Injury Prevention through Data Linkage Phase 3: The linkage of hospital inpatient data to police road traffic accident reports in East and West Sussex by Colin Cryer Sylvia Westrup Adam Cook CHSS at Tunbridge Wells May 2000
Summary Injury Prevention through Data Linkage Phase 3 CONTENTS Key Findings..iii Summary... v 1. Introduction... 1 1.1 Background... 1 1.2 The value of the project... 3 1.3 Aims and methods... 6 2 Linkage Performance... 9 2.1 Linkage study results... 9 2.2 Possible reasons why links were not made... 13 2.3 Characteristics of the unlinked cases... 15 2.4 Discussion in relation to other work... 18 3 Accuracy of the data... 21 3.1 Linkage study results... 21 3.2 Others work... 27 3.3 Implications... 30 4. Bias and the linked database... 31 4.1 Estimated proportion of cases in the linked database... 31 4.2 Investigation of bias... 32 4.3 Bias and linked serious and non-fatal injury cases... 36 5. Epidemiology... 47 5.1 All injury all road users... 48 5.2 Serious injury, all road users... 59 5.3 Occupants of motor vehicles... 70 5.4 Motorcyclists... 80 5.5 Pedal cyclists... 88 5.6 Pedestrians... 96 5.7 Child casualties... 104 5.8 Older road users... 114 5.9 Young drivers aged 17-24... 124 6. Principal Findings and Conclusions... 133 References... 141 i Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 ii Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Key Findings Serious injury resulting from road traffic crashes is an important health problem. Information derived solely from police road traffic accident reports (STATS19) can misinform accident prevention work Information on serious injuries, derived from hospital admission cases linked to police road traffic accident reports, is more robust than that from STATS19 alone. There were an estimated 1,340 non-fatal serious injuries that occurred on the roads in East and West Sussex during the period 1 April 1995 to 31 March 1998. The pattern of occurrence of non-fatal serious injury showed the following: All road users- There were high rates for people aged 15-24 and aged 75 and over. 37% were occupants of motor vehicles, 32% pedestrians, 21% motorcyclists, and 10% pedal cyclists. Half occurred on A-roads, most frequently on the A259, A23, and A27. Occupants of motor vehicles- Approximately equal proportions of men and women were seriously injured. 60% occurred on A-roads, most frequently on the A259 and A27. Motorcyclists- A high proportion occurred at junctions; he/she was making no manoeuvre in a high proportion of junction accidents. A high proportion of serious crash injuries involved side impacts to the motorcycle or the other vehicle involved. A high proportion of serious injuries were to the leg or foot (60%), and a low proportion to the head (7%); relatively few had short stays in hospital Pedal cyclists- He/she was making no manoeuvre in a high proportion of junction accidents. A high proportion resulted in serious head injuries (30%). Pedestrians- Highest rates were to those aged 75 and over. 25% of serious pedestrian injury occurred in Brighton. On average, there was at least one pedestrian injury in Sussex per month resulting in over 28 days stay in hospital. Child casualties- The majority seriously injured were pedestrians or cyclists. The average length of stay in hospital was less than for other user groups. Older casualties- Half were pedestrians, and a quarter were seriously injured when driving. The average length of stay in hospital was higher than for other road users. On average, there was at least one injury to an older person in Sussex per month resulting in over 28 days stay in hospital. Young drivers- Two thirds were to men; and two thirds occurred in West Sussex. A higher proportion occurred on A-roads, most often the A27, A29, and A259. The pattern of injury suggests a greater proportion of high velocity crashes. iii Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 iv Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Injury Prevention through Data Linkage Phase 3: The linkage of hospital inpatient data to police road traffic accident reports in East and West Sussex Summary The scope of the report This executive summary gives an overview of the road crash data generated by linking hospital inpatient data to police road traffic accident reports in the combined study area of East Sussex, Brighton & Hove, and West Sussex. (Following this, separate tables and figures will be produced for (1) East Sussex (excluding Brighton and Hove), (2) East Sussex Brighton & Hove, and (3) West Sussex.) The size of the problem Accidental injury is an important health problem, and road traffic crashes are one of the most important causes of accidental injury. In the period 1 April 1995 to 31 March 1998, there were 3,166 casualties recorded as Serious or Fatal on the police s road traffic accident database (STATS19) and there were 2,666 admissions to hospital recorded as road traffic crashes in the combined study area. The need for data linkage v Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 These data sources