costs and Nursing, its costs are relatively
greater than other MTHs for Laboratory and Theatre costs. The only exception
is Ninewells, which appears to have consistently higher cost-ratios in every cost
category.
♦ There is also considerable variation between the additional costs by Health Board
and within Health Board. It is not clear why the cost ratios of MTHs that are of
similar size should vary, nor why hospitals within the same Health Board should
have significantly different cost ratios.
42
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
7. CONCLUSIONS
7.1. Direct cost modelling
The direct costing exercise has collated together – for the first time – a vast database
of comparable information on activities and costs across all four Medical Schools.
This is a significant achievement given the complexity of the curricula and the variety
in delivery between Schools.
This information now makes possible much more detailed analysis of individual cost
structures, NHS inputs and overhead recharges that has hereto been possible.
Section 4.5 of this report has described several (out of many) ways in which the
results from the direct cost modelling might be used in resource allocation. It posits
four options all of which require top-slicing of the aggregate quantum of funding
identified.
Option 1: Make allocations to Boards on the basis of the number of students
supported in the local NHS at a single Scotland-wide average cost per student across
all years
Option 2: A similar approach to option1 – except that a different Scotland-wide cost
is applicable to students in each year of the curriculum.
Option 3: Delivery Unit costing
In which information is collected from the Medical Schools on the volume of teaching
activity delivered by each institution (teaching and non-teaching hospital) including
that delivered in primary care settings. Scottish wide costs are then applied to each
activity to
determine a ‘budget’ for each institution, which are then summed to give a Board
level total.
Option 4: Allocations by student week
In which Scottish average costs are derived for a student week in different teaching
settings as the basis of a compensation mechanism to all Boards providing
undergraduate teaching.
All four models are described in detail in Section 4.5, which also includes a set of
specimen calculations using the first option. With this option, although the quantum
of funding has significantly reduced, the share of resources under the proposed
mechanism is, for each teaching Board, within 2% points of the ACT shares.
Options for direct cost compensation are currently under further development by
Analytical Services Division, Scottish Executive Health Department.
43
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
7.2 Estimating overall cost differences and indirect costs
The most innovative feature of this project has been the separation of costs associated
with teaching into direct and indirect components – a division that might be carried
into procedures for resource allocation.
This approach differs from the most obvious comparison case, the SIFT (Service
Increment for Teaching) mechanism for allocating funds for undergraduate medical
placements to English teaching hospitals. The derivation of SIFT concentrated on the
overall cost differences between teaching and other hospitals and analysed these
differences by specialty. In so doing, it found great variability in the case costs of
specialties at teaching hospitals and there have been criticisms in the way that these
diverse costs were “averaged” in the development of a funding mechanism.
A similar approach was explored by this project, using specialty case costs for both
Scottish and English hospitals. We found that the average differences in specialty case
costs (after adjusting for case-mix) between teaching and non-teaching hospitals in
England, when applied to Scottish teaching hospitals, predicted an overall cost of
teaching that is similar to present levels of ACT minus estimated direct costs. This
result supports the assumption that Scottish and English teaching hospitals have a
similar cost base because the English cost differences are based on figures that should
already exclude most direct teaching costs.
Building on these results, the specialty level analyses were applied directly to the
Scottish hospital cost data, albeit with the English HRG costs used to control for casemix. The recurring difficulties with such work, as has been noted by commentators on
SIFT, is the high variability in specialty level data, the small number of teaching
hospitals available for analysis and the absence of comparable data for those
specialties that tend to be concentrated in teaching hospitals. Despite these
difficulties, a few regression analyses were applied to the various components of
specialty costs presented in the Scottish hospital data. However, the problems of high
unexplained variability and small numbers of cases led to this line of investigation
being abandoned.
Subsequently, ASD has taken the lead in analysing overall cost differences at the
hospital level, leading to the results presented in Section 5 of this report. Parallel
analyses of the Scottish hospital costs, conducted by MSA-Ferndale, have produced
very similar results. Moreover, further analyses of the English cost data have
identified close parallels to the Scottish results, notably very similar economy of scale
effects to those found for the non-teaching hospitals in Scotland.
The shift in the level and methods of analyses was accompanied by something of a
shift in objectives. These became: firstly, to establish the overall difference in cost
between the two groups of hospitals; secondly, to remove identifiable elements such
as direct costs and research income from these differences and, thirdly, to explore the
causes of the differences between the MTHs.
