Cost-Outcome Analysis of Booster Seats for Auto Occupants Aged 4 to 7 Years

METHODS

We conducted 2 cost-outcome analyses in this study: a benefit-cost and a cost-utility analysis. In the benefit-cost analysis, the costs and benefits of booster seat use were both measured in monetary units. In the cost-utility analysis, the outcome unit was a QALY. A QALY is a health outcome measure that values a year in perfect health at 1.0, values death at 0.0, and allows fates worse than death. QALYs merge some aspects of quality of life with quantity (or length) of life into a single measure. The number of life years (quantity of life) gained as a result of booster seat use is combined with published estimates of the quality of those life years^5–7 to calculate the number of QALYs gained. Quality of life is a multidimensional concept measuring the physical, emotional, and social aspects that are relevant and important to a person's well-being.

The 2 alternatives evaluated in this study were use of a lap-shoulder belt with or without a booster seat. Given that a child should use a booster seat for 4 years, benefits accrue each year. Benefits in future years were discounted to present value using a 3% discount rate (with sensitivity analysis at 4%). This rate is recommended for reference case comparisons worldwide by the US Panel on Cost-Effectiveness in Health and Medicine.⁸ Statistical tables are available that provide the present value factors of a given monetary amount received each year. A present value factor is the “multiplier” that, when multiplied by the amount received each year, gives the present value of the cash flows. The present value factor for 4 years and a discount rate of 3% is 3.8286.⁹

The estimated cost of a booster seat came from Hendrie et al.¹⁰ They found that a low-end booster seat costs $20. Their US booster seat costs were ascertained from the www.shopping.com Web site and verified by visiting 2 national retailers. A backless seat was readily available for $18 to $22 (including sales tax or shipping), a model with back for $30, and premium models for $40 to $100. The available booster seat effectiveness data are predominantly for a sample of boosters with backs, so the analysis uses a $30 cost. In sensitivity analysis, the cost of time spent buying, installing, and maintaining a booster seat was estimated at 4 hours per year. This time was priced at the $10.88 hourly wage (in 2000 dollars) for an installation, maintenance, and repair helper, from US government statistics.¹¹ We ignored the early retirement of some booster seats, most often ones involved in crashes. All of the costs and benefits were measured in year 2000 dollars.

Following previous studies,^12–14 for laws and regulations we use Downing's¹⁴ estimates that the costs of approving mandates average 2.9% to 7.1% of the direct costs imposed on the public, with public implementation and administration costing another 4.2% to 4.6%. That means that the costs of a law mandating use of booster seats, on average, would add 9.4% to the direct cost of a booster seat.

To estimate the costs of injuries for belted and booster-seated injury victims, 3 components were used: the distribution of serious injuries for these 2 types of victims, the probability of a serious injury for a child aged 4 to 7 years who travels belted or in a booster seat, and injury costs. Any reduction in minor injuries associated with the use of booster seats was assumed to be offset by a rise in minor injuries that would have been severe without the booster seat.

The first component, the distribution of serious injuries, came from Durbin et al⁴ That source provides the only available effectiveness data on the impacts of booster seats on the number of serious injuries (fatalities, facial lacerations, and injuries that pose a moderate to critical risk of death). The probability of injury could not be computed from the same data, because no one knows the total number of children traveling in motor vehicles insured by State Farm Insurance (the claims file used in Durbin et al⁴). Therefore, the probability of a serious injury for a child aged 4 to 7 years who travels belted was estimated using the annual number of injuries from the National Highway Traffic Safety Administration's (NHTSA) 1993–1999 Crashworthiness Data System (CDS)¹⁵ and the average number of children from US Census data. We chose these years because booster seat use seems to have expanded rapidly starting in 2000. Our data are for the most recent period when booster seat use was virtually nonexistent.

The mix of injuries reported in Durbin et al⁴ was costed with published injury costs by body part and severity.^15,16 These costs include the monetary value of the loss of health plus all of the other costs resulting from a serious injury.

We used the published cost estimates,^15,16 because they mirror the official values used in regulatory analysis by NHTSA but computed at a 3% rather than a 4% discount rate. They draw on data from 1992–1994 Civilian Health and Medical Program of the Uniformed Services data for physician and emergency department fees, 1994–1995 data on hospital costs in Maryland and New York (the only 2 states where costs, not charges or payments, were known), and 1987 National Medical Expenditure Survey and 1979–1987 National Council on Compensation Insurance data on the percentage of costs that occur >6 months postinjury.

The published estimates based short-term parental work loss on the assumption that the lowest paid parent would stay home to care for an injured child on any day that an adult with a comparable injury would not work. Information on the probability that an employed person would lose work for a specific injury came from the CDS 1993–1999. Information on the days of work lost per person who lost work came from the US Bureau of Labor Statistics 1993 Survey of Occupational Injury and Illness.¹⁷ Mean probabilities of work loss were estimated from just those CDS records that had the relevant information, which frequently was missing. Sample size considerations drove the decision to pool several years of CDS data. Long-term productivity loss by diagnosis was based on 1979–1987 National Council on Compensation Insurance Detailed Claims Information data on the probability that injuries would cause permanent partial/total disability and 1997 Detailed Claims Information data on the percentage loss of earning power for partially disabled injury victims.