individually have limitations for the investigation of cause and the identification of potential methods of prevention. It was hypothesised that the linkage of hospital inpatient data records to STATS19 would provide a richer source of data for these purposes. STATS19 data is inaccurate As a result of data linkage, problems were identified with the accuracy of the STATS19 data. The magnitude and nature of these problems could result in injury prevention planning being misinformed. The linked database is incomplete. The linked database is also potentially biased; it is estimated that only 50% of RTAs admitted to hospital are included in the linked database. The most complete linkage was achieved for pedestrians, and car drivers, the least complete for pedal cyclists. but the evidence suggests that the linked database is superior to STATS19 for describing non-fatal serious injury The linked database is probably more suitable than STATS19 for investigating cause and prevention of RTAs resulting in non-fatal serious injuries since it appears to be less biased than STATS19 data. Reassuringly, the patterns of occurrence of non-fatal serious injuries from the linked database were similar to those for the hospital inpatient data, when tabulated by age, sex, and road user group. The patterns of occurrence for the STATS19 data was different to both of these. Serious injuries resulting from RTAs There were 1072 non-fatal serious injury cases identified from the admissions data. These were defined as road traffic crashes with a length of stay in hospital of 4 days or more, or for which a transfer to another hospital occurred, and which took place in the study area over this 3 year period. Of these, 668 were vi Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 included on the linked database and 64% were males. The linkage rate for these non-fatal serious injury cases was 62% over all road user groups including pedal cyclists. What follows within this executive summary are a number of figures and tables, that could only be produced from the linked database, which show the pattern of occurrence of non-fatal serious injury cases. For more detailed analysis, please see the main body of the report. These analyses have been repeated for the following subgroups: occupants of motor vehicles, motorcyclists, pedal cyclists, pedestrians, child casualties, older road users, and young drivers. Details of this work can be found in sections 5.3 to 5.9. Please note that in the tables and figures that follow, the rates and numbers that have been generated from the linked database are likely to be approximately one half those of the true rates and numbers of casualties with serious non-fatal injuries defined as casualties with a length of stay in hospital of 4 or more days or as transfers to another hospital. vii Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Age-specific Rates of Serious Injury as Estimated from the Admissions and Linked Databases 1995/6-1997/8 Rate per 100,000 130 120 110 100 90 80 70 60 50 40 30 20 10 0 00-14 15-24 25-34 35-64 65-74 75+ Age Band Admissions Linked - There were high rates for people aged 15-24 and aged 75 and over. - For pedestrians, the highest rates were for people aged 75 and over. - Across all road users, approximately two thirds of the serious casualties were male. - Approximately equal proportions of men and women occupants of motor vehicles were seriously injured. - A greater rate of serious injury was apparent in young men than young women drivers. - The vast majority of serious injuries cases amongst motorcyclists were men. - 70% of serious pedal cycle casualties were men or boys. viii Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Crashes resulting in Serious Injury by Road Type - 1995/6-1997/8 Unclassified 16% C Road 17% A Road 51% B Road 16% - Half of the non-fatal serious injuries occurred on A-roads, most frequently on the A259, A23, and A27. - For occupants of motor vehicles, 60% occurred on A-roads. - The highest proportion that occurred on A-roads was for young drivers aged 17-24, most often on the A27, A29, and A259. - For pedal cyclists, a lower proportion occurred on A-roads (36%), and a higher proportion on unclassified roads (30%). - For child casualties, there was a lower proportion of serious injuries on major roads than for other road users. ix Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Crashes Resulting in Serious Injury by District - 1995/6-1997/8 Brighton Eastbourne Hastings Hove Lew es Rother Wealden Adur Arun Chichester Craw ley Horsham Mid-Sussex Worthing 0 10 20 30 40 50 60 70 80 90 100 Number - 25% of serious pedestrian injury occurred in Brighton - Two thirds of non-fatal serious injuries to young drivers (aged 17-24) occurred in West Sussex. x Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Crashes Resulting in Serious Injury by Road User - 1995/6-1997/8 Pedestrian 0-4 5-9 10-15 16-29 30-44 45-59 60-74 75+ Pedal Cyclist 0-15 16+ Motorcycle Riders Passengers Car/Taxi Driver U17 Driver 17-24 Driver 25-34 Driver 35-44 Driver 45-59 