2015 Annual Report of A TV-Related Deaths and Injuries January 2017 CPSAJ,I..I!H1l kW\RED for PUBJIC{ M ~ MFRS/PRVTLBLRS OR PRODUCTS IDENTIFIED EXCEPTED BY: PETITION RULEMAKING ADMIN. PRCOG WITH PORTIONS REMOVED: - -- John Topping , M.S. Directorate for Epidemiology Division of Hazard Analysis U.S. Consumer Product Safety Commission 4330 East West Highway Bethesda, MD 20814 This analysis was prepared by CPSC staff. It has not been reviewed or approved by, and may not necessarily reflect the views of, the Commission . CPSC Hotline: 1-800-638-CPSC (2772) CPSC's Website: http://www.cpsc.gov * 0/'11 I 13 Executive Summary ATV-Related Fatalities  As of December 31, 2015, CPSC staff received reports of 14,129 ATV-related fatalities occurring between 1982 and 2015. CPSC staff received reports of 340 ATV-related fatalities occurring in 2015, 547 occurring in 2014, and 581 occurring in 2013. Reporting for the years 2013 through 2015 is ongoing; these numbers are expected to increase in future reports.1  From 1982 through 2015, CPSC staff received reports of 3,163 ATV-related fatalities of children younger than 16 years of age. This represents 22 percent of the total number of reported ATV-related fatalities (14,129).  Of the 3,163 reported ATV-related fatalities of children younger than 16 years of age, 1,380 (44 percent) were younger than 12 years of age.  In 2012, the most recent year where reporting is considered complete, 68 (12 percent) of the reported 573 ATV-related fatalities were children younger than 16 years of age.  The estimated number of ATV-related fatalities is 674 for 2014 and 657 for 2013. Reporting for 2013 and 2014 is ongoing; thus, these numbers are expected to change in future reports. For the year 2015, there are not yet sufficient data to estimate fatalities.  In 2012, the most recent year where reporting is considered complete, the estimated number of ATV-related fatalities is 655. The estimated number of four-wheel ATV-related fatalities in 2012 is 638. ATV-Related Emergency Department-Treated Injury Estimates  In 2015, there were an estimated 97,200 ATV-related, emergency department-treated injuries in the United States. An estimated 28 percent of these involved children younger than 16 years of age.  The increase in the estimated number of ATV-related, emergency department-treated injuries from 2014 to 2015 (93,700 in 2014 to 97,200 in 2015) is not statistically significant.  The only age group with statistically significantly different injuries between 2014 and 2015 is the 55+ age group (an increase from an estimated 5,300 injuries in 2014 to 7,100 in 2015). Other groups also had increased estimates (children younger than 16, 16-24 and 45-54) or decreased estimates (25-34 and 35-44), but none of these reflect statistically significant differences between 2014 and 2015.  Of the 97,200 estimated ATV-related, emergency department-treated injuries for all ages in 2015, a majority are categorized as treated and released (85 percent).  A plurality of the 2015 estimated ATV-related, emergency department-treated injuries were diagnosed as contusions/abrasions or fractures (23 percent and 25 percent, respectively). 1 To illustrate the increase in the number of reported deaths over time, at the time of data collection cut off for the 2014 report (December 31, 2014), 385 ATV-related deaths had been reported for 2014. Similarly for the 2013 report, 426 ATV-related deaths had been reported for 2013 at the time of data collection cutoff (December 31, 2013). 2  The 2015 estimated ATV-related, emergency-department treated injuries primarily affected the following body parts: the arm (the shoulder down), the head or neck, leg, and the torso (29, 29, 20, and 21 percent, respectively).  Prior reports have already characterized statistically significant increases during the period 2001 to 2007. More recent years reflect a change in the trend direction for ATV-related injury estimates. This report focuses the injury discussion on the years 2007 through 2015 during which the number of injuries per year has gone through a statistically significant decline.  During the years 2007 through 2015 ATV-related injury estimates generally decrease each year. Although not every individual year-to-year decrease is statistically significant, CPSC staff did find statistically significant year-to-year decreases in the estimates between 2007 and 2008 and between 2009 and 2010. Additionally, CPSC staff found a statistically significant decrease in the estimates when comparing just the first and last years of the period (comparing 2007 against 2015). When considering the entire nine years together (2007−2015), CPSC staff found a statistically significant decreasing linear trend. 3 Table of Contents Executive Summary ................................................................................................................................................. 2 Introduction .............................................................................................................................................................. 5 Background .............................................................................................................................................................. 5 ATV-Related Fatalities ............................................................................................................................................. 6 Reported Deaths ................................................................................................................................................. 6 Reported Deaths by State ................................................................................................................................... 8 Reported Deaths of Children ............................................................................................................................. 10 Three-Wheel Versus Four-Wheel ATVs ............................................................................................................ 13 Estimated ATV-Related Deaths, 1985 to 2014 ................................................................................................. 13 ATV-Related Injuries.............................................................................................................................................. 15 ATVs with Three, Four, and Unknown Number of Wheels ............................................................................... 15 Four-Wheel ATVs .............................................................................................................................................. 21 Discussion ............................................................................................................................................................. 22 1982–1998 ........................................................................................................................................................ 22 1999–2012 ........................................................................................................................................................ 