This article was downloaded by: [New York University] On: 23 July 2015, At: 13:55 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place, London, SW1P 1WG
Traffic Injury Prevention Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gcpi20
The Economic Impact of Helmet Use on Motorcycle Accidents: A Systematic Review and Meta-analysis of the Literature from the Past 20 Years a
b
c
d
b
Chang-Yeon Kim , Daniel H. Wiznia , Leon Averbukh , Feng Dai & Michael P. Leslie a
Yale School of Medicine, New Haven, Connecticut
b
Department of Orthopaedics and Rehabilitation, Yale–New Haven Hospital, New Haven, Connecticut c
New York Institute of Technology College of Osteopathic Medicine, Old Westbury, New York
d
Yale Center for Analytical Sciences, Yale School of Medicine, New Haven, Connecticut Published online: 28 Apr 2015.
Click for updates To cite this article: Chang-Yeon Kim, Daniel H. Wiznia, Leon Averbukh, Feng Dai & Michael P. Leslie (2015) The Economic Impact of Helmet Use on Motorcycle Accidents: A Systematic Review and Meta-analysis of the Literature from the Past 20 Years, Traffic Injury Prevention, 16:7, 732-738, DOI: 10.1080/15389588.2015.1005207 To link to this article: http://dx.doi.org/10.1080/15389588.2015.1005207
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Traffic Injury Prevention (2015) 16, 732–738 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2015.1005207
The Economic Impact of Helmet Use on Motorcycle Accidents: A Systematic Review and Meta-analysis of the Literature from the Past 20 Years CHANG-YEON KIM1, DANIEL H. WIZNIA2, LEON AVERBUKH3, FENG DAI4, and MICHAEL P. LESLIE2 1
Yale School of Medicine, New Haven, Connecticut Department of Orthopaedics and Rehabilitation, Yale–New Haven Hospital, New Haven, Connecticut 3 New York Institute of Technology College of Osteopathic Medicine, Old Westbury, New York 4 Yale Center for Analytical Sciences, Yale School of Medicine, New Haven, Connecticut
Downloaded by [New York University] at 13:55 23 July 2015
2
Received 20 October 2014, Accepted 5 January 2015
Objective: The incidence and cost of motorcycle accidents are projected to increase. Motorcycle helmets are accepted as an effective strategy for reducing the morbidity and therefore the cost of motorcycle accidents. Despite this, states have continued to repeal helmet laws in the past 20 years. In addition, variations in the methodologies and outcomes of published reports have contributed to uncertainty regarding the health care dollars saved due to motorcycle helmet use. The purpose of this systematic review and meta-analysis is to clarify the economic impact of motorcycle helmet use. Methods: Our primary source was Medline. Search terms included “motorcycle,” “motorbike,” “motorcycle helmet,” “head protective devices,” and “cost and cost analysis.” The review only included articles that were primary studies, written in English, evaluations of periods after 1994, and published in a peer-reviewed journal. Two independent authors extracted data using predefined data fields. Meta-analysis was done using the R-metafor package. Results: Twelve papers met the criteria for inclusion. Meta-analysis demonstrated that nonhelmeted patients required $12,239 more in hospital costs per patient. Nonhelmeted patients also required more postdischarge care and were more likely to use publicly funded insurance. Studies also found lower injury severity and better hospital course in the helmeted population. Study limitations included selection bias, unclear statistical assumptions, lack of precision measures, confounding variables, and lack of standardization to a common year. Meta-analysis demonstrated an I 2 of 67%, attributing a significant proportion of outcome variation to study differences. Conclusions: Motorcycle helmet use reduces morbidity and contributes to significant health care cost savings. Continuing antihelmet legislation will impose a substantial economic burden to the health care system, the government, and the public. Keywords: motorcycle, systemic review, meta-analysis, motorcycle helmet, head protective devices, costs and cost analysis
Introduction Riding a motorcycle remains a dangerous activity. On a per crash basis, motorcyclists are more than 5 times as likely to be killed as passenger car drivers (NHTSA 2013). The severity of motorcycle-related injuries causes considerable economic loss, costing the United States an estimated $14 billion in 2010. Costs are only expected to rise, because both motorcycle use and incidence of motorcycle accidents have steadily increased since the 1990s (NHTSA 2011). Motorcycle helmets are regarded as an important strategy for reducing the morbidity and therefore the cost of motorcyManaging Editor David Viano oversaw the review of this article. Address correspondence to Dr. Michael P. Leslie, DO, Yale Orthopedics, P.O. Box 208071, New Haven, CT 06519. E-mail: [email protected]