44
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Results relating to the first of these objectives have been reported in Section 5 of this
report. The main findings are that:
♦ Overall, major teaching hospitals are shown to have costs that are greater than
would be expected for hospitals of a similar size. Across the 8 MTHs, this is
estimated to be around £124.8 million.
♦ Complexity of case-mix accounts for around £34 million of the additional costs.
Adjusting for this results in an estimated additional cost of teaching hospitals at
£90.4m.
These additional costs of £90M include components that can be separately estimated,
such as the direct costs of teaching and some forms of research funding. A separate
paper will be presented to the ACT Sub-Group by ASD discussing the implications of
subtracting the costs of these components from the £90M and approaches to
explaining the remaining additional cost.
In relation to resource allocation, the outstanding problems are how to fund these
additional costs and whether it is possible to develop any sort of formula for hospital
level allocations. Several mechanisms have been suggested for using the results from
the direct costs model as the basis for allocating resources to cover direct costs (see
above and Section 4.5). Although these differ in detail, their common feature is that
cost estimates will be more or less proportional to numbers and lengths of placements.
The relation between indirect costs and teaching load is far less clear. On the one
hand, it can be argued that these costs relate to infrastructure, services and other
activities that are relatively independent of student numbers. On the other hand, there
is some evidence that additional (indirect) costs are correlated with the numbers of
student placements amongst the 8 MTHs as well as in the other Scottish hospitals that
host student placements. The relationship is statistically significant, but is not
sufficiently robust to use as the sole, or possibly even a major, basis for resource
allocation.
ASD have been examining other factors that might explain the cost differences
between the 8MTHs, but many of these are more obviously related to the type and
quality of care than the needs of undergraduate teaching. Whether it is more
appropriate to fund them via mechanisms other than ACT is one of the issues raised in
the separate ASD paper to the Group.
45
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
References
Anderson. (1987) Teaching hospitals in the USA: institutions under stress.
Financial Accountability and Management 1987:3 117-128
Australian Universities Commission AUC (1965).
Second Report of the Committee on teaching Costs of Medical Hospitals.
Bevan G (1999) The medical service increment for SIFT: a £400 anachronism for the
English NHS? BMJ 1999:319:90811 (Oct 2nd)
Bevan G (1987) SIFT - an exposition and critique
Financial Accountability and Management 1987:3 147-160
Bevan G and Rutter (1987) Organisation and functions of teaching hospitals in
different countries. Financial Accountability and Management 1987:3 77-115
Bevan G. (1987) Funding English teaching hospitals by capitation.
Financial Accountability and Management 1987:3 161-73
Busby DD James C Leming I Merlin I (1972) Unidentified educational costs in a
university teaching hospital: an initial study. Jou Medical Education 47.4 April 1972
243-253
Culyer A Drummond MF (1978) Financing medical education: inter-relationship
between medical school and teaching hospital expenditure. In Culyer A et al (eds)
Economic Aspects of Health Services. London Martin Robertson 1978
Culyer,A.J., Wisenam,J.,Drummond,M.F.,West, P.A. (1978) What Accounts for the
Higher Costs of Teaching Hospitals? Soc and Econ Admin 12 1 Spring (1978)
Detsky AS Abrams HB Ladha L Stacey SR (1986) Global budgeting and the teaching
hospital in Ontario. Medical Care 1986:24(1) 89-94
Easthaugh SR. (1979) 'Cost of Elective Surgery and Utilization of Ancillary
Services in Teaching Hospitals' Health Services Research, 1979, Volume
14, Number 4, pp290-308.
Foster (1987) Literature review: on estimating the costs of the products of teaching
hospitals. Financial Accountability and Management 1987:3 175-192
Foote G Hurst J and Sondheimer P (1988) Technical paper on the service increment
for teaching (review of the resource allocation working party formula) London
Economic Advisors Office, DHSS
Garg ML, Gliebe WA and Elkhatib MB. (1978) 'Diagnostic Testing as a cost factor
46
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
in teaching hospitals' Hospitals, 1978, Volume 52, Number 14, pp97-100.
Gaynor M and Anderson GF (1995) Uncertainty demand and structure of hospital
costs, and the cost of empty beds. Jou OF Health Economics 1995:14:291-317
Gonzalez-Lopez Valcarcel B and Barber P (1996) Changes in public Spanish hospital
efficiency after the program contracts. Investigaciones Economicas 20 3-20
Grannemann T Brown R and Pauly M (1986) Estimating hospital cost, a multiple
output analysis. Jou Health Economics 5: 7-127
Griner PF and Liptzin B. (1971) 'Use of the Laboratory in a Teaching Hospital'
Annals of Internal Medicine, 1971, Volume 75, pp157-163.