23 2013–2015 ........................................................................................................................................................ 24 Appendix A: Estimation Methodologies ................................................................................................................. 25 ATV-Related Deaths .......................................................................................................................................... 25 In-Scope ATV-Related Fatalities .................................................................................................................... 25 ICD-9 Versus ICD-10 Coding ......................................................................................................................... 25 Estimation of ATV-Related Fatalities (1999–Present) ................................................................................... 25 Estimation of ATV-Related Fatalities (1985–1998) ........................................................................................ 26 Estimation of Fatalities Associated with Four-Wheel ATVs ........................................................................... 27 ATV-Related Injuries ......................................................................................................................................... 28 Estimation of Emergency Department-Treated Injuries Associated with ATVs ............................................. 28 Coefficients of Variation ................................................................................................................................. 28 Estimation of Emergency Department-Treated Injuries Associated with Four-Wheel ATVs ......................... 29 Appendix B: Historical Estimated Number of Emergency Department-Treated Injuries 1985–2015 .................... 30 References ............................................................................................................................................................ 31 List of Tables Table 1: Reported Fatalities by Year 1982–2015 .................................................................................................... 7 Table 2: Reported Fatalities by State 1982–2015 ................................................................................................... 9 Table 3: Reported ATV-Related Fatalities for Children Younger than 16 and 12 Years of Age 1982–2015 ........ 11 Table 4: Annual Estimated ATV-Related Fatalities 1985-2014 ............................................................................. 14 Table 5: Annual Estimates of ATV-Related, Emergency Department-Treated Injuries 2007–2015 ..................... 15 Table 6: Estimated Number of Emergency Department-Treated Injuries by Age Group 2007–2015 ................... 19 Table 7: Estimated Number of Emergency Department-Treated Injuries 2007–2015 .......................................... 21 Table 8: Historical ATV-Related Emergency Department-Treated Injuries Estimates for ATVs with 3,4, or Unknown Number of Wheels and Four-Wheel ATVs from 1985 to 2015 ..................................................... 30 List of Figures Figure 1: Number of Reported ATV-Related Fatalities by State (1982–2012) ...................................................... 10 Figure 2: Reported ATV-Related Fatalities by Age Group (2001–2012)............................................................... 12 Figure 3: Annual ATV-Related Hospital Emergency Department-Treated Injury Estimates for All Ages and Less Than 16 Years (2007–2015) ........................................................................................................................ 16 Figure 4: Disposition, Diagnosis, and Body Part ATV-Related Emergency Department-Treated Injuries for All Ages, 2015 ................................................................................................................................................... 18 Figure 5: Annual ATV-Related Hospital Emergency Department-Treated Injury Estimates by Age Group (2007–2015) ................................................................................................................................................. 20 4 Introduction This report presents the 2015 annual update of information collected by U.S. Consumer Product Safety Commission (CPSC) staff on deaths and injuries related to the use of all-terrain vehicles (ATVs). The update includes information on ATV-related deaths, based on data available to CPSC staff as of December 31, 2015, as well as information on ATV-related injuries treated in hospital emergency departments from January 1, 2015 through December 31, 2015. The report begins with a brief background section. This is followed by a summary of ATV-related fatality counts (reports) derived from CPSC staff data, along with a discussion of reported ATV-related deaths involving children younger than 12 and 16 years of age. Then, based on the counts of ATV-related fatalities reported to CPSC staff, annual estimates of ATV-related deaths are presented. Next, the report provides estimates of ATV-related, hospital emergency department-treated injuries. The report concludes with a short discussion of the observed patterns of ATV-related deaths and injuries over time. Background CPSC staff considers an ATV to be an off-road, motorized vehicle having three or four low-pressure tires, a straddle seat for the operator, and handlebars for steering control. CPSC staff does not categorize off-road motor vehicles having steering wheels and either bench or bucket seats (e.g., golf carts, dune buggies, recreational off-highway vehicles (ROVs), and certain types of utility vehicles) as ATVs. Consequently, fatalities and injuries associated with these types of vehicles are not addressed in this report. CPSC staff first began collecting ATV-related incident data in the early 1980s to provide information on the numbers of deaths and injuries associated with three-wheel ATVs. In the late 1980s, the major ATV distributors agreed to stop distributing three-wheel ATVs (U.S. CPSC, 2006). Consequently, although some older three-wheel ATVs continue to be used by consumers, nearly all ATVs in use today are fourwheel ATVs. In 2005, the CPSC issued an advance notice of proposed rulemaking, followed by a 2006 notice of proposed rulemaking, for ATVs. Section 232 of the Consumer Product Safety Improvement Act of 2008 (CPSIA) included provisions that directed the CPSC to make the voluntary standard a mandatory standard. The CPSIA imposed certain additional requirements on ATVs through ATV action plans, and the statute prohibited the importation and distribution of three-wheel ATVs.2 The mandatory standard was published in late 2008, and the mandatory standard became effective in April 2009. In 2012, CPSC revised the mandatory standard to reflect changes in the voluntary standard. 2 The provisions relating to ATVs are at 16 C.F.R. part 1420. See: http://www.ecfr.gov/cgi-bin/text-idx?SID=dbc1bddfdc8b6fa15f91bd3bda47acd6&tpl=/ecfrbrowse/Title16/16cfr1420_main_02.tpl . 