A. J. Wells, F. M. Cocks, C. Cryer and L. Davidson
It has been suggested that whilst injuries are a major public health problem, the amount of research with respect to other forms of morbidity and mortality is small. This situation is not due to a lack of interest with local authorities, county councils, ministry of transport, emergency services and numerous charities actively interested. It is more likely to be due to limited funding, no main organising body and lack of integration between data collection services (Cryer et al., 1993). Furthermore, whilst numerous fields within health care management are beginning to benefit from the influence of geographic information systems (GIS), the implications for accident distribution analysis have still to be fully realised. The aim of this paper is to offer a broad definition of the role of accident distribution analysis to those with an understanding of GIS and to assess the theoretical potential of GIS within this area. The present structure of research between the University of Greenwich and the South East Institute of Public Health is outlined and discussion offered on the present pilot study. It is concluded that GIS offers unique opportunities for the visualisation of accident distribution data and future enhancement of present relationship analysis and spatial modelling. However, incomplete and incompatible data sets severely restrict the accuracy of quantitative output whilst increasing the time required for analysis. Whilst this is not in itself a new conclusion the serious effects on analysis through GIS offer further reason for a major assessment of data collection and processing within this field.
World-wide, injuries are the leading cause of death during half
of the human lifespan (Barss et al., 1991). Within the
United States, whilst deaths induced by injuries are low in relation
to other causes such as cardiovascular diseases and cancer, the
average age of those who die from these diseases is far older
than those who are fatally injured. If measured in terms of lost
potential years of life, these injuries were found to claim more
years of life annually than cardiovascular disease and cancer
combined (Rice et al., 1989). If non-fatal injuries are
included, the conservative estimate of the lifetime cost of injuries
since 1984 has exceeded one trillion dollars within the US alone
(Baker et al., 1992). For the United Kingdom, a total of
4,000 deaths and three million people visiting hospital per year
due to injury at home or at leisure (Department of Trade and Industry,
1992) has lead to a suggested cost of 1% of gross national product,
equating to £2500m (Vipulendran, et al., 1988). The
personal effect of such injuries is impossible to quantify. However,
conclusions as to possible implications can be drawn from research
in the UK that found accidents caused three children to die every
day and resulted in 10,000 children being permanently disabled
every year (Child Accident Prevention Trust, 1989).
Injury rates and their associated causes are not constant throughout
a resident population. (Cryer et al.) (1993) examined injuries
within South East England and concluded that the leading causes
of death were motor vehicle traffic crashes, suicides and falls
which accounted for 61% of all accidental deaths. The leading
cause of hospitalisation were falls, accounting for 41% of accident
admissions with especially high rates for falls within the over
65's community. Two further variables, the place of accident occurrence
and type of injury, are also assessed by injury epidemiologists.
Nationally, for example, accidents in the home are the biggest
cause of injury mortality and morbidity (Secretary of State for
Health, 1991).
Discussion still continues within the field of accident distribution
analysis on the definition of the terms injury and accident (Bijur,
1995; Avery, 1995). The Oxford English Dictionary implies that
an accident is without apparent cause, a product of chance, an
Act of God, or a random event which is out of control of the individual
or society. This fatalistic attitude implies that accidents are
inevitable and there appears no point in the attempted prevention.
However, within health preventative programmes which focus on
accidents, the definition of an accident is:
"An unintentional damaging occurrence resulting from sequences of chains of events. Property damage or personal damage (injury or disease), or both can occur."
(Low, 1988)
This definition suggests a cause-effect relationship for accidents,
a view which is supported by Cryer (1995) who concludes that all
accidents can be attributed to a set of causal factors which interact
in a way that leads to injury or property damage. These factors
are dependent on the nature of the accident. For example, Cwikel
and Fried (1992), stated that research within the last decade
has shown that falls within the elderly do not happen by chance
but result from a combination of medical, environmental, psychological
and social factors that interact with physiological age-related
changes.
