Visual simulation in hospitals: A managerial or a political tool?

Visual simulation in hospitals: A managerial or a political tool?

European Journal of Operational Research 29 (1986) 167-177 North-Holland 167 Visual simulation in hospitals: A managerial or a political tool? L y n...

980KB Sizes 0 Downloads 10 Views

European Journal of Operational Research 29 (1986) 167-177 North-Holland

167

Visual simulation in hospitals: A managerial or a political tool? L y n M. J O N E S a n d A n d r e w J. H I R S T

University of Edinburgh, Department of Community Medicine, Usher Institute, Medical School, Teviot Place, Edinburgh, EH8 9/1 G, United Kingdom

Abstract: In hospitals, managerial change is often a matter of negotiation and discussion between medical, administrative and nursing disciplines. Any O R intervention should therefore: (a) be easily understandable by all concerned; (b) aid discussion of options. Visual interactive simulation meets these requirements. A complex of surgical services comprising several wards and operating theatres has been modelled, in such a way that movements of patients, variation in waiting lists, bed occupancy etc. can all be seen on a visual display screen. A simultaneous study of the managerial environment of the surgical unit has revealed the importance of internal politics, and the role which the simulation model may play in bids for resources. It is suggested that it is normal for O R projects to be used in this way. The implications for design of O R projects are discussed. Keywords: Health services, simulation, organisational politics of O R

The use of OR: Action research

This project is unusual for an O R study. It is an applied, practical study, involving the application of an O R technique, simulation, to help guide managerial choice, but this is not its fundamental purpose. We were more interested in exploring why managers should seek to have their choices guided in such a manner at all. What induces them to enlist O R support? Under what circumstances might they do so? This is an issue of general interest, of course (or so one would suppose), to O R scientists in any industry. It is of particular interest, and particular difficulty, in the Health Services, because in the Health Services it is not even clear who the managers are, or even if there are any managers at all. We had become aware of a technical improvement in O R capabilities which seemed, according to earlier reports, to make some aspects of communication between O R specialists and others very much easier. This was visual simulation - the

Received July 1985

ability to display the progress of a simulation on a computer video terminal, dynamically, pictorially, and in considerable detail. This seemed a good opportunity for some action research into the way O R gets used. We resolved to try visual simulation out on a Health Service problem which seemed, prima facie, to be suitable. At the same time, we would study carefully the organisational context in which we were operating. H o w were decisions made? Who was involved? Did people habitually appeal to facts or to quantitative analysis of any kind? If so, under what circumstances did they do so? What were their reasons? We could also study the actual use of the visual simulation model that we constructed. At the start, we knew the managerial problems it would address, in their general outline. What we could not predict was the particular questions it might be called on to answer: the options it would be asked to assess, what variables would be assumed controllable or uncontrollable, what detailed output would be called for, and so on. We are therefore keeping a careful record of the model's progress, and of the decisions it may affect. This is, as stated above, action research. That

0377-2217/87/$3.50 © 1987, Elsevier Science Publishers B.V. (North-Holland)

168

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

is, we are making atl intervention which is not totally under our control: other people and events will influence its history. Moreover, as it proceeds, we ourselves will adjust and modify what we do to keep it as relevant as possible. At the same time, we stand back from and observe what we are doing and the effect it seems to be having. This model of research is borrowed from the fields of social science and organisational studies. It clearly brings methodological problems, as it fails to conform to the austere, though largely illusory, paradigm of "objective" scientific enquiry. On the other hand, it often a rich source of fruitful hypotheses and suggestive evidence. Examples in the Health Services include the extensive studies by Revans et al (1,2). Cope (3) gives a useful discussion of the approach. The work reported here is still in progress. This is therefore an interim report.

Layout of paper The remainder of this paper is laid out as follows. The next section ('Relevant organisational theory') reviews briefly the organisational theory on Health Service Management in so far as it is relevant to this project and also the (rather meagre) theory relevant to understanding the role of OR in organisations. The potential impact of visual simulation is then discussed, in the particular context of Health Services. The following section describes the modelling part of the project. Next, the methods and interim results of the research into the organisational context of the modelling project are described. The paper ends with a general discussion of findings so far.

Relevant organisational theory For our purposes it is useful to distinguish two streams of thought about organisations, the 'rational-actor' stream and the 'organisational politics' stream. These are similar to those identified by many previous writers, e.g. Allison [4-6]. Hunter [7] discusses their application to the NHS. The rational-actor stream In its simplest form, the rational-actor stream assumes that an organisation has clear objectives.