Hadley, J (1983) Teaching and Hospital Costs. Jou Health Econ 1983; 2: 75-9
Horn, Horn RA and Moses H. (1986a) 'Profiles of Physician Practice and Patient
Severity of Illness' American Journal of Public Health, 1986, Volume 76,
pp532-535.
Horn, Horn RA and Moses H. (1986b) 'Reliability and Validity of the Severity of
Illness Index' Medical Care, 1986, Volume 24, pp159-178.
Institute of Medicine (1974) Report of a study: Costs of Education in the health
professions: Parts I and II (Nat Academy of Science Washington DC).
Lave JR (2001) Reflections on "a longitudinal study of the effects of graduate medical
education on hospital operating costs" (Editorial] Health Services Research
35(6):1203-6
Lehner L Burgess J (1995) Teaching and hospital production: the use of regression
estimates. Health Economics 1995 4: 113-25
Linna,M., Hakkinen,U., Linnakko, E. (1998) An Econometric Study of Costs of
Teaching and Research in Finnish Hospitals. Health Econ 7: 291-305 (1998)
Linnakko E and Linna M (1995) Costs and reimbursements of teaching and clinical
research . Paper presented at the 3rd European Conference on Health Economics
Stockholm Aug 1995
Lopez-Casasnovas,G., Saez,M. (1999) The Impact of Teaching Status on Average
Costs in Spanish Hospitals Health Econ 8: 641-651 (1999)
Martz EW Ptakowski R (1978) Educational costs to hospitalised patients.
Jou Medical Education 53.5 May 383-6
47
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Mechanic R Coleman K Dobson A (1998) Teaching hospital costs: implications for
acadeic missions in a competitive market. JAMA 280(11) 1015-9
Milne R Abebe A Torsney B (1989) The impact of teaching on hospital costs: a
budgetary approach to non-market institutions. Jou of Operational Research Society
1989; 40:1089-98
Morey R, Ozcan Y Retlaff-Roberts D Fine D (1995) Estimating the hospital wide cost
differentials warranted for teaching hospitals. An alternative approach to regression
aprroaches Medical Care 1995; 33:531-47
NHS Executive (1995) Health Service Guidelines HSG(95)59
Service Increment for Teaching: Operational Guidance
NHS Executive (1995) Health Service Guidelines HSG(95)60
Guidance to NHS Trusts on Costing for SIFT Contracts
Palmer G Aisbett C Fetter R et al (1991) Estimates of costs by DRG in Sydney
teaching hospitals: an application of the Yale cost model.
Australian Health Review 14(2) 127-136
Perrin (1987) Financial Accountability and Management 1987:3 209-230
Perrin and Magee (1982) The Cost, Joint Products and Funding of English Teaching
Hospitals WPIBA No8 CRIBA Univ of Warwick
Schroeder S.A. O’Leary D.S. (1977) Differences in laboratory use and length of stay
between university and community hospitals.
Jou Medical Education 52.5 May 418-420
Sloan, F., Feldnam,R. Steinwald,A. (1983) Effects of Teaching on Hospital Cost. Jou
Health Econ 1983; 2:1-28
Sloan F Perrin JM Valvona J (1985) The teaching hospitals growing surgical
casseload. JAMA 254(3) 376-382
Raynor M (1985) Teaching and non-teaching hospitals: case and activity comparisons
London DHSS Operations Service 1985 (ORZ1159/2)
Rogowski, j. Newhouse J (1992) Estimating the Indirect Costs of Teaching Jou
Health Econ 1992 11:153-71
Sheldon TA (1990) The Leicester University study of undergraduate clinical teaching.
Leicester: Trent Regional Health Authority and Leicester University (1990)
Sheldon TA Clarke M Woods J (1991) The student diary survey: a method of
monitoring hospital based medical education. Med Education 1991 25 213:3
48
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Sheldon TA (1991) The NHS Review and the funding of teaching hospitals.