The definition of injury (based on those provided by Waller (1985)
and Robertson (1992) is quite distinct from accident and can be
considered as
"Tissue damage resulting from either the acute transfer to individuals of one of the five forms of physical energy (kinetic or mechanical, thermal, chemical, electrical or radiation) in amounts above the tolerance of human tissue, or from the sudden interruption of normal energy patterns necessary to maintain life processes."
(Cryer, 1995)
The contribution of knowledge to our understanding of injuries
comes from many sources including microbiology, sociology, psychiatry
and biomechanics. However, an increasing desire to consider the
spatial distribution of events has led to an increased use of
injury epidemiology. Epidemiology, i.e. the scientific study of
the distribution and determinants of health and disease in human
population can be split into two divisions (Robertson, 1992).
Descriptive epidemiology includes the identification of injuries
and quantifies its incidence in relation to time, space or person
characteristics within a population. Analytic epidemiology attempts
to identify the causes of the injury in relation to a specific
factor, or more often multiple, interacting causes.
The field of injury epidemiology is an important facet of injury
control, accident prevention or reduction of the seriousness of
injury. Whilst the spectrum of research is beyond explanation
within this paper, three broad categories can be exemplified.
Some researchers have examined the theoretical approach to accident
analysis offering structure and improved analysis technique to
future research. For example, Haddon (1972) introduced the concept
of a three phase system, pre-accident phase, event phase and post-accident
phase which has been used to classify the direction of accident
prevention.
Research has also used descriptive epidemiology to improve present
understanding of injury patterns (Whitelegg, 1987) or to increase
the efficiency of resource management. Finally analytical epidemiology
has been used to assess hypothetical relationships between selected
variables and the specific type of injury (Durkin et al., 1994).
The use of GIS within health care management is more recent concept.
The initial impetus for a broader acknowledgement of the possible
role of GIS came from the Consultation on European Environment
and Health Systems meeting (Frankfurt, 1989), where the World
Health Organisation (WHO) stated that:
"Geographical Information Systems are of value in the compilation and presentation of environmental health outcome data related to the impact and use of health services at national and regional levels."
(WHO, 1989)
Whilst this statement concentrates on environmental factors such
as air quality, water quality, etc., a working group was formed
to determine the role of GIS and enhance its capabilities within
the field of health care management. WHO concluded that the role
of GIS would be as a hypothesis generator or persuasive tool rather
than a confirmation mechanism (WHO, 1992). Furthermore, whilst
it was acknowledged that GIS offers theoretical benefits for analysis
within this field, the demonstration of operational GIS packages
would offer the highest opportunities for advancement of GIS within
this field. Past research within the field of GIS and health care
management has followed two paths. The first is an examination
of the research methodologies concerned with data manipulation,
processing and visualisation. Work by Austin (1992) linked police
and hospital road accident information in order to maximise the
quantity of information concerning a single incident. Others have
concentrated on the statistical processes through which patterns
of disease or accidents can be examined (Jones et al., 1995).
Researchers have also assessed the need for the visualisation
of results to be dependent on the data and analysis. Pukkala (1992)
concluded that if the occurrence rate for specific diseases is
very low, this precludes the use of normal visualisation techniques.
In a case study using cancer incidence, the boundaries associated
with political and regional factors were found to limit the available
visual information. Instead it was suggested that smoothing and
interpolation techniques be used.
The second area concerns practical applications of GIS within health care management. The outcome of this research has been focused on one or more of the phases listed in the last section. Within the pre-accident phase, further work by Austin (1994) assesses the use of GIS as a producer of information for the prevention of accidents. By measuring the variation between observed and expected accident rates for road networks, it was concluded that GIS would improve mistake identification in accident report forms, increase the efficiency of road repairs management and identify the home location of casualties. This research later concentrated on the use of GIS in the analysis of child safety during journeys to and from school (Austin et al.,1995).