cle accidents. Helmets reduce the chance of fatal injury by 37% (Blincoe et al. 2014) and the risk of significant brain injuries (Abbreviated Injury Score 3–5) by a factor of 3 (Ouellet and Kasantikul 2006). This reduced injury profile has been associated with substantial economic cost savings (Blincoe et al. 2014). Despite the benefits of helmet use, repeals of universal mandatory helmet laws have gained renewed support in the past 2 decades. Efforts at antihelmet legislation were restarted after the Gingrich Revolution of 1995, which empowered the anti-motorcycle helmet lobby. As a result, in 1997 Arkansas became the first state in 14 years to repeal its universal mandatory helmet law (Jones and Bayer 2007). Arkansas was followed by Texas (1997), Kentucky (1998), Louisiana (1999), Florida (2000), Pennsylvania (2003), and Michigan (2012). Currently, only 19 states and the District of Columbia enforce universal mandatory helmet laws. Repeals have contributed to
Downloaded by [New York University] at 13:55 23 July 2015
Impact of Helmet Use on Motorcycle Accidents substantial economic losses, much of which was preventable (Blincoe et al. 2014). In addition, there has been debate regarding how much money that helmets actually save the health care system (Heller and Jacoby 2005). Nonhelmeted populations have increased alcohol and drug use (Subramanian 2005), a factor cited to explain their higher injury severity (Kasantikul et al. 2005). Although some models consider the effect of such variables (Weiss 1992), few studies directly compare their significance against helmet use. Furthermore, differences in analytic strategies have contributed to broad variation in reported costs (Hyder et al. 2007), complicating efforts to create a coherent economic policy. The rising cost of motorcycle accidents, in the context of continuing antihelmet laws, calls for a thorough examination of the economic benefits of helmet use. Such an examination is especially relevant with the passage of the Affordable Care Act (ACA), which has significantly changed the economic structure of the U.S. health care system. In this systematic review and meta-analysis, we have assessed studies that evaluate the economic impact of motorcycle helmets. Our goal with this review was 2-fold: (1) infer a single estimate for the cost of decreased helmet use in the past 20 years and (2) describe the methodological variations and limitations in helmet use reports. With this review, we clarify the economic benefits of motorcycle helmet use and analyze strategies for improving the quality of future economic evaluations in this field.
Methods Criteria for study inclusion and analysis were specified in advance in a protocol (Supplemental File 1, see online supplement) derived from the Community Guide for Economic Review (Community Preventive Services Task Force 2010). We then examined studies that analyzed the impact of helmets on hospital costs from motorcycle trauma. Articles needed to be in English, a primary study, an evaluation of periods in the last 20 years, and published in a peer-reviewed journal. Studies were identified by searching Medline (1946 to present), with the last search conducted on August 5, 2014. We used the following search words: “motorcycle,” “motorbike,” “motorcycle helmet,” and “head protective devices.” We then combined our search with “cost and cost analysis” (search strategy in Supplemental File 2, see online supplement). Eligibility assessment and data extraction were performed independently by 2 reviewers. Disagreements were arbitrated by a third reviewer and then resolved by consensus. From each retrospective study, we gathered information on (1) study design (time period of evaluation, study size, data source, inclusion criteria); (2) type of intervention (helmet vs no helmet, pre-repeal vs post-repeal of mandatory helmet laws); and (3) type of outcome measure (hospital charges/cost, insurance status, discharge location, injury profile). Study validity was assessed by evaluating selection bias, validity of statistical analysis, confounding variables, transparency regarding missing data, and standardization of costs to a base year. Publication bias and selective reporting were also analyzed. The cost outcome from each study was pooled to calcu-
733 late a weighted average, using a preestablished methodology (Supplemental File 3, see online supplement). Meta-analysis of data was performed using R metafor package (Viechtbauer 2010). We only considered studies that evaluated helmet use, becau studies of law repeal only measured the cost of head injuries. The pooled effect was defined as the difference in the mean hospital costs between helmeted and nonhelmeted patients. To account for heterogeneity among various studies, the DerSimonian and Laird random-effects statistical model (DerSimonian and Laird 1986) was applied. I 2 statistic was used to indicate study heterogeneity. An alpha value of less than .05 was considered statistically significant. For studies that did not report the precision (i.e., standard deviation) of estimated hospital costs, we used values from studies that did have the precision information. For the 8 regional studies, the standard deviations from Bledsoe et al. (2002; a regional study) were used. For the 2 national studies, the standard deviations from Eastridge et al. (2006; a national study) were used. Under this assumption, the computed values for the combined effects would still be valid, although their standard errors were likely not precise.