Moreover, each subdivision of the organisation, and the individuals within it, also has clear objectives, each of which contributes to the overall objectives. Conflicts of interest are either supposed not to exist, or to be aberrations occasioned by failures in communication, personality defects, etc.. Problems and difficulties do still face the inhabitant of such an organisation, however, because of the sheer uncertainty and complexity of the world. The consequences of actions are not known with certainty; neither are environmental conditions, present or future; it is not even necessarily clear what actions are possible. The organisational politics stream This stream takes a pluralistic view. It sees organisations as coalitions of individuals and groups, each having its own objectives. The careerist may pursue promotion and money, the insecure may seek a sense of affiliation, the technical perfectionist may seek excellence in product design; Departments collectively may seek aggrandizement or independence; and so on. All accept that they may only pursue these goals as long as they abide by certain rules of the game, which include some contribution (or appearance of contribution) to the objectives of hierarchically higher levels. In practice, then, decisions and actions within such an organisation represent individual and group compromises between various sets of partly overlapping, partly conflicting interests. Life in such an organisational world is necessarily highly 'political', and even if the 'objective' world were orderly and predictable, there would be much uncertainty in predicting the actions of other actors. Health Service organisations There is no doubt that Health Services, at least in the UK, conform much more closely to the second model than to the first [7-10]. One fundamental consideration is that the work of the Service depends crucially on the decisions and actions of doctors. Doctors, as independent professionals, simply do not acknowledge the organisational hierarchy as having authority over their work, except in certain limited ways. It is probably true

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

that doctors tend to view the Health Service as existing to help them do their work, rather than the other way around. Those administrators, civil servants, and politicians who see themselves at the apex of the Health Service Hierarchy have always found this difficult to accept, and have attempted, by a series of massive and complex restructuring, to assert a more unitary model. To date they have had little success. OR in organisations

Relatively little organisational theory seems to address itself to the role of 'Management Services' of any kind. Mintzberg [11], however, supplies a helpful framework. He sees organisations as consisting of a core of operational units - - those who actually produce the goods or services - - supervised and coordinated by a layer of junior and middle management, who are responsible in turn to the 'organisational apex' - - the executive directors for example. This pyramidical picture is of course totally conventional. Flanking this pyramid are two other sets of organisational functions (see Figure 1). One consists of 'service' activities - - those which are essential to the organisation but not part of its directly productive activity. These may include the works canteen, the Personnel Department, the Legal Services office, and many other such functions. The other set is what Mintzberg calls the technostructure, and this is where OR comes in.

~, S'I"RATEGIC APEX 1

169

The technostructure is the means whereby the organisational apex seeks to control the way productive work is done - - at the operational level and at junior, middle and even senior management levels. Of course, the ordinary line of management, through the setting of objectives, providing incentives, and regular supervision does in a general way control productive activity. For example if the Production Director believes that assembly work in the factory is inefficient, he or she may exhort and threaten the line managers to improve it. They may reorganise, invest in new plant, or exhort and threaten their assembly workers. However, the Production Director has another resort: he or she may establish a Work Study division, and direct that it be used. This gives an independent channel of control, and one which reaches right down to the shop floor. The way in which work at the shop floor is carried out then becomes to some degree standardised, under central control, rather than at the discretion of the individual worker. This much is also familiar territory. However, OR can be looked upon as an exactly similar part of the technostructure, except that its activities are addressed to various levels of management rather than to the shop floor. Just as the existence of Work Study says to the operative (in effect): "we think you may be unable, or unwilling, to discover how to discharge your work in the most efficient manner", so the existence of OR says the same to the manager. OR scientists do not as a rule see themselves in such a light, of course: they see themselves in a helping collaborative relationship with management. While this may be true at a personal level, Mintzberg's argument would be that the underlying structural rationale is one of power and control. OR in Health Services

/d

_~

MANAGER'~

Figure 1. Relationship of technostructure and support services to line management. (After Mintzberg [11].)

In Health Services, some shop floor operatives may be Work Studied; including laundry workers, cleaners, ambulance workers and even nurses. Other shop floor workers, chiefly doctors, may not. However much the 'organisational apex' might wish to do so, professional medical automony is usually much too powerful to allow it. In Mintzberg's terms, this particular group of operatives can shrug off the technostructure. It is commonplace to remark that doctors, par-

