Article Text

Download PDFPDF

Lost working days, productivity, and restraint use among occupants of motor vehicles that crashed in the United States
  1. B E Ebel1,3,
  2. C Mack3,
  3. P Diehr1,2,3,
  4. F P Rivara1,3
  1. 1Department of Pediatrics, University of Washington, Washington State, USA
  2. 2Department of Health Services, University of Washington
  3. 3Harborview Injury Prevention & Research Center, Seattle, WA, USA
  1. Correspondence to:
 Dr Beth E Ebel
 Division of General Pediatrics, University of Washington, Harborview Injury Prevention & Research Center, 325 Ninth Avenue, Box 359960, Seattle, WA 98104;


Background: In 2001, 6.3 million passengers were involved in motor vehicle crashes. This study aimed to determine the number of work days lost as a result of motor vehicle crashes and factors that influenced people’s return to work.

Methods: This was a retrospective, population based cohort study of occupants in motor vehicles involved in crashes from the 1993–2001 Crashworthiness Data System produced by the National Highway Traffic Safety Administration. The sample population of people aged 18–65 years included two groups: occupants who survived and were working before the crash and occupants who were injured fatally and were estimated to have been working before the crash. Multivariate linear regression was used to analyze the impact of restraint use and injury type on return to work.

Results: Overall, 30.1% of occupants of vehicles that crashed missed one or more days of work. A crash resulted in a mean 28.0 (95% confidence interval 15.8 to 40.1) days lost from work, including losses associated with fatalities. The 2.1 million working occupants of vehicles that crashed in 2001 lost a total of 60 million days of work, resulting in annual productivity losses of over $7.5 billion (2964 to 12 075). Unrestrained vehicle occupants accounted for $5.6 billion in lost productivity.

Conclusions: Motor vehicle crashes result in large and potentially preventable productive losses that are mostly attributable to fatal injuries.

  • AIS, Abbreviated Injury Scale
  • ISS, injury severity score
  • Motor vehicle injury
  • accidents/traffic
  • time lost from work
  • economic analysis
  • productivity
  • seat belts

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Over 6.3 million adults were involved in motor vehicle crashes reported by police in 2001, resulting in more than 3 million injuries and health care visits.1,2 Although most of these crashes did not result in permanent disability, one of the most important measures of the functional outcome of injuries is the amount of time lost from work. We wanted to determine how many work days were lost because of motor vehicle crashes and whether people who were properly restrained in vehicles returned to work earlier than those who were unrestrained occupants in vehicles.

Previous research examined the ability of people to return to work and disability after severe injuries that required hospitalization,3–5 lower extremity fractures,6–8 spinal cord injury, and traumatic brain injury.9,10 These estimates have been based on specific diagnoses and generally represent more severely injured patients treated in trauma centers. Miller and colleagues estimated that non-fatal injuries in 1988 resulted in 0.7 years of lost functional capacity per injury.11

This study examines data from the crashworthiness data system from 1993 to 2001 to determine American population estimates for days lost from work for occupants of motor vehicle crashes.1 This study updates analyses that are now nearly 20 years old and considers the impact of injury type and restraint use on return to work.


Sample population

The people considered in this analysis were a nationally representative sample of 65 060 people aged 18–65 years who were occupants in motor vehicle crashes from 1993 through 2001. These people represent 31 151 190 crash occupants nationally. From this sample, we chose the 32 748 people who were estimated to have been working before the crash (representing 18 228 327 million people nationally). The sample population included two groups: surviving occupants who were working before the crash and fatally injured occupants estimated to have been working before the crash. The people or their families were contacted about 1–4 weeks after the crash. Only respondents who had been working at least half time immediately before the crash were asked to estimate how many work days they lost as a consequence of the crash. Because we were interested in examining the effect of seat belts on time lost from work, we did not restrict the population to injured or surviving people, as use of seat belts presumably might not only reduce the severity of injury but also the risk of injury itself.

