Personne! planning in health care:
an example in the field of rheumatology
Th.J.B.M. Postma, State University of Groningen, The Netherlands*
M.T. Smits, Tilburg University, The Netherlands
S. Terpstra, University Hospital, Groningen, The Netherlands
C.A.Th. Takkenberg, Tilburg University, The Netherlands
ABSTRACT
This paper describes a methodology for manpower planning in health care. The
methodology is applied in the field of rheumatology. This methodology uses the
concept of political rationality: different actors with different mental models,
goals, languages and power interact in a bargaining process with incomplete and
imperfect information. A Group Decision Support Systems approach is advocated
where interactive mode! building stimulates shared meaning and communication. In
health care important decisions usually have a multi-level and multi-actor
character. A bottom up procedure, starting at the detailed level gives a justification
when aggregating to an higher level. Consequently the project was started with
discete event models before apptying continuous simulation like system dynamics.
Besides the modelling and communication processes the creation of a network of key
decision makers in health care applying this approach is seen as a major product.
1. Introduction
In the Netherlands, but also in other western countries,
manpowerproblems in the field of health care are regular items
on the agenda of policy makers. At the one hand structural
elements determine the demand and supply of manpower. At the
other hand more cyclical phenomena can be observed. Examples
are the supply of dentists and ophtalmologists in the
Netherlands. The Ministry of Welfare, Health and Culture
therefore initiated a research project evaluating the existing
manpower planning methodologies for the health care sector.
About the same time the National Board for Hospital Facilities,
an advisory body of the Dutch government, was confronted with
a question implying a new standard for the number of medical
specialists, especially rheumatologists, for The Netherlands.
For both purposes the authors of this paper were independently
approached and decided to integrate both projects.
The authors have the opinion that the use of simulation models
in the context of Group Decision Support Systems (GDSS) will
offer good possibilities to support policy makers in the field of
health care and give arguments for it in this paper.
Economics Faculty, P.O. Box, 9700 AV Groningen, The Netherlands
= 515.4
:
The twofold formulated research question is:
-How to design and develop a methodology for decision making
in the field of manpowerplanning for health care purposes?
-How to design and develop a group decision support system for
capacity decisions (the number of rheumatologists) with
respect to rheumatic diseases? This research question was also
put forward by The Dutch Society of Rheumatologists.
The first question is aimed at obtaining an generic approach to
be used for manpowerplanning of different medical specialist
groups. The second question can be interpreted as a pilot study
for the first question. This paper concentrates on the first
question.
The research approach chosen consists of a combiredicn of a
Delphi study, a workshop and a scenario approach, and is based
on a_ study of recent publications in the field of
manpowerplanning (Postma et al., 1992). The Delphi study and
workshop are interactive means to generate consensus on
needed models and data. The scenario approach is based on
simulation models. The scientific challenge of this part of the
project is the determination of the necessary level of
aggregation of the simulation models and the relation between
these levels. We started the project at the regional or local
level with an object oriented approach (individual patients and
medical specialists), using discrete event simulation. At the
national level, however, a system dynamics approach seems to
be better. At this time we expect that a combined approach is
needed. One of the goals of this project is to develop a group
decision support system, in which gaming elements are included.
This: fits the way policy decisions are made or prepared in the
Netherlands, namely negotiations and interactions at different
levels of decision making.
The above mentioned elements will be worked out in the
following sections of this paper. We will start with some
characteristics of the Dutch health care system.
- 516 -
system
necessary to gain more insight in health care
participants.
2.1. Levels
levels of decision making; see table 1.