5 ATV-Related Fatalities This section provides details on the numbers of ATV-related death reports received by CPSC staff on or before December 31, 2015, and discusses the estimates associated with ATV-related deaths. The reported numbers of deaths are totaled in Table 1, listed by state in Table 2, and categorized by age group in Table 3. The estimated numbers of deaths associated with ATVs having three, four, or an unknown number of wheels are reported in Table 4, along with the estimated numbers of four-wheel ATV-related fatalities. Reported Deaths By December 31, 2015, CPSC staff had received reports of 14,129 ATV-related deaths that occurred between 1982 and 2015 (Table 1). The number of fatality reports increased by 512 since the December 31, 2014 tabulation CPSC staff prepared (U.S. CPSC, November 2015). Data collection for the years 2013 through 2015 is ongoing. As a result, the numbers of reported deaths for 2013 through 2015 are expected to increase before the next annual report is prepared. 6 Table 1 Reported ATV-Related Fatalities (by Year) ATVs with 3, 4, or Unknown Number of Wheels Reported for the Period January 1, 1982 through December 31, 2015 Reported Number of Deaths Difference Since Last Update (12/31/2014*) 2015 14,129 340 +512 +340 2014 547 +162 2013 581 +34 2012 573 -1 2011 622 -4 2010 650 -4 2009 720 -1 2008 759 +1 2007 832 -1 2006 831 -1 2005 799 0 2004 747 -6 2003 651 +1 2002 547 -3 2001 517 -3 2000 445 -2 1999 396 -1 1998 253 +1 1997 240 0 1996 249 0 1995 200 0 1994 198 0 1993 183 0 1992 220 0 1991 230 0 1990 235 0 1989 230 0 1988 250 0 1987 264 0 1986 300 0 1985 250 0 1984 156 0 1983 85 0 1982 29 0 Year Total Source: U.S. Consumer Product Safety Commission: Directorate for Epidemiology/Division of Hazard Analysis. Note: Italics indicate that reporting is ongoing for the years 2013–2015. Note: The heavy line marks the change from death certificate mortality coding under the Ninth Revision of the International Classification of Diseases (ICD-9) to coding under the Tenth Revision (ICD-10). * Comparison is relative to counts of reported death in previous year’s report which was based on incidents reported to CPSC by 12/31/2014. Further review of historic records revealed information supporting adjustments to date of death, vehicle classification (i.e., ATVs with 4 wheels for which the number of wheels was not previously determined or out of scope vehicles such as utility vehicles), and scope. Application of alternative search schemes identified records that could be further associated as multiple reports of the same incident. . 7 In Table 1, the counts presented for 1999 and later (i.e., the values above the heavy line) reflect a revised mortality data classification system from the system used before 1999. Specifically, the heavy line marks the change from death certificate mortality coding under the Ninth Revision of the International Classification of Diseases (ICD-9), to coding under the Tenth Revision (ICD-10), a transition that allows CPSC staff to gather more accurate mortality data for a number of consumer products, including ATVs. This change was implemented by the National Center for Health Statistics (NCHS) in January 1999 (NCHS, 2007). Since the implementation of ICD-10 coding, all ATV-related fatalities, including incidents involving traffic accidents on public roads, are grouped under a single set of mortality codes. Because of the use of different coding systems between the two periods (i.e., before 1999, versus 1999 and later), comparisons of numbers between these periods should be undertaken with caution. The ICD-10 transition and related methodological issues are discussed more fully in Appendix A. Reported Deaths by State Table 2 gives the numbers of reported ATV-related deaths for each state, the District of Columbia, and Puerto Rico. States are listed in descending order of the number of ATV-related fatalities reported for the years 1982 through 2012 (i.e., the years for which data collection is considered complete). Reported deaths that occurred during these years are tabulated in the second column. The following states had the highest numbers of reported ATV-related deaths occurring in this period: Texas (698), California (654), West Virginia (641), Pennsylvania (635), and Kentucky (580). Together, these five states accounted for 3,208 deaths or 25 percent of all reported ATV-related deaths in the United States for the years 1982 through 2012 (n =12,661), as shown in the third column of Table 2. When reviewing state death counts for the period 1982 through 2012, two points deserve note:  Consistent with CPSC staff’s previous annual reports on ATV-related deaths and injuries, the counts shown in Table 2 have not been adjusted for demographic characteristics (e.g., total population, age structure of population).  Also consistent with previous CPSC staff reports, these counts reflect the state in which the death occurred, rather than the state where the ATV incident occurred. This approach allows the most accurate matching of death certificates to other types of incident reports received by CPSC staff. As medical transport capabilities (e.g., helicopter transport) and trauma care have advanced in recent years, some states with major trauma centers have ATV-related fatalities included in their reported counts for incidents that did not occur within their state boundaries. Similarly, some states have reported counts that do not fully capture all of the ATV-related fatality incidents that occurred within their state boundaries. The fourth column of Table 2 presents, by state, the number of ATV-related deaths reported to CPSC staff as of December 31, 2015, for the period 2013 through 2015. These counts should not be used for comparisons between states because data collection for this period is ongoing and because data collection for some states is more complete than for other states for these years. Each state’s total number of reported deaths since 1982 is listed in the fifth column. These counts include information for the years that have ongoing reporting, as well as information for the years where data collection is considered complete. 8 Table 2 Reported ATV-Related Fatalities (by State) ATVs with 3, 4, or Unknown Number of Wheels Reported for the Period January 1, 1982 through December 31, 2015 State TEXAS CALIFORNIA WEST VIRGINIA PENNSYLVANIA KENTUCKY FLORIDA TENNESSEE NEW YORK NORTH CAROLINA MICHIGAN GEORGIA OHIO ARKANSAS MISSOURI MISSISSIPPI WISCONSIN MINNESOTA ALABAMA LOUISIANNA ILLINOIS ARIZONA VIRGINIA INDIANA UTAH OKLAHOMA OREGON IDAHO COLORADO ALASKA WASHINGTON SOUTH CAROLINA IOWA KANSAS MAINE NEBRASKA NEW MEXICO MONTANA NEVADA MARYLAND NEW JERSEY NORTH DAKOTA SOUTH DAKOTA MASSACHUSETTS NEW HAMPSHIRE VERMONT WYOMING CONNECTICUT HAWAII DELAWARE RHODE ISLAND WASHINGTON, D.C. PUERTO RICO Cumulative Percent of U.S. Reported Deaths Reported Deaths 1982–2012 1982–2012 698 654 641 635 580 512 492 433 428 405 400 389 369 358 357 316 300 297 287 271 257 239 235 215 213 211 197 179 164 161 158 157 149 143 123 114 108 106 96 91 84 80 75 65 65 62 46 19 9 9 5 4 6% 11% 16% 21% 25% 29% 33% 37% 40% 43% 46% 49% 52% 55% 58% 61% 63% 65% 68% 70% 72% 74% 75% 77% 79% 80% 82% 83% 85% 86% 87% 89% 90% 91% 92% 93% 94% 94% 95% 96% 97% 97% 98% 98% 99% 99% 100% 100% 100% 100% 100% 100% Reported Deaths (Ongoing Reporting) 2013-2015* Total Reported Deaths* (Including Ongoing Reporting) 82 49 84 62 72 64 64 51 32 35 17 53 18 61 41 56 26 39 34 18 14 29 28 17 41 20 37 30 29 27 23 23 16 15 14 26 28 10 14 10 5 13 10 8 13 2 2 2 1 1 1 1 780 703 725 697 652 576 556 484 460 440 417 442 387 419 398 372 326 336 321 289 271 268 263 232 254 231 234 209 193 188 181 180 165 158 137 140 136 116 110 101 89 93 85 73 78 64 48 21 10 10 6 5 Source: U.S. Consumer Product Safety Commission: Directorate for Epidemiology/Division of Hazard Analysis. Note: State rankings are based on ATV-related fatality counts for the period 1982–2012. *Italicized data (columns 4 and 5) denote the years for which reporting is ongoing (2013-2015). 