Crashworthiness data system

The crashworthiness data system collects detailed information on an annual sample of about 5000 crashes reported by police that involve involving passenger vehicles each year in the United States.1 The crashworthiness data system uses a three stage sampling procedure to sample crashes that result in tow away damage to make the data representative of all crashes reported by police. Data are weighted to generate national estimates.

A field investigator determined the injuries by examining the medical, hospital, police, emergency medical service, and autopsy records. Injuries were grouped into body regions using the Abbreviated Injury Scale (AIS).12 The AIS-90 codes were scored by crash investigators who assigned AIS injury scores to each of the eight AIS-90 body regions. These body regions were collapsed into six regions for calculation of injury severity scores (ISSs). The ISSs were grouped into the following categories: 0, 1–8, 9–12, 13–15, 16–24, and ⩾25.

Crash investigators determined data on use of seat belts after they collected information from interviews, police reports, and examination of the vehicle. Restraint use was determined by analyzing the type of manual or automatic seat belt available for use in the vehicle and indications of whether the seat belt was properly used. The large number of possible configurations was collapsed into three possibilities: seat belt not used (whether available or not), seat belt used improperly (if the manual or automatic seat belt was improperly used), and seat belt used properly (if seat belts installed in the vehicle were used properly). Missing data on restraint use were imputed according to methods described below. In our final analyses, the “improper restraint use” and “no restraint use” categories were combined as “unrestrained” because of similarities in injury rates and lost work time.

Imputation of seat belt use

Data on use of seat belts was missing for 22.5% of occupants of motor vehicles that crashed. On average, the characteristics of occupants for whom seat belt use was missing more closely resembled those of unrestrained occupants. If these observations were ignored, selection bias could be introduced into the data analysis, which could result in incorrect estimates of confidence intervals.13 We used multiple imputation methods to predict missing data on use of seat belts on the basis of the following predictor variables: occupant age and sex, airbag deployment, seat location (front v rear), ejection, injury severity score, death, vehicle’s model year, occupant role (driver v passenger), and whether or not working days were lost.14 We used linear discriminant analysis for the people with complete data to estimate the probability of them being in the three groups: seat belts used properly, seat belts used improperly, and no seat belts. We created three imputed seat belt values for each person with missing data. We generated three random numbers for each person and used these to assign a seat belt status to be used in the multiple imputation, according to the discriminant analysis probability.

We compared our data analysis with imputed use of restraints to an analysis in which surviving people with missing data were omitted. The number of days lost was 10% greater when imputed data was used compared with when unimputed data was used (mean 2.0 days v 1.8 days), largely because of estimates of lost work time for people with “⩾61 days” of lost work. Use of seat belts was more likely to be missing when occupants were young, male, more severely injured, driving older vehicles, and passenger rather than driver. Occupants without recorded use of seat belts were more likely to be working and were less likely to have reported missed days of work, but if missed work was reported, they were likely to miss more days than occupants for whom restraint use was recorded. We therefore decided to use imputed restraint use in our analyses to more accurately explore associations between restraint use and lost time from work.

Estimation of time lost from work

The crashworthiness data system survey asked respondents who were working at the time of the crash to state the exact number of days of work lost. For people who missed more than two months of work, all responses were coded as ⩾61 days. Although the proportion of people who missed this much work was relatively low, they contributed a disproportionate share of the productivity losses caused by motor vehicle crashes. The probability of returning to work decreased in the first six weeks of follow up but was roughly constant at 0.09 for each week of weeks 7–12. We assumed that the number of additional days lost for those who had not returned to work at 60 days had a negative binomial distribution, with a probability of 0.09. For each person with ⩾61 days of work loss, we drew three random numbers from this distribution and added that to 60 to give estimates of and standard errors for the total work days lost for each person to provide three estimated values for the multiple imputation. This added, on average, 1/0.096 = 10.5 weeks (52 days) for these people.

Time lost from work or productive activity for occupants who were not working at the time of the crash was not measured in the crashworthiness data system. Although many of these people lost productive time from their normal activities as a result of the crash (opportunity cost), they are not included in this analysis, which is restricted to lost work time.