Research concerned with the post-event phase of an accident has
either examined the improvement of response times for ambulances,
or been used to organise the distribution of facilities within
the community. Research by Jones (1993) examined the role of GIS
in the routing and timing of ambulance response dependent on variables
associated with the road network. Whilst suggesting that the quality
and accessibility of data limit analysis, the process did offer
opportunities for improvement in resource allocation planning.
For the counties of Somerset, Devon and Cornwall, GIS has been
implemented at the heart of the information network used during
an ambulance response incident (Ward, 1994). The purpose was to
increase the information available during a call-out and maximise
the routing efficiency to any particular incident.
The formation of an internal market within the NHS during the
early 1990s increased the need for the strategic planning of money
and resources for regional (RHA) and district (DHA) health authorities,
general practitioners and primary health care facilities. Due
to the size of these administrative areas and the possibility
of merging selected DHAs, data manipulation and analysis of the
population within a selected area is difficult. With the introduction
of GIS, numerous researchers have offered information on the catchment
area of a selected health care provider. This organisation of
the catchment system and the order and size of the hierarchical
system for health care was been examined using GIS (Bullen et
al., 1994). Through the use of information concerning official
boundaries, shopping centres, roads, railways, flows and social
and demographic profiles, a proposed hierarchy of local community
areas for health planning was composed for the West Sussex area.
Work by Dawson and Aspinall (1994) introduced the user, in this
case the South Thames Regional Health Authority, to a system for
the easy access of census information including, deaths, fertility
rates, age, sex and socio-economic status. Whilst it was suggested
that the analytical component of this system was limited, the
tool was useful for descriptive epidemiology and offered much
in the way of future development. This availability of information
is also of use to general practitioners. Hirshfield et al.
(1993; 1995) offered the same form of socio-economic information
and demographic data for catchment areas of GPs in order to assess
for differences between the availability of primary health care.
However, instead of direct comparison to variables supplied by
the Office of Census and Population Survey (OCPS), a super profile
ranking system (Brown, 1990) was employed to delineate between
areas of differing affluence. This locality profiling was also
employed by Wain (1993) in the Walsall area to improve the geographic
correlation between areas of high disease, such as coronary heart
disease, and the required services.
This rationale of extending the information available concerning
the provision of health care has been further extended to the
pharmacies (Hirshfield et al., 1994). Information concerning
the locality in relation to the GPs, type and size of the community
served and the type of services available including information
as to their individual role as advice-givers was offered as useful
information for the future siting of pharmacies.
As well as general facilities, GIS has been used to site more
specialised equipment. In the Wirral Health Authority, GIS was
used to site a second oncology centre (Todd et al., 1994),
using decision parameters based on patterns of morbidity, mortality,
population and environmental factors.
The present dearth in research on accident distribution analysis
within the UK does not reflect the applicability of GIS to this
subject. Whilst numerous statistical packages and databases have
been employed with much success, the visualisation of information
and therefore the ability of the system to completely fulfil the
role of a hypothesis generator and communication too is somewhat
lacking. The visualisation power of GIS, combined with the availability
of data processing, manipulation and analysis, offers the accident
analyst possible solutions to a range of problems. These problems
were summarised in a report on the collection and dissemination
of accident data (Child Accident Prevention Trust) and include:
i) Tailoring data management, analysis and especially visualisation of results to the requirements of the user.
The range of agencies requiring information is broad and purposes to which the information applied varies. At a national level, these include government, academics and researchers, organisations working in the field, the private sector and the media. At a regional scale interested agencies include Local Health Authorities (LHAs), for health promotion, planning and preventative work, voluntary sector agencies, inter-agency groupings and local practitioners.
ii) Collating information on the geographic distribution of potential socio-economic, demographic, behavioural and environmental risk factors for comparison with accident distribution.