2. Some characteristics of the Dutch health
Decision making in Dutch health care is an activity in which
many participants on different levels play their roles. Decision
making is not just a “top down" process, but is also initiated and
implemented “bottom up" and “middle out". Therefore
levels and
Organizations and institutions can be classified according to
LEVELS ORGANIZATIONS/INSTITUTIONS
INTERNATIONAL
International professional associations
Government
Departments
Advisory boards
#
National professional associations
REGIONAL Provincial boards
Sickfunds
Cooperation of hospitals or physicians
Hospitals
Universities
Departments
Physicians
Table 1: Levels, Organizations and Institutions in Dutch health care
+ 517 =
This table gives only an impression of the different relevant
levels and organizations/institutions. For instance, if we zoom
in and have a more detailed look at the Dutch advisory boards,
we observe more than five relevant boards.
2.2. Participants
Because of the multitude of possibilities it is of no use to
enumerate all types of participants. One can better think of
networks or relation patterns of participants. For example, in
the field of reumotology, at the hospital level more than 10
different participants work together in one way or another and
have a very close relation.
Given the discussion untill now, we conclude that there are
many levels and actors in the field of health care. These, actors
participate in different networks on different levels. The
interests of these actors and networks can be more or less
opposite but also more or less compatible. This might result in
Political activities such as power plays, lobbying and possibly
manipulation of communication and information. The methaphor
of the "garbage can" could be a valid one (Cohen, et al., 1972).
The Dutch health care system therefore can be characterized as
a multi-level and multi-actor system, with coalitions, and
negotiation and political decision making. Moreover the context
of the health care sector at the macro level is dynamic and
sometimes even turbulent: developments in politic, economics,
law, technology, medical-ethics, demographics, and
epidemiology will be taken into account.
3. Changing visions on rationality
It is a well known fact that there is a gap between economic
theory and the way decisions are taken in practice. In micro
economic theory the paradigm of unbounded rationality is still
alive (Douma and Schreuder, 1991). However, axiomatic
foundations like perfect foresight, knowledge of all alternatives
and the ability to select the best alternative are not realistic
when studying processes of real problem solving of individuals
and groups. Especially when looking at the Dutch health care
system, the “garbage can" character as described in section 2
gives a poor resemblance with the traditional economic way of
looking at reality. The last 40 years, visions on rationality have
- 518 -
been changing and the gap between theory and practice has been
closing slightly.
Newell and Simon (1972) concentrate on individual problem
solving. Human problem solving is restricted by the nature of
the cognitive system. Individuals are not able to optimize and in
general use simple heuristic rules to come to a_ satisfactory
solution of a problem.
Simon introduced "bounded rationality" and this opened the way
to many other approaches of rationality. Simon (1976) mentions
“substantive rationality" and “procedural rationality". The tatter
tefers to the use of heuristic rules and search for satisficing
solutions. “Substantial rationality" resembles in some ways the
properties of the homo economicus and gives a link with the
phenomenon of the modern computer. Computers made it
possible to define very large sets of alternatives and provide
impressing processing power when searching for good
alternatives. In many cases the efficiency of this search
Process is influenced by the quality of the selecting algorithm.
Therefore “substantive rationality" is almost an “algorithmic
rationality” (Takkenberg, 1983) and the homo economicus is
transformed into the "homo informaticus". We find examples in
applications of mathematical programming and game theory.
When studying planning and policy making, we nowadays notice
the use of computers and software for decision support that
relax cognitive restrictions of man in solving semi-structured
problems.
In practice, looking at the ill-structured problems of health
care manpower planning, we notice that the paradigms of
Procedural rationality and substantive rationality need
extensions to give a sufficient explanation of the actual
decision making processes. Therefore we will introduce
"political rationality’ (Baakman, 1990), enabling the description
of decision making in a situation where different actors with
different mental models, goals and languages play a role: muiti-
actor situations. Planning situations in health care will be
considered from this political paradigm. Power relations have a
strong influence. In general, planning in health care often
resembles bargaining with incomplete and imperfect knowledge.