Lost time from work for occupants who died

The crashworthiness data system did not include time lost from work for people who died as a result of the crash, and no information was recorded to indicate whether the deceased person had been working before the crash. We estimated workforce participation rates for fatally injured people aged 18–65 years by age and sex on the basis of workforce participation of occupants of similar crashes who survived and had reported whether they were working before the crash. We assumed that age and sex matched workforce participation rates before the crash did not differ between surviving and fatally injured occupants, and we allowed labor participation rates to change with age over the projected working lifetime of each deceased person, as predicted by participation rates of surviving people in our sample.

Each fatally injured person thus contributed a number of working days lost on the basis of the probability that a living person of the same sex would be working at each age. Deceased people who were not working before the crash were assumed to have no working days lost. An annual discount rate of 3% was used to determine the present value of future lost days from work and lost wages, as suggested by the Panel on Cost-effectiveness in Health and Medicine.15 About 10% of deceased people were coded in the database with an ISS of zero; these people were assumed to be injured rather than uninjured because they died as a result of the crash and they were included in all analyses except that of lost days of work by ISS group.

Estimates of national productivity losses due to motor vehicle crashes

Crashworthiness data system data from years 1993–2001 were used to determine the mean time lost from work after a motor vehicle crash. We used these data to estimate annual time lost from work by restraint use category for the 3462 previously working occupants of vehicles that crashed in 2001, representing 2 110 572 crash occupants nationally. The median weekly earnings for full time workers aged >16 years was used to estimate the daily productivity loss, which was $123 per person day in the second quarter of 2003.16 These estimates of lost work time were used to estimate lost productivity with crash data from 2001, calculated in year 2003 dollars

Data analysis

We used survey adjusted programs in Stata software (version 7; Stata Corporation, College Station, TX) to tabulate sample descriptives from population weighted crashworthiness data system variables. We calculated mean and total work days lost by age, sex, injury severity (ISS group), and injured region (AIS regions) with appropriate confidence intervals. We linearly regressed days lost on seat belt use, controlling for occupant age and sex, airbag deployment, vehicle model year, whether the occupant was the driver or a passenger, and seat position (front versus rear). Although the fact that only 30% of people who had been working actually missed days of work theoretically might prompt a concern about whether linear regression would result in biased estimates, for large samples such as ours the central limit theorem guarantees that linear regression models produce valid estimations of means, parameter estimates, and test statistics, even for highly skewed distributions.17 We imputed missing data on restraint use and lost time from work >60 days as described above. We performed regression analyses three times—once for each set of imputed values—and we calculated the mean and standard deviation of the three regression coefficients by the usual multiple imputation methods.13


Characteristics of study population

Table 1 describes the weighted study sample of motor vehicle crash occupants aged 18–65 years. The mean age of occupants was 33 years, and 55% of the occupants were men. Most occupants were restrained properly with a seat belt (58.3%), although 2.5% were restrained improperly and 16.8% unrestrained at the time of the crash. Most crash occupants were uninjured (57%), with the remainder most likely to suffer relatively minor injuries. Fatally injured people comprised 0.6% of all occupants of vehicles that crashed. Injuries were most common to the head, face, or neck, followed by the lower and upper extremities and spine. Most surviving occupants (58.2%) were reported as working before the crash. Among people who had been working, 30.1% missed days of work after the crash.

Table 1

 Characteristics of occupants of motor vehicles that crashed in 1993–2001 (95% confidence interval)

Mean working days lost by age and sex

Table 2 shows estimated work time lost because of the crash for surviving and fatally injured occupants. Survivors lost an average of two (95% confidence interval 1.9 to 2.2) work days, which rose to a mean of 4.0 (3.7 to 4.3) days among injured occupants and 2570 (2371 to 2770) days among fatally injured occupants. When we considered surviving and fatally injured occupants together, the mean days lost from work was 28 days per crash victim. Women survivors who were working before the crash missed more work days than men, but total working days lost were significantly higher for men than women (35.8 v 18.3 days) because of the higher number of deaths for men.