It has been suggested that there are gaps in the data available concerning causational factors (behavioural and environmental) which lie behind the occurrence of accidents as well as information linking accidents with socio-economic profiles and characteristics. Furthermore, local studies have shown wide differences between accidents occurring in different districts, which can be explained geographically, environmentally and sociologically.
iii) Integration of differing base denominator data
Problems in analysis have been caused by incompatibility of coding systems, use of different populations and denominators and lack of temporal continuity. Standardisation is required to gain a truer picture of trends in casualties (numbers of casualties per unit population) and type of accident (e.g. number of road accidents per unit traffic volume).
iv) Temporal analysis for intervention management
Within accident analysis it is considered essential to review trends and plot changes over time. This is required to examine whether intervention measures are successful and to manage resources for future prevention schemes.
v) Modelling
Estimates of the overall picture can often only be made by extrapolating
the findings of local studies. Whilst this is sometimes the only
option in the assessment of the picture of the nation, it can
be dangerous if the region used is not a typical area. Furthermore,
modelling offers information for future injury prevention planning
and resource allocation.
It has long been established that the correct visual presentation
of such spatial analysis is as crucial as the data analysis which
produced them (Cassettari, 1991). The effectiveness of GIS as
a communication tool offers a distinct advantage over the statistical
packages and databases which are presently employed within the
field of accident analysis. This is especially the case due to
the large number of end-users of the information, many of which
have differing information requirements. Through a structured
design, GIS also offers the possibility of information access
to users ranging from the information specialist, to the information
enquirers (Worrall and Rao, 1991), increasing the versatility
of data dissemination. The ability of GIS to map environmental
considerations spatially is also well documented (Brown, 1991)
with software structured to allow the comparison of more than
one factor concurrently (e.g. age, sex, socio-economic status,
housing density etc.). Use of GIS with data produced by the OCPS
(Martin, 1991) also allows the standardisation required to calculate
rates of injury for areal comparison. Finally, the requirement
to measure change over time using geographic information systems
has already been proved to be important for spatial analysis (Langran,
1991). Whilst present studies concerning injury analysis do not
contain this component, it may prove useful in the assessment
of preventative measures in the future.
At present, the pilot study is at an early stage. Due to methodological
problems associated with the analysis of medical data, the results
produced to date can not be included within this paper. However,
a discussion can be offered concerning the study area and the
use of hospital in-patient information within a GIS.
The area available for study was covered by the South Thames Regional
Health Authority (STRHA) and contained district health authorities
from Kent, E. Sussex and Inner and Outer London. Within this area,
37,286 hospital admission records were made available, covering
almost all accidents requiring hospital treatment during 1991.
Of this figure, 34,068 were resident and were treated in the project
area whilst the other 2,602 records relate to residents of the
area who were admitted to hospitals elsewhere in the country.
Each case offered information on the spatial distribution of residence,
the type of person injured and the type of accident I injury occurring
(Table 1).
The second data source, originating from the OCPS, contained the
age structure and boundary information for the area based on the
1991 census. Due to the relative size of the area and limitations
set by STRHA because of data confidentiality, the data was analysed
at a ward level.
The accident information offered two distinct series of problems,
those pertaining to the data formats and those associated with
the data collection process. The original data was collected at
hospitals throughout the country and organised by a series of
regional computing centres. An extract from this data concerning
the first episodes of cases whose principal diagnosis is an injury,
within the South Thames (East) region, was then transferred to
SEIPH. At present this is stored and manipulated using SAS, an
in-house statistical software system from which data was transferred
to the GIS. Whilst this process of data transferral is under review,
there are three causes for concern.
i) Data format
At present, the number and type of value delineators offers distinct processing problems. It is suggested that most existing problems will be solved through the optimisation of the data transfer format. However, because the sequence of accident code and injury code was not constant throughout the file, an intermediate processing stage will have to be implemented before the data can be used within a GIS.
ii) Area coding
The coding system, used to site the accident information was different to that used by the OCPS. Whilst the basic areal constituents were the same (1991 census boundaries), conversion tables are required before the two sets of information could be merged.
iii) Missing values
Due to the variable nature of the information being coded, the
number of columns contained within a particular case varied throughout
the data set. There was no specific pattern to this variation
with some cases having no patient identification code, more than
one injury or accident code, or no spatial information. Routines
are currently being written which will automatically correct for
missing data or spurious characters.