170

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

ticularly hospital specialists, have a managerial as well as a shop floor role. They effectively control the disposition of large amounts of resource, within their own spheres of activity; and furthermore they have a powerful influence on the planning of future facilities and services. In the NHS this influence is well documented [7-10]. It is exerted through both formal and informal channels. The formal ones include the professional advisory machinery, which is an arrangement of committees at specialty level, feeding into a committee which represent all medical specialties at District level. The point here is that this managerial function of doctors is also beyond the reach of the technostructure. Clinical doctors are no more more willing to have their managerial work subjected to OR than they are to have Work Study officers stop-watching appendicectomies in the operating theatre. (This is not a reaction specific to doctors - - i t is characteristic of most professionals. Academics are notorious for it. Are OR workers themselves any different?) The consequences for OR are as follows. Where there is a hierarchical organisation within the Health Service which does not involve doctors, as for instance there is in relation to most hotel functions in the U K Health Service, OR can be deployed as part of the technostructure. Higher levels of management can also use OR to help them formulate strategic plans, embracing clinical and non-clinical facilities and resources, but these plans cannot then be imposed: they must be negotiated through elaborate internal machinery. What, then of managerial decisions involving clinicians at a more technical level? For example, given the existing stock of hospital beds: how many beds should specialty X have, and in which hospitals? What operating sessions should specialty Y have, and on what days of the week? Should an extra consultant be appointed to specialty Z? These are decisions which directly affect patient care and the efficiency of service: and they are the kinds of decision where the influence of the technostructure is apt to be seen as directly challenging to professional autonomy, and likely to be strongly resisted. Duncan and Curnow [12] put forward a related argument in respect of attempts to model resource allocation in health services for the elderly. The actual process of allocation, they observed, was

the aggregate of individual professional decisions, based on the values of, and information available to, professional workers. Any attempt to construct an optimising model, even one used in a sophisticated 'what-if' manner, such as the DHSS's Balance of Care model [13], would be impracticable both to build (because the relevant data did not exist) and to implement (because it could not reflect local variations and preferences). Subsequent reports [14,15] on the Balance of Care project seem to confirm this view at least partially, because the model seems to have been drastically simplified in practice; one major change has been to remove the optimising element and use it interactively with teams of local professionals. In at least one case, clinicians rejected its use outright. Another case in point is the conduct of out-patient clinics. It has always been well known that a large proportion of out-patient clinics in the NHS are organised in such a way as to inflict quite unnecessarily lengthy waiting time on patients. The scale of this imposition is vast: one recent study estimated that the equivalent of three and a half million working days are lost by people sitting waiting in clinics in England and Wales each year [16]. This problem has been given the attention of OR and Work Study specialists in the NHS many times, over many years. The problem remains intractable because many doctors see the problem differently, and exercise their managerial prerogative to resist change. Rosenhead [17] has commented critically on attempts by OR analysts to contribute both to tactical and strategic health planning. He found that OR studies in the health care field typically: (1) attempted to optimise single objectives; (2) depended on wholesale (and implausible) quantification of the social world; (3) were defined and executed in a manner devoid of political content; (4) assumed a single decision-maker; and (5) assumed a static environment. In the view of the present authors, those criticisms have some justification. Many studies with these characteristics have been undertaken. In effect, they represent attempts by a technostructure to plan services ' t o p - d o w n ' , according to some overarching set of priorities - - either failing to recognise, or seeking to alter, the existing political truth that power in the health services resides largely at the periphery in the hands of the profes-

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

sionals. Rosenhead implies that this approach was misguided and would be ineffective, and expresses his preferences for an approach that would, among other characteristics, avoid an attempt at 'a technocratic abolition of politics'. One thread (though not the principal one) in Rosenhead's argument was that any approach which was more politically realistic, and favoured participation, would be technically simple and make relatively few data demands. Here we believe technical progress has overtaken his conclusions. Techniques are now available which are internally complex, and make heavy demands on data, but which are transparent to the untutored user.

The potential role of visual simulation If OR work can genuinely by controlled by its subject--i.e, the manager to whose work it is being applied--then the power clash between the two dissolves. Most OR work in large organisations, I should guess, is not subject-controlled in this sense, but neither is it totally externally imposed. Decision Support Systems (DSS), for instance, often operate at the level of the supervisor or junior manager; this person may (or may not) welcome the D S S - - b u t it is probably the idea of someone senior. As Duncan and Curnow (12) pointed out, the particular OR technique used has an important bearing here. Their argument was that constraints of time, and the degree of local variability in problems, made it impracticable to use complex and sophisticated models in the Health Services. There is another, perhaps even more important consideration. If the technique and its reasoning are opaque to the subject, he or she will inevitably find it difficult to control and therefore threatening. If its reasoning is clear and comprehensible, it will be easier to control, and therefore less threatening. Visual simulation models [18,19] have the merit of virtually instant comprehensibility. In the present study, a visual simulation of part of a hospital has been shown to several different groups of people, mostly doctors and nurses, with no previous exposure to OR. On none of these occasions was it necessary to give more than the most cursory explanation of what a model was, or what a simulation was, or how it could be used: the

171

moving image was virtually self explanatory. (Other reports [18,19] confirm this experience.) Intelligent questions about assumptions, sources of data etc. reveal that the understanding is genuine. This is a tremendous change from much earlier OR experience with simulations relying on printed alpha numeric output. It was our contention that this property of visual simulation might be particularly helpful in hospital-level studies involving clinicians. The corollary, however, would be that the way it was used by clinicians would be unpredictable. If it was controllable, it would be controlled. Thus, the doctors would ask questions of it that they wanted to ask - which are not necessarily those that senior Health Service management want asked. Naturally, senior Health Service management too will seek to control the tool, and they too will find this relatively easy. At best, this could make it the instrument of genuine communication and negotiation. At worst, both sides might become disillusioned and reject it. We believe that a model carefully built up, with genuine clinical cooperation, would tend towards the better outcome.