Table 2

 Working days lost for occupants of motor vehicles that crashed in 1993–2001 who were working before the crash, by age, sex, and injury status (95% confidence interval)

Median time lost from work

The distribution of work loss was highly skewed. Overall, 50% of vehicle occupants who had been working before the crash missed no work days and 90% of all crash occupants missed ⩽5 days. Among injured occupants, 50% missed ⩽1 day of work. For people who lost some amount of work time, the median days lost was three and the 90th percentile was 17 days of lost work.

National estimate for lost work days and lost productivity

An estimated 2.1 million crash occupants aged 18–65 years who had been previously working existed in 2001—the latest year for which crashworthiness data are available. These people cumulatively lost an estimated 60.8 million days of work, resulting in direct annual productivity losses of over $7.5 billion (table 3).

Table 3

 Population weighted data representative of crashes in the United States showing lost work days and lost productivity estimates for people who survived a crash and people fatally injured in a crash (95% confidence interval)

Lost work days, lost productivity, and restraint use

Use of a seat belt, which has been shown to reduce the frequency and severity of injury, led to a reduction in lost days of work. Table 4 indicates that unrestrained occupants lost an average of 96 work days relative to an average of 10 work days lost by occupants who were properly restrained in a seatbelt. In a multiple linear regression analysis (not shown), controlling for age, sex, airbag deployment, model year, occupant role (driver versus passenger), and seat position (front versus back), unrestrained occupants had a mean excess of 81.5 (53.5 to 109.4) lost work days more than restrained occupants, with a total productive value estimated at $5.6 billion in 2001.

Table 4

 Population weighted data representative of crashes in the United States showing lost time from work and lost productivity and restraint use (95% confidence interval)

Lost work days and lost productivity by injury severity

People with more severe injuries were more likely to miss more work (table 5). People who were reported as having no injuries lost a mean of 0.3 (0.2 to 0.4) work days, while the most severely injured group (which includes most of the fatal injuries) lost a mean of 2202 (2079 to 2326) work days. Injuries to the abdomen, thorax, or spine resulted in the greatest time lost from work per person. Because head, face, and neck injuries were more prevalent, however, they resulted in the greatest total losses of productivity. People with lower extremity fractures missed more work than those with upper extremity fractures.

Table 5

 Population weighted data representative of crashes in the United States showing work days lost and productivity loss for injured occupants of vehicles that crashed who were working before the crash, by category of injury severity score and body region injured (95% confidence interval)


Motor vehicle crashes in the United States result in significant lost work time and lost productivity, amounting to over $7.5 billion in 2001. Most occupants, however, have minor injuries and miss only a few days of work. Fatally injured occupants, though a relatively small proportion of crash occupants, are often young and lose a lifetime of productive potential. Although this analysis was unable to estimate the opportunity cost of motor vehicle crashes for occupants who were not working before the crash, or were younger or older than our study population, the economic losses are considerable. Seat belts, which previously have been shown to reduce the risk of injury and death in a crash,18–25 dramatically reduce the number of work days lost, and result in over $5 billion dollars a year in potentially avoidable productive work loss.

Lost time from work correlated significantly with severity of injury, although even low levels of injury severity resulted in lost work time. Even people who were not reported as injured lost work time, perhaps as a result of the time needed to address insurance issues and arrange transportation or from emotional trauma.

Study limitations

Our analysis has a number of limitations. Some crashes are not reported to police and therefore are not included in the crashworthiness data system, although these crashes generally are less severe and less likely to result in injury. The crashworthiness data system did not gather data on time lost from work >60 days, which we have estimated for our models. These estimates raise the average days lost from 1.8 to 2.0 among survivors. Although our estimates of additional days were based on probabilities of return in the data, they could be inaccurate. Although this is a limitation, the data are the best nationally representative data available, and the 60 day period represents longer follow up than data from the National Health Interview Survey (where “six or more days” is the top category) and that from the Bureau of Labor Statistics for private employers (where “30 or more days” is the top category). Nonetheless, people who are out of work for extended periods contribute largely to the economic costs of injury.