The problems associated with the collection of information are
dependent on the nature of the accident and the body which collects
the data. Whilst researchers have discussed the problems associated
with the use of accident data collected from bodies such as the
police on road accidents (Gooder and Charny, 1993), the data under
consideration here is hospital in-patient information.
The main problem for spatial analysis is that of the incident
siting. The spatial information defined within each case relates
to the place of residence of the injured party. No information
is present on where the injury occurred. This may artificially
increase or reduce the number of injuries occurring within an
area. Other problems are present due to the diversity of information
required and the complicated nature of errors within the data.
These can be split into those which can be defined and those which
may be more difficult to measure. In the case of missing data,
whilst this may cause problems in the compilation and analysis
of files (as discussed above), the influence on results can usually
be quantified.
The effect of case transferral between consultants is more difficult
to define. In a number of cases, the injury has required more
than one consultant for treatment and therefore could be counted
more than once. Whilst this could be overcome by using just the
first consultant episode within each case, other problems such
as single cases being admitted numerous times within a four week
period are more difficult to assess. Further problems may occur
in the coding of cause of injury. The specific cause may fall
into more than one category, producing a requirement to decide
which code to use. Incorrect coding or preference towards a particular
code may cause variance from the true pattern, although coding
rules are published which may provide a qualitative description
of the effect.
For the time being, most of the research will be concerned with
data formats and compatibility between coding systems. However,
once completed for STRHA, the processes could be transferred to
other health authorities within the U. K. For the longer term,
age and sex distributions combined with the socio-economic data
held by the OCPS or local authorities, offers an almost limitless
field of hypothesis generation for GIS. However particular projects
have been considered of particular interest.
i) Visualisation of the areal distribution of burns
This will be used as a specific example of the use of GIS for standardisation of data using the base population figures to assess whether the distribution of burn injuries is dependent on selected factors. At present these include, the socio-economic background of the area, the percentage ethnic population, the standard of housing and the role of local authorities in targeting accident prevention.
ii) The analysis of falls within the elderly community
Falls are the principal cause of injury among older people in
the S. Thames (East) area, with 50% of falls resulting in some
form of injury (South East Institute of Public Health, 1995).
The spatial analysis of these falls may offer an opportunity to
reduce injury levels in line with the Heath of the Nation guidelines
(Secretary of State for Health, 1992). Data taken from hospital
in-patient data is based on the residential address of the person
involved. For many types of accident this may make spatial analysis
problematic as the place of accident occurrence is critical. However,
as a large percentage of falls within the elderly community occur
in the home, a direct link is formed between the place of residence
and the place of injury occurrence. Past research has suggested
numerous factors which increase the risk of falls within the elderly.
Some of these are environmental hazards associated with the dwelling
and can only be assessed if an inspection of the dwelling is carried
out. However, visualisation of the distribution of falls within
the elderly coupled with OCPS information on socio-economic background
and local authority information concerning the state of housing
may offer improvements in the targeting of future inspections
to reduce the risk of falls.
Research within the UK concerning the role of GIS within accident
distribution analysis is extremely limited. Three hypotheses can
be suggested for this apparent lack of work in this area. The
GIS community may not be aware of the accident information present,
a hypothesis confirmed by the omission of accident information
(except local authority road accident reports) from the list of
major data sets of spatially referenced information contained
within the Chorley report. Secondly, those with specialist GIS
knowledge may have considered the data format to be unsuitable
for use with the available software. Finally, the accident analysis
specialists may not be aware of the possible implications that
GIS may have on epidemiological studies.
Whether one or all of these suggestions are correct, the solution is to begin discussion between the interested parties on the role of GIS within accident distribution analysis. This paper offers a beginning to this discussion. However, as discussed here, the way forward is the production of working examples that offer the user of the information (accident epidemiologists) proof that GIS can offer benefits within the field which are unique to the particular system. It is hoped that future work will begin to offer analyses concerned with those project noted earlier which appear in theory to offer scope for GIS involvement.