The modelling project The hospital and surgical unit The research was carried out in a District General Hospital ( D G H ) in Scotland, which provides acute services for a population of over 200000. The hospital includes the usual range of acute specialties. Patients requiring highly specialised services are sent to a major teaching centre some 30-40 miles away. The hospital has been built in stages within the last 25 years. The area served is partly urban and partly rural, the rural hinterland extending up to 30 miles or so from the DGH. Within this area there are 3 small hospitals, 2 of them staffed mainly by General Practitioners. All 3 have operating facilities. In addition, there is a convalescent hospital close to the D G H . These four 'peripheral' hospitals constitute a valuable potential resource. The surgical unit is based in the DGH. It has 94 adult beds plus the use of a small number of beds in the paediatric wards. Twenty of the adult beds are reserved for the use of the urology spe-

172

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

cialty; the remainder are used for patients in general surgery. There are three consultant 'specialist' general surgeons, supported by junior medical staff. The general surgeons and urologists share the use of two operating theatres. In addition, they have access to a daybed ward in another part of the hospital, which has its own theatre. Some operating is also carried out in the peripheral hospitals. This study concentrated on general surgery only. The specialty admits about 4000 patients annually, most of them to the D G H but about 300 to the peripheral hospitals. (About 700 of the patients admitted to the D G H are subsequently transferred to one of the other hospitals.) The case mix is fairly typical of general surgery. Of the 4000 admissions, about half are emergencies and half are for elective surgery. At the start of the project, the waiting list for elective surgery was over 1000 in total and growing slowly. Bed occupancy was running at about 80%. The perception of medical, nursing and administrative staff was that the unit was running at or near to capacity, and there was concern about the waiting list. There were many suggestions as to how the throughput could be increased. These included: - more use of facilities in peripheral hospitals, - more use of day surgery, - more surgical staff, reorganised theatre timetables, - more flexible use of beds. There was no prospect of new beds or theatres. Indeed, there were very tight financial constraints on the Health Authority which had a policy in any case of giving priority to the nonacute specialties such as psychiatry, geriatrics and mental handicap, and to community facilities. -

of stay. The patient is then allocated a bed and a surgeon. As soon as possible, the patient is taken to the operating theatre-(if an operation is required), then returns to the bed for a post-operative stay. After this the patient is discharged. Elective patients are sampled daily and go on to a waiting list, along with an indication of whether major or minor surgery is required, the surgeon who is treating the patient, and the anticipated hospital of admission. Each week, the waiting list is scanned, and suitable numbers of patients scheduled for admission and for operation in the subsequent week. The scheduling takes account of the anticipated availability of beds and operating time, leaving space for emergencies. On the scheduled admission day, the patient moves into a bed allocated to his/her surgeon, and in due course is operated on and later discharged. These movements are all depicted on screen. (See Figure 2). Emergency patients are shown in red, elective patients in blue, as small coloured rectangles which move about the screen. All the beds have a separate, marked position on the screen, and the operating theatres are likewise shown in diagrammatic form. Patient admissions are shown by the appearance of a patient symbol at the corner of the screen, which then moves to the allocated bed position, later to the theatre, back to the bed, and in due course (on discharge) off another corner of the screen. Each of the five hospitals in the area is simulated, the simulations proceeding simultaneously, but with only one on the screen at a time. The progress of the simulation can be interrupted at any time, and switched to display a different hospital, or to separate displays showing statistical summaries of bed occupancy, waiting lists and so on.

Simulation modal

Data

The simulation model was built using the See-Why system developed by British Leyland Systems Ltd. (now Istel) [18]. (see Appendix.) It deals with each patient as an individual entity. Emergency arrivals are sampled randomly. On admission, further random sampling determines the patient's sex, whether adult or child, whether h e / s h e will require an operation (40% of emergency admissions do not), and his/her length

Scotland is fortunate in having a good system of data collection by in-patient stay. Details of age, sex, diagnosis, operative procedure, dates of admission and discharge, consultant in charge, source of admission and destination of discharge are all recorded, along with various other details. Figures for three full years were available to us for analysis. These figures have been used as the basis for all input concerning patients.