- 519 -
In our view choice and the implementation of decisions have to
rely on a certain consensus among participants and has to be
based on the use of validated models to gain confidence (Smits,
1990). The related methodology as sketched in the introduction
will be described in more detail in section 5 and aimes at the
construction of validated conceptual and empirical models that
enable communication between participants. Communication
will probably lead to the required consensus. In this respect
"Group Decision Support Systems" can be considered as an
adequate support instrument. DeSanctis and Gallupe (1987)
describe GDSS as follows: “Group Decision Support systems
combine computer, communications, and decision technologies
to support problem finding, formulation and solution in group
meetings”. Recently research for a fundamental basis of G@Dss
was published by Scheper (1991).
We will now first pay attention to a problem mentioned” before,
namely the level of aggregation in modelling.
4. Choosing the level of aggregation in simulation
Practice
Reality can be viewed very globally or more detailed. At first
sight, manpower planning for health care at the national level
requires an aggregate model, but these models may be too
abstract. At a global level, variables of the detailed level are
frequently lost. In our opinion the choice of the level of
description and analysis in simulation is an important one.
When using detailed models, we in general have to restrict
ourselves to a partial view of a system. This is caused by both
technical constraints of the computer support system and
cognitive constraints of the participants. For example, it is
Practically impossible to construct empirical (operational)
models of health care at the national level when flows of
individual patients and detailed decisions of medical specialists
have to be taken into account.
Even if such an outstanding technical DSS could be constructed,
the analysis of its dynamic behaviour would be hampered by our
cognitive system. When using aggregate models, we often
assume that the model is sufficiently representative for all
subsystems. This assumption has to be tested and forces the
researcher to analyze a sample of subsystems in more detail.
- 520 -
Therefore we recommend a methodology with a bottom
up approach: start with detailed studies of a sample of
subsystems and later on take a justified step to a more
aggregate level.
This recommendation will be sustained by the results of the
Delphi study in the case of rheumatology. We concluded that the
decision making and practice of rheumatology in the regions
studied were rather different. Therefore a higher level of
modelling (a national model) generates additional requirements
to assure a sufficient resemblance between model and reality.
This approach also includes different modes of simulation: on a
detailed level discrete event simulation with an object oriented
approach will be used, and later on continuous models based on
system dynamics will be tried. Following these steps the
consistency between the detailed models and the aggregate
models will be checked.
5. The methodoloy used in the rheumatology case
Decision making in the field of health care policy is a complex
process: many participants representing different parties try to
transform and influence the decision processes and the ultimate
decisions. All participants are driven by particular, specific,
hidden and/or private goals. It is difficult to arrive at policy
decisions that are supported by all parties involved (consensus).
This is particularly true for manpower planning in health care.
The methodology chosen and further developed in this study is of
the policy-delphi type, that is, using a Delphi approach is
combined with a structured workshop to construct a conceptual
model of the problem situation (Vennix, 1990). Experts in model
building as well as several experts in the field of health care
are involved in these activities. The steps in table 2 describe
the recommended methodology. Relatively new in this approach
is the modelling on two fevels of aggregation and therefore
combining discrete event simulation and continuous simulation;
the motivation was given in section 4.
= 521 =
1. Formation of a project group and a steering committee.
Literature study.
2. Definition of a preliminary conceptual model.
Development of a questionnaire for Delphi interviews.
3. First Delphi round with a sample of specialists.
4. Regional analysis of several selected geographical areas.
Collecting data and constructing data bases.
Development of questionnaire for Delphi studies at regional levels.
GDSS based on Discrete Event simulation of regional models.
7. Combining the regional modets into a national model, using GDSS.
with System Dynamics or other continuous simulation methods.
8. Applications of the GDSS in negotiations. with parties
involved in the health care process concerned.
Table 2: GDSS methodology recommended for health care
The analysis of the actual organization of rheumatological care
in the Netherlands is first performed on a regional level. In each
region the actual demand and supply of rheuma care are analyzed
and a prognosis is given of the desired care in the coming 20
years. This analysis is used for the building of a formalized
empirical model of rheuma care on a_ regional basis, using
discrete event simulation. Modern software like EXTEND, using
the graphical user interface of Apple Macintosh, enables
interactive modelling and stimulates communication.
