- Email: [email protected]
Inpatient costs of injury due to motor vehicle traffic crashes in New Zealand
Accid. Anal. nnd Prao. Vol. 25. No, 5. pp. 585-592. Printed in the U.S.A.
1993
Owl-4575193 $6.00 + .oo 0 1993 Pergamon Press Ltd.
INPATIENT COSTS OF INJURY DUE TO MOTOR VEHICLE TRAFFIC CRASHES IN NEW ZEALAND JOHN D. LANGLEY,'DAVID
PHILLIPS,~and STEPHEN W. MARSHALL'
‘Injury Prevention Research Unit, University New Zealand; *British Development (received
8 November
of @ago, Medical School, P.O. Box 913, Dunedin, Division-Pacific, Private Bag, Suva, Fiji
1991; in revised form 20 N#ve~~er
1992)
Abstract-The hospital inpatient costs for the treatment of injury were obtained for Dunedin Hospital for a two-year period using the Resource Utilisation System. These data were used as a basis for estimating national inpatient costs for motor vehicle traffic crashes (MVTCs). The results show that injuries were on average more expensive to treat than non-injuries ($3,115 vs. $2,749 per case). At a mean cost of $5,253 per case, injuries due to MVTCs were the most expensive class of injury event to treat. The average cost of treating a case varied by the class of road user, nature of injury, and severity of injury. Pedestrians were on average, twice as costly to treat as motor vehicle occupants. Costs for treating the late effects of motor vehicle crashes were higher, with a mean cost of $5,695. Internal injuries were on average 4.5 times more expensive to treat than intracranial injuries, the least expensive group of injuries to treat. On average, critical injuries cost 15 times more to treat than minor injuries. The cost of treating pedestrians was disproportionately higher than that for other road users. Nationally pedestrians accounted for 10% of all hospitalised road users, but they contributed 18% of the total cost. Internal injuries had the highest disproportionate cost accounting for 7% of cases and 16% of the costs. Similarly, severe and critical injuries combined accounted for 6% of the national cases but contributed 19% of the national costs. Overall, the leading contribution to the total national cost were made by occupants (430/o), lower limb fractures (24%), and moderate injuries (37%). It w-as estimated that the annual national inpatient cost for treating victims of MVTCs was between $43.0 million and $47.4 million, with the lower figure being considered the more accurate. In addition to this was the estimated national cost of treating the late effects of motor vehicle crashes, which was estimated to be $9.1 million.
INTRODUCTION
MVTC-related injuries require operative or other surgical intervention, which contributes to longer lengths of stay and greater resource utilisation (Guria 1991). The introduction into many Area Health Boards in New Zealand in recent years of a financial management system known as the Resource Utilisation System (RUS) has permitted the description and analysis of hospital costs in a manner hitherto not possible. The principal aim of this study was to determine the inpatient costs attributable to MVTCs in total, by class of road user and by nature and severity of injury.
The societal costs of injuries due to motor vehicle traffic crashes (MVTCs) are an essential element in cost benefit analyses of roading projects. The cost of medical treatment accounts for a significant proportion of the cost of injury. The most expg,nsive component of the medical costs are those attributable to hospital inpatient treatment. Recently revised estimates for New Zealand for injuries resulting in hospitalisation indicate that inpatient costs account for 65% of medical costs (Guria 1991). The revised project evaluation manual of Transit New Zealand (1991) incorporates the recently estimated societal costs (Guria 1991) mentioned above. As in the previous manual (Bone 1986), the estimates relating to medical treatment have several limitations; one of the principal ones is that they are based on the average cost of hospitalisation per day for all admissions. It is very likely that this is an underestimate, since, relative to disease and other classes of injury, a disproportionate number of
METHOD Inpatient costs were obtained for the principal secondary care facility in the Otago region namely Dunedin Hospital. Dunedin Hospital is a 450-bed teaching hospital attached to a university and provides the full spectrum of specialist management 585
5X6
J. 1). LANGLEY
including neurosurgery. Dunedin has a population of approximately 106,000 persons. There are approximately 4,000 injury admission per year. The comparable figures for New Zealand as a whole are 3.5 million and 55,000, repectively. For the period I April 1988 to 3 I March 1990 all admissions to Dunedin Hospital with a principal diagnosis of injury (ICD-O-CM codes 800-999) were identified from the electronic database maiilt~ined by the hospital. All cases of medical misadventure (Supplementary ICD codes: E870-E879~ and adverse reaction to drugs (E930-E949) were excluded from the analyses if the principal diagnostic code was not a late effect of injury (90.5-909). The period of two years was used to enhance precision in the estimates, especially in those areas where there were relatively few events. MVTCs were defined as those events that had an E code in the range E8 IO-E819 (“motor vehicle traffic accidents”). Late effects* of motor vehicle crashes (E929.0) were also included but these were treated separately because it could not be determined definitely what proportion were traffic crashes and there was an absence of information on class of road user and severity of injury. The information on costs was obtained from the RUS system operating in Dunedin Hospital. The RUS system develops the cost of a specific patient admission by converting the resources of the hospital (staff, materials, etc.) into specific patient services. These are divisible into fixed and variable, direct and indirect costs. Fixed costs are those not affected by volume changes, whilst variable costs are those that change in relation to volume consumed. Direct costs are those specifically associated with a given procedure or service, whilst indirect costs are those not so associated. These costs’once derived are processed through a computer model giving a “library” of costs for a given department or service. These are then applied to any inpatient admission. The only elements not incorporated into the equation are those relating to capital items and depreciation (Ayling 1988). Depreciation and interest on loans is estimated to be in the region of 75%10%. Goods and services tax (12.5%) has not been included in any estimates. The inpatient costs exclude accident and emergency costs for those cases who were admitted subsequent to a visit to an accident and emergency department. Costs include both costs for initial acute care as well as rehospitali-
* Late effects of accidental injury is used ‘.. to indicate accidental injury as the cause of death or disability from late effects, which are themselves classifiable elsewhere. The “late effects” include conditions reported as Tuch or occurring a$ sequelae one year or more after accidental in.jtlry” (WHO 1977).
et al.