Results Study Design A total of 12 studies were identified for inclusion, as identified in our search flow diagram (Supplemental File 4, see online supplement). The initial search on Medline yielded 78 articles. From that selection, we excluded 51 studies based on the inclusion criteria. We then examined the full text of the remaining 27 citations and excluded an additional 15 studies. Eleven studies selected for review were retrospective cohort studies. One study, by Hotz et al. (2002), evaluated both prospective and retrospective cohorts. All were published in English-language peer-reviewed journals (Supplemental File 5, see online supplement), with evaluation periods ranging from 1994 to 2012. Eight studies used data from single institutions, 2 from state registries, and 2 from national databases. A total of 30,248 participants were involved in the economic evaluation, with study sizes ranging between 146 to 9,764 patients. The main inclusion criterion was admission to a hospital after a motorcycle accident, with known helmet status. Outcome reports were stratified into 2 major categories: (1) helmeted versus nonhelmeted patients and (2) patients before versus after repeal of mandatory helmet laws. The primary outcome was total cost from acute hospital care. All studies but one (Brandt et al. 2002) used hospital charges as a proxy for hospital costs. Brandt et al. (2002) provided actual hospital costs for analysis. With the exception of Eastridge at al. (2006), studies only evaluated patients who were admitted to a hospital. Eastridge et al. (2006) was unique in including the number of nonhospitalized crash victims. This allowed Eastridge et al. (2006) to calculate the average hospital charge per crash rather than per hospitalization. None of the studies included costs of rehabilitation, long-term medical costs, short-term work loss, long-term disability, and lost quality of life.
734
Downloaded by [New York University] at 13:55 23 July 2015
Analysis of Methodological Techniques Selection bias was a consistent issue (Supplemental File 6, see online supplement), because the majority of studies only evaluated patients who were injured severely enough to require hospital admission. This process excluded patients who were self-treated at the crash scene or discharged from the emergency room. Eastridge et al. (2006) attempted to address this issue by evaluating the average charge per crash. This value was calculated by dividing the total hospital charges by the total number of crash victims, regardless of hospitalization. A second methodological trend was the possible use of imprecise statistical techniques. The choice of statistical tests depends on the characteristics of the data; in particular, use of a t test would be incorrect to compare hospital costs, which are typically not normally distributed. Six papers addressed this issue by using nonparametric statistical tests or logarithmic normalization methods to resolve nonnormal distributions. However, 4 papers utilized t tests without making clear statements about the data distribution. Another 3 did not evaluate the statistical significance of their cost differences, so it was unclear how much of their analysis was affected by statistical assumptions. A related statistical trend was the absence of measures of precision, such as standard deviations or interquartile ranges. These parameters are necessary for gauging the spread and shape of quantitative data. They are also important for comparing different studies to generate pooled quantitative estimates, such as meta-analysis. Among the 12 papers included in this review, only 3 (Bledsoe et al. 2002; Eastridge et al. 2006; Hotz et al. 2002) provided standard deviations with their cost measures. In addition, we found that the majority of studies did not control for alcohol and drug abuse, 2 potential confounders in helmet effectiveness analysis. Out of the 12 studies, 3 papers (Brown et al. 2011; Hundley et al. 2004; Philip et al. 2013) examined admission levels of alcohol and drugs in crash victims. Among these, only Hundley et al. (2004) analyzed the influence of alcohol/drugs on patient outcomes (mortality, hospital charges, Injury Severity Score [ISS], hospital course) compared to helmet use. A final potential source of bias was the evaluation of costs from different time periods without standardizing to a base year. Because study evaluations ranged from 2 to 10 years, the true value of outcomes from the beginning of the study are not equal to those from the end. In our analysis, only 3 studies (Eastridge et al. 2006; Mertz et al. 2008; Ulmer and Northrop 2005) explicitly adjusted their hospital charges to a common year. We did not detect significant evidence of publication or reporting bias, because all studies reported measured outcomes regardless of statistical significance.