Austin, K., 1992. A linked police and hospital road accident database for Humberside. Traffic Engineering and Control, December: 674-683.
Austin, K., 1994. An advanced computer system for road safety analysis. Highways and Transportation, November: 16-19
Austin, K. P., Tight, M. R. and Kirby, H. R., (In press). An advanced system for the study of children's safety on the journeys to and from school.
Avery, J. G., 1995. Accident prevention-injury control-injury prevention-or whatever? Injury Prevention, 1:10-11.
Baker, S. P., O'Neill, B., Ginsburg, M. J. and Li, G., 1992. The Injury Fact Book, 2nd edition. Oxford University Press, New York.
Barss, P., Smith, G. S., Mohan, D. and Baker, S. P., 1991. Injuries to adults in developing countries: Epidemiology and policy. The World Bank, Washington D C :1 - 132.
Bijur, P. E., 1995. What's in a name? Comments on the use of the terms 'accident' and 'injury'. Injury Prevention, 1: 9
Brown, P. J. B., 1991. Exploring geodemographics. In : Masser and Blakemore (Editors) Handling geographical information : methodology and potential applications. J. Wiley and Sons, New York : 221-258.
Bullen, N., Moon, G. and Jones, K., 1994. Defining communities: a GIS approach to delivering better health care. Mapping Awareness, March : 22-25.
Cassettari, S., 1993. Introduction to integrated geo-information management. Chapman and Hall: 169-193.
Child Accident Prevention Trust, 1989. Basic principles of child accident prevention : A guide to action.
Child Accident Prevention Trust, 1993. The collection and dissemination of accident data.
Cryer, C., 1995 The epidemiology of work-related injury. In: Slappendel, C. (Editor) Health and safety in New Zealand workplaces Dunmore Press, Palmerston North.
Cryer, C., Davidson, L. and Styles, C., 1993. Injury epidemiology in the South East:
identifying priorities for action. South East Thames Regional Health Authority, HOR paper No.6.
Cwikel, J and Fried, A. V., 1992. The social epidemiology of falls among community-dwelling elderly: guidelines for prevention. Disability and Rehabilitation, 14(3):113-121.
Dawson, I. and Aspinall, P., 1994. A suitable case for treatment: using GIS and census data in the NHS. Mapping Awareness, October: 24-27.
Department of Trade and Industry, 1992. Home and accident research: sixteenth annual report of the Home Accident Surveillance System. Consumer Safety Unit.
Durkin, M. S., Davidson, L. L., Kuhn, L., O'Connor, P. and Barlow, B., 1994. Low-income neighbourhoods and the risk of severe paediatric injury: a small-area analysis in northern Manhattan. American Journal of Public Health, 84(4): 587-592.
Haddon, W., 1972. A logical framework for categorising highway safety phenomena and activity. Journal of Trauma, 12 :193-207.
Hirshfield, A., Brown, P. J. B. and Bundred, P., 1993. Doctors, patients and GIS. Mapping Awareness, 7(9) 9-12.
Hirshfield, A., Brown, P. J. B. and Bundred, P., 1995. The spatial analysis of community health services on Wirral using geographic information systems. Journal of Operational Research Society, 46 :147-159.
Hirshfield, A., Wolfson, D. J. and Swetman, S, 1994. Location of community pharmacies: a rational approach using geographic information systems. The International Journal of Pharmacy Practice, October : 42-52
Jones, A., 1993. Using GIS to link road accident outcomes with health service accessibility. Mapping Awareness, 7(8): 33-37.
Jones, A. P., Langford, I. H. and Bentham, G, (In Press). The application of K function analysis to the geographical distribution of road traffic accident outcomes in Norfolk, England.
Kerner, J. F., Andrews, H., Zauber, A. and Struening, E., 1988. Geographically-based cancer control: methods for targeting and evaluating the impact of screening
interventions on defined populations. Journal of Clinical Epidemiology, 41(6): 543-553.