Management in Medicine 1991 5 617
Sheldon TA (1999) letter to BMJ
Snijders EHM Sol JCA Schepers J Stevens FCJ Groot LMJ (1987) The Allocation of
Costs of University Hospitals: an overview of the methods used in the NEtherlands
Financial Accountability and Management 1987:3
Thorpe K (1988) The Use of Regression Analysis to Determine Hospitals Payment:
the case of Medicare’s indirect teaching adjustment. Inquiry 1988; 25: 219-31
Wagstaff A and Lopez-Casasnovas (1996) Hospital costs in Catalonia: a stochastic
frontier analysis Applied economic Letters 3:471-4
Weinberg E. O'Sullivan P. Boll AG. Nelson TR (1994) The cost of third-year
clerkships at large non-university teaching hospitals. JAMA 272(9):669-73
Welsch W (1987) Do all teaching hospitals deserve an add-on payment under the
prospective payment system? Inquiry 1987 24: 221-32
Zuckerman S Hadley J and Iezzoni L (1994) Measuring hospital efficiency with
frontier cost functions Jou Health Economics 13:255-80
Zuidema GD. (1980) 'Problem of Cost Containment in Teaching Hospitals - John
Hopkins Experience' Surgery, 1980, Volume 1, pp41-45.
49
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Annex 1
(a) Hospitals used in study
MTHs
Aberdeen Royal Infirmary
Edinburgh Royal Infirmary
Glasgow Royal Infirmary
Ninewells
Raigmore
Southern General
Western/Gartnavel
Western General, Edinburgh
DGHs
Borders General
Crosshouse Hospital
D & G Royal Infirmary
Dr Grays, Elgin
Falkirk Royal Infirmary
Hairmyres, East Kilbride
Inverclyde Royal Hospital
Monklands
Perth Royal Infirmary
Queen Margaret Hospital
Royal Alexandra Hospital
St John’s
Stirling Royal Infirmary
Stobhill
The Ayr Hospital
Victoria Infirmary
Victoria Kirkcaldy
Wishaw
50
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
(b) Specialties used in study
A&E
Coronary Care Unit
Dermatology
Ear, Nose & Throat
General Surgery
Geriatric Assessment
Gynaecology
Haematology
Medical
Medical Paediatrics
Nephrology
Obstetrics Specialist
Ophthalmology
Oral Surgery & Medicine
Orthopaedics
Pain Relief
Plastic Surgery & Burns
Rehabilitation Medicine
Respiratory Medicine
Rheumatology
Special Care Baby Unit
Urology
(c) Specialties excluded from study
(i) Those specialties which are performed in Major Teaching Hospitals, but in very
few/no DGHs. These are cardiothoracic surgery, communicable diseases,
neurosurgery, surgical paediatrics and spinal paralysis.
(ii) Those specialties which are carried out in District General Hospitals, but not in
MTHs. These are acute other and general practice.
(iii) Those specialties which have significant numbers of outpatients, but a very
limited number of inpatients. These are radiotherapy, neurology and dental.
51
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Annex 2
Costs are analysed between Direct Costs and Allocated Costs with, in some cases, a
sub-analysis of Direct Costs. Details of the contents of each are shown below.
Medical and Dental
Medical and Dental staff
Nursing
Nursing staff
Pharmacy
Pharmacy staff and direct supplies, i.e. drugs, dressings,
instruments and sundries, TSSU and CSSD.
PAM
PAM staff directly involved in patient care and direct
supplies,
i.e. radiography, physiotherapy,
occupational therapy, industrial
therapy, chiropody and
any other P & T departments,
paramedical
equipment purchase, rental and repair.
Other Direct Care
appliances,
rental and repair.
Other direct care staff and supplies, i.e. surgical
medical/surgical equipment purchase,
Theatre
Theatre staff and theatre supplies
Laboratories
account
staff and supplies and
costs, heat, light and
Laboratory costs are likely to emanate from a trading
which will include the costs of direct
allocated costs such as premises
depreciation.
Allocated costs
All other costs not included as direct costs,
i.e.
Administration
Nurse teaching
Catering – patients and staff
Bedding and Linen
Patients Clothing
Uniforms
Laundry
Portering
Residences
Waste Disposal
Transport and Travel
Property maintenance
Cleaning
Heating
Rent and Rates
52
Research on the Additional Costs of undergraduate medical Teaching in NHSScotland
A Joint Report for the Standing Committee of Resource Allocation in NHSScotland by Secta
Consulting & Analytical Services Division, SEHD
Furniture and other equipment purchase,
rental and repairs
Depreciation
Notional interest
Miscellaneous
53
Download