Driver 60-74 Driver 75+ Passenger 1 3 9 14 18 19 18 24 26 29 28 33 34 32 40 39 55 57 83 106 0 10 20 30 40 50 60 70 80 90 100 110 Number - 37% of non-fatal serious injuries on Sussex roads were occupants of motor vehicles, 32% pedestrians, 21% motorcyclists, and 10% pedal cyclists. - There were a large number of non-fatal serious injuries to motorcyclists relative to their road use. - The majority of seriously injured child casualties were pedestrians or pedal cyclists. - Half of older casualties of RTAs were seriously injured as pedestrians, and a quarter whilst driving. xi Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Crashes Resulting in Serious Injury by Manoeuvre 1995/6-1997/8 Ahead (no manoeuvre) 324 Ahead - right hand bend 62 Ahead - left hand bend Overtaking stationary - offside 39 75 Overtaking moving - offside Turning right 31 50 Turning Left Starting Stopping Reversing Other 19 19 11 12 26 0 50 100 150 200 250 300 350 Number - In many cases, no manoeuvre was involved; but where it was, it often involved overtaking, turning right or turning left. - A high proportion of motorcycle and pedal cycle accidents resulting in serious injury occurred at junctions; he/she was making no manoeuvre in a high proportion of these accidents. - A high proportion of serious crash injuries to motorcyclists involved side impacts to the motorcycle or the other vehicle involved. - A high proportion of serious injuries to young drivers, compared with other occupants of motor vehicles, occurred on right turns. xii Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Crashes Resulting in Serious Injury by Junction Type - 1995/6-1997/8 No junction 315 Roundabout 28 Mini roundabout 1 T or staggered junction 212 Y junction 15 Slip road 7 Crossroads 38 Multiple junction 12 Private drive 35 Other 5 0 50 100 150 200 250 300 350 Number - Approximately half of road users seriously injured were injured at junctions, and in the majority of cases these were T-junctions. - A high proportion of motorcycle and pedal cycle accidents resulting in serious injury occurred at junctions. - Seventy percent of non-fatal serious RTA injuries to older people occurred at junctions. xiii Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Serious Injuries to Pedestrians by Type of Movement - 1995/6-1997/8 Crossing - driver's nearside 91 Crossing - driver's nearside masked 19 Crossing - driver's offside 48 Crossing - driver's offside masked 14 In carriageway - stationary 5 In carriageway - stationary masked 1 Walking in carriageway - facing traffic 6 Walking in carriageway - back to traffic 8 Unknown/Other 13 0 10 20 30 40 50 60 70 80 90 100 Number - Most pedestrians were injured whilst crossing the road. xiv Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Serious Injury to All Road Users by Nature of Injury 1995/6-1998/8 No injury code recorded Head Neck Thorax Abdomen, lower back, lumbar spine & pelvis Shoulder & upper arm Elbow & forearm Wrist & hand Hip & thigh Knee & lower leg Ankle & foot Multiple body regions Unspecified body part Poisonings Complications of trauma 1 1 3 7 14 13 12 20 31 42 59 60 89 128 188 0 20 40 60 80 100 120 140 160 180 200 Number - Across all road users, the body site of injury was the lower leg (28%), head (19%), hip and thigh (13%), chest (9%), and abdomen, back or pelvis (9%). - A high proportion of serious injuries to motorcyclists were to the leg or foot (60%) - A relatively high proportion of serious pedal cycle injuries were head injuries (30%). - A large proportion of serious injuries to pedestrians were to the legs and feet (almost 60%). - The pattern of injuries for young drivers suggests a greater proportion of high velocity crashes. xv Prepared by CHSS at Tunbridge Wells
Summary Injury Prevention through Data Linkage Phase 3 Non-fatal Injuries (All Severities) to All Road Users by Length of Stay in Hospital - 1995/6-1997/8 0 178 1 417 2 200 3-4 201 5-7 173 8-14 196 15-28 124 >28 80 0 50 100 150 200 250 300 350 400 450 Number - For all road users combined, discharge from hospital was the same or the following day in almost 40% of cases (all severities), was over a week in 26%, over a fortnight in 13%, and over a month in 5%. - Relatively few seriously injured motorcyclists had a short length of stay. - On average, there was at least one pedestrian injury and at least one injury to an older person in Sussex per month that resulted in over 28 days stay in hospital. - The average length of stay in hospital for child casualties was less, and for older casualties was greater, than for other road user groups. xvi Prepared by CHSS at Tunbridge Wells