Synthesis of Results Table 1A summarizes hospital costs stratified by helmet status and Table 1Bsummarizes hospital costs stratified by helmet law repeal. To normalize for different evaluation periods, all dollar outcomes were adjusted to year 2014 values using
Kim et al. the medical component of the Consumer Price Index, provided by the U.S. Bureau of Labor Statistics (Supplemental File 7, see online supplement). For studies that did not provide an explicit base year, we used the final year of the study evaluation period. The extent of the cost savings varied widely, ranging from a cost ratio of 1.08 to 1.98 and hospital cost differences of $3,560 to $26,332. The hospital costs, charge difference, and cost ratios from studies involving single institutions were generally higher than those from studies that evaluated multiple institutions. The weighted average cost of treating a nonhelmeted patient was $7,059 higher than that for a helmeted patient, representing a cost ratio of 1.14. The weighted average cost of repealing universal mandatory helmet laws was $20,065 for the treatment of head injuries, representing a cost ratio of 1.25. Meta-analysis demonstrated a significant difference of $12,239 (95% confidence interval, 5,678–18,800, P < .001) in average hospital cost between the nonhelmeted and helmeted groups. The analysis also detected a high level of statistical heterogeneity (I 2 = 67%, P < .01) among studies. This outcome suggested that a significant 67% of the variability in the effect estimates was due to study differences (heterogeneity). This was visually evident from the wide scatter of estimates in their confidence intervals (Figure 1). Although all studies found that nonhelmeted and postrepeal patients cost more health care dollars, not all differences were statistically significant. Among studies that analyzed helmet use, 2 did not report P values. Among studies that analyzed law repeal, Bledsoe et al. (2002) reported a difference that was not statistically significant, and the others did not report P values. Six studies examined the insurance profile (Table 2) of helmeted and nonhelmeted patients, and another 6 evaluated their discharge disposition (Table 3). Although statistical significance was not commonly reported, all studies showed that nonhelmeted patients were more likely to require posthospital rehabilitation and to have publicly funded insurance or be uninsured. Ten studies (Supplemental File 8, see online supplement) examined admission and hospital course variables. All found that nonhelmeted patients had a higher Injury Severity Score (ISS), lower Glasgow Coma Scale (GCS), longer hospital length of stay, and longer intensive care unit length of stay.
Discussion Outcomes In 1991, congress passed a law that authorized a 3% decrease in highway funding in states without a universal mandatory helmet law. A conservative congressional majority in 1995, however, empowered the national motorcycle lobby to successfully overturn that penalty (Jones and Bayer 2007). As a result, in 1997 Arkansas became the first state in 14 years to repeal its universal mandatory helmet law. Other states soon followed. Repeals have coincided with a rise in the cost of motorcycle accidents, calling for a thorough reaffirmation of
Impact of Helmet Use on Motorcycle Accidents
735
Table 1. Cost of motorcycle accidents.
Downloaded by [New York University] at 13:55 23 July 2015
Values used for cost calculation
Helmeted (# of patients)
A. Helmet vs no helmet Nonhelmeted (# of patients) Difference (ratio)
Hospital costs Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges per injury Hospital charges per crash
$51,772 (174) $39,589 (98) $50,781 (978) $4,668 (101) $80,071 (522) $47,999 (152) $22,153 (153) $ 68,642 (88) $48,725 (5,484) $55,129 (3,783) $13,254 (128,445)
$62,005 (42) $64,204 (69) $57,672 (760) $8,236 (256) $106,203 (473) $74,090 (366) $34,381 (39) $91,478 (47) $52,444 (2,269) $59,766 (1,560) $18,743 (69,163)
calculation
Values used for cost (# of patients)
Hospital charges Hospital charges for head injury Hospital charges for head injury
$10,234 (1.98) $24,624 (1.62) $6,892 (1.41) $3,569 (1.76) $26,332 (1.33) $26,091 (1.54) $12,228 (1.55) $22,837 (1.33) $3,719 (1.08) $4,637 (1.08) $5,490 (1.41)
P value
Authors
Not reported
Traffic Injury Prevention Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gcpi20
The Economic Impact of Helmet Use on Motorcycle Accidents: A Systematic Review and Meta-analysis of the Literature from the Past 20 Years a
b
c
d
b
Chang-Yeon Kim , Daniel H. Wiznia , Leon Averbukh , Feng Dai & Michael P. Leslie a
Yale School of Medicine, New Haven, Connecticut
b
Department of Orthopaedics and Rehabilitation, Yale–New Haven Hospital, New Haven, Connecticut c
New York Institute of Technology College of Osteopathic Medicine, Old Westbury, New York
d
Yale Center for Analytical Sciences, Yale School of Medicine, New Haven, Connecticut Published online: 28 Apr 2015.
