Scand J Soc Med 3: 5-11, 1975

An Evaluation of the Official Swedish Statistics on Seriously Injured in Road Traffic Accidents! Jens Sande and Jan Thorson From the Department of Social Medicine, Karolinska Sjukhuset, Stockholm and the Central Bureau of Statistics, Stockholm

An evaluation of the official Swedish statistics 011 seriously illjured ill road traffic accidents. Sande, J. and Thorson, J. (Central Bureau of Statistics, Stockholm and Dept. of Soc. Med., Karolinska Sjukhuset, Stockholm). Scand J Soc Med 1975, 1 (5-11).

Only 32 % of the category "seriously injured" in road traffic accidents were correctly registered as such in the official road traffic accident statistics. Furthermore, 25 % were registered, though wrongly, as "slightly injured". Consequently almost one-half of the seriously injured were not registered at all, and they constituted a drop-out which contributes a bias to the statistics obtained. False registration is primarily associated with short-term hospitalization and absence of surgical operations during the hospital stay, i.e. slight injuries, and low age. The material analysed was comprised of 2689 individuals from hospital or road traffic accident statistics covering the Uppsala hospital region during 1966.

The general picture of the magnitude of the road, accident injury problem is described annually in the publication "Road Traffic Accidents", published by the Swedish Central Bureau of Statistics. These statistics are elaborated in accordance with principles developed within ECE in "1965 (6). These principles imply that reports from the police are the basis of the statistics and that the definition of seriously injured includes those who have been admitted to hospitals or have suffered fractures or crush injuries. It is supposed that the latter category, i.e, seriously injured not institutionally treated, constitute's a small group. Thus, statistics on in-patients might be used as criteria for evaluation of the comprehensiveness of official statistics concerning "seriously injured" persons. This was the main aim of the study now reported. A preliminary report was presented at the 3rd International Congress for Traffic Medicine", New York, 1969.

I

REVIEW OF THE LITERATURE As regards fatalities in road traffic accidents there are two sources for comparisons: the official road traffic accident statistics and the official statistics on causes of "death. The discrepancies between them are small. Each contains I 200 to 1300 cases annually. The limitation of the official road traffic accident statistics to fatally injured persons dying within 30 days after the time of accident is consequently immaterial. This condition concerning fatalities, i.e, that the police-reported material is complete, is not thought to be valid in the case of non-fatalities. However, the magnitude of the drop-out number and its qualities had not been very thoroughly investigated until Roosmark's & Fraki's study was undertaken (12). By interviewing a sample of the Swedish population they found that 51 % of the road traffic accidents causing injuries were not investigated by the police and not reported to the Central Bureau of Statistics. It was also estimated that 40 % of accidents causing injuries were not reported to any insurance company. The definition of road traffic accident with injury means that the trauma occurred on a public road, that at least one motor vehicle was involved and that a doctor was consulted. The material was a sample of 2 998 individuals, born between 1895 and 1948, from the whole of Sweden, and the investigation concerned accidents in the period 1963-1964; the drop-out was 269. Hansson (8) found that 43 % of seriously injured in-patients were correctly registered, 40% of injured in-patients were registered by the police; among out-patients the percentage was 29. Hansson analysed patients treated at Halmstad hospital, Sweden, in 1970. Further information on the "magnitude" of Scand J Soc Med 3