Part 1: Introduction Injury Prevention through Data Linkage Phase 3 Injury Prevention through Data Linkage Phase 3: Linkage of Hospital Inpatient Data to Police Road Traffic Accident Reports in East and West Sussex Part 1. Introduction 1.1 Background Accidental injury is an important health problem. The prevention of accidental injury is a national priority, and it is one of four key areas in the Government s health strategy Saving Lives: Our Healthier Nation. Reducing the burden of injury is essential since: accidental injuries are responsible for 10,000 deaths a year across England accidents are the greatest single threat to life for children and young people and are responsible for more admissions to hospital amongst children than any other cause treating injuries costs the NHS in the region of 1.6 billion each year (Secretary of State for Health 1999). Injury has been reported to be the fifth leading cause of death, the third leading cause of potential years of life lost, and the fourth most common cause of hospital bed utilisation (Cryer et al. 1993) Road traffic accidents are one of the most important causes of accidental injury. There were almost a third of a million road accident casualties in 1998 in England of whom more than 3,500 died, half of these on rural roads (Secretary of State for Health 1999). Motor vehicle traffic accidents are the leading cause of injury death and the second leading cause of hospital bed utilisation for injury (Cryer et al 1993, Marchant et al 1997). Information has an important role to play in the prevention of accidents and injury: to accurately describe 1 Prepared by CHSS at Tunbridge Wells
Part 1: Introduction Injury Prevention through Data Linkage Phase 3 the distribution of accidents and injury across the population; for monitoring trends; to examine the longterm effects of accidents and injuries; to influence the setting of priorities; to influence the allocation of resources; to give clues to the causes of injury, and the events that lead to them; and to suggest methods of prevention. Readily accessible and relevant data systems available within the NHS are the hospital inpatient data and the OPCS registrations of deaths. These data have good information on the consequences of accidents, but contain limited data on the personal characteristics of the injured person, where the event occurred, and how it occurred. Data sources from other agencies also have their strengths and weaknesses. For example the data collected by the police on road traffic accidents have a wealth of information on the circumstances of the accident, but have limited information on its consequences. Given the limitations of any one source to provide comprehensive information relevant to accident prevention and injury control, it was postulated that if an accident database could be linked with the hospital inpatient data, the resultant database would provide a much more powerful tool for injury control. Phase 1: The assessment of data sources for data linkage Phase 1 of this project (Cryer and Aspinall, 1994) was a review of accident data sources in order to identify and describe the candidate accident data sources for linkage and these included: road traffic (STATS19); home and leisure (HASS and LASS); work (HSE RIDDOR system); school; and fire statistics. Phase 1 focussed on data sources available in East Sussex. Other sources of data related to injury prevention were also identified and described briefly but these were not amenable for data linkage within East Sussex. The information collected suggested that any one of three sources could be considered for linkage: STATS19, RIDDOR, or fire statistics. However, the report concluded that there was a greater probability of success (fewer difficulties) and the likelihood of greater benefits of linkage between STATS19 and inpatient data. Phase 2: Data linkage pilot: 2 Prepared by CHSS at Tunbridge Wells
Part 1: Introduction Injury Prevention through Data Linkage Phase 3 The linkage of police road traffic accident reports to hospital admissions to Eastbourne DGH and Conquest Hospital, Hastings. This work (Cryer et al. 1995) investigated the feasibility and utility of linking the police road traffic accident reports to hospital inpatient data within East Sussex. It concluded the following. - Within East Sussex, the linkage of STATS19 and hospital admission data can only be successfully achieved through the use of names and addresses of the casualties, and that as the data are currently organised, a successful matching procedure involves manually searching police registers and records. Linkage was particularly useful in highlighting inaccuracies in the Police RTA database which is currently used by the County Council for planning road safety and accident prevention initiatives. - The data linkage process provides a more reliable group of 'severe' motor-vehicle traffic accident cases than STATS19. For those cases that could be linked, there is more comprehensive data on the circumstances of the accident than would be available otherwise. This can provide insight into what is causing accidents that result in hospital admission, and so guide prevention. To make real progress in using linked data for accident prevention and injury control, a greater number of cases are required than were available in the pilot study. Consequently, it was recommended that several years data be linked in order to provide a suitable database to describe the epidemiology of injury resulting from road traffic accidents. 