L.M. Jones, A.J. Hirst / Visual simulation in hospitals iiiiiiiii~!ii:i!~Mh:~:Yiiiiiiiii!ii!!!~Uh~a;~a~i~ja~b~e~!i!i!i!iiiiii:!~!ii:i: ] Monday

I!

A DM I SSIONS SCHEDULED

~

.

.

.

.

.

.

.

.

.

I

fi IL

Jr e j e cPatiepts ted/1000 E merg WL 5.1 27" U Patients

:i:i:i:i:i:i:CCM I

! iiiiiiiiiiiiiCM• 1 i ili~ ~I ~ '!iiii" " " ~

rHUR

].....

iI ,

~1

on4 7WL

i

I

FRI

O Day no 64

i l i i i i i : i i ! i l i ~ ................................. ]i::i]ili::Interact ing

.

SAT

11

173

.!.!!ii!iii)?CF I i~ I'"'"""~1L'~]

;i ?i i!i

i CF~ """" !iii CM--M--[

BedUtilisation50.5% Utilisation 1 I BedCurrent 7~-8%

;:i![::i::!::iii

male child Total female Admissions Emergency I

~~B

iiiiiiiiiiii

1OOElective 74 24 I

ii:i:i:!:ii!

[ 115

I ~'~

~E~

~'~I

i:i:!i]ii:i:i

AF AM

=

::i::::::::)::::::: _I

,,, OlON ICC , ICM *,

-!!i:i:i:i:i:

:::::::::::::::::::::

~!i:i!!:!!i:!:lC C J

82

19, ]

!ii:i:i:i:ii:

Elective Patients

::::::::::::::::::::::~

malel "~ female2.ochild2.7 I

~

No Op Emerg WL

PITA ~:~:~.ii:i:i:i:):)i:!:i:!:i:::i:!:!:!:2:i:i:!:!:}~i~)~i~[~)i:!:!...::~:i:!:i:i:i:i:i:i:i:i!)::).~:. i);i)ii!ilili!i::iiiiiiii!i!!!}!ii?i})!?iii?iiii}ii!!!!::i::i::i::iiiiiiii:! H0.• ....... 1

ii!iii!i;i~.;~:~.~:.~i:i:iii:i:iiii~i!:i:i:~:i:!:i:~:`.:~:.~.~..~a:}~!!!!`:i:i:i:~iii;ii:!:!i.~.-:~:.:;~.~:i:!!!:i:!:i:i:!:!:!:i:i:i:!:i:..:~.~a:~iiiii!ii:!:i:i:i:ii~ii:i!i: ', A = MMR.Y R. X . * * **..L. . . . o H ,___ ~;~wAR~i~iiii~ii~!!!!iiii~ABp;~!ii~iiii~i~i~i:wAR~i~3i~i~i~iii~i!!i!!~!~!~!~i!iw`~ R:~!~i~l

....

I

Ave rage

iiii!iiii!:i~

~i:il:i:~i i i Ii!ii{;:):i

J

-I~

Movement of Elective Patient Movement of Emergency Patient

I ~

--Patient

1st letter = I 2nd letter

/

M ~. Z

Consultant Male/Female/Child

1.3rdSymbol;~=Major

Case

Figure 2. Layout of simulation screen showing patients scheduled for admission, patients in wards, operating theatres, and summary statistics. Typical paths of patient movement on admission are shown; subsequently, patients are shown moving to theatre, back to the ward, and off the screen on discharge. Colour is used to distinguish elective patients, day cases and emergencies. Instructions to control the simulation are entered via the 'interaction box', top right.

Case-mix Cases were d i v i d e d into 5 t r e a t m e n t groups, the case being a l l o c a t e d a c c o r d i n g to w h e t h e r it was: - e m e r g e n c y or elective, - d i d or d i d not result in an o p e r a t i o n , - if elective, a n d r e q u i r i n g a n o p e r a t i o n , w h e t h e r this was m a j o r o r minor. The distinction between 'major' and 'minor' was a p r a g m a t i c one. C e r t a i n m o r e c o m p l e x elective o p e r a t i o n s were carried o u t only at the D G H ; s i m p l e r p r o c e d u r e s were carried out at all of the four hospitals with theatres. ' M a j o r ' p r o c e d u r e s

were d e f i n e d as those carried out only at the D G H ; all others were classified ' m i n o r ' . I n a d d i t i o n , the following characteristics are all r e c o r d e d for each p a t i e n t : - sex, - whether child or adult, - c o n s u l t a n t in charge, - hospital of admission. T h e sex of child p a t i e n t s is not noted, however, so the n u m b e r of distinct categories used is 5 ( t r e a t m e n t group) x 3 ( a g e / s e x ) x 3 (consultant) × 4 (hospital of a d m i s s i o n ) = 180.