Click for updates To cite this article: Chang-Yeon Kim, Daniel H. Wiznia, Leon Averbukh, Feng Dai & Michael P. Leslie (2015) The Economic Impact of Helmet Use on Motorcycle Accidents: A Systematic Review and Meta-analysis of the Literature from the Past 20 Years, Traffic Injury Prevention, 16:7, 732-738, DOI: 10.1080/15389588.2015.1005207 To link to this article: http://dx.doi.org/10.1080/15389588.2015.1005207
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Traffic Injury Prevention (2015) 16, 732–738 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2015.1005207
The Economic Impact of Helmet Use on Motorcycle Accidents: A Systematic Review and Meta-analysis of the Literature from the Past 20 Years CHANG-YEON KIM1, DANIEL H. WIZNIA2, LEON AVERBUKH3, FENG DAI4, and MICHAEL P. LESLIE2 1
Yale School of Medicine, New Haven, Connecticut Department of Orthopaedics and Rehabilitation, Yale–New Haven Hospital, New Haven, Connecticut 3 New York Institute of Technology College of Osteopathic Medicine, Old Westbury, New York 4 Yale Center for Analytical Sciences, Yale School of Medicine, New Haven, Connecticut
Downloaded by [New York University] at 13:55 23 July 2015
2
Received 20 October 2014, Accepted 5 January 2015
Objective: The incidence and cost of motorcycle accidents are projected to increase. Motorcycle helmets are accepted as an effective strategy for reducing the morbidity and therefore the cost of motorcycle accidents. Despite this, states have continued to repeal helmet laws in the past 20 years. In addition, variations in the methodologies and outcomes of published reports have contributed to uncertainty regarding the health care dollars saved due to motorcycle helmet use. The purpose of this systematic review and meta-analysis is to clarify the economic impact of motorcycle helmet use. Methods: Our primary source was Medline. Search terms included “motorcycle,” “motorbike,” “motorcycle helmet,” “head protective devices,” and “cost and cost analysis.” The review only included articles that were primary studies, written in English, evaluations of periods after 1994, and published in a peer-reviewed journal. Two independent authors extracted data using predefined data fields. Meta-analysis was done using the R-metafor package. Results: Twelve papers met the criteria for inclusion. Meta-analysis demonstrated that nonhelmeted patients required $12,239 more in hospital costs per patient. Nonhelmeted patients also required more postdischarge care and were more likely to use publicly funded insurance. Studies also found lower injury severity and better hospital course in the helmeted population. Study limitations included selection bias, unclear statistical assumptions, lack of precision measures, confounding variables, and lack of standardization to a common year. Meta-analysis demonstrated an I 2 of 67%, attributing a significant proportion of outcome variation to study differences. Conclusions: Motorcycle helmet use reduces morbidity and contributes to significant health care cost savings. Continuing antihelmet legislation will impose a substantial economic burden to the health care system, the government, and the public. Keywords: motorcycle, systemic review, meta-analysis, motorcycle helmet, head protective devices, costs and cost analysis
Introduction Riding a motorcycle remains a dangerous activity. On a per crash basis, motorcyclists are more than 5 times as likely to be killed as passenger car drivers (NHTSA 2013). The severity of motorcycle-related injuries causes considerable economic loss, costing the United States an estimated $14 billion in 2010. Costs are only expected to rise, because both motorcycle use and incidence of motorcycle accidents have steadily increased since the 1990s (NHTSA 2011). Motorcycle helmets are regarded as an important strategy for reducing the morbidity and therefore the cost of motorcyManaging Editor David Viano oversaw the review of this article. Address correspondence to Dr. Michael P. Leslie, DO, Yale Orthopedics, P.O. Box 208071, New Haven, CT 06519. E-mail: [email protected]
cle accidents. Helmets reduce the chance of fatal injury by 37% (Blincoe et al. 2014) and the risk of significant brain injuries (Abbreviated Injury Score 3–5) by a factor of 3 (Ouellet and Kasantikul 2006). This reduced injury profile has been associated with substantial economic cost savings (Blincoe et al. 2014). Despite the benefits of helmet use, repeals of universal mandatory helmet laws have gained renewed support in the past 2 decades. Efforts at antihelmet legislation were restarted after the Gingrich Revolution of 1995, which empowered the anti-motorcycle helmet lobby. As a result, in 1997 Arkansas became the first state in 14 years to repeal its universal mandatory helmet law (Jones and Bayer 2007). Arkansas was followed by Texas (1997), Kentucky (1998), Louisiana (1999), Florida (2000), Pennsylvania (2003), and Michigan (2012). Currently, only 19 states and the District of Columbia enforce universal mandatory helmet laws. Repeals have contributed to