1.2 The value of the project Reported here are the results of the linkage of hospital admissions data to police road traffic accident reports for the whole of East and West Sussex and for the years: 1995/6, 1996/7 and 1997/8. The use of a linked data source by the road safety teams within the County Council Departments of Highways and Transport could highlight priorities for their work. In some instances, their current plans are influenced by the accident severity gradings provided by the police. The police use a crude system to code the severity of the casualty (ie. Fatal, Serious, Slight ) which provides only limited information of questionable accuracy on the seriousness of the injury. Through linkage to hospital data, more detail of the injury, 3 Prepared by CHSS at Tunbridge Wells
Part 1: Introduction Injury Prevention through Data Linkage Phase 3 length of hospital stay and discharge destination is available, and so the most seriously injured (non-fatal) casualties can be identified. In the pilot it was found that these crude STATS19 severity codes are quite inaccurate, and so it was anticipated that analysis of more reliable data resulting from data linkage would result in some changed priorities. The Health Authorities envisage that the more detailed cause of accident information from the police road traffic accident reports linked with the hospital information, which includes diagnosis, length of stay, and operative procedure, would provide valuable information for the development of health promotion and injury prevention programmes. Work commissioned by the Department of Health and carried out by the Public Health Information Strategy Group (1993) identified a number of data items required for an accident database (shown in Table 1). It can be seen from this table that STATS19 provides information on the accident characteristics, some personal characteristics, but very little on the consequences of the accident. On the other hand, the hospital inpatient data has much less on the accident characteristics, much more on the personal characteristics, as well as some data items relating to the consequences. Within a linked database, therefore, the majority of the recommended data items for an accident database would be available. Consequently, the expectation was that such a linked database would give a much more powerful base for injury control. Furthermore, it was felt that linkage would: - increase information available for injury prevention; - identify crashes that are not ascertained within the STATS19 database but which result in hospital admission; - identify errors in the databases: ie. where there is a mismatch in the information between STATS19 data and hospital in-patient data; - identify sources and magnitude of bias resulting from inaccurate data and missing cases. This knowledge would help to improve the data on both systems, and hence improve their usefulness. 4 Prepared by CHSS at Tunbridge Wells
Part 1: Introduction Injury Prevention through Data Linkage Phase 3 Table 1: Accident information requirements Structure Items STATS19 Hospital Inpatient Data Accident Characteristics Place of occurrence Geographic identifier of location Y Y Type Y Y Circumstances Y Personal Characteristics Age & sex Y Y Area of residence Y Socio-economic?Y Ethnic Group Y Activity Y Predisposing factors? Consequences Nature of injury Y Severity of injury??y Health service impact Y Outcome?? Key: Y = Recorded on the database?y = Indicator can be generated from the database? = Only limited information available <blank> = No information available 5 Prepared by CHSS at Tunbridge Wells
Part 2: Linkage performance Injury Prevention through Data Linkage Phase 3 1.3 Aims and Methods The aims of the project are: - to describe the epidemiology of non-fatal serious injury resulting from road traffic accidents using a database created by linking hospital inpatient data to police road traffic accident reports; - to identify crashes that are not ascertained within the STATS19 database but which result in hospital admission; - to identify errors in the databases. Population The population of interest are those who had a road traffic accident in East and West Sussex during the period 1 April 1995 to 31 March 1998. Selection of STATS19 cases In theory, all cases admitted to hospital should be classified as 'Serious' on the STATS19 database. A large proportion of admitted cases, however, are classified as 'Slight' - and so all casualties on the STATS19 database that fell within the catchment area were considered. Selection of hospital cases Cases, admitted to hospitals as a result of motor-vehicle traffic accidents sited in Brighton & Hove, East and West Sussex, during the period 1 April 1995 to 31 March 1998, were provided to CHSS at Tunbridge Wells by the Information Managers from the following acute hospital Trusts: Brighton, Eastbourne, Hastings & Rother, Mid Sussex (Haywards Health), Queen Victoria (East Grinstead), Royal West Sussex (Chichester), Crawley Horsham, and Worthing & Southland. 6 Prepared by CHSS at Tunbridge Wells