6

J. Sande and J. Thorson

the problem was collected from the international literature: Bya "Mikrozensus" in April 1965 in West Germany it was found that about 25 % of children injured in road traffic accidents were not registered by the police. No methodological data were apparent (16). . Slatis (13) observed that the number of accident victims seen by doctors exceeded the number reported to the "civil authorities" by 26%. Data on out- and in-patients collected in 1962 in Finland were analysed. Within a material of in- and out-patients in Oslo, Norway, three-quarters were registered with the police, according to Be (5). Nordentoft et al. (II) investigated the extent of police enquiries into traffic accidents resulting in 2055 victims who comprised a Danish hospital material. Only 803 (39.1 %) of these injured were known to the police. As regards "seriously injured" the percentage was 72; as regards "slightly injured" it was 26. The hospital had registered 99.4 % of the total material. The drop-out from the police material was mainly single vehicle accidents, slightly injured, road user categories other than car occupants, and patients who required a short period of treatment at hospital or attended the hospital long after the accident. There was a positive correlation between the intensity of the traumas and registration with the police. The material consisted of people who consulted the hospital in Odense from the February 1st to the November 30th, 1971, with the exclusion of observation cases. The definition of "seriously injured" used was almost completelyin agreement with that of ECE, and it was expressed as admission to hospital. "Slight injury" implied consultation as an out-patient.. A study with an approach similar to that of Nordentoft et al. was carried out for Greater Copenhagen by Kamper-Jorgensen (9). He found that in November 196860% of those "seriously injured" and 40% of the "slightly injured" were reported to the police. The definitions of serious and slight were arbitrary and in accordance with the nature of the injuries; fractures, lacerations being "serious", wounds and dislocations being "slight". The rule of the ECE on this matter of classification was referred to above. It could be criticized, and it has not been accepted for the official statistics in Denmark (18). Nordentoft et al. (II) used a similar definition, however. It would ScandJSocMedJ

be valuable if a common definition were introduced in order that like could be compared with like and a difference such as the one between the latter two projects would be prevented or at least supplemented. Ekstrom et al. (7) found that only 29% of those children who were admitted to a clinic for pediatric surgery in Stockholm during a to-year period were registered with the police. It was also found that 33 % of those who were killed in traffic had been admitted to hospital. It was concluded that the police-registered material is unsuitable for morbidity studies of traffic injuries and that hospital materials would be unsuitable for mortality analysis. MATERIAL AND METHODS Medical service in Sweden is organized into seven selfsupporting areas. One of them, the Uppsala region. was used in this investigation. It contains one-sixth of the Swedish population, or 1.3 million inhabitants. Patient statistics from all hospitals within this area are dealt with in Stockholm at the National Board of Health. A more extensive description of Uppsala hospital region appears in (15), (19). All in-patients who were admitted to hospitals in the region because of traffic injuries in 1%6 were chosen for this project. The material consisted of 2536 persons represented by discharges from the hospitals which occurred as items in a list produced by the National Board of Health and Welfare. Because the admissions of 1%6 were chosen, discharges in 1967 were also checked with regard to date of admission, and those admitted in 1%6 were included. The definition of a traffic item in the lists was limited by the then-existing codes for external causes of injury, a Swedish modification (10) now abandoned. Nevertheless, there might have been a considerable risk of misclassification of traffic injured by similar codes, e.g. "late effects of motor vehicle accidents", "vehicle accidents n.u.d." or "machinery accidents". These groups were checked, and the importance of the misclassification was regarded as small (see (19». The police material was taken from the Central Bureau of Statistics in Sweden and covered the same geographic area as the hospital statistics. The comparison was primarily carried out on the basis of dates and birth registration numbers, which should occur in both materials, though not for those classified as "slightly injured" by the police. Nor for all seriously injured could this criterion be fulfilled. Concordance regarding date of birth. sex, place of the accident, or possibly on the basis of a birth registration number with two figures transposed were also accepted as criteria of identity. Also concordance as to year of age in addition to sex, date and place of the accident was accepted as a criterion ranked four in quality. The distribution of the hospital material according to this scale of criteria is

Swedish statistics on road traffic accidents presented in Table I. The method of computer was used as regards those with complete birth registration numbers in both registers. The remainder of the study was performed manually. Seventy-seven percent of those "seriously injured" in the traffic accident material (RASM) appeared in the hospital material (HM) according to identity of birth registration numbers. On the other hand, the opposite quality was predominant as regards correspondence between "slightly injured" in the RASM and the HM. A check of "scene of accident" was performed because of the fact that the accidents were categorised as road traffic accidents in the sense of the official statistics, only if they had happened on roads: Two materials of in-patients were specially examined by questionnaire or by telephone interview to check the scenes of accidents. One material consisted of children, i.e. in-patients aged less than IS at the time of accident in 1965. Another material consisted of adults whose accidents happened in 1%6. Because of another concomitant project (9) and for. financial reasons, children injured in 1%5 and not those of 1966 were studied in this respect. The children were a total material from the Uppsala hospital region of 1%5; the adults a sample of 20% of the corresponding stratum of 1966. Seventy-seven percent of the accidents involving children and 91% of the accidents involving adults had occurred on the roads (19).