Downloaded by [New York University] at 13:55 23 July 2015
Impact of Helmet Use on Motorcycle Accidents substantial economic losses, much of which was preventable (Blincoe et al. 2014). In addition, there has been debate regarding how much money that helmets actually save the health care system (Heller and Jacoby 2005). Nonhelmeted populations have increased alcohol and drug use (Subramanian 2005), a factor cited to explain their higher injury severity (Kasantikul et al. 2005). Although some models consider the effect of such variables (Weiss 1992), few studies directly compare their significance against helmet use. Furthermore, differences in analytic strategies have contributed to broad variation in reported costs (Hyder et al. 2007), complicating efforts to create a coherent economic policy. The rising cost of motorcycle accidents, in the context of continuing antihelmet laws, calls for a thorough examination of the economic benefits of helmet use. Such an examination is especially relevant with the passage of the Affordable Care Act (ACA), which has significantly changed the economic structure of the U.S. health care system. In this systematic review and meta-analysis, we have assessed studies that evaluate the economic impact of motorcycle helmets. Our goal with this review was 2-fold: (1) infer a single estimate for the cost of decreased helmet use in the past 20 years and (2) describe the methodological variations and limitations in helmet use reports. With this review, we clarify the economic benefits of motorcycle helmet use and analyze strategies for improving the quality of future economic evaluations in this field.
Methods Criteria for study inclusion and analysis were specified in advance in a protocol (Supplemental File 1, see online supplement) derived from the Community Guide for Economic Review (Community Preventive Services Task Force 2010). We then examined studies that analyzed the impact of helmets on hospital costs from motorcycle trauma. Articles needed to be in English, a primary study, an evaluation of periods in the last 20 years, and published in a peer-reviewed journal. Studies were identified by searching Medline (1946 to present), with the last search conducted on August 5, 2014. We used the following search words: “motorcycle,” “motorbike,” “motorcycle helmet,” and “head protective devices.” We then combined our search with “cost and cost analysis” (search strategy in Supplemental File 2, see online supplement). Eligibility assessment and data extraction were performed independently by 2 reviewers. Disagreements were arbitrated by a third reviewer and then resolved by consensus. From each retrospective study, we gathered information on (1) study design (time period of evaluation, study size, data source, inclusion criteria); (2) type of intervention (helmet vs no helmet, pre-repeal vs post-repeal of mandatory helmet laws); and (3) type of outcome measure (hospital charges/cost, insurance status, discharge location, injury profile). Study validity was assessed by evaluating selection bias, validity of statistical analysis, confounding variables, transparency regarding missing data, and standardization of costs to a base year. Publication bias and selective reporting were also analyzed. The cost outcome from each study was pooled to calcu-
733 late a weighted average, using a preestablished methodology (Supplemental File 3, see online supplement). Meta-analysis of data was performed using R metafor package (Viechtbauer 2010). We only considered studies that evaluated helmet use, becau studies of law repeal only measured the cost of head injuries. The pooled effect was defined as the difference in the mean hospital costs between helmeted and nonhelmeted patients. To account for heterogeneity among various studies, the DerSimonian and Laird random-effects statistical model (DerSimonian and Laird 1986) was applied. I 2 statistic was used to indicate study heterogeneity. An alpha value of less than .05 was considered statistically significant. For studies that did not report the precision (i.e., standard deviation) of estimated hospital costs, we used values from studies that did have the precision information. For the 8 regional studies, the standard deviations from Bledsoe et al. (2002; a regional study) were used. For the 2 national studies, the standard deviations from Eastridge et al. (2006; a national study) were used. Under this assumption, the computed values for the combined effects would still be valid, although their standard errors were likely not precise.