9 Figure 1 represents the first two columns of Table 2. The intensity of the color in each state represents the number of reported ATV-related fatalities from 1982 through 2012. Figure 1: Number of Reported ATV-Related Fatalities by State (1982–2012) Source: U.S. Consumer Product Safety Commission: Directorate for Epidemiology/Division of Hazard Analysis. Graph generated using SAS®. Note: This figure corresponds to the first two columns of Table 2. Reporting for 2013–2015 is ongoing, and Figure 1 does not include data for these years. Reported Deaths of Children A review of the reported ATV-related fatalities indicates that 3,163 decedents (22 percent of the 14,129 total) between 1982 and 2015 were younger than 16 years of age, and 1,380 (10 percent) were younger than 12 years of age. Forty-four percent of ATV-related child fatalities (i.e., children under 16 years of age) were children younger than 12 years of age. Table 3 gives the total number of reported fatalities, by year, of children younger than 16 years of age; the corresponding percentage of the total number of reported fatalities, by year; the total numbers of fatalities by year for children younger than 12 years of age; and the corresponding percentage relating to all ATV-related fatalities of children younger than 16 years of age. Figure 2 displays the total number of reported ATV-related fatalities, by year and age group, from 2001 to 2012. 10 Table 3 Reported ATV-Related Fatalities for Children Younger than 16 and 12 Years of Age ATVs with 3, 4, or Unknown Number of Wheels Reported for the Period January 1, 1982 through December 31, 2015 Year Younger Than 16 Younger Than 16: Percent of Total Younger Than 12 Younger Than 12: Percent of Children* Total 3,163 22% 1,380 44% 2015 58 17% 32 55% 2014 71 13% 40 56% 2013 70 12% 28 40% 2012 68 12% 39 57% 2011 80 13% 38 48% 2010 89 14% 39 44% 2009 96 13% 41 43% 2008 109 14% 53 49% 2007 136 16% 58 43% 2006 142 17% 63 44% 2005 162 20% 77 48% 2004 179 24% 69 39% 2003 153 24% 69 45% 2002 133 24% 45 34% 2001 133 26% 58 44% 2000 122 27% 50 41% 1999 90 23% 34 38% 1998 82 32% 30 37% 1997 79 33% 38 48% 1996 87 35% 40 46% 1995 64 32% 26 41% 1994 54 27% 20 37% 1993 59 32% 18 31% 1992 71 32% 32 45% 1991 68 30% 40 59% 1990 81 34% 27 33% 1982-1989 627 40% 276 44% Source: U.S. Consumer Product Safety Commission: Directorate for Epidemiology/Division of Hazard Analysis. Note: Italics denote the period for which reporting is ongoing. *Percent of total ATV-related fatalities of children younger than 16 years of age. Note: Reporting is ongoing for 2013–2015. Percentages for these years should be interpreted with caution because the rate at which deaths are reported may not be consistent for each of the years indicated. The percentage of victims younger than 16 years of age appears to have generally declined over time, but it is possible that adult deaths are underreported during the period 1982 to 1998 as well as in the most recent years for which reporting is not complete (2013, 2014, and especially 2015). Because of coding limitations for ATV-related fatalities under the old ICD-9 system (see Appendix A), CPSC staff generally was not able to gather reports of deaths on public roads during the years 1982 to 1998. If adults were more likely than children to have been involved in ATV-related fatality incidents on public 11 roads, then, for the years before 1999, the calculated percentages of deaths involving children, shown in Table 3, may be higher than the true proportion of ATV-related fatalities involving children. The most recent year, 2015, is especially dependent upon investigations or incidents that were reported from public news sources, and child ATV fatalities may be more likely to be reported in the news than adult fatalities. In future reports the proportion of child and adult deaths for the most recent year (2015) may be especially likely to exhibit a higher proportion of adult deaths as reports of ATV deaths from states and other non-news related sources are provided to CPSC.3 Figure 2: Reported ATV-Related Fatalities by Age Group (2001–2012) Less than 12 Age 12-15 Age 16+ 900 832 831 799 800 759 747 720 700 650 651 622 600 573 547 517 500 568 637 689 696 650 498 400 624 561 300 384 414 542 505 200 100 75 58 2001 0 88 84 110 85 79 78 56 55 50 45 69 69 77 63 58 53 41 39 42 38 29 39 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Source: U.S. Consumer Product Safety Commission: Directorate for Epidemiology/Division of Hazard Analysis. Note: This figure corresponds to the data reported in Tables 1 and 3. Reporting for 2013–2015 is ongoing; thus, Figure 2 does not display these years. 3 From past reports the percentage of child fatalities for a given year tends to be highest when it is the most recent year with incomplete reporting and declines after reporting for that year is complete. For example, the 2012 report indicated 15% of reported fatalities for the year 2012 to involve children, but now that reporting for 2012 is completed the percentage of child fatalities for the year 2012 is only 12%. Although reporting is still ongoing for the years 2013 and 2014, the percentage of children for those years was also higher in previous reports. Children represented 15% of year 2013 fatalities in the 2013 report and 16% of year 2014 fatalities in the 2014 report (compared with 12% for 2013 and 13% for 2014 in this year’s report). 