sation for further treatment and rehabilitation. No outpatient costs have been included. Dunedin Hospital was used as the principal data source because Dunedin is one of the original sites for the RUS project in New Zealand. and consequently the database covers a relatively long time period and is more complete than in other centres. The Abbreviated Injury Scale (AIS) was used as a measure of severity of injury. The AIS is the most widely recognised severity scoring syatcm based on anatomic descriptors. ICDiAlS scores were calculated using the ICD-CM-9 to AIS computer mapping programme ICDMAP (The Johns Hopkins Health Services Research and Development Center et al. 1988). Scores were based on the principal injury requiring admission, which is actually recorded upon discharge. Validation studies show that while the computerised conversion is not perfect, it provides reasonably good information on severity, which would otherwise not be available for the large injury populations described here (MacKenzie, Steinwachs, and Shankar 1989). Since injury diagnoses from multiple body regions are rarely recorded in the discharge summaries, no attempt was made to compute ISS for multiple injuries. National estimates of costs were based on the overall mean cost of a MVTC for Dunedin. Assessment of the precision in this estimate was sought by also providing estimates based on the mean costs attributable to different classes of road user, nature of injury, and severity of irrjury. All national inpatient numbers produced in this report were obtained from the 1988 Health Statistical Service morbidity file, The method for selecting cases was the same in all respects to that employed for the Dunedin Hospital cases. In 1983 New Zealand moved from a bulk grant system to a population-based funding formula for funding public health services in New Zealand. As a result some area health boards found they were overfunded, and central government sought to gradually reduce their funding to bring them in line with the formula. The Otago Area Health Board was a case in point. Starting in 1989 the board moved to reduce its expenditure by $25 million over a fouryear period. As a consequence greater efficiencies were achieved. The mean cost of treating inpatients was substantially reduced in 1989. This was possibly due to higher efficiency. However, a portion of it could be due to achange in the distribution of injuries as well, particularly where the number of cases is small. The 1988-1989 period of study should, therefore, be considered a transitional period during which efficiency improved. Since similar changes were taking place in all hospital systems and there
Costs
of injury
due to MVTCs
existed a large variation in the level of efficiency, an average over the two-year period was considered appropriate for estimating the national average. The national estimates can thus be treated as at January 1989. The approach adopted here has been to provide costs for an unadjusted figure for the two-year period and, where appropriate, costs for each year and thus provide the opportunity for readers to use the figure they consider appropriate for their purposes. RESULTS Earlier work undertaken by the authors has demonstrated that injury cases are on average more expensive to treat than noninjury cases. The data are presented here in order to place MVTCs in the wider perspective of all injury admissions and noninjury admissions. For the year ending 30 June 1990, the mean cost per injury was $3,115, 13% higher than that for noninjury ($2,749). The mean cost per day for noninjury was $43 1, whereas that for injury was $428 indicating that most of the extra cost attributable to injury was due to the longer stay in hospital (6.3 vs. 6.6 days). Admission for injury accounted for 10% of cases and 11% of all inpatient costs (Phillips, Langley, and Marshall 1993). For the period 1 April 1988-31 March 1990, a total of 3,905 injury cases were identified. Costing information was not present for 214 (6%) cases and these were deleted from all subsequent analyses. The deleted cases, compared to the nondeleted cases, were younger (mean age 26.1 years vs. 33.1 years), contained fewer women (31% vs. 38%), and spent less time in hospital (3.3 days vs. 7.8 days). Deleted cases sustained more intracranial injury (34% vs. 12%) but fewer lower limb fracture (8% vs. 20%). The mean costs for treating all injury for the 1988 and 1989 financial years were $3,114 and $3, I 13, respectively. The distribution of costs was highly skewed, and it should be borne in mind that in this situation the sample means become responsive to changes in the upper ranges of the cost distribution. The median cost for the two years was $2,2 13 and the mode $140. Table 1 shows the Dunedin costs for the leading classes of injury event (in terms of total cost) according to major E-code categories for 1988 and 1989. Categories with fewer than 20 cases are not presented. Of the categories listed, MVTCs (584 cases) had the highest average mean cost at $5,253 over the two-year period. The mean cost per day for a MVTC
in New Zealand
587
was $422, whereas that for all injury was $399, suggesting that the relatively high mean cost for MVTCs is largely attributable to the relatively longer length of stay. The average cost of a MVTC was 1.7 times that for all injury cases. The table also includes national frequencies of hospital admissions together with estimates of total costs based on the Dunedin averages. The results show that in terms of the distribution of events, the Dunedin experience was similar to that of the country as a whole. MVTCs accounted for approximately 18% of all injury cases nationally and 29% of the estimated national costs. Table 1 also includes comparable data for late effects of motor vehicle crashes (E929.0) and demonstrates that these are second only to MVTCs in terms of mean cost and account in large part for the relatively high mean cost of all types of late effect (E929). They also represent a significant national cost at an estimated $9.1 million. The mean cost for a MVTC for the 1989 financial year at $4,543 tended to be lower than that for the 1988 financial year, namely $5,906, although this drop was not statistically significant (Wilcoxon ranksum test, p = 0.21). Extrapolating from 1988 Dunedin data the estimate of total national cost was $53.3 million whereas the comparable figure using 1989 Dunedin data was $41 million. The variability in estimates is due to a small number of individuals who contributed disproportionately to the cost. For example, in 1989 in Dunedin one individual was admitted for a period of 213 days due to the late effects of a motor vehicle crash. Similarly the national series also contained cases with high bed-days stay. Costs by type of road user are shown in Table 2. “Other road users” includes other specified road users (e.g. horse riders) and unspecified persons. The most striking finding was the variation in average costs. Injuries to pedestrians were on average twice as costly to treat as those to occupants. Costs per day were, however, not so different. Pedestrians had a cost per day of $415, whereas, in the case of all road users, it was $422. A significant proportion of the mean pedestrian cost can thus be accounted for by the relatively high average length of stay for pedestrians, which was nearly twice that of all other road users. Table 2 also provides an estimate of national inpatient costs. The table shows that in terms of the distribution of road users, the Dunedin Hospital experience was not markedly different from that of the country as a whole. Pedestrians accounted for 10% of the national cases and 18% of the national costs. Overall, however, occupants contributed the greatest (43%) to the total costs.
588
J. D.
LANGLEY
et
al.
Table I. Inpatient costs by type of injury event: Dunedin Hospital 1988-1989 and national estimates (1988) DUNEMN ECoda
810-819
Motor
880-688
Falls
929
NATIONAL
1 Freq. 1 % 1Tot.LOS 1Av.LOSI
1Description
trattlc
crashes
12
1% 1
$3,067,597
1
Av.Cost
Freq.
1 46 1
Est.Tot.Cost
584
16
7.272
27
$5,253
9,026
21
30
10,154
9
$3.664.121
32
53,325
14,497
3p
$46,202,525
36
Late Effects
305
6
2,069
7
$1.018.317
9
53,339
4,241
10
$14.160.699
11
920
Sharp objects
313
6
1,060
3
5564,930
5
51,605
3,428
8
$6.187.540
5
950-959
Suicide 8 selfinfficted
193
5
1,561
6
$547,906
5
$2,839
3.061
7
$8.690.179
7
924
Hot substance
$4.558.600
3
86
2
1,249
15
$426,130
4
$4,955
920
All other events
1,692
46
12,707
8
55,270,600
46
$3,115
16,360
All Injury
3,691
100
28,800
8
100
$3,114
r 05
3
1Late
effects
of
motor
vebicfe
1
1
986
/
9 )
$11,492,004
5597,932
5 1 S&695
42,507
1 7599
S47,413,578
1 %
1.102
l929.0
vehicle
Tot.Cost
2 381 1001 4
36
$50.961.593 $132,761,136
(
$9,?ff5,659
1crashes (Est.Tot.Cost based
Mean costs tended to vary between years for motorcyclists and pedal cyclists, although these differences were not statistically significant (Wilcoxon rank-sum test, p = 0.010, p = 0.010, respectively). The small number involved preclude any meaningful interpretation. Extrapolating from the 1988 Dunedin data the estimate of the total national cost was $54.2 million, whereas the comparable figure using 1989 Dunedin data was $4 I .2 million. Table 3 shows the costs of treating cases according to the leading injury categories (according to total cost). The six injury categories listed were also the highest ranking injuries in terms of average cost. Average costs varied significantly with the treatment of internal injury incurring on average 4.5 times the cost of treating intracranial injury. Skull fracture had the highest cost per day at $580 followed by internal injury at $553 compared with the overall cost per day of $422. Internal injury had the highest disproportionate cost with 7% of the cases nationally and 16% of the costs. This was followed closely by skull fractures, for which the respective percentages were 9% and 18%. Overall, however, fractures of the lower limb contributed the greatest (24%) amount to the total cost. This result is of particular relevance to motorcyclists, who suffered a disproportionate number of lower limb fractures (Table 4).
on sum of sub totals.!