Data from the crashworthiness data system do not include measures for physical or mental function or measures of quality of life. For more severe injuries, such as traumatic brain injury, even people who return to work often are unable to function at their previous level or suffer physical or mental impairments. The economic costs from lost work time alone thus are underestimates. In addition, data on loss of productivity were not collected on people who said they were not working at least half time before the crash, yet these people have an economic impact as well, as they may be unable to perform their usual jobs and may need care from other household members. This analysis also did not include the productivity of older people, uncompensated work done in the home, or the potential lost productivity of injured children. We were unable to control for other variables known to be important, such as insurance status, or litigation. Finally, this analysis was restricted to people aged 18–65 years. As up to a quarter of men and 16% of women aged 65–69 years currently are employed, our analysis underestimates the lost work time for these and older people. Despite these limitations, this population based crash database offers a unique opportunity to measure the lost work time and lost productivity resulting from crashes.

The largest contributor to lost work days was the group of people who were fatally injured in a crash. Nonetheless, non-fatal crashes are commonplace, and our study found that previously working surviving people lost an average of two working days in a crash. Even with our conservative estimates of time lost from work after the initial 60 days, severely injured people lost over $3.5 billion in productivity, while people listed as uninjured lost only $49 million as a consequence.

Previous researchers have studied factors associated with return to work in hospitalized crash victims. These studies were based on patients admitted to trauma centers and found persistent disability with high economic costs. Mackenzie et al found that among previously working patients hospitalized at a trauma center for lower extremity fractures, only 48% had returned to work by six months.6 For another group of hospitalized patients, only 56% were able to return to full time work by one year. As was expected, fewer people with severe head or spinal cord injuries returned to work.3 A study in London of survivors of major trauma found that 69% of people had returned to work by one year.26 A study of patients with traumatic brain injury at a level I trauma center found that 28% were unable to return to their level of productive activity at one year.9

Miller et al estimated that non-fatal injury in 1988 resulted in 0.7 years of lost functional capacity per injury.11 The authors based their analysis on data on the first 60 days after a crash from the 1982–88 National Automobile Sampling System and used data from the National Crash Severity Study (a 1977–79 data collection instrument that is no longer maintained) to estimate work loss up to 180 days after injury combined with workman’s compensation charges estimated by body region for the most severe injury. As return to work was not quantified for people who reported losing >60 days of work, we predicted the likelihood that a person would return to work in one week given that they had not been working in the previous week to estimate mean return to work time, as described in the methods section above.

Key points

  • Motor vehicle crashes resulted in a mean 28.0 days lost from work per occupant of a vehicle that crashed.

  • Nearly one third of all occupants of vehicles that crashed lost some time from work.

  • In 2001, the 2.1 million previously working occupants of vehicles that crashed lost 60 million days of work, resulting in annual productivity losses of over $7.5 billion.

  • Unrestrained occupants of vehicles that crashed accounted for $5.6 billion in lost productivity.


Our study examined a population based database that was not restricted to hospitalized trauma patients and allowed national estimates of productivity loss to be determined. We found considerable economic impact, even from less severely injured people, most of whom were not hospitalized for their injuries.

The magnitude of the economic costs, which are borne by employers (in terms of absenteeism and productivity) as well as workers and their families (lost leave time and lost wages), raises questions about how these losses might be prevented. Previous work suggested that factors such as health before injury, socioeconomic status, social support, ongoing litigation, and stress also contribute to persistent disability.7,8,27 The universal use of seat belts would potentially save $5.6 billion dollars of lost productivity. Effective strategies to increase seat belt use, such as primary enforcement seat belt laws, thus would yield considerable economic gains.


Supported in part by Centers for Disease Control and Prevention and a grant from National Highway Traffic Safety Administration for the Crash Injury Reconstruction and Engineering Network.