Steps 6 to 8 of the methodology consist of building formalized
(empirical, operational) models for several regions. A second
Delphi round will be executed to evaluate the regional models.
For building the national model the continuous simulation mode
of EXTEND or the use of the STELLA package is quite adequate.
Once we have arrived at validated models (Kleijnen, 1992), the
final step will be the exploration of the future through scenario
analysis.
One of the main factors to be analyzed will be the influence of
demographic evolution. In this respect a fascinating question
concerns the tuning of the demand for physician manpower to
the supply, a population with more aged people will be taken
into account. When sketching the methodology, it seems that the
modelling cycles play a central role. This is true, but in our
opinion the process as a whole -where parties involved are
stimulated to develop a common mode! and shared meaning- is
of essential importance. This will be the ingredients for real
communication and the basis for practicable decisions.
6. Conclusions
Manpower planning in health care is a process to be viewed with
political rationality. This means that different actors have
different mental models, goals, languages and power. The
decision process resembles a negotiation process with many
aspects, based on incomplete and imperfect knowledge.
The GDSS approach aimes at solid mode! building as well as the
stimulation of communication between parties. Object oriented
modern software using a graphical user interface plays a major
role in interactive model building and exploration of the future
by a scenario method. Discrete event simulation is used on the
detailed levels where as systems dynamics (continuous
simulation) is applied at the aggregate level.
In the setting of healthcare manpower planning we also consider
the construction of a network of people involved as a major
product.
Modelling complex manpower decision processes in the field of
health care requires insight at both a detailed and at a more
aggregated level. Therefore manpower planning research in this
field needs to be aware of several levels of aggregation and the
relation between levels. Our research is still ongoing and not
yet finished. The foundations for an implementable GDSS,
however, have been laid.
= 523)=
LITERATURE
Baakman, N.AA., (1990), Kritiek van het Openbare Bestuur, Doctoral
dissertation, Thesis Publishers, Amsterdam.
Cohen, M.D., J.G. March and J.P. Olsen, (1972), A garbage can model of
Organizational Choice, Adm. Science Quarterly, 17,1, pp. 1-25.
DeSanctis, G., R.B. Gallupe, (1987), A foundation for the study of Group Decision
Support Systems, Management Science, Vol. 33, No. 5, May, pp. 589-609.
Douma, S., H. Schreuder, (1991), Economic Approaches to Organizations, Prentice
Hall, New York.
Kleijnen, J.P.C., (1992), Verification and Validation of simulation Models,
Research Memorandum of the School for management and economics, Tilburg
University.
Newell, A. HA. Simon, (1972), Human Problem Solving, Prentice Hall,
Englewood Cliffs.
Postma,Th.J.B.M., M.T. Smits, S. Terpstra, C.A.Th. Takkenberg, (1992),
Mankrachtraming in de gezondheidszorg, Stuurgroep Toekomstscenario's
Gezondheidszorg (STG), Rijswijk.
Scheper, W.J., (1991), Group Decision Support Systems, Doctoral dissertation,
Tilburg University.
Simon, H.A., (1976), From substantive to procedural rationality, \n:
Kastelein,T.J., S.K. Kuipers (eds), 25 years of economic theory, retrospect and
Prospect, Martinus Nijhoff, Leiden.
Smits, M.T., (1990), The quality of information systems, an inquiry into
complexity, Working Paper Series of Washington University, St. Louis,
Missouri,.Ed. C. Hartog, Vol. 3, no. 4.
Takkenberg, C.A.Th., (1983), Planning en Methode van Onderzoek, Doctoral
dissertation, Groningen University.
Vennix, J.A.M., (1990), Mental models and computer models, Doctoral
dissertation, Nijnegen University.
= 524 -