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

174

Length of stay Analyses were carried out to discover whether lengths of stay varied significantly between categories, using analysis of variance. In summary, the results showed that there were significant differences ( p < 0.05) between: - adult and child lengths of stay, most pairs of treatment groups, - consultants. No significant differences at this level could be demonstrated between: - male and female adult patients, - hospitals. Separate lengths of stay distributions are therefore used for adults and children, in each treatment group, for each consultant - - 2 × 5 x 3 = 30 distributions. A common form of distribution was sought which would match the observed distributions as closely as possible. The lognormal seems to give an adequate fit, although it tends to underestimate very long lengths of stay.

Interim conclusions of feasibility We have now reached a point where some real positive conclusions have emerged. Principally, we have established: 1. It is possible to produce a highly flexible, visual simulation of a surgical unit and peripheral resources, using See-Why. The specification of the model is so flexible as to promise relatively easy adaptation to other surgical units, even if their resources, routines and case-mix are substantially different. 2. The visual model has instant face credibility to professional Health Service staff, including medical staff. The potential for application to other surgical units is highly significant. There are some 27 units of comparable size in Scotland, and over ten times that number in the UK. Prima facie, there seems no reason why the model could not be developed into a system which could be applied to a large proportion of these with relatively little in the way of skilled re-programming.

Organisational

context:

research

methods

This, however, begs the main question to which the research was addressed: can such a system

have a useful impact on decisions, given the organisational context of the NHS? To explore this organisational context, we have used two methods: interviews, and documentary analysis. Prior to model construction and use, interviews have been held with a large number of the significant actors in and close to the situation - - t h o s e who have, potentially, an influence on the kind of decision a simulation model might purport to affect. Another series of interviews will be held with the same people at the end of the project. The initial interviews were unstructured but covered a predetermined range of topics: what managerial decisions were there to be taken at the level of the surgical unit; who would influence such decisions; what would be the process by which they were taken; and what part would facts, figures and rational analysis play. To give clearer focus to these questions, an actual decision affecting the unit which had been taken in the recent past was identified. All the above questions were asked, not only in general terms, but in relation to this specific decision. (This decision is referred to hereafter as 'Decision D'.) Decision D was the subject of the other investigative method, documentary analysis. The progress of D was traced through a large number of documents on the r e c o r d - - m e m o r a n d a , papers, committee minutes etc.. It proved possible in this way to build up a detailed history of D. Neither the interview method nor the documentary method are without defects, of course. Interviewees all have their own beliefs and prejudices, variable memories, and private views which they may wish to press on the interviewer. However, it is the individual's perceptions of the organisational structures and processes which matter, so some at least of these 'biases' are themselves objects of interest. Interviewer bias is another matter, of course: one means used to limit this was to tape record interviews rather than rely on notes. Documentary analysis is a little more 'objective' of course, but suffers from being a very selective view of events, and failing (usually) to reflect the interplay of personalities. In the event, both approaches have proved workable (as Barnard et al. [20] found previously) and have provided much relevant and interesting material.

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

Organisational context: interim findings The first thing to emerge was confirmation of the notorious unwieldiness of NHS decision-making. It is complex and slow--sometimes almost unbelievably slow. Decision D, for instance, took twelve years to make, even though the new resources involved were relatively m o d e s t - roughly some £25 000 per year at 1983 prices. It was discussed at least 30 times in various Committees over that period, and we located over 100 documents concerned with it. Furthermore, it became clear that some of the main participants in the process did not understand it fully themselves. For example, Decision D concerned a proposed enhancement of services, involving extra resources. The system demands, first, that such a change be approved as feasible and desirable by the affected professional groups and Board officials; second, that the professional groups accord it a high priority; and third, that the Health Board alocates money for it. Nowhere is this three-part process laid down or explained. Some of the participants clearly confused the different stages, not realising that each had to be taken up and pursued in turn. The reason for this elaborate and apparently inefficient procedure has to be sought in the organisational dynamics of the Health Service. The Service tends to be exceedingly conservative in its methods. In general, changes only come about as a consequence of increased resources. Proposals concerning the potential use of increased resources appear constantly, not as a result of synoptic planning at a Board or hospital level, but from individuals or Departments throughout the Service. These surface eventually as a shopping list of 'bids', totally disparate in size and nature, at Health Board level. (The process is described in detail in Hunter [7].) Typically, bids exceed available resources many times over. This shopping list is virtually impossible to rank by any rational criteria, not only because the bids are so varied, but because there is no overall framework in which to fit them. Their chances of success are therefore heavily influenced by the political weight of their respective sponsors and the sponsors' skill in steering them through. Senior Board Officers, and the Board itself, of course do have their own opinions on priorities, but their authority is heavily constrained.