RESULTS Table II sets out the main results. The hospital material (HM) covers 94 % of all those who were seriously injured, according to the ECE-definition. The material from the road traffic Table I. The quality of estimates of the identity between items in the RMa and the RASMa Percentages within parentheses Injured according to the RASM Quality

Seriously No. (%)

Slightly No. (%)

l. Identical personal

registration numbers 2. Identical dates of birth; sex; time and scene of the accident 3. As 2 but difference between two figures in the dates of birth (mostly transposition) 4. Identity of age; sex; time and scene of the accident Total

479 (77) 69 (II)

58 (10)

70 (11)

25 (4)

2 (0)

620 (100)

524 (86) 607 (100)

HM: Hospital material. RASM: Road accident statistics material.

a

7

Table II. The completeness of the road accident statistics on "seriously injured" Original data

Corrected data"

No.

No.

%

%

l. Reported as seri-

773 ously injured 2. Institutionally' 2536 treated 3. Reported as seriously injured and institutionally treated 620 4. Not reported as seriously injured but institutionally treated 1916 5. Reported as seriously injured but not institutionally treated 153 6. Reported as slightly injured and institutionally 607 treated 7. Not reported at all but institutionally treated 1309 8. Total number of seriously injured ("2+5") 2689

29

773

32

94

2251

94

23

620

26

71

1 631

68

6

153

6

23

607

25

49

1024

43

100

2404

100

"Corrected data" are original data minus non-traffic accidents (see above control studies).

a

accident statistics (RASM) covers 32 %, according to data which were corrected with respect to estimated proportions of those who had not been injured on the roads: 9% of the adults, 191 persons, and 23 % of the children, 94 persons, or 285 persons altogether. The total, 12404, should be regarded as a minimum, because a drop-out, however vague, corresponding to those 153 cases who were only registered with the "RASM", is thought to exist. Slightly injured amounted to 607 or 25 % of the 2 404. Totally unknown to the police were I 024, or 43%. Thus the drop-out from the official statistics on road accidents as regards seriously injured was 68%, if only those who were correctly registered with the RASM are considered. Also considering those registered as slightly injured, the drop-out rate in the RASM is 43 %. Three-quarters of the children not registered with the RASM were injured in places not intended for traffic, e.g, parks, play-grounds. A good half of Scand J Soc

M~d 3

8

J. Sande and J. Thorson

Table III. The HM distributed according to category ofregistration and type of accident in the RASM Reported within the RASM as

I. Accidents with cars

seriously injured No. (%)

slightly injured No. (%)

545 (30)

490 (27)

772 (43)

I 807 (100)

51 (16)

74 (23)

202 (62)

327 (100)

12 (4) 12 (19) 620 (24)

32 (9) 11(17) 607 (24)

295 (87) 40 (63) 1 309 (52)

339 (100) 63 (100) 2536 (100)

Not reported No. (%)

Total No. (%)

2. Accidents with motor cycles

(not with cars) 3. Single-vehicle accidents with pedal

cycles 4. Miscellaneous accidents

Total

the children within this latter category were riding a bicycle at the moment of the accident (21). The incorrect classification of in-patients as slightly injured implies a reduction of the frequency for that category in the RASM. The reported number of slightly injured was about 2700 for the region under study. However, 607 of them were institutionally treated and should rightly have been reported as seriously injured. The correct estimate for slightly injured would thus be around 2100. However, the number of slightly injured is only approximate-only those reported by the police are known with certainty. Table III is a distribution according to categories within the RASM and types of accident according to the classification of the hospital statistics. Those 153 who were registered within the RASM alone are excluded. The concordance of seriously with slightly injured is better than that of any of those within the third category, those who only existed within the HM. Thirty percent of accidents involving cars were correctly reported. The corresponding results were 16% for motor cycles and 4 % for· single vehicle accidents with pedal cycles. Because of the

then-existing classification of accidents (10) the classes of accidents were very crudely defined. Pedestrians were distributed according to vehicle involved; single vehicle accidents only occurred for pedal cycles.' Table IV illustrates the distribution of the HM according to categories within the RASM and length of hospital care. The median value was 9 days for those who were correctly reported and 4 days for both the other categories. It is remarkable that 30% of those not reported were treated for more than one week. Age and sex were also studied with relation to the categories of registration within the RASM. The findings regarding age are shown in Table V. The extent of correct registration (19%) among children was low. Single-vehicle accidents involving pedal cycles were common in this group: 46% of children aged 5-9 and 39% of those aged 10-14 at the time of accident were pedal cyclists who 1

This method of coding has been criticized e.g. in

(17). The rules were changed in the subsequent issue of the Swedish Classification of Diseases and Injuries (14).