The average cost for all injury categories was lower in 1989. The greatest difference was for skull fracture where the average cost halved. In this case the average length of stay was reduced by 40%. Extrapolating from 1988 Dunedin data the estimate of total national cost was $47.9 million. whereas the comparable figure using 1989 Dunedin data was $39.5 million. Table 4 shows the distribution of the leading types of injury (according to injury type) by class of road user. Pedestrians and pedal cyclists had 92% and 93%, respectively, of their total costs accounted for by the six most expensive injuries to treat (i.e. those listed in Table 3). This contrasted with other road users, for whom the range was from 72%-83%. Table 5 shows the average costs according to severity as measured by the ICDIAIS. On average, costs approximately double with each increment in severity. The table also provides the distribution of AIS scores for the country as a whole and provides estimates of national costs extrapolating from the Dunedin data. Apart from the minor (AIS-I) injuries. the Dunedin Hospital severity case mix is not markedly different from that of the country as a whole. Although severe (AIS-4) and critical (AIS-5) injuries represented only 6% of the case load, they contributed 19% to the national costs. Overall, moderate
Table 2. MVTC inpatient costs by type of road user: Dunedin Hospital 1988-1989 ar Id national estimates (1988)
I
[Occupants
I I I
NATIONAL
DUNEDIN Freq. 291
I x
I I 501
Tot.LOSI
2,925l
I
Av.LOS
1 I
Tot.Cost
I % 1 I
Av.Cost
I
421
$4,471
21
$846,626
281
$5,100
1 1
31
$110,898
41
$7,393
j
101
$1,301,154
Freq.
1 % 1 I
4528
Est.Tot.Cost
I %
5C
234
39
Est. Totxost
based on sum of sub
totals.
589
Costs of injury due to MVTCs in New Zealand
Table 3. MVTC inpatient cost by leading classes of injury: Uunedin Hospital 1988-1989 and national estimates (1988)
I
I
I
/All injury
i
iOO/
f,272/
121
(AK-2) injuries contributed the greatest proportion (37%) to total national costs. With the exception of serious injuries (AIS-3), the average cost in 1989 was lower for all levels of severity. The most marked difference was for severe (AIS-4) injuries (1988: $12,604; 1989: $5,950); however, the size of this difference may be largely attributable to the small number of cases. extrapolating from 1988 Dunedin data the estimate of total national cost was $47.6, whereas the comparable figure using 1989 Dunedin data was $39.5 million. Table 6 provides a summary of four estimates of the national inpatient costs described above. In addition, an kstimate of national MVTCs based on the mean cost for a11injury is included. This shows that extrapolation from mean data for all injury underestimates the national cost of MVTCs by a sizable amount. DISCUSSION The inpatient costs presented here are for public hospital treatment. Because of shortcomings in coding, MVTC cases for private hospitals could not be identified. Private hospital costs have been estimated to be $3.5 million for the year ending 3 1 March 1990 (Langley, Phillips, and Marshall 1993). The results show that motor vehicle traffic crashes consumed 27% of local and 36% of national
Table 4. MVTC admissions.
I
I
I
584
$3,0§7,597
iOOl
I 55,253
I
100 IOOj t43,905,520 Est.Tot.cost based on sum of sub totals.
1
9,026
costs for the treatment of injury victims. The data presented here combine acute and readmission costs. An analysis of data presented by Rice et al (1989) shows 26% of the “initial hospitalisation”, “‘rehospitalisation”, and “inpatient rehabilitation” costs combined for injury victims were attributable to motor vehicles. The national estimate of costs based on the overall average cost for Dunedin for 1988-1989 was $47.4 million (Table I). The estimate based on type of road user (Table 2) was very similar ($47.3 miilion). Those based on nature of injury (Table 3) and severity of injury (Table 5), however, were significantly lower at $43.9 million and $43.0 million, respectively. The latter result is probably more accurate, since it appears from Table 5 that Dunedin has a relatively low percentage of minor injuries, thus the mean treatment cost would be higher than that for the country as a whole. What is not known is the degree to which the costs incurred in Dunedin are similar to those of the rest of the country. The major restructuring of the health sector in New Zealand has already meant that it will be dif~cult to obtain similar data in the future because of commercial sensitivities. The drop in the mean cost of treating a MVTC casualty between the two financial years is probably largely attributable to the reduction in mean stay from 15 days in 1988 to 10 days in 1989. This may
Leading classes of injury for each road user: Dunedin Hospital 1988-1989
J. D.
LANCXEY
Table 5. MVTC inpatient costs by severity of injury:
et al.
DunedinHospital 1988-3989and
national
Est.Tot.Cost
be an artefact of variation in case mix. A higher proporti~~n of occupants, who had the shortest mean stay, were admitted in 1989 compared to 1988 (53% vs. 47%). The results confirm Guria’s (1991) criticism of using the average cost of all hospitalisations as a proxy for the cost of motor vehicle-related trauma. The average cost of a motor vehicle traffic crash was shown to be approximately 1.9 times the average for all admissions, irrespective of cause. Moreover, the cost of motor vehicle traffic injuries was I.7 times that of the average cost for the treatment ofall injury, being on average the most expensive class of injury event to treat (Table 2). As the results show this difference is largely attributable to the longer stays associated with such crashes. Extrapolating from U.S.A. and Australian research, Guria recently reestimated the hospital/ medical costs of a “serious” (hospitalised) injury as $3,614 in June 1990 prices (Guria 1991). As has been shown here the figure is more likely to be 1.5 times that value, or approximately $5,250. The results also demonstrate that average costs vary significantly according to class of road user. Comparable research is virtually nonexistent.
Tabte 6. Summary of estimates of national inpatient costs of MVTCs in New Zealand Basis of Estimates
1
AII injury mean All MVTCs mean
Road User
i
Year 1990 1988 1989 1988189 1988
1 Estimate(m)
I 1
$28.8 $53.3 $41 .o $47.4 $54.2
estimates
(1988)
based on sum of sub totals.