Langran, G., 1991. Time in Geographic Information Systems. Taylor and Francis, London.
Low, I., 1988. An introduction to accident damage control in the workplace. Part 2: The accident phenomenon. Journal of Occupational Safety and Health - Australia and New Zealand, 4 : 49-55.
Martin, D., 1991. Geographic Information Systems and their Socio-economic Applications. Routledge, New York.
Multi-Agency Working Group on Accidents, 1995. Prevention of falls and their Sequelae amongst older people. South East Institute of Public Health.
Pukkala, E., 1992. Cancer maps of Finland: an example of small area-based mapping.
Presented at WHO meeting of the geographical information systems for health care management meeting, Helsinki.
Rice, D. P., MacKenzie, E. J., Jones, A. S., et al., 1989. Cost of injury in the United States A Report to Congress. San Francisco: Institute for Health and Ageing, University of California.
Robertson, L., 1992. Injury Epidemiology. Oxford University Press, New York.
Secretary of State for Health, 1991. The Health of the nation. A consultative document for health in England. Prepared by the Department of Health. London: HMSO.
Todd, P., Bundred, P. and Brown, P., 1994. The demography of demand for oncology services : A health care planning GIS application. AGI 94:17.1.1-17.1.7
Vipulendran, V., Mason, A. R. and Sunderland, R., 1988. Cost to the NHS of accidents to children in the West Midlands. British Medical Journal, 296 : 611.
Wain, R., 1993. The use of a geographical information system in locality profiling.
Mapping Awareness, 7(8): 20-22.
Ward, A., 1994. Saving lives in the West Country, using GIS to improve ambulance response times. Mapping Awareness, April : 36-37.
Whitelegg, J., 1987. A geography of road traffic accidents. Transactions of the Institute of British Geographers, 12 :161-176.
World Health Organisation, 1988. Consultation on environmental health information systems in the European region. Summary report, Berlin (West), EURIICP/CEH 074 A(S).
World Health Organisation, 1992. Geographical information systems for health care management. Report on a WHO working group, Helsinki, EURIICPJHST 149.
Worrall, L. and Rao, L., 1991. The Telford urban policy information
system project. In: Worrall (Editor), Spatial Analysis and Spatial
Policy using Geographic Information Systems. Belhaven Press, London.
TABLE 1 HOSPITAL INPATIENT CASE DATA FORMAT CONCERNING ADMISSIONS
FOR INJURIES
| Variable | Length and type | Variable | Length and type |
| District of Treatment | 2 (C) | Hospital provider | 5 (A-N) |
| Admission source | 2 (N) | Admission method | 2 (N) |
| Discharge method | 1 (N) | Discharge destination | 2 (N) |
| Patient identification | 10 (A-N) | District of Treatment | 3 (A-N) |
| Sex | 1 (N) | Postcode | 7 (A-N) |
| Local authority code | 4 (A-N) | Electoral ward | 2 (A-N) |
| Referring G. P. code | 8 (A-N) | Registered G. P. code | 8 (A-N) |
| Speciality of treatment | 5 (A-N) | Diagnoses 1 - 7 | 7 (A-N) |
| Date of hospital admission | Y-M-D | Date of discharge or death | Y-M-D |
| Age | 3 (N) | Century of birth - 8 or 9 | 1 (N) |
| Date of Birth | Y-M-D |
(C) Character
(N) Numeric
(A-N) Alpha Numeric
Y-M-D Year, month, day
Wells, A. J
ERDAS UK Ltd,
Telford House,
Fulbourn,
Cambridge,
CB1 5HB
Cocks, F. M
School of Earth Science,
University of Greenwich, Medway Towns Campus
Chatham Maritime,
Chatham,
Kent,
ME4 4AW
Cryer, C
South East Institute of Public Health,
David Salomons' Estate,
Broomhill Road,
Tunbridge Wells,
Kent,
TN3 OXT
Davidson, L
Lambeth, Southwark and Lewisham Health Commission,
1 Lower Marsh
London
SE1 7NT