12 Three-Wheel Versus Four-Wheel ATVs As noted in the Background section above, CPSC staff began collecting reports of ATV-related deaths and injuries in the early 1980s to assess incidents associated with three-wheel ATVs. However, in the late 1980s, the major distributors agreed to stop distributing three-wheel ATVs. Currently, the ATVs distributed in the United States are nearly all four-wheel models (U.S. CPSC, 2006).4 The percentage of reported fatalities involving four-wheel ATVs increased from 7 percent or less, before 1985, to 99 percent in 2015, based on 2015 fatalities reported to CPSC staff as of December 31, 2015. Although data collection for 2015 is ongoing, this percentage is not expected to change greatly as additional reports of 2015 fatalities are received. From the incident reports, it is not always possible to ascertain whether the ATV involved in the incident has three wheels or four wheels. In these cases, the vehicle is coded as an ATV having an unknown number of wheels. For the estimates of ATV-related deaths described below, ATVs having an unknown number of wheels were apportioned between three-wheel and four-wheel ATVs, using the methods described in Appendix A. Estimated ATV-Related Deaths, 1985 to 2014 Death reports received by CPSC staff represent a minimum count of ATV-related deaths because not all ATV-related fatalities may be reported. To account for unreported deaths, CPSC staff estimated annual ATV-related fatalities for the period 1985 to 2014, using a statistical estimation method called capture-recapture (Hook and Regal, 2004). See Appendix A for details on this estimation method. Table 4 shows both the annual reported counts and the estimated number of ATV-related deaths involving ATVs having three, four, or an unknown number of wheels. Due to the low data collection completion rate (35%) as of December 31, 2015, estimates for 2015 were not calculated for this report but will be estimated in future reports. The heavy line between 1998 and 1999 in Table 4 demarcates the switch from mortality data collection under ICD-9, to collection under ICD-10. Because mortality coding under ICD-10 allows CPSC staff to gather more complete data on ATV-related deaths, some of the increase in estimated deaths observed between 1998 and 1999 is probably a result of the ICD-9 to ICD-10 transition. Although the magnitude of the effect of the coding change is unknown, it follows that the death estimates calculated for the years before 1999 may have been underestimated. Through the 2012 ATV Annual Report, Column 5 of Table 4 gave annual estimates of the number of four-wheel ATVs in use. According to CPSC staff’s All-Terrain Vehicle 2001 Injury and Exposure Studies, about 5.6 million three- and four-wheel ATVs were in use, and about 86 percent of these were four-wheel ATVs (Levenson, 2003a). Since that study, CPSC staff had relied on annual ATV sales information, together with survival analysis models, to derive estimates of the number of four-wheel ATVs in use each year. However, these techniques assume that the ATV market and ATV lifespans measured in the original study likely reflect the ongoing ATV market and ATVs’ lifespans. Such assumptions are only reasonable for a relatively finite number of years, and CPSC does not have data to update or maintain these assumptions. A discussion of the methodologies used to estimate ATV-related deaths is given in Appendix A. 4 There are some six-wheel models in distribution, but such are not counted in the scope of this report. 13 Table 4 Annual Estimates of ATV-Related Fatalities 1985–2014 Based on Fatality Data Available as of December 31, 2015 *Reported Deaths Estimated Deaths Associated with ATVs with 3, 4, or Unknown Wheels Estimated Deaths Involving 4Wheel ATVs 2014 547 674 663 2013 581 657 642 2012 573 655 638 2011 622 684 663 2010 650 724 712 2009 720 793 772 2008 759 845 826 2007 832 895 865 2006 831 898 871 2005 799 922 882 2004 747 836 798 2003 651 752 717 2002 547 597 560 2001 517 586 542 2000 445 540 491 1999 396 521 473 1998 253 291 248 1997 240 289 235 1996 249 268 209 1995 200 276 213 1994 198 243 167 1993 183 210 143 1992 220 242 159 1991 230 251 150 1990 235 252 152 1989 230 257 152 1988 250 285 151 1987 264 282 126 1986 300 348 95 1985 250 293 55 Year 5 Source: U.S. Consumer Product Safety Commission: Directorate for Epidemiology/Division of Hazard Analysis. Note: Italics denote the period for which reporting is ongoing. *Due to adjustments noted under Table 1, fatality estimates may differ from the estimates from last year’s report even for some years in which the number of reported deaths is the same. 5 Reporting is ongoing for 2013–2014. Reporting for 2015 is too preliminary to credibly estimate deaths. 14 ATV-Related Injuries ATVs with Three, Four, and Unknown Numbers of Wheels Table 5 shows estimates of ATV-related injuries treated in hospital emergency departments nationwide between January 1, 2007, and December 31, 2015. These estimates were generated based on the CPSC’s National Electronic Injury Surveillance System (NEISS), a probability sample of U.S. hospitals with 24-hour emergency departments with more than six beds. In this analysis, the current estimates were compared to the estimates from the previous year (2014), as well as to a base year. The base year chosen for comparison was 2007.6 The existence of a possible trend in injuries associated with ATVs with three, four, or an unknown number of wheels is also considered, based on trend analysis methods developed by CPSC staff (Schroeder, 2000). For historical estimates of the number of ATVrelated, emergency department-treated injuries, see Appendix B. Also included in Table 5 are the estimated number of ATV-related, emergency department-treated injuries for children younger than 16 years, with the corresponding percent of total injuries. Estimates for those younger than 12 years are provided; these estimates are a subset of the younger than 16 age group. That is, the estimated injuries for those younger than 12 years are also included in the younger than 16 years age group estimates. Figure 3 displays the information for the estimated injuries for all ages and for the younger than 16 age group. Year 2015 2014 2013 2012 2011 2010 2009 2008 2007 Table 5 7 Annual Estimates of ATV-Related, Emergency Department-Treated Injuries ATVs with 3, 4, or Unknown Number of Wheels January 1, 2007 through December 31, 2015 Younger than 12 Years: Percent Estimated Younger than 16: of Injuries to All Number of Younger than 16 Percent of Total Younger than 12 Children Injuries: Years: Estimated (All-Ages) Years: Estimated Younger than 16 All Ages Number of Injuries Injuries Number of Injuries Years 97,200 26,700 28% 13,400 50% 93,700 24,800 26% 11,400 46% 99,600 25,000 25% 13,100 52% 107,900 26,500 25% 12,200 46% 107,500 29,000 27% 15,100 52% 115,000 28,300 25% 14,100 50% 131,900 32,400 25% 15,500 48% 135,100 37,700 28% 19,800 53% 150,900 40,000 27% 19,800 50% Source: U.S. Consumer Product Safety Commission: National Electronic Injury Surveillance System. Note: The coefficients of variation (CVs) for the all-ages injury estimates range from 9 percent to 12 percent. During this same period, CVs for injury estimates among the younger-than-16 age group and among the younger-than-12 age group range from 9 percent to 15 percent. See Appendix A for an explanation of the use and calculation of CVs. 6 7 See the methodology section in Appendix A for a discussion of the rationale for choosing 2007 as the base year. Estimates have also been adjusted to account for cases that are out of scope for this report. See Appendix A for additional discussion. 