Stutts, Rutledge, and Martell (1991) showed that hospital costs for motorcyclists tended to be similar to those for all other road users combined. A reanaiysis of the data presented here (Table 2) results in a similar conclusion. The mean cost for motorcyclists was $5,100, whereas that for all other road users combined was $5,313. However, as Table 2 shows. mean cost varies significantly within the nonmotorcyclist grouping. The mean cost of treating a motorcyciist is only 54% of that for treating a pedestrian. This highlights the inappropriateness of using average injury costs based on all MVTC victims in costbenefit studies that focus on a specific intervention for a specific class of road user. This is particularly important for pedestrians, whose medical costs are likely to make up a substantial proportion of the total costs of a crash, when there is minimal property damage. Pedestrians had a high mean cost and contributed disproportionately to the total cost. A longer-than-average length of stay was identified here as an explanation. Related to this is the severity of injury. Langley and Marshall (1993) have shown that pedestrians tend to have more severe injuries. The findings show that costs double on average for each increment of severity. This contrasts with a recent U.S.A. report that showed medical costs per person increased eightfold between minor and moderate injury, approximately threefold between moderate and serious, and serious and severe, and fourfold between severe and critical (Urban Institute 1991). It should be noted, however, that different ascertainment methods were used for cases and costs. and the numbers of severe and critical injuries used for the estimate in the present study are small. Perhaps what is of more interest is that minor and moderate injury contributes substantially to overall costs. It should be emphasised that the inpatient analyses were confined to motor vehicle traffic crashes and as such are an underestimate of the costs of all traffic crashes on public roads. Excluded from the road vehicle accidents analyses are “other
591
Costs of injury due to MVTCs in New Zealand
(E826-E829)“, that is, crashes that do not directly involve vehicles with motors. The most significant subgroup among these are bicycle crashes. Collins et al (1993) have estimated that on public roads there are approximately 1,100 bicycle crashes that result in inpatient treatment and that approximately 65% of these do not involve motor vehicles. Most of the deaths and the majority of the more serious injuries, however, are associated with motor vehicles. Given the relatively large numbers involved, and the interest in promoting helmet use, costs for all bicycle crashes on public roads should be estimated. The mean costs for specific subgroups were, in some cases, based on small numbers (e.g. pedal cyclists: IS-Table 2; critical injuries: 12-Table 5). Thus the national estimates based on them need to be treated with caution. Given the relative rarity of such events, an extra year’s data for Dunedin is unlikely to significantly improve the precision of these estimates. Significant gains in precision may be obtained for other moderately sized injury subgroupings (e.g. pedestrians: 66-Table 2; most of the injuries in Table 3; and severe: 27 and minor injuries: 48 in Table 5). An extra year of RUS data has become available since the commencement of the analyses, and it is thus recommended that the analyses undertaken here be repeated using the three year’s data. The data for late effects of motor vehicle crashes is not directly comparable with those for motor vehicle traffic crashes since in the former instance there would be a number of nontraffic cases. As previously noted it was not possible to identify these cases. It is assumed that the error introduced by these cases would be relatively small. This assumption is based on two findings. First, motor vehicle nontraffic crashes represent approximately 8% of all motor vehicle crashes (National Health Statistics Centre 1988). Second, nontraffic crashes tend to be less severe than traffic crashes (e.g. Begg et al. 1991). Langley et al (1991) have estimated that the total medical cost of treating all motor vehicle traffic crash-related injury is approximately $80 million. That estimate includes hospital accident and emergency and Accident Rehabilitation and Compensation Insurance Corporation (ACC) costs. The ACC meets most of the medical costs associated with injury victims treated by genera1 practitioners, specialists, paramedics, dentists, private hospitals, and conveyance for medical treatment. With one or two minor exceptions it does not pay for public hospital costs. At 64% of the total medical cost public hospital inpatient costs are a substantial portion of the economic burden. The estimate of $80 million is an
underestimate since it excludes bicycle crashes not involving motor vehicles and some public hospital costs. Significant among the latter are the costs incurred in providing community services to injury victims (e.g. domiciliary care). The estimate also excludes injury victims’ use of public hospital outpatient services (e.g. visits to the eye clinic). Of particular interest are the costs for “old” injury. Langley et al. (1991) have shown that $8.3 million (36%) of the ACC medical costs relate to injuries that occurred prior to the 1989 financial year. This figure, combined with that from inpatient analysis presented here ($9.1 million), demonstrates that the late effects of motor vehicle traffic crashes are substantial at $17.4 million. This estimate does not include the costs of inpatient treatment in the acute phase and within the first year of admission. These data suggest that the identification and prevention of the types of crashes that lead to late effects need to be a priority if the costs associated with medical treatment are to be substantially reduced. The results reported here provide a previously unavailable and important dimension to be considered when determining priorities for motor vehicle traffic crash prevention. A more accurate appraisal of priorities is largely dependent on: (ij disaggregation of ACC medical costs according to the areas of interest (e.g. road user, nature of injury), and (ii) RUS data from another large hospital, (iii) determining costs for public hospital outpatient treatment of injury, and (iv) determining what treatment costs are borne directly by patients.
Acknowledgements-This research was funded by Land Transport. The advice of Ellen MacKenzie in the planning of this study is appreciated. We wish to acknowledge the stimulation, enthusiasm, and assistance of Jagadish Curia in all phases of this research. The assistance and advice of the following are also acknowledged: David Chalmers, Bill Coppin, Nancy Devlin, Terri Green, Paul Hansen, Russell Pugh, Janet Seats, and Richard Whitney.
REFERENCES Ayling, J. Resource management in New Zealand public hospitals. New Zealand Health Review 8:32-26; 1988. Begg, D. J.; Langley, D. J.; Reeder, A. I. Motorcycle crashes resulting in death and hospitalisation: A report prepared for the Accident Compensation Corporation, Injury Prevention Research Unit. Dunedin, Accident Compensation Corporation, 1991. Bone, I. H. The economic appraisal of roading improvement projects. RRU Technical Recommendation TR9. Wellington: National Roads Board; 1986. Collins. B. A. et al. Bicycle crashes in New Zealand. (in preparation) Dunedin: Injury Prevention Research Unit, University of Otago; 1993.
592
.I.
D.
f..ANGl.tY
Curia, J. Estimates of social costs of accidents and injuries. Wellington: Land Transport; 1991. The Johns Hopkins Health Services Research and Development Center and the Maryland Institute for Emergency Medical Services System. ICDMAP-Determinseverity for hospital discharges: A ing injury programme to map ICD-9-CM diagnoses into AIS and ISS severity scores. Baltimore, MD: Author; 1988. Langley, J. D.; Marshall, S. W. Severity of road traffic crashes resulting in hospitalisation (under consideration 1993). Langley, J. D.; Phillips, Lt.; Marshall S. W. The public hospital inpatient costs of injury due to motor vehicle traffic crashes. Unpublished final report to Land ‘l’ransport Division, Ministry of Transport. Wellington: Ministry of Transport; 1991. MacKenzie, E.; Steinwachs, D.; Shankar B. Classifying trauma severity based on hospital discharge diagnoses. Medical Care 27(4):412-422; 1989. National Health Statistics Centre. Hospital and selected morbidity data 1988. Wellington: Department of Health; 1988. Phillips, D.; L,angley, J. D.; Marshall, S. W. Injury-the
et al.
medical and related costs; New Zealand 1993. New Zealand Medical Journal (in press). Rice, D. P.; MacKenzie, E. J.; and associates. Cost of injury in the United States: A Report to Congress. San Francisco, CA: Institute for Health and Aging, University of California and Injury Prevention Center, The Johns Hopkins University: 1989. Stutts. J. C.; Rutledge, R.; Martell, C. An Analysis of Injury Outcome and Insurance Status of Hospitalized Motorcyclists. 35th Annual Proceedings, Association for the Advancement of Automotive Medicine, Toronto Canada, October 7-9, 1991. Transit New Zealand. Project evaluation manual: Volume 1: Simplified procedures: Volume II: Full procedures. Wellington: Author: 1991. Urban Institute. The costs of highway crashes. Final Report to United States Department of Transportation. Washington, DC: U.S. Department of Transportation: 1991. WHO (World Health Organisation). International classification of diseases-ninth revision 1977. Geneva: WHO; 1977.