Table IV. The HM distributed according to category ofregistration and length ofinstitutional care Length of institutional care (days)

Reported as seriously injured and institutionally treated Reported as slightly injured but institutionally treated Not reported but institutionally treated Total Scand J Soc Med 3

0-2

3-7

8-14

15-21

22-35

36-

No. (%)

No. (%)

No. (%)

No. (%)

No. (%)

No. (%)

73 (12)

210 (34)

120 (19)

52 (8)

65 (10)

100 (16)

620(100)

204 (34)

231 (38)

81 (13)

30 (5)

32 (5)

29 (5)

607(100)

403 (31)

511(39)

207 (16)

58 (4)

51 (4)

79 (6)

1 309(100)

680 (27)

952 (38)

408 (16)

140 (6)

148 (6)

208 (8)

2536 (100)

Total No. (%)

Swedish statistics 011 road traffic accidents Table V. The HM distributed according to category of registration and age Reported within the RASMas seriously injured No. (%)

slightly injured No. (%)

5-14 15-19 20--29 30--54 55-74 75-

II (17) 65 (19) 112 (23) 121 (28) 153 (25) 131 (26) 27 (30)

10 (16) 74 (21) 144 (30) III (26) 143 (23) 106 (21) 19 (21)

43 (67) 207 (60) 231 (47) 198 (46) 314 (51) 271 (53) 45 (49)

64 (100) 346(100) 487(100) 430 (100) 610(100) 508 (100) 91 (100)

Total

620 (24)

607 (24)

I 309 (52)

2536 (100)

Age

M

Not reported No. (%)

Total No. (%)

were injured in falls (Table VI). Those aged 2029 were reported to a relatively high degree, which

might be explained by the fact that as many as 84% of this age group were injured in car accidents (Table VI). Old age too is associated with a comparatively frequent registration; 30% were correctly reported. Presumably this is explained by an increase in severity of injury in those aged over 50 or 60 for every trauma. The interpretation of Table V may, as exemplified, benefit from a knowledge of the distribution according to type of accident and age. This is presented as Table VI. Accidents involving cars were especially numerous among small children and adults aged 20-54. Motor cycle accidents showed a sharp peak in the 15-19 age group, showing just how serious is the problem of moped accidents (I, 3); however,

9

the present illustration is somewhat biased because of the crudeness of the classification system. Single-vehicle accidents involving pedal cycles are of greater importance in the age groups 5-14 and 55 upwards than in any other group. The distribution in Table VI is adequately explained by exposure, though the inclusion of all collisions with cars in the first category exerts a serious bias. It is estimated that half the mopedists were injured in collisions with cars, as well as one-third of pedal cyclists aged less than 15 and half of pedal cyclists aged 15 upward. These estimates are based on the above supplementary materials to the present study (19) and another analysis of mopedists aged 15-16 based upon the hospital statistics from the Uppsala Region for 1965; it was accounted for in (2). In Table VII, the distribution according to category of registration and sex is presented. The differences are very small, which is surprising, as the accident patterns are not equal for the two sexes. This was established by the study of the above control materials and the following results were obtained (19, 22): No c1earcut difference was observed for boys and girls with regard to road user categories, pedal cyclists comprised about half of the injured of both the sexes. However, girls were more frequently injured by simple falls; boys more frequently in collisions with motor vehicles. As regards adults, car occupants made up 57 % of the males, 61 % of the females. Differences occurred for other road users: e.g, 15% of the males and 3 % of the females had been riding a moped at the time of the accident. The reverse sex distribu-

Table VI. The HM distributed according to type ofaccident and age (columns 1-5: percentages]

Age

Accidents with cars

Accidents with motor cycles (not 1)

Singlevehicle ace. with pedal cycles

aneous ace.