Results Study Design A total of 12 studies were identified for inclusion, as identified in our search flow diagram (Supplemental File 4, see online supplement). The initial search on Medline yielded 78 articles. From that selection, we excluded 51 studies based on the inclusion criteria. We then examined the full text of the remaining 27 citations and excluded an additional 15 studies. Eleven studies selected for review were retrospective cohort studies. One study, by Hotz et al. (2002), evaluated both prospective and retrospective cohorts. All were published in English-language peer-reviewed journals (Supplemental File 5, see online supplement), with evaluation periods ranging from 1994 to 2012. Eight studies used data from single institutions, 2 from state registries, and 2 from national databases. A total of 30,248 participants were involved in the economic evaluation, with study sizes ranging between 146 to 9,764 patients. The main inclusion criterion was admission to a hospital after a motorcycle accident, with known helmet status. Outcome reports were stratified into 2 major categories: (1) helmeted versus nonhelmeted patients and (2) patients before versus after repeal of mandatory helmet laws. The primary outcome was total cost from acute hospital care. All studies but one (Brandt et al. 2002) used hospital charges as a proxy for hospital costs. Brandt et al. (2002) provided actual hospital costs for analysis. With the exception of Eastridge at al. (2006), studies only evaluated patients who were admitted to a hospital. Eastridge et al. (2006) was unique in including the number of nonhospitalized crash victims. This allowed Eastridge et al. (2006) to calculate the average hospital charge per crash rather than per hospitalization. None of the studies included costs of rehabilitation, long-term medical costs, short-term work loss, long-term disability, and lost quality of life.
734
Downloaded by [New York University] at 13:55 23 July 2015
Analysis of Methodological Techniques Selection bias was a consistent issue (Supplemental File 6, see online supplement), because the majority of studies only evaluated patients who were injured severely enough to require hospital admission. This process excluded patients who were self-treated at the crash scene or discharged from the emergency room. Eastridge et al. (2006) attempted to address this issue by evaluating the average charge per crash. This value was calculated by dividing the total hospital charges by the total number of crash victims, regardless of hospitalization. A second methodological trend was the possible use of imprecise statistical techniques. The choice of statistical tests depends on the characteristics of the data; in particular, use of a t test would be incorrect to compare hospital costs, which are typically not normally distributed. Six papers addressed this issue by using nonparametric statistical tests or logarithmic normalization methods to resolve nonnormal distributions. However, 4 papers utilized t tests without making clear statements about the data distribution. Another 3 did not evaluate the statistical significance of their cost differences, so it was unclear how much of their analysis was affected by statistical assumptions. A related statistical trend was the absence of measures of precision, such as standard deviations or interquartile ranges. These parameters are necessary for gauging the spread and shape of quantitative data. They are also important for comparing different studies to generate pooled quantitative estimates, such as meta-analysis. Among the 12 papers included in this review, only 3 (Bledsoe et al. 2002; Eastridge et al. 2006; Hotz et al. 2002) provided standard deviations with their cost measures. In addition, we found that the majority of studies did not control for alcohol and drug abuse, 2 potential confounders in helmet effectiveness analysis. Out of the 12 studies, 3 papers (Brown et al. 2011; Hundley et al. 2004; Philip et al. 2013) examined admission levels of alcohol and drugs in crash victims. Among these, only Hundley et al. (2004) analyzed the influence of alcohol/drugs on patient outcomes (mortality, hospital charges, Injury Severity Score [ISS], hospital course) compared to helmet use. A final potential source of bias was the evaluation of costs from different time periods without standardizing to a base year. Because study evaluations ranged from 2 to 10 years, the true value of outcomes from the beginning of the study are not equal to those from the end. In our analysis, only 3 studies (Eastridge et al. 2006; Mertz et al. 2008; Ulmer and Northrop 2005) explicitly adjusted their hospital charges to a common year. We did not detect significant evidence of publication or reporting bias, because all studies reported measured outcomes regardless of statistical significance.