1993
Owl-4575193 $6.00 + .oo 0 1993 Pergamon Press Ltd.
INPATIENT COSTS OF INJURY DUE TO MOTOR VEHICLE TRAFFIC CRASHES IN NEW ZEALAND JOHN D. LANGLEY,'DAVID
PHILLIPS,~and STEPHEN W. MARSHALL'
‘Injury Prevention Research Unit, University New Zealand; *British Development (received
8 November
of @ago, Medical School, P.O. Box 913, Dunedin, Division-Pacific, Private Bag, Suva, Fiji
1991; in revised form 20 N#ve~~er
1992)
Abstract-The hospital inpatient costs for the treatment of injury were obtained for Dunedin Hospital for a two-year period using the Resource Utilisation System. These data were used as a basis for estimating national inpatient costs for motor vehicle traffic crashes (MVTCs). The results show that injuries were on average more expensive to treat than non-injuries ($3,115 vs. $2,749 per case). At a mean cost of $5,253 per case, injuries due to MVTCs were the most expensive class of injury event to treat. The average cost of treating a case varied by the class of road user, nature of injury, and severity of injury. Pedestrians were on average, twice as costly to treat as motor vehicle occupants. Costs for treating the late effects of motor vehicle crashes were higher, with a mean cost of $5,695. Internal injuries were on average 4.5 times more expensive to treat than intracranial injuries, the least expensive group of injuries to treat. On average, critical injuries cost 15 times more to treat than minor injuries. The cost of treating pedestrians was disproportionately higher than that for other road users. Nationally pedestrians accounted for 10% of all hospitalised road users, but they contributed 18% of the total cost. Internal injuries had the highest disproportionate cost accounting for 7% of cases and 16% of the costs. Similarly, severe and critical injuries combined accounted for 6% of the national cases but contributed 19% of the national costs. Overall, the leading contribution to the total national cost were made by occupants (430/o), lower limb fractures (24%), and moderate injuries (37%). It w-as estimated that the annual national inpatient cost for treating victims of MVTCs was between $43.0 million and $47.4 million, with the lower figure being considered the more accurate. In addition to this was the estimated national cost of treating the late effects of motor vehicle crashes, which was estimated to be $9.1 million.
INTRODUCTION
MVTC-related injuries require operative or other surgical intervention, which contributes to longer lengths of stay and greater resource utilisation (Guria 1991). The introduction into many Area Health Boards in New Zealand in recent years of a financial management system known as the Resource Utilisation System (RUS) has permitted the description and analysis of hospital costs in a manner hitherto not possible. The principal aim of this study was to determine the inpatient costs attributable to MVTCs in total, by class of road user and by nature and severity of injury.
The societal costs of injuries due to motor vehicle traffic crashes (MVTCs) are an essential element in cost benefit analyses of roading projects. The cost of medical treatment accounts for a significant proportion of the cost of injury. The most expg,nsive component of the medical costs are those attributable to hospital inpatient treatment. Recently revised estimates for New Zealand for injuries resulting in hospitalisation indicate that inpatient costs account for 65% of medical costs (Guria 1991). The revised project evaluation manual of Transit New Zealand (1991) incorporates the recently estimated societal costs (Guria 1991) mentioned above. As in the previous manual (Bone 1986), the estimates relating to medical treatment have several limitations; one of the principal ones is that they are based on the average cost of hospitalisation per day for all admissions. It is very likely that this is an underestimate, since, relative to disease and other classes of injury, a disproportionate number of
METHOD Inpatient costs were obtained for the principal secondary care facility in the Otago region namely Dunedin Hospital. Dunedin Hospital is a 450-bed teaching hospital attached to a university and provides the full spectrum of specialist management 585
5X6
J. 1). LANGLEY
including neurosurgery. Dunedin has a population of approximately 106,000 persons. There are approximately 4,000 injury admission per year. The comparable figures for New Zealand as a whole are 3.5 million and 55,000, repectively. For the period I April 1988 to 3 I March 1990 all admissions to Dunedin Hospital with a principal diagnosis of injury (ICD-O-CM codes 800-999) were identified from the electronic database maiilt~ined by the hospital. All cases of medical misadventure (Supplementary ICD codes: E870-E879~ and adverse reaction to drugs (E930-E949) were excluded from the analyses if the principal diagnostic code was not a late effect of injury (90.5-909). The period of two years was used to enhance precision in the estimates, especially in those areas where there were relatively few events. MVTCs were defined as those events that had an E code in the range E8 IO-E819 (“motor vehicle traffic accidents”). Late effects* of motor vehicle crashes (E929.0) were also included but these were treated separately because it could not be determined definitely what proportion were traffic crashes and there was an absence of information on class of road user and severity of injury. The information on costs was obtained from the RUS system operating in Dunedin Hospital. The RUS system develops the cost of a specific patient admission by converting the resources of the hospital (staff, materials, etc.) into specific patient services. These are divisible into fixed and variable, direct and indirect costs. Fixed costs are those not affected by volume changes, whilst variable costs are those that change in relation to volume consumed. Direct costs are those specifically associated with a given procedure or service, whilst indirect costs are those not so associated. These costs’once derived are processed through a computer model giving a “library” of costs for a given department or service. These are then applied to any inpatient admission. The only elements not incorporated into the equation are those relating to capital items and depreciation (Ayling 1988). Depreciation and interest on loans is estimated to be in the region of 75%10%. Goods and services tax (12.5%) has not been included in any estimates. The inpatient costs exclude accident and emergency costs for those cases who were admitted subsequent to a visit to an accident and emergency department. Costs include both costs for initial acute care as well as rehospitali-
* Late effects of accidental injury is used ‘.. to indicate accidental injury as the cause of death or disability from late effects, which are themselves classifiable elsewhere. The “late effects” include conditions reported as Tuch or occurring a$ sequelae one year or more after accidental in.jtlry” (WHO 1977).
et al.