175

To summarise, all significant change involves bids for resources; and bidding for resources is a highly political process. The reasons for the cumbersome decision machinery now become clear; it is a way of sifting bids from below, not through a managerial or an analytical filter, but through a political one. It is a means of ensuring that only bids with good political backing get through for consideration at the higher levels. Consequently, delays and referrals back are actually helpful, in many cases, to senior management, because the immediate pressure of competing demands is thereby reduced. Indeed, they could hardly operate without these apparent inefficiencies, It comes as something of a surprise, therefore, to find that facts and figures are widely used in this political process. The documents relating to decision D made frequent reference to figures. However, these were almost wholly limited to routinely collected statistics, of patient throughput and waiting lists. Occasionally, comparisons of these would be made with the corresponding figures in other parts of Scotland. Remarkably, nowhere in the whole twelve year saga of decision D was any estimate made of the increased patient throughput that would be made possible by the proposed enhancement--nor was any such estimate asked for. Nor was the logistical feasibility of the enhancement examined until the very final stage, after the funds were c o m m i t t e d - - a n d then only at the request of a body external to the Board whose approval was required. Throughout, when figures were used it was as instruments of persuasion rather than analysis. Never was there any attempt to assess options analytically, for instance, because the system does not work that way. Options were sometimes put forward, but the decision between them was made not as a result of 'objective' comparisons, but according to which found most favour with the key professional groups. Facts and figures could be used, however, to demonstrate existing pressure on a service, and hence to bolster a case. In other words, they were seen as instruments to persuade others, not to enlighten oneself.

Discussion Any discussion on the basis of these findings must be extremely tentative, because this is only

176

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

one study half complete. It must also be confined to the particular kind of decision context examined here, namely decisions at the level of the surgical unit which involve either no, or relatively small, increments of resource input. At first sight, the indications are not promising for the use of OR. On the evidence of this study, it is apparently rare for anyone at the level of the clinical unit itself to consider significant changes in its organisation or running except in the context of a bid for increased resources. Objective or analytical appraisal of options is not therefore called for. Bids for new resources are likely to be made on the basis of perceived need by professionals; this perception may arise because of feelings of overload, or new technical opportunities noticed, or a wish to extend the service. Whatever its origin, fulfilment of the perceived need is likely to be seen as desirable ab initio. The bidders are therefore unlikely themselves to undertake O R work on it. The higher echelons with power to approve or reject the bid are also, in the current scene, unlikely to submit it to analytical scrutiny, either as a single project or in comparing it to other resource bids. This is because, as argued above, resource allocation is essentially a political, rather than a managerial or analytical process. There are two kinds of change to this picture which might alter the prospects for OR. First, the introduction of a new element, like the easy availability of the visual simulation, might of itself cause a shift in the organisational dynamics. Second, external changes to organisational structures and processes could bring a new constellation of forces and motivations. Taking the first of these, it is conceded immediately that a small technical innovation like this is not going to shift the fundamentals of power in the Health Service. Nevertheless, small changes do take place even in such monolithic structures, sometimes just locally, sometimes cumulatively. The key observation here is that, even in the present context, facts and figures play a very important part. They are used p o l i t i c a l l y - i.e. to convince others that the change y o u desire is in their interest. For instance, figures are used to demonstrate that the service currently provided is overloaded, under-resourced compared with like services elsewhere, and leading to longer waiting lists; therefore more resources are required. Is it

possible, than, that an accessible tool like visual simulation might be seized on for similar purposes? 'Conventional' O R techniques are difficult to deploy in this way, since they are so much under the control of the O R specialist, and hence seen as part of the technostructure; but a visual simulation is potentially controllable by the nonspecialist. It is interesting that the first interest shown in using the simulation concerns a bid for new resources--in fact, a new Consultant Surgeon. It is an interesting philosophical point whether such use of an O R tool is desirable. The present authors believe it is, not because it would render resource negotiations any less political in essence, but because it would ground them more firmly in fact. To gain the political 'backing' of the model for a bid, would entail much closer prior examination of the way the new resources would fit into existing patterns of work, and estimating their effect on patient throughput etc.. This is likely to lead to more considered bids. External changes are potentially much more powerful. Major changes are currently underway in the British NHS, including both a heavy squeeze on resources and introduction of general managers at Health Board and hospital level. They are likely to enhance the role of the technostructure in general, and O R in particular. Rosenhead [17] puts forward criteria for what he considers would be more acceptable and effective O R in health services. He favours approaches which: (1) make reduced demands on data; (2) reject optimisation in favour of co-ordination; (3) accept uncertainty and try to keep options open; (4) are not restricted to hierarchical deduction, but facilitate participation; (5) do not attempt a technocratic abolition of politics. Visual simulation seems to meet all but the first. No objective is specified--several measures of performance may be calculated, but it is left to the user to decide what significance to place on them. Options may also be specified by the user, though the model of course does place some constraints. It is explicitly designed for participation - - b y professionals. It seems to fit into, rather than seek to abolish the existing political system.