Part 2: Linkage performance Injury Prevention through Data Linkage Phase 3 Within hospitals, the principal diagnosis at admission is placed in the first diagnosis field (DIAG1) on the computer record. If this was an injury, it should take a code between S00-S99, or T00-T98 within the International Classification of Diseases 10th revision (ICD10) codes (WHO 1992). Each injury diagnosis should be accompanied by an external cause code (E-code), usually in the next available diagnosis field (DIAG2), but occasionally will be held in the field DIAG3 to DIAG7. Transport accidents were selected for which DIAG1 had a code between S00-T98 inclusive, and where the first E-code following this principal diagnosis was in the range V01-V89 (ICD10). Matching procedure The schema shown in Figure 1 indicates how these cases were linked to the STATS19 data. The STATS19 records do not include names and addresses. Using names and addresses from the hospital data, the police s accident reference number for the casualty was found from the police paper records: their accident registers and accident reports. This police accident reference number was added to the inpatient record and so provided the link between the hospital records and the STATS19 computer record, which includes the accident reference number. Hospital Inpatient Records Link to STATS19 STATS19 Transcribe Accident Reference Number to Hospital Inpatient Record Match Police Accident Register: Manual search using name and date of accident Figure 1: A schema of the manual data linkage procedure In more detail, the following procedure was followed. The inpatient data was obtained from these hospitals and was loaded on to the secure computer at CHSS at Tunbridge Wells. Road traffic accidents were selected from these hospital records, and were ordered by name, in a list that included date of admission, and hospital where admitted. Research staff (SW) visited each police station in the two counties and manually matched these details to those held in the police's road accident register (held in books). A manual search of the police accident registers was carried out for names and dates of accidents 7 Prepared by CHSS at Tunbridge Wells
Part 2: Linkage performance Injury Prevention through Data Linkage Phase 3 that matched, approximately, to the names and dates of admission on each of the hospital records. The details from the inpatient record were kept confidential from the police during this procedure. Once a match was found, the police accident reference was written against the details from the hospital record, and then these were transcribed on to the inpatient data held at CHSS for the next phase of the matching. These records were then linked to the relevant STATS19 records using the accident reference number as the linking variable. All the links, and any multiple matches, were checked by comparing the information on age, sex, type of road user, and date of accident / admission between the two electronic databases. Where a mismatch on these variables existed, these cases were investigated through a manual search of the more extensive police accident (paper) files. A summary of the number of cases considered and linked is shown in Figure 2. Figure 2: Hospital inpatients linked to STATS19 casualties 5313 Transport accidents admitted to hospital 2666 Road traffic accident casualties admitted to hospital 1625 Cases linked to STATS19 casualties 1041 Cases unlinked 1567 Non-fatal injury cases admitted to hospital and linked to STATS19 8 Prepared by CHSS at Tunbridge Wells
Part 2: Linkage performance Injury Prevention through Data Linkage Phase 3 Part 2: Linkage Performance 2.1 Linkage study results A breakdown of the linkage performance is shown in Table 2 and Figure 3. Of the original 2,666 admissions coded as road traffic accidents, 1625 of these cases (61%) were linked to the registers and paper records in police offices and subsequently to the STATS19 record. The linkage rates were higher for admissions to Brighton (65%), Eastbourne (66%), Hastings (66%), and Worthing (65%). It was less for Mid-Sussex (54%), and Crawley Horsham (48%). Queen Victoria at East Grinstead is a tertiary referral hospital, ie. other hospitals (many of which are outside Sussex) refer their cases to this hospital for super-specialist care. This is likely to be a major contributor to the very low number of cases (n=33) and the very low linkage rate for these cases (6%). Table 2: Linkage performance by hospital and police station STATS19 Records Hospital Brighton Haywards Heath Crawley Eastbourne Hastings Worthing Chichester Unlinked Grand Total Brighton 328 51% 11 2% 81 13% 224 35% 644 Eastbourne 2 1% 207 63% 7 2% 113 34% 329 Hastings & 3 1% 242 65% 126 34% 371 Rother Mid-Sussex 156 50% 10 3% 144 46% 310 Queen 2 6% 31 94% 33 Victoria St.Richards 22 4% 320 57% 219 39% 561 Crawley & 109 48% 116 52% 225 Horsham Worthing 30 16% 73 38% 22 11% 68 35% 193 Grand Total 330 12 % 169 6% 139 5% 301 11 % 249 9% 95 4% 342 13% 1041 39% 2666 9 Prepared by CHSS at Tunbridge Wells