sation for further treatment and rehabilitation. No outpatient costs have been included. Dunedin Hospital was used as the principal data source because Dunedin is one of the original sites for the RUS project in New Zealand. and consequently the database covers a relatively long time period and is more complete than in other centres. The Abbreviated Injury Scale (AIS) was used as a measure of severity of injury. The AIS is the most widely recognised severity scoring syatcm based on anatomic descriptors. ICDiAlS scores were calculated using the ICD-CM-9 to AIS computer mapping programme ICDMAP (The Johns Hopkins Health Services Research and Development Center et al. 1988). Scores were based on the principal injury requiring admission, which is actually recorded upon discharge. Validation studies show that while the computerised conversion is not perfect, it provides reasonably good information on severity, which would otherwise not be available for the large injury populations described here (MacKenzie, Steinwachs, and Shankar 1989). Since injury diagnoses from multiple body regions are rarely recorded in the discharge summaries, no attempt was made to compute ISS for multiple injuries. National estimates of costs were based on the overall mean cost of a MVTC for Dunedin. Assessment of the precision in this estimate was sought by also providing estimates based on the mean costs attributable to different classes of road user, nature of injury, and severity of irrjury. All national inpatient numbers produced in this report were obtained from the 1988 Health Statistical Service morbidity file, The method for selecting cases was the same in all respects to that employed for the Dunedin Hospital cases. In 1983 New Zealand moved from a bulk grant system to a population-based funding formula for funding public health services in New Zealand. As a result some area health boards found they were overfunded, and central government sought to gradually reduce their funding to bring them in line with the formula. The Otago Area Health Board was a case in point. Starting in 1989 the board moved to reduce its expenditure by $25 million over a fouryear period. As a consequence greater efficiencies were achieved. The mean cost of treating inpatients was substantially reduced in 1989. This was possibly due to higher efficiency. However, a portion of it could be due to achange in the distribution of injuries as well, particularly where the number of cases is small. The 1988-1989 period of study should, therefore, be considered a transitional period during which efficiency improved. Since similar changes were taking place in all hospital systems and there
Costs
of injury
due to MVTCs
existed a large variation in the level of efficiency, an average over the two-year period was considered appropriate for estimating the national average. The national estimates can thus be treated as at January 1989. The approach adopted here has been to provide costs for an unadjusted figure for the two-year period and, where appropriate, costs for each year and thus provide the opportunity for readers to use the figure they consider appropriate for their purposes. RESULTS Earlier work undertaken by the authors has demonstrated that injury cases are on average more expensive to treat than noninjury cases. The data are presented here in order to place MVTCs in the wider perspective of all injury admissions and noninjury admissions. For the year ending 30 June 1990, the mean cost per injury was $3,115, 13% higher than that for noninjury ($2,749). The mean cost per day for noninjury was $43 1, whereas that for injury was $428 indicating that most of the extra cost attributable to injury was due to the longer stay in hospital (6.3 vs. 6.6 days). Admission for injury accounted for 10% of cases and 11% of all inpatient costs (Phillips, Langley, and Marshall 1993). For the period 1 April 1988-31 March 1990, a total of 3,905 injury cases were identified. Costing information was not present for 214 (6%) cases and these were deleted from all subsequent analyses. The deleted cases, compared to the nondeleted cases, were younger (mean age 26.1 years vs. 33.1 years), contained fewer women (31% vs. 38%), and spent less time in hospital (3.3 days vs. 7.8 days). Deleted cases sustained more intracranial injury (34% vs. 12%) but fewer lower limb fracture (8% vs. 20%). The mean costs for treating all injury for the 1988 and 1989 financial years were $3,114 and $3, I 13, respectively. The distribution of costs was highly skewed, and it should be borne in mind that in this situation the sample means become responsive to changes in the upper ranges of the cost distribution. The median cost for the two years was $2,2 13 and the mode $140. Table 1 shows the Dunedin costs for the leading classes of injury event (in terms of total cost) according to major E-code categories for 1988 and 1989. Categories with fewer than 20 cases are not presented. Of the categories listed, MVTCs (584 cases) had the highest average mean cost at $5,253 over the two-year period. The mean cost per day for a MVTC
in New Zealand
587
was $422, whereas that for all injury was $399, suggesting that the relatively high mean cost for MVTCs is largely attributable to the relatively longer length of stay. The average cost of a MVTC was 1.7 times that for all injury cases. The table also includes national frequencies of hospital admissions together with estimates of total costs based on the Dunedin averages. The results show that in terms of the distribution of events, the Dunedin experience was similar to that of the country as a whole. MVTCs accounted for approximately 18% of all injury cases nationally and 29% of the estimated national costs. Table 1 also includes comparable data for late effects of motor vehicle crashes (E929.0) and demonstrates that these are second only to MVTCs in terms of mean cost and account in large part for the relatively high mean cost of all types of late effect (E929). They also represent a significant national cost at an estimated $9.1 million. The mean cost for a MVTC for the 1989 financial year at $4,543 tended to be lower than that for the 1988 financial year, namely $5,906, although this drop was not statistically significant (Wilcoxon ranksum test, p = 0.21). Extrapolating from 1988 Dunedin data the estimate of total national cost was $53.3 million whereas the comparable figure using 1989 Dunedin data was $41 million. The variability in estimates is due to a small number of individuals who contributed disproportionately to the cost. For example, in 1989 in Dunedin one individual was admitted for a period of 213 days due to the late effects of a motor vehicle crash. Similarly the national series also contained cases with high bed-days stay. Costs by type of road user are shown in Table 2. “Other road users” includes other specified road users (e.g. horse riders) and unspecified persons. The most striking finding was the variation in average costs. Injuries to pedestrians were on average twice as costly to treat as those to occupants. Costs per day were, however, not so different. Pedestrians had a cost per day of $415, whereas, in the case of all road users, it was $422. A significant proportion of the mean pedestrian cost can thus be accounted for by the relatively high average length of stay for pedestrians, which was nearly twice that of all other road users. Table 2 also provides an estimate of national inpatient costs. The table shows that in terms of the distribution of road users, the Dunedin Hospital experience was not markedly different from that of the country as a whole. Pedestrians accounted for 10% of the national cases and 18% of the national costs. Overall, however, occupants contributed the greatest (43%) to the total costs.
588
J. D.
LANGLEY
et
al.
Table I. Inpatient costs by type of injury event: Dunedin Hospital 1988-1989 and national estimates (1988) DUNEMN ECoda
810-819
Motor
880-688
Falls
929
NATIONAL
1 Freq. 1 % 1Tot.LOS 1Av.LOSI
1Description
trattlc
crashes
12
1% 1
$3,067,597
1
Av.Cost
Freq.
1 46 1
Est.Tot.Cost
584
16
7.272
27
$5,253
9,026
21
30
10,154
9
$3.664.121
32
53,325
14,497
3p
$46,202,525
36
Late Effects
305
6
2,069
7
$1.018.317
9
53,339
4,241
10
$14.160.699
11
920
Sharp objects
313
6
1,060
3
5564,930
5
51,605
3,428
8
$6.187.540
5
950-959
Suicide 8 selfinfficted
193
5
1,561
6
$547,906
5
$2,839
3.061
7
$8.690.179
7
924
Hot substance
$4.558.600
3
86
2
1,249
15
$426,130
4
$4,955
920
All other events
1,692
46
12,707
8
55,270,600
46
$3,115
16,360
All Injury
3,691
100
28,800
8
100
$3,114
r 05
3
1Late
effects
of
motor
vebicfe
1
1
986
/
9 )
$11,492,004
5597,932
5 1 S&695
42,507
1 7599
S47,413,578
1 %
1.102
l929.0
vehicle
Tot.Cost
2 381 1001 4
36
$50.961.593 $132,761,136
(
$9,?ff5,659
1crashes (Est.Tot.Cost based
Mean costs tended to vary between years for motorcyclists and pedal cyclists, although these differences were not statistically significant (Wilcoxon rank-sum test, p = 0.010, p = 0.010, respectively). The small number involved preclude any meaningful interpretation. Extrapolating from the 1988 Dunedin data the estimate of the total national cost was $54.2 million, whereas the comparable figure using 1989 Dunedin data was $4 I .2 million. Table 3 shows the costs of treating cases according to the leading injury categories (according to total cost). The six injury categories listed were also the highest ranking injuries in terms of average cost. Average costs varied significantly with the treatment of internal injury incurring on average 4.5 times the cost of treating intracranial injury. Skull fracture had the highest cost per day at $580 followed by internal injury at $553 compared with the overall cost per day of $422. Internal injury had the highest disproportionate cost with 7% of the cases nationally and 16% of the costs. This was followed closely by skull fractures, for which the respective percentages were 9% and 18%. Overall, however, fractures of the lower limb contributed the greatest (24%) amount to the total cost. This result is of particular relevance to motorcyclists, who suffered a disproportionate number of lower limb fractures (Table 4).
on sum of sub totals.!