Synthesis of Results Table 1A summarizes hospital costs stratified by helmet status and Table 1Bsummarizes hospital costs stratified by helmet law repeal. To normalize for different evaluation periods, all dollar outcomes were adjusted to year 2014 values using
Kim et al. the medical component of the Consumer Price Index, provided by the U.S. Bureau of Labor Statistics (Supplemental File 7, see online supplement). For studies that did not provide an explicit base year, we used the final year of the study evaluation period. The extent of the cost savings varied widely, ranging from a cost ratio of 1.08 to 1.98 and hospital cost differences of $3,560 to $26,332. The hospital costs, charge difference, and cost ratios from studies involving single institutions were generally higher than those from studies that evaluated multiple institutions. The weighted average cost of treating a nonhelmeted patient was $7,059 higher than that for a helmeted patient, representing a cost ratio of 1.14. The weighted average cost of repealing universal mandatory helmet laws was $20,065 for the treatment of head injuries, representing a cost ratio of 1.25. Meta-analysis demonstrated a significant difference of $12,239 (95% confidence interval, 5,678–18,800, P < .001) in average hospital cost between the nonhelmeted and helmeted groups. The analysis also detected a high level of statistical heterogeneity (I 2 = 67%, P < .01) among studies. This outcome suggested that a significant 67% of the variability in the effect estimates was due to study differences (heterogeneity). This was visually evident from the wide scatter of estimates in their confidence intervals (Figure 1). Although all studies found that nonhelmeted and postrepeal patients cost more health care dollars, not all differences were statistically significant. Among studies that analyzed helmet use, 2 did not report P values. Among studies that analyzed law repeal, Bledsoe et al. (2002) reported a difference that was not statistically significant, and the others did not report P values. Six studies examined the insurance profile (Table 2) of helmeted and nonhelmeted patients, and another 6 evaluated their discharge disposition (Table 3). Although statistical significance was not commonly reported, all studies showed that nonhelmeted patients were more likely to require posthospital rehabilitation and to have publicly funded insurance or be uninsured. Ten studies (Supplemental File 8, see online supplement) examined admission and hospital course variables. All found that nonhelmeted patients had a higher Injury Severity Score (ISS), lower Glasgow Coma Scale (GCS), longer hospital length of stay, and longer intensive care unit length of stay.
Discussion Outcomes In 1991, congress passed a law that authorized a 3% decrease in highway funding in states without a universal mandatory helmet law. A conservative congressional majority in 1995, however, empowered the national motorcycle lobby to successfully overturn that penalty (Jones and Bayer 2007). As a result, in 1997 Arkansas became the first state in 14 years to repeal its universal mandatory helmet law. Other states soon followed. Repeals have coincided with a rise in the cost of motorcycle accidents, calling for a thorough reaffirmation of
Impact of Helmet Use on Motorcycle Accidents
735
Table 1. Cost of motorcycle accidents.
Downloaded by [New York University] at 13:55 23 July 2015
Values used for cost calculation
Helmeted (# of patients)
A. Helmet vs no helmet Nonhelmeted (# of patients) Difference (ratio)
Hospital costs Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges Hospital charges per injury Hospital charges per crash
$51,772 (174) $39,589 (98) $50,781 (978) $4,668 (101) $80,071 (522) $47,999 (152) $22,153 (153) $ 68,642 (88) $48,725 (5,484) $55,129 (3,783) $13,254 (128,445)
$62,005 (42) $64,204 (69) $57,672 (760) $8,236 (256) $106,203 (473) $74,090 (366) $34,381 (39) $91,478 (47) $52,444 (2,269) $59,766 (1,560) $18,743 (69,163)
calculation
Values used for cost (# of patients)
Hospital charges Hospital charges for head injury Hospital charges for head injury
$10,234 (1.98) $24,624 (1.62) $6,892 (1.41) $3,569 (1.76) $26,332 (1.33) $26,091 (1.54) $12,228 (1.55) $22,837 (1.33) $3,719 (1.08) $4,637 (1.08) $5,490 (1.41)
P value
Authors
Not reported