The published estimates included a variety of other direct costs. Among them were emergency services, insurance claims administration, legal and court costs, and workplace disruption costs. These estimates used insurance data and data from previous NHTSA studies.

Following earlier studies,^5,18 the published estimates based quality of life loss (QALY) on physicians' estimates of the functional capacity lost over time by injury diagnosis and a systematic review of the survey literature on the loss in value of life that results from different functional losses. These losses were costed based on a meta-analysis examining what people pay for small changes in fatality risk and surveys on what they state they are willing to pay. Recognizing that monetizing the value of quality of life is controversial, we provide analyses with both unmonetized and monetized QALYs.

The savings from booster seat use have 2 components: injury prevention and injury severity reduction. We valued injury prevention by multiplying the probability of a serious injury for a child aged 4 to 7 years who travels belted (.000704 annually) by the sum of booster seat effectiveness (59%) times the average cost per seriously injured belted child estimated using the published unit costs ($851745). We valued severity reduction at 41% (100%–59%) of the difference in the average cost of a serious injury to a child in a belt versus a child in a booster seat ($851745 vs $402139). In compact form, the calculation was as follows:

Annual costs averted = 0.000704 × [0.59 × $851745 + 0.41 × ($851745 − $402139)]

RESULTS

Each booster seat in use can avert $484 {0.000704 × [0.59 × $851745 + 0.41 × ($851745 − $402139)]} in injury costs annually. The present value of savings over 4 years is 3.8286 times this amount or $1854 (Table 1). The savings include $245 in medical spending, $161 in other resource costs, $433 in work losses, and QALYs valued at $1015. The estimated net savings per booster seat is $1824 ($1854 − $30). The benefit-cost ratio for a booster seat is 61.8 ($1854/$30). The savings in medical and other resource costs from a booster seat exceed its purchase price by $376. Thus, it offers net cost savings of $348543 per QALY. If parental time expenditures are included, the estimated net savings per booster seat is $1657 ($1854 − $197), the benefit-cost ratio for a booster seat is 9.4 ($1854/$197), and the savings in medical and other resource costs from a booster seat exceed its cost by $209. In addition, if the costs of mandating use are included, the estimated net savings per booster seat is $1638 ($1854 − $216), the benefit-cost ratio for a booster seat is 8.6 ($1854/$216), and the savings in medical and other resource costs from a booster seat still exceed its cost by $190.

In sensitivity analysis, a 4% discount rate was used, and booster seat effectiveness was set at a minimum of 35% and a maximum of 70%. The resulting variance in injury cost savings was −25% to +10%.

DISCUSSION

Benefit-cost and cost-utility analyses support comparisons between injury interventions when the unit of outcome varies between alternatives. Unlike cost-effectiveness analyses that measure outcomes using a generic unit, such as life years saved, they account for differences in quality of life that result from intervention. This makes them appropriate for comparing programs that are primarily life saving with those that focus on reducing nonfatal injury. Booster seats and laws mandating their use both offer net resource cost savings, which places them among the top tier of preventive interventions.¹⁹

An unquantified benefit in this study is that booster seats make safety belts more comfortable and make it difficult to slip the shoulder harness behind the shoulder. This may affect children's willingness to wear restraints and improve their positioning and behavior in the car. Thus, injury costs averted can be higher than estimated.

Other limitations of this analysis include: (1) only 1 study of seat effectiveness in actual use for all crash directions is available; (2) the injury unit costs used are not child specific; (3) some of the data used in estimating the other resource costs and QALY losses are old; (4) only modest validation has been done on the QALY loss estimates despite ∼15 years of use in regulatory analysis; (5) the CDS data profile serious injuries (Abbreviated Injury Severity 2 and above) in tow-away crashes, whereas the State Farm data profile serious injuries (Abbreviated Injury Severity 2 and above) in all crashes with insurance claims, meaning the injury cost savings might describe more serious injuries by body region than the effectiveness study and be overestimated; (6) no allowance is made for early seat retirement; and (7) some parents buy a child a booster seat for each family car, raising total seat cost per child. Nevertheless, in sensitivity analysis, when both booster seat effectiveness and injury unit cost estimates were lowered, although the savings were reduced by 25%, a booster seat law still offered net resource cost savings of $274 per seat. The net resource cost savings would be $279 per booster seat if seats were retired 1 year early and $173 per seat if 2 seats were purchased per child. Thus, the finding that booster seats are excellent investments is robust.

Despite these limitations, the benefits from using a booster seat clearly exceed the costs. Booster seats are a worthwhile investment from a societal perspective. The cost savings and safety provided by booster seats support the passage of booster seat use laws nationwide. Experience with child seat laws suggests that a booster seat law and related publicity should raise booster seat use by 40%.^20,21 Seat laws will save both lives and money.