The average cost for all injury categories was lower in 1989. The greatest difference was for skull fracture where the average cost halved. In this case the average length of stay was reduced by 40%. Extrapolating from 1988 Dunedin data the estimate of total national cost was $47.9 million. whereas the comparable figure using 1989 Dunedin data was $39.5 million. Table 4 shows the distribution of the leading types of injury (according to injury type) by class of road user. Pedestrians and pedal cyclists had 92% and 93%, respectively, of their total costs accounted for by the six most expensive injuries to treat (i.e. those listed in Table 3). This contrasted with other road users, for whom the range was from 72%-83%. Table 5 shows the average costs according to severity as measured by the ICDIAIS. On average, costs approximately double with each increment in severity. The table also provides the distribution of AIS scores for the country as a whole and provides estimates of national costs extrapolating from the Dunedin data. Apart from the minor (AIS-I) injuries. the Dunedin Hospital severity case mix is not markedly different from that of the country as a whole. Although severe (AIS-4) and critical (AIS-5) injuries represented only 6% of the case load, they contributed 19% to the national costs. Overall, moderate
Table 2. MVTC inpatient costs by type of road user: Dunedin Hospital 1988-1989 ar Id national estimates (1988)
I
[Occupants
I I I
NATIONAL
DUNEDIN Freq. 291
I x
I I 501
Tot.LOSI
2,925l
I
Av.LOS
1 I
Tot.Cost
I % 1 I
Av.Cost
I
421
$4,471
21
$846,626
281
$5,100
1 1
31
$110,898
41
$7,393
j
101
$1,301,154
Freq.
1 % 1 I
4528
Est.Tot.Cost
I %
5C
234
39
Est. Totxost
based on sum of sub
totals.
589
Costs of injury due to MVTCs in New Zealand
Table 3. MVTC inpatient cost by leading classes of injury: Uunedin Hospital 1988-1989 and national estimates (1988)
I
I
I
/All injury
i
iOO/
f,272/
121
(AK-2) injuries contributed the greatest proportion (37%) to total national costs. With the exception of serious injuries (AIS-3), the average cost in 1989 was lower for all levels of severity. The most marked difference was for severe (AIS-4) injuries (1988: $12,604; 1989: $5,950); however, the size of this difference may be largely attributable to the small number of cases. extrapolating from 1988 Dunedin data the estimate of total national cost was $47.6, whereas the comparable figure using 1989 Dunedin data was $39.5 million. Table 6 provides a summary of four estimates of the national inpatient costs described above. In addition, an kstimate of national MVTCs based on the mean cost for a11injury is included. This shows that extrapolation from mean data for all injury underestimates the national cost of MVTCs by a sizable amount. DISCUSSION The inpatient costs presented here are for public hospital treatment. Because of shortcomings in coding, MVTC cases for private hospitals could not be identified. Private hospital costs have been estimated to be $3.5 million for the year ending 3 1 March 1990 (Langley, Phillips, and Marshall 1993). The results show that motor vehicle traffic crashes consumed 27% of local and 36% of national
Table 4. MVTC admissions.
I
I
I
584
$3,0§7,597
iOOl
I 55,253
I
100 IOOj t43,905,520 Est.Tot.cost based on sum of sub totals.
1
9,026
costs for the treatment of injury victims. The data presented here combine acute and readmission costs. An analysis of data presented by Rice et al (1989) shows 26% of the “initial hospitalisation”, “‘rehospitalisation”, and “inpatient rehabilitation” costs combined for injury victims were attributable to motor vehicles. The national estimate of costs based on the overall average cost for Dunedin for 1988-1989 was $47.4 million (Table I). The estimate based on type of road user (Table 2) was very similar ($47.3 miilion). Those based on nature of injury (Table 3) and severity of injury (Table 5), however, were significantly lower at $43.9 million and $43.0 million, respectively. The latter result is probably more accurate, since it appears from Table 5 that Dunedin has a relatively low percentage of minor injuries, thus the mean treatment cost would be higher than that for the country as a whole. What is not known is the degree to which the costs incurred in Dunedin are similar to those of the rest of the country. The major restructuring of the health sector in New Zealand has already meant that it will be dif~cult to obtain similar data in the future because of commercial sensitivities. The drop in the mean cost of treating a MVTC casualty between the two financial years is probably largely attributable to the reduction in mean stay from 15 days in 1988 to 10 days in 1989. This may
Leading classes of injury for each road user: Dunedin Hospital 1988-1989
J. D.
LANCXEY
Table 5. MVTC inpatient costs by severity of injury:
et al.
DunedinHospital 1988-3989and
national
Est.Tot.Cost
be an artefact of variation in case mix. A higher proporti~~n of occupants, who had the shortest mean stay, were admitted in 1989 compared to 1988 (53% vs. 47%). The results confirm Guria’s (1991) criticism of using the average cost of all hospitalisations as a proxy for the cost of motor vehicle-related trauma. The average cost of a motor vehicle traffic crash was shown to be approximately 1.9 times the average for all admissions, irrespective of cause. Moreover, the cost of motor vehicle traffic injuries was I.7 times that of the average cost for the treatment ofall injury, being on average the most expensive class of injury event to treat (Table 2). As the results show this difference is largely attributable to the longer stays associated with such crashes. Extrapolating from U.S.A. and Australian research, Guria recently reestimated the hospital/ medical costs of a “serious” (hospitalised) injury as $3,614 in June 1990 prices (Guria 1991). As has been shown here the figure is more likely to be 1.5 times that value, or approximately $5,250. The results also demonstrate that average costs vary significantly according to class of road user. Comparable research is virtually nonexistent.
Tabte 6. Summary of estimates of national inpatient costs of MVTCs in New Zealand Basis of Estimates
1
AII injury mean All MVTCs mean
Road User
i
Year 1990 1988 1989 1988189 1988
1 Estimate(m)
I 1
$28.8 $53.3 $41 .o $47.4 $54.2
estimates
(1988)
based on sum of sub totals.