L.M. Jones, A.J. Hirst / Visual simulation in hospitals

Nevertheless, it is probably not quite what Rosenhead had in mind. There is a strong sense in his paper that he seeks participation by patients and the public, not by the already-powerful medical profession. He criticises models, like this kind of simulation, which treats patients as 'passive objects' to be modelled, rather than individuals capable of 'co-operative decision-making'. However, in seeking the latter is it not Rosenhead who seeks, if not to abolish politics, then to transform them utterly? These authors share his desire for radically greater citizen and patient participation in health care planning. An OR study, however, seems an unlikely route for achieving it. The aims here are more modest: to render the existing machinery a little more flexible, a little more capable of utilising resources well, and to inject rationality in a form tailored to suit the political processes and balances which we find in existence. Whether such a tactic contributes to or obstructs progress towards wider reform is a much bigger and more complex problem.

Appendix The visual simulation has been constructed using See-Why simulation software, supplied by ISTEL Ltd., Tribox House, Sandy lane, Littlemore, Oxford, OX4 5LB. This runs on a Cromemco System 1H computer with 512k of RAM and a 20 megabyte hard disk, connected to an Intecolor VDU. (The system is now also available on an IBM AT microcomputer, suitably configured.) The Intecolor VDU gives a large (33 cm x 22.5 cm, 48 lines x 80 characters), 8-colour display. The software is a suite of Fortran sub-routines providing the facility to construct a 3-phase simulation, and a wide range of utilities. Many of these are concerned with constructing the display. Entities can be displayed on this at will, and their movements watched as the simulation proceeds.

177

References [1] Revans, R.W. (Ed.) (1972), Hospitals: Communication, Choice and Change, Tavistock, London. [2] Wieland, G.F. (1981). Improving Health Care Management: Organization Change. Health Administration Press, Ann Arbor, Michigan. [3] Cope, D.E. (1981), Organisation Development and Action Research in Hospitals, Gower, Farnborough. [4] Allison, G.T. (1971), Essence of Decision, Little, Brown & Co., Boston. [5] Cyert, R.M. and March, J.G. (1963), A Behavioural Theory of the Firm, Prentice-Hall, Englewood Cliffs, NJ. [6] Hall, R.H. (1977), Organisations: Structure and Process, Prentice-Hall, Englewood Cliffs, NJ. [7] Hunter, D.J. (1980), Coping with uncertainty Policy and politics in the National Health Service, Research Studies Press, Chichester. [8] Lee, K. and Mills, A. (1982), Policy-Making and Planning in the Health Sector, Croom Helm, London. [9] Ham, C. (1981), Policy.making in the National Health Service, Macmillan, New York. [10] Klein, R. (1984), "Who makes the decision in the NHS?" British Medical Journal 288, 1706-1708. [11] Mintzberg, H. (1979), The Structuring of Organizations, Prentice-Hall, Englewood Cliffs, NJ. [12] Duncan, I.B. and Curnow, R.N. (1978), "Operational research in the health and social services", Journal of the Royal Statistical Society A141(2), 153-94. [13] McDonald, A.G., Cuddeford, G.C. and Beale, E.M.L (1974), "Mathematical models of the balance of care", British Medical Bulletin 30, 262-270. [14] Borley, R.G., Taylor, S.H. and West, C.R. (1981), "Balance of care - - A user's view of a new approach to joint strategic planning", Omega 9, 473-479. [15] Nicholls, I.G. (1981), "Joint planning in Dudley - - The role of balance of care", Omega 9, 501-508. [16] Department of Health and Social Security Operational Research Service (1985), "Reducing waiting time in out-patient departments", Department of Health and Social Security, London. [17] Rosenhead, J. (1978), "Operational research in health services planning", European Journal of Operational Research 2, 75-85. [18] Fiddy, E., Bright, R.G. and Hurrion, R.D. (1981), "See-Why: Interactive simulation on the screen", paper presented at the Conference of the Institute of Mechanical Engineers. [19] Hollocks, B. (1983), "Simulation and the micro", Journal of the Operational Research Society 34, 331-343. [20] Barnard, K., Lee, K. and Reynolds, J. (1980), Tracing Decisions in the National Health Service, King's Fund, London.