Total

Total numbers

(I)

(2)

(3)

(4)

(5)

(6)

M 5-9 10-14 15-19 20-29 30-54 55-74 75-

83 48 54 65 84 81 70 59 71

6 3 4 28 II 8 13 13 13

6 46 39 7 4 8 14 22 13

5 3 3 0 2 3 3 6 3

100 100 100 100 100 100 100 100 100

64 184 162 487 430 610 508 91 2536

Allages

Miscell-

Scand J Soc Med 3

IO

J. Sande and J. Thorson

Table VII. The HAl distributed according to category ofregistration and sex Reportedwithin the RASM as seriously slightly injured injured No. (%) No. (%) Males Females Total

Not reported No. (%)

Total No. (%)

424 (25) 196 (24)

417(24) 190 (23)

883 (51) 426 (52)

I 724 (\00) 812 (\00)

620 (24)

607 (24)

I 309 (52)

2536 (\00)

tion was observed for pedestrians and pedal cyclists. In Table VIII, the material is distributed according to category of registration and body area, of all diagnoses. Injuries to the head and face were frequently associated with registration as slightly injured, especially so regarding internal injuries. Also a great part of this group was not reported at all. Certainly, the internal head injuries are often slight in a medical sense, and several patients were institutionally treated for observation only. An analysis of the specificity of the diagnosis "commotio cerebri" showed that about half of these cases within the above control materials from the Uppsala Hospital Region were not based on incontestable facts and 1/7 of them had no basis at all (20). Best reported were injuries to the throat or thorax and the lower extremities. Among the 2536 of the HM there were 1089 who had been surgically operated on. The operations appeared as code numbers according to a special classification of the National Board of Health and Welfare. Those who were operated on had been registered correctly, i.e. as seriously injured, more frequently than the others. The above associations, duration of hospital care, age of 2029 and high age respectively with frequent registration as seriously injured, and the present findings, support the concept that patients with severe injuries are more often reported than those with slight injuries (see II). This is also supported by the fact that those who suffer long-term effects of trauma are more frequently reported as seriously injured than those who recover within 5 years: the respective proportions were 50% respectively 38 % according to the follow-up of the control material of children (19). Regarding adults, the findings Scand J Soc Med 3

were that 36% of those who suffered long-term effects were correctly reported as seriously injured, and of those who recovered 20% were (19).

DISCUSSION Because of the concomitant change in definition of seriously injured in road traffic accidents and the availability of the statistics on in-patients in the Uppsala Hospital Region, a good possibility of a check-up on an important group in the official statistics on road traffic accidents was offered. It has been established that the drop-out problem is considerable, even with respect to the category "severely injured". Of course it is of doubtful value to run statistics which contain only about 32% of the items which by definition should have been included. Furthermore, the fact that further 25% of the cases were investigated by the police is nevertheless of no use: this group was reported as slightly injured and was thus confused with an indistinct category, not clearly distinguished from the "uninjured". The material analysed was valid for the year 1966, the first year of the present ECE-definition of seriously injured being used in Sweden. The

Table VIII. The HAl distributed according to category ofregistration and body area; all diagnoses Reportedwithin the RASM as Body area

seriously injured No. (%)

Head, 589 (29) face Throat, 108 (38) chest Neck, 44 (28) back Abdomen, 54 (33) pelvis Upper extr.

slightly injured No. (%)

Not reported No. (%)

Total No. (%)

527 (26)

940 (46)

2056(\00)

60 (21)

113 (40)

281 (100)

39 (25)

74 (47)

157 (100)

40 (25)

69 (42)

163 (100)

147 (31)

108 (23)

213 (46)

468 (100)

262 (34) 62 (26)

156 (21) 67 (28)

342 (45) 107 (45)

760 (\00) 236 (100)

1266(31)

997 (24)

I 858 (45)

4 121 (100)

Lower extr.