Stutts, Rutledge, and Martell (1991) showed that hospital costs for motorcyclists tended to be similar to those for all other road users combined. A reanaiysis of the data presented here (Table 2) results in a similar conclusion. The mean cost for motorcyclists was $5,100, whereas that for all other road users combined was $5,313. However, as Table 2 shows. mean cost varies significantly within the nonmotorcyclist grouping. The mean cost of treating a motorcyciist is only 54% of that for treating a pedestrian. This highlights the inappropriateness of using average injury costs based on all MVTC victims in costbenefit studies that focus on a specific intervention for a specific class of road user. This is particularly important for pedestrians, whose medical costs are likely to make up a substantial proportion of the total costs of a crash, when there is minimal property damage. Pedestrians had a high mean cost and contributed disproportionately to the total cost. A longer-than-average length of stay was identified here as an explanation. Related to this is the severity of injury. Langley and Marshall (1993) have shown that pedestrians tend to have more severe injuries. The findings show that costs double on average for each increment of severity. This contrasts with a recent U.S.A. report that showed medical costs per person increased eightfold between minor and moderate injury, approximately threefold between moderate and serious, and serious and severe, and fourfold between severe and critical (Urban Institute 1991). It should be noted, however, that different ascertainment methods were used for cases and costs. and the numbers of severe and critical injuries used for the estimate in the present study are small. Perhaps what is of more interest is that minor and moderate injury contributes substantially to overall costs. It should be emphasised that the inpatient analyses were confined to motor vehicle traffic crashes and as such are an underestimate of the costs of all traffic crashes on public roads. Excluded from the road vehicle accidents analyses are “other
591
Costs of injury due to MVTCs in New Zealand
(E826-E829)“, that is, crashes that do not directly involve vehicles with motors. The most significant subgroup among these are bicycle crashes. Collins et al (1993) have estimated that on public roads there are approximately 1,100 bicycle crashes that result in inpatient treatment and that approximately 65% of these do not involve motor vehicles. Most of the deaths and the majority of the more serious injuries, however, are associated with motor vehicles. Given the relatively large numbers involved, and the interest in promoting helmet use, costs for all bicycle crashes on public roads should be estimated. The mean costs for specific subgroups were, in some cases, based on small numbers (e.g. pedal cyclists: IS-Table 2; critical injuries: 12-Table 5). Thus the national estimates based on them need to be treated with caution. Given the relative rarity of such events, an extra year’s data for Dunedin is unlikely to significantly improve the precision of these estimates. Significant gains in precision may be obtained for other moderately sized injury subgroupings (e.g. pedestrians: 66-Table 2; most of the injuries in Table 3; and severe: 27 and minor injuries: 48 in Table 5). An extra year of RUS data has become available since the commencement of the analyses, and it is thus recommended that the analyses undertaken here be repeated using the three year’s data. The data for late effects of motor vehicle crashes is not directly comparable with those for motor vehicle traffic crashes since in the former instance there would be a number of nontraffic cases. As previously noted it was not possible to identify these cases. It is assumed that the error introduced by these cases would be relatively small. This assumption is based on two findings. First, motor vehicle nontraffic crashes represent approximately 8% of all motor vehicle crashes (National Health Statistics Centre 1988). Second, nontraffic crashes tend to be less severe than traffic crashes (e.g. Begg et al. 1991). Langley et al (1991) have estimated that the total medical cost of treating all motor vehicle traffic crash-related injury is approximately $80 million. That estimate includes hospital accident and emergency and Accident Rehabilitation and Compensation Insurance Corporation (ACC) costs. The ACC meets most of the medical costs associated with injury victims treated by genera1 practitioners, specialists, paramedics, dentists, private hospitals, and conveyance for medical treatment. With one or two minor exceptions it does not pay for public hospital costs. At 64% of the total medical cost public hospital inpatient costs are a substantial portion of the economic burden. The estimate of $80 million is an
underestimate since it excludes bicycle crashes not involving motor vehicles and some public hospital costs. Significant among the latter are the costs incurred in providing community services to injury victims (e.g. domiciliary care). The estimate also excludes injury victims’ use of public hospital outpatient services (e.g. visits to the eye clinic). Of particular interest are the costs for “old” injury. Langley et al. (1991) have shown that $8.3 million (36%) of the ACC medical costs relate to injuries that occurred prior to the 1989 financial year. This figure, combined with that from inpatient analysis presented here ($9.1 million), demonstrates that the late effects of motor vehicle traffic crashes are substantial at $17.4 million. This estimate does not include the costs of inpatient treatment in the acute phase and within the first year of admission. These data suggest that the identification and prevention of the types of crashes that lead to late effects need to be a priority if the costs associated with medical treatment are to be substantially reduced. The results reported here provide a previously unavailable and important dimension to be considered when determining priorities for motor vehicle traffic crash prevention. A more accurate appraisal of priorities is largely dependent on: (ij disaggregation of ACC medical costs according to the areas of interest (e.g. road user, nature of injury), and (ii) RUS data from another large hospital, (iii) determining costs for public hospital outpatient treatment of injury, and (iv) determining what treatment costs are borne directly by patients.
Acknowledgements-This research was funded by Land Transport. The advice of Ellen MacKenzie in the planning of this study is appreciated. We wish to acknowledge the stimulation, enthusiasm, and assistance of Jagadish Curia in all phases of this research. The assistance and advice of the following are also acknowledged: David Chalmers, Bill Coppin, Nancy Devlin, Terri Green, Paul Hansen, Russell Pugh, Janet Seats, and Richard Whitney.
REFERENCES Ayling, J. Resource management in New Zealand public hospitals. New Zealand Health Review 8:32-26; 1988. Begg, D. J.; Langley, D. J.; Reeder, A. I. Motorcycle crashes resulting in death and hospitalisation: A report prepared for the Accident Compensation Corporation, Injury Prevention Research Unit. Dunedin, Accident Compensation Corporation, 1991. Bone, I. H. The economic appraisal of roading improvement projects. RRU Technical Recommendation TR9. Wellington: National Roads Board; 1986. Collins. B. A. et al. Bicycle crashes in New Zealand. (in preparation) Dunedin: Injury Prevention Research Unit, University of Otago; 1993.
592
.I.
D.
f..ANGl.tY
Curia, J. Estimates of social costs of accidents and injuries. Wellington: Land Transport; 1991. The Johns Hopkins Health Services Research and Development Center and the Maryland Institute for Emergency Medical Services System. ICDMAP-Determinseverity for hospital discharges: A ing injury programme to map ICD-9-CM diagnoses into AIS and ISS severity scores. Baltimore, MD: Author; 1988. Langley, J. D.; Marshall, S. W. Severity of road traffic crashes resulting in hospitalisation (under consideration 1993). Langley, J. D.; Phillips, Lt.; Marshall S. W. The public hospital inpatient costs of injury due to motor vehicle traffic crashes. Unpublished final report to Land ‘l’ransport Division, Ministry of Transport. Wellington: Ministry of Transport; 1991. MacKenzie, E.; Steinwachs, D.; Shankar B. Classifying trauma severity based on hospital discharge diagnoses. Medical Care 27(4):412-422; 1989. National Health Statistics Centre. Hospital and selected morbidity data 1988. Wellington: Department of Health; 1988. Phillips, D.; L,angley, J. D.; Marshall, S. W. Injury-the
et al.
medical and related costs; New Zealand 1993. New Zealand Medical Journal (in press). Rice, D. P.; MacKenzie, E. J.; and associates. Cost of injury in the United States: A Report to Congress. San Francisco, CA: Institute for Health and Aging, University of California and Injury Prevention Center, The Johns Hopkins University: 1989. Stutts. J. C.; Rutledge, R.; Martell, C. An Analysis of Injury Outcome and Insurance Status of Hospitalized Motorcyclists. 35th Annual Proceedings, Association for the Advancement of Automotive Medicine, Toronto Canada, October 7-9, 1991. Transit New Zealand. Project evaluation manual: Volume 1: Simplified procedures: Volume II: Full procedures. Wellington: Author: 1991. Urban Institute. The costs of highway crashes. Final Report to United States Department of Transportation. Washington, DC: U.S. Department of Transportation: 1991. WHO (World Health Organisation). International classification of diseases-ninth revision 1977. Geneva: WHO; 1977.