15 Figure 3 Annual ATV-Related, Hospital Emergency Department-Treated Injury Estimates for ATVs with 3, 4 or Unknown Number efWheele [200? - 2015] Total Under {L7l C3 Cl C3 C3 200T 2000 2000 2010 2011 2012 2013 2014 2015 Seuree: US. Ceneumer Preduet Safety,r Netienel Eleetrenie Injuryr Suweillenee System. Year 16 The 2015 emergency department-treated injury estimate for all ages reflects an increase of 4 percent from the 2014 estimate. However, this increase is not statistically significant (p-value =0.44).8 The overall decrease of 36 percent between the estimated number of injuries in 2007 and 2015 is statistically significant (p-value < 0.0001). In addition, trend analysis indicates that for ATVs having three, four, or an unknown number of wheels, there is a statistically significant downward trend in emergency department-treated injuries for all ages, collectively, during the years 2007 through 2015 (pvalue =0.0108). The 2015 emergency department-treated injury estimate for children younger than 16 years of age represents a 8 percent increase over the 2014 estimate, although this is not a statistically significant increase (p-value =0.28). The comparison of the 2007 to the 2015 estimated numbers of emergency department-treated injuries for children younger than 16 years of age shows a 33 percent decrease; these two estimates are statistically different (p-value < 0.0001). On average over the period 2007-2015, children younger than 12 years of age represent an estimated 13 percent of emergency department-treated injuries affecting all ages, but in the year 2015 children younger than 12 represent an estimated 14 percent of the injuries (i.e., 134,400/1,038,800 and 13,400/97,200, respectively). About 50 percent of estimated injuries to children under 16 involve victims under the age of 12 for the year 2015 and for the overall period 2007-2015 (134,400/270,400). Figure 4 provides the 2015 estimated number of ATV-related, emergency department-treated injuries broken down by disposition, diagnosis, and body part for all ages. Of the 97,200 estimated ATV-related, emergency department-treated injuries for all ages in 2015, a majority are categorized as treated and released (85 percent). Fourteen percent of the estimated emergency department-treated injuries for all ages are categorized as either treated and admitted or treated and transferred. The remaining percentages of treatments are spread across several categories, such as left without being seen, held for observation, fatalities, and unknown. A plurality of the 2015 estimated ATV-related, emergency department-treated injuries for all ages were diagnosed as contusions/abrasions or fractures (23 percent and 25 percent, respectively). The remaining diagnoses are distributed into categories like lacerations, sprains/strains, internal organ injuries, and other (which includes concussions). The majority of the 2015 estimated ATV-related, emergency-department treated injuries were located on the arm (the shoulder down) or the head or neck, or (29 percent and 29 percent, respectively). 8 The p-value represents the probability of observing results as extreme as, or more extreme than, the results obtained if there is no difference in the two reported estimates. A small p-value (e.g., p-value < 0.05) suggests that an observed difference is likely to reflect a difference between the two estimates that is associated with some factor or factors beyond the inherent variability in the estimates themselves. P-values listed in this report have not been corrected for multiple comparisons. 17 Figure 4: Disposition, Diagnosis, and Body Part ATV-Related Emergency Department-Treated Injuries for 9 All Ages, 2015 ATV-related Emergency Department-Treated Injury Disposition Estimates: All Ages, 2015 13,300, 14% 1,200, 1% Treated and Released Treated and Transferred OR Treated and Admitted Other 82,700, 85% ATV-related Emergency Department-Treated Injury Diagnosis Estimates: All Ages, 2015 18,000, 19% 22,400, 23% 12,500, 13% 24,600, 25% Contusion/Abrasion Fracture Laceration Internal Organ Injury Sprain/Strain Other 9,500, 10% 10,200, 10% ATV-related Emergency Department-Treated Injury Body Part Estimates: All Ages, 2015 *, 1% 28,100, 29% 28,300, 29% 19,600, 20% 20,200, 21% 9 Head and Neck Torso Leg Arm Other Totals may not sum to the 2015 annual estimate due to rounding. Percentages are based on the rounded estimates. Note: Sample size is too small in 2015 to estimate injuries associated with the “other “ body part category. 18 For the 26,700 estimated ATV-related emergency department-treated injuries in 2015 for children younger than 16 years of age, the majority were treated and released (88 percent). Eleven percent of children younger than 16 years of age were either treated and admitted or treated and transferred. The remaining treatments were spread across several categories, such as left without being seen, held for observation, fatalities, and unknown. In 2015, children younger than 16 years of age were diagnosed with contusions/abrasions 29 percent of the time for ATV-related, emergency-department treated injuries and 25 percent of the time with fractures. The remaining diagnoses are distributed into categories such as lacerations, sprains/strains, internal organ injuries, and concussions. Table 6 breaks down the estimated numbers of ATV-related, emergency department-treated injuries from 2007 to 2015 by age group, while Figure 5 gives the corresponding graph of Table 6. Table 6 Annual Estimates of ATV-Related, Emergency Department-Treated Injuries by Age Group 2007–2015 Age Group Year Under 16 2015 16–24 25–34 35–44 45–54 55 + Total 26,700 24,200 19,600 11,700 7,700 7,100 97,200 2014 24,800 22,700 19,800 13,700 7,400 5,300 93,700 2013 25,000 26,400 21,600 12,300 8,000 6,500 99,600 2012 26,500 28,000 24,600 14,000 8,600 6,300 107,900 2011 29,000 27,700 23,000 14,000 8,000 5,900 107,500 2010 28,300 29,500 25,900 15,200 9,000 7,000 115,000 2009 32,400 36,400 30,200 16,600 10,100 6,200 131,900 2008 37,700 35,000 30,700 18,000 8,000 5,800 135,100 2007 40,000 45,800 31,200 17,800 9,600 6,500 150,900 Source: U.S. Consumer Product Safety Commission: National Electronic Injury Surveillance System. Note: Rows may not sum to the annual totals presented elsewhere in this report due to rounding. Analysis of information in Table 6 (see page 17) and Figure 5 supports the following:  Between 2014 and 2015 the only age group with a statistically significant difference in injuries was the 55+ age group, with a significant increase (p-value= 0.0252). However, none of the previous years 2007 through 2013 have estimates for 55+ that are significantly different from the 2015 estimates and so this may just reflect random variation for the year 2014.The younger than 16, 16-24, and 45-54 age groups’ injury estimates also increased from 2014 to 2015, but with no statistical significance (p-value = 0.28, 0.44, and 0.74, respectively). The remaining age groups 25-34 and 35-44 have injury estimates that decreased from 2014 to 2015, but with no statistical significance (p-value = 0.16 and 0.50, respectively).  When comparing the base year 2007 against the 2015 injury estimates, we find statistically significant decreases in injuries for all except the eldest two age groups. While the 45-54 age group’s injury estimate is less in 2015 than in 2007, this is not a statistically significant decrease (p-value=0.16). The 55+ age group is the only age group for which the injury estimate is greater 19 in 2015 than in 2007; however this is not statistically significant increase (p-value = 0.50). For each of the youngest four age groups (younger than 16, 16-24, 25-34, and 35-44) the 2015 estimate is significantly less than the 2007 estimate (p-value <0.0001 for the younger than 16, 16−24, and 25-34 age groups and p-value = 0.0018 for the 35-44 age group). 20 Four-Wheel ATVs Table 7 shows estimates of four-wheel ATV-related, emergency department-treated injuries for the years 2007 to 2015. In 2015, four-wheel ATV injuries constituted 98 percent of the total injury estimate for ATVs having three, four, or an unknown number of wheels (i.e., 95,500/97,200). The four-wheel ATV emergency department-treated injury estimate for 2015 represents a 4 percent increase over the 2014 estimate; however, this decrease is not statistically significant (p-value =0.55). There is a statistically significant difference between the 2007 and 2015 estimates (p-value < 0.0001). There is also a statistically significant linear trend in emergency department-treated injuries associated with fourwheel ATVs from 2007 to 2015 (p-value =0.0120). Table 7 Estimated Number of 4-Wheel ATV-Related Emergency Department-Treated Injuries January 1, 2007 through December 31, 2015 Year 2015 2014 2013 2012 2011 2010 2009 2008 2007 4-Wheel ATV-Related 10, 11 Injury Estimate 95,500 92,200 97,600 105,000 105,500 111,900 128,600 131,700 146,500 Sources: U.S. Consumer Product Safety Commission: National Electronic Injury Surveillance System, the Directorate for Epidemiology/Division of Hazard Analysis. Note: The coefficients of variation (CVs) for four-wheel ATV injury estimates (column 2) all range from 8 to about 11 percent. (See Levenson, (2005) and Garland (2011) for CVs for ATV-related injury estimates). 10 Estimates have also been adjusted by factors to account for cases that are out of scope for this report. Appendix A provides additional detail. 11 Estimates have been rounded. 21 Discussion In analyzing deaths and injuries associated with ATVs, it is useful to consider three distinct periods, the boundaries of which are determined primarily by fatality data availability and by the completeness of the available data. By considering these three periods separately, we can compare years within periods, and thereby control, at least in part, for changes in fatality data availability, as well as for possible changes in the ATV marketplace (see Appendix A). While the boundaries of the periods considered here are defined by factors involving the collection of mortality data, it is also useful to consider the injury estimates within the same periods. The periods selected for discussion were defined as follows:  The first period, from 1982 to 1998, begins with the first year of CPSC staff’s reported ATVrelated death counts (see Table 1); and the first period ends with the ICD-9 to ICD-10 transition for classification of mortality data.  The second period, from 1999 to 2012, begins with the transition to ICD-10 coding and ends with the most recent complete year of death data collection.  The third period, from 2013 to 2015, spans the period of ongoing mortality data collection by CPSC staff. ATV-related deaths and injuries occurring in each of these three periods are reviewed below. 1982–1998 In the first period (1982–1998), reported deaths reached a high of 300 in 1986 (Table 1). These reported deaths were largely associated with three-wheel ATVs, which were still being manufactured and sold. During the mid-1980s, three-wheel ATVs were still heavily in use, and four-wheel ATVs were only beginning to gain in popularity. As previously noted, CPSC staff’s ability to gather death reports during the first period (1982–1998) was limited by the ICD-9 reporting codes and by ICD-9 reporting requirements, which made it difficult for CPSC staff to obtain death certificates for ATV-related fatality incidents occurring on public roads (see Appendix A). Consequently, the death estimates for this period are likely to be underestimated. However, because data collection methodologies were substantially consistent throughout the first period, general comparisons among the annual death estimates within the first period may still be made, if the degree of underestimation is similar from year to year. Other than the ICD-9 coding, CPSC staff is not aware of any factors that would have contributed to an underestimation of ATV-related fatalities in this period. CPSC staff is also not aware of any factors that would have caused differences in the degree of ICD-9-related underestimation in different years. With these cautions in mind, a review of Table 4 suggests that, during the first period, the estimated number of deaths associated with all ATVs (i.e., ATVs having three, four, or an unknown number of wheels) likely peaked around 1986. This peak was followed by a decline in estimated ATV-related fatalities until the early-to-mid-1990s. Then, a general increase in the estimated deaths appears to have occurred from the mid-1990s to the end of this period (1982−1998). As previously mentioned, this reporting period used the ICD-9 reporting requirements, and thus, the death estimates for this period are likely to be underestimated. The reader should use caution when generalizing in this period, due to the impact of this underestimation on the magnitude of the estimates. A similar pattern can be observed in the estimated number of emergency department-treated injuries associated with ATVs having three, four, or an unknown number of wheels (see Table 8, page 29 for 22 injury estimates for this period). That is, the estimated number of ATV-related, emergency departmenttreated injuries appeared to peak during the years 1985 and 1986, when injuries were above 100,000. This was followed by a decline in injury estimates until the early-to mid-1990s, and then by an increase in estimated injuries until the end of the period. The similarities between death and injury data suggest that the pattern seen in the estimated number of deaths is not simply an artifact of the fatality data. 1999–2012 Because of the transition to ICD-10 mortality coding, the second period (1999–2012) reflects several years during which CPSC staff had a greater opportunity to collect comprehensive data on ATV-related fatalities than had been possible prior to 1999. ATV-related regulatory activity began in this reporting period. Consequently, any effect of heightened media exposure on data collection began to be a factor in this period. This effect could have started in 2002, with a petition submitted to the CPSC that requested the Commission to issue a rule banning the sale of adult-size four-wheel ATVs sold for the use of children under the age of 16 years. This effect could have continued throughout this time period due to the exposure resulting from CPSC issuing an advance notice of proposed rulemaking in 2005, and a notice of proposed rulemaking in 2006. Comparing the estimated numbers of deaths associated with ATVs having three, four, or an unknown number of wheels, there is an increase of about 77 percent between 1999 to 2005, followed by a decrease of about 29 percent between 2005 to 2012 (Table 4). While this reflects an increase followed by a decrease within the 1999 to 2012 period, there is an overall 26 percent increase between 1999 and 2012 in estimated deaths associated with ATVs having three, four, or an unknown number of wheels. The estimated numbers of emergency department-treated injuries associated with ATVs having three, four, or an unknown number of wheels increased 40 percent from the 1999 estimate of 82,000 to the 2007 estimate of 150,900 before declining each year thereafter down to an the 2011 estimate of 107,500 (Table 8 in Appendix B, page 29). The 2012 estimate of 107,900 emergency department-treated injuries is slightly higher than the corresponding 2011 estimate, but this does not reflect a statistically significant increase. In 2008, the CPSIA became law. Section 232 of the CPSIA included provisions directing the CPSC to make the voluntary standard, the American National Standard for Four Wheel All-Terrain Vehicles, developed by the Specialty Vehicle Institute of America (American National Standard ANSI/SVIA 1 – 2007), a mandatory standard. The mandatory standard was published in late 2008, and became effective in April 2009. In addition, all companies importing and distributing ATVs in the United States were required by the CPSIA to have action plans12 approved by, and on file with, the Commission. The CPSIA also banned the importation and distribution of three-wheel ATVs. One likely result of the regulatory focus on ATVs may have been an increase in media attention on ATV-related fatalities; and this, in turn, may have resulted in the collection of more complete and timely death reports during these periods. Additionally, during the latter half of this period, and continuing into the next period, the CPSC launched a campaign to increase awareness of ATV safety via television and radio public service announcements (PSAs), created www.ATVSafety.gov, and partnered with organizations and officials dedicated to promoting ATV safety. In 2006, the CPSC’s website, www.ATVSafety.gov, was launched and includes information on ATV safety, state laws and regulations for ATVs, and fatality and injury data. The CPSC’s Office of Information and Public Affairs (OIPA), now the Office of Communications (OCM), initiated a Rapid Response program in April 2007, to respond to ATV-related deaths and injuries. The Rapid Response program is triggered when there is a report of an ATV death or injury. Working with media in the affected area, the CPSC’s OCM publishes information on ATV safety via 12 The term “ATV action plan” means a written plan or letter of undertaking that describes actions the manufacturer or distributor agrees to take to promote ATV safety, including rider training, dissemination of safety information, age recommendations, other policies governing marketing and sale of the ATVs, the monitoring of such sales, and other safety-related measures, and that is substantially similar to the plans described under the heading, “The Undertakings of the Companies in the Commission Notice” published in the Federal Register on September 9, 1998 (63 FR 48199–48204).’ 23 radio and television PSAs, or through news stories. By raising awareness of ATV safety, this campaign was designed to reduce the number of deaths and serious injuries associated with ATVs (U.S. CPSC OIPA, 2006). CPSC staff does not have the data to ascertain the causes of any trend in injuries. Thus, no conclusions can be made to the causes of any trends for any period of time. 2013-2015 The third period (2013–2015) contains 3 years of incomplete death data. As of December 31, 2015, death certificate completion was 95 percent for 2013, 71 percent for 2014, and 35% for 2015. (This is not a percentage of death certificates versus actual deaths; rather, it is the percentage of the combination of states and months where CPSC staff has received at least one death certificate.) The number of reported deaths for these years will increase as CPSC staff receives additional reports. For this reason, the 2013 through 2014 estimated numbers of deaths will require revision for these years in future CPSC staff reports. Initial estimates of deaths for 2015 will be provided in next year’s report. During the third period, emergency department-treated injury estimates for all ages do not exhibit significant change, both for ATVs having three, four, or an unknown number of wheels (Table 5), and for four-wheel ATVs alone (Table 7). For ATVs having three, four, or an unknown number of wheels, the overall decrease from 2013 to 2015 is not statistically significant (99,600 and 97,200, respectively; p-value = 0.77). Similarly, the decrease, when comparing the estimated number of injuries associated with four-wheel ATVs from 2013 to 2015, is also not statistically significant (97,600 and 95,500, respectively; p-value = 0.79). However, the data-collection process supporting the derivation of the injury estimates is complete for all reported years, including this period (2013–2015). Thus, the injury estimates for 2013 through 2015 are not expected to require revision in future CPSC staff reports. In the period 2013−2015, the CPSC continued maintenance of the ATVSafety.gov website and deployment of the Rapid Response media awareness program. CPSC also continued to promote ATV safety awareness with annual news releases aimed to coincide with the Memorial Day holiday weekend, which has high rates of ATV-related deaths and injuries and marks the beginning of the recreational riding season. In 2013, the ATVSafety.gov website was redirected into CPSC’s main website as the “ATV Safety Information Center.” Additionally, CPSC published new ATV safety educational products, including a Web-based image and a print-ready poster presenting injury and fatality statistics with graphic illustrations. 24 Appendix A: Estimation Methodologies This appendix describes the methodologies used to estimate ATV-related deaths and injuries and other information to develop the report analyses. ATV-Related Deaths In-Scope ATV-Related Fatalities ATV-related fatalities considered in scope in this report include any unintentional incident involving an ATV, whether or not the ATV was in operation at the time of the incident. Because of the difficulties inherent in distinguishing between occupational and nonoccupational use, occupational fatalities are included in both the death counts and the death estimates. For example, it may be difficult to classify a fatality that occurs when a victim is riding next to a fence on a ranch to examine the fence, and subsequently is involved in an ATV-related fatality incident while deviating from his work to take a recreational ride up a nearby hill. ICD-9 Versus ICD-10 Coding In 1999, CPSC staff began collecting death certificates for all fatalities in which an external cause of death listed on the death certificate was reported to involve an ATV, as coded under ICD-10. ICD-10 