Misc. Total

Swedish statistics

proportion of seriously injured among survivors in road traffic accidents has risen from 22 % in that year to 33 % in 1973, according to the official Swedish statistics on road traffic accidents.! The motives for the official statistics of road traffic accidents should be defined. Is the task limited to a description of the magnitude of a problem by number of individuals in a "suitable" number of classes? Or is it also aimed at presenting a basis for prevention of injuries and/or accidents? Obviously, neither objective can be well served by the present statistics: besides their incompleteness the bias of the material available is an indeterminate factor. For the official statistics to be of use in nation-wide epidemiological studies or for international comparisons, as accepted by Bull (4), the problems raised in the present study will have to be solved. Perhaps a good solution would be to use the present patient statistics as one source in order to assess the magnitude of the problem. Subsequently, samples from it could be investigated to elucidate special conditions of interest from the preventive point of view. Some of the resources which are now available for the country-wide routine statistics would be better devoted to sampling applied to the hospital or the official statistics. The advantages of sampling versus total studies, especially if the latter are incomplete, appear pertinent to the present situation: less risk of bias, smaller costs, better quality of the observations, results obtained rapidly. REFERENCES I. AIdman, B.: Personskadeolyckor i trafiken. Lakartidningen 62: no. 19. 1965. 2. AIdman. B. & Thorson. J.: Pediatric injuries. /11 (ed, K. M. Brinkhous), Accident Pathology. Proc. Int. Conf., Washington D.C., June 1968. US Government Printing Office, Washington. D.C .• 1970. 3. - Motorization and Traffic Mortality in Sweden. Accident Analysis and Prevention. 3. 215, 1971. (Uikartidningen 68: 1863, 1971 in Swedish». 4. Bull, J. P.: International comparisons of road ac-

I

Sande, J.: Personal communication.

011

road traffic accidents

11

cident statistics. Accident Analysis and Prevention /: 293,1969. S. Be, 0.: Road Casualties. An Epidemiological 'Investigation. Universitetsforlaget, Oslo, 1972. 6. ECE: Statistics of road traffic accidents in Europe 1966. United Nations. New York, USA, 1968. 7. Ekstrom, G., Gastrin, U. & Qvist, 0.: Trafikolycksfall bland bam. Lakartidningen 63: 2886, 1966. 8. Hansson. P.: Road traffic casualties in a surgical department. Acta Chir Scand, Suppl. 442. 1974. 9. Karnper-Jargensen, F.: Palideligheden af den officielJe trafikulykkesstatistik. Ugeskr La-ger /34: 1005.1972. 10. Medicinalstyrelsen, Kungl.: Klassifikation av sjukdornar, del I, systematisk forteckning, Stockholm, 1965. II. Nordentoft, E. L., Dally. T., Lagoni, P., Adler Petersen, E. & Thaarup, C.: Rapportering av trafikulykker med personskade. Radet fOr trafiksikkerhetsforskning, Copenhagen. Rapport 13, 1972. 12. Roosmark, P. O. & Friiki, R.: Antal trafikolyckor, deras kostnader och de polisrapporterade olyckornas representativitet Ar 1964. Statens Vaginstitut. Specialrapport 60. Stockholm. 1968. (Interview investigation of road traffic accidents. Accident Analyses and Prevention I: 279, 1969.) 13. Slatis, P.: Injury patterns in road accident traffic. An analysis of 5291 cases recorded in Finland during three months 1962. Ann Chir Gynecol Fenn 56: Suppl. ISO, 1967. 14. Socialstyrelsen: Klassifikation av sjukdomar. Stockholm, 1968. IS. Socialstyrelsen: Patientstatistik I. Stockholm, 1969. 16. Statistisches Bundesamt: Wirtschaft und Statistik. Heft 10,649, 1966. 17. Thorson, J.: Klassifikation av trafikolyckor for patientstatistik. Socialmedicinsk Tidskrift 46: 43, 1969. 18. - Kallrnaterial for beskrivning av personskador i trafiken. Liikartidningen 66: 1736, 1969. 19. - Long-term Effects of Traffic Accidents. The annual incidence of permanently disabled in-patients in the Uppsala Hospital region, Sweden. Diss. Karolinska institutet. Stockholm. 1973. Hakan Ohlssons forlag, Lund, Sweden. 1975. In press. 20. Injury diagnoses within morbidity statistics. Scand J Soc Med /: 109, 1973. 2 J. - Pedal cycle accidents with special reference to the prevention of injuries caused by falls. Scand J Soc Med 2: 121, 1974.

Address for reprints: Jan Thorson, M.D. Nockebyvagen 9 S-16140 Bromma Sweden

Scand J Soc Med J

An evaluation of the official swedish statistics on seriously injured in road traffic accidents.

Only 32 per cent of the category "seriously injured" in road traffic accidents were correctly registered as such in the official road traffic accident...
406KB Sizes 0 Downloads 0 Views