OUTLIERS
Missouri identified an outlier charge of $914,331 for a belted, truck driver. Beltedness for this driver was confirmed in all three files (crash, EMS, HSCIT) and the high charge was
determined to be unusual for belted patients. It had the effect of distorting the average charge for belted patients by 7 percent. Two other victims were involved in the same crash. Both were unbelted and suffered fatal injuries but incurred no hospital charges. These three records reveal the difficulty in using hospital charges as an outcome variable to assess safety belt effectiveness. When the outlier was dropped from the analyses, the 95 percent confidence interval decreased around the average patient charge for belted patients when drivers of cars and trucks were included together. When trucks were analyzed separately, exclusion of the outlier decreased the average charges from $20,247 to $10,934! However, since the data, though an outlier,
represented a valid linked record for a belted driver, including it in the analyses was perceived as being less objectionable.
TYPES OF RESTRAINTS AND CRASHES
Hawaiian belt usage (representing one of the highest rates in the United States) was studied in relationship to injuries among crash-involved front seat occupants of motor vehicles.
systems. Lap-shoulder systems afforded the greatest level of protection, followed by lap only, shoulder only. No restraints increased the likelihood of a fatal or incapacitating injury as did rollover or head-on crashes.
ALCOHOL AND DRUG USE
Hawaii developed a structural model to explain the relationships between certain driver characteristics and behaviors, crash types, and injury severity once a crash has occurred. The model clarified the role of driver characteristics and behaviors in the causal sequence leading to more severe injuries. The effects of various factors were studied to determine how much each factor increased or decreased the odds of more severe crash types and injuries. The results indicated that driver behaviors of alcohol or drug use and lack of safety belt use greatly increase the odds of more severe crashes and injuries. Driver errors were found to have a small effect, while personal characteristics of age and sex were generally insignificant. The study failed to find any strong association between age, sex, and driver behavior suggesting that the young males involved in crashes are not much more likely to be engaging in these negligent behaviors than anyone else involved in crashes. As a result, they are not much more likely to be involved in the more severe crashes and injuries as one might expect from a disproportionately high rate of negligent behavior.
COST OF CARE
The New York CODES mandated model failed to establish a direct relationship (R-square = 0.02) between crash parameters and cost of care. Separate consideration of the crash
parameters was studied to determine the relationship between the type and severity of injury with variations in the cost of care. The injury severity for drivers' admitted as inpatients, measured by the maximum abbreviated injury scale (MAIS) score, was shown to be a fair (R-square = 0.25) predictor of case level cost as measured by hospital charges. However, there was wide variance in cost by body region ($3,623 for external injuries to $15,874 for abdominal injuries) within the same severity level. Since AIS severity level assignments reflect only a single injury and a crash victim may be treated for several injuries, the Injury Severity Score (ISS) was substituted to improve the cost model. ISS measures cumulative injury severity across multiple body regions. As expected, inclusion of the multiple injuries per victim caused approximately 14 percent of all cases to shift to higher severity levels with ISS versus the MAIS. In addition, many average hospital charge values changed significantly. The progressive increase of charges by severity level was much more consistent across all body regions with ISS severities. More consistent average charge values at the injury category level did not change the estimated total charges significantly, and did not improve the situation at the case level. High variances within injury categories mean that low confidence cases still contribute approximately one-third of the estimated total charges. Case level models using ISS severity showed only slight improvements over models using MAIS severity. These study results suggest that case level cost prediction using diagnosis codes and computer algorithms for translation to injury severity, body region, and average cost may be an
effective approach. However, model predictions should be coupled with medical record data that identify transfers, extraordinary treatment cases, or other outliers.
Wisconsin used linked crash and Medicaid claims data to obtain information on health outcomes and service utilization for each Medicaid-injured crash occupant for one year following the crash. The Wisconsin Medicaid program paid $6.5 million to cover crash-related costs incurred during the year following the crash, for eligible beneficiaries who were in crashes any time during 1991. In fiscal year 1991, $6.5 million was paid by Wisconsin Medicaid to cover crash-related costs incurred during the year following the crash for eligible beneficiaries. Over $4 million was paid by Medicaid for inpatient hospital services, slightly less than $1 million was paid for physician services, and one-half million dollars was paid for outpatient services. Over $3.5 million was paid on behalf of beneficiaries involved in passenger car crashes, with payments for crash-involved pedestrians and motorcyclists totaling approximately $1 million for each group. Costs were incurred immediately for (1) individuals who were Medicaid-eligible prior to the crash, and (2) individuals who became eligible immediately following the crash as a result of severe injuries requiring expensive care. Many others became eligible after their casualty insurance reached its maximum limits. Although this study tracked Medicaid costs for only the first year following each occupant’s crash date, some severely-injured occupants become eligible for Medicaid two or three years after the crash when the payments from their liability settlements finally run out. An algorithm was developed to identify the crash-related health care specifically for physician, hospital, long term care, and other services provided to Medicaid beneficiaries. The Medicaid population of drivers, more likely to be female and younger, suffered higher injury severity possibly because of a higher reported rate of alcohol use and a lower rate of safety belt use.
RELATIONSHIP OF CRASH INJURY AND AGE
New York used the linked data to investigate whether crash related injuries differed between older and younger drivers hospitalized as the result of a motor vehicle crash. All drivers were assigned to groups covering a span of 10 years (16-24, 25-34, etc.). Only 68 percent of the drivers hospitalized age 16-54 were belted compared to 81 percent of the drivers age 55 and older. Almost 90 percent of the crashes were defined as low speed. Older drivers hospitalized were more likely to sustain a wide spectrum of crash related injuries (internal injuries, fractures, and contusions or superficial injuries) than younger drivers. Length of stay was longer (10 days versus 6 for younger drivers) and hospital charges greater (26 percent higher for the age group 65-74). Although belted and driving more slowly, elderly drivers are likely to suffer more serious injuries than younger drivers. Thus programs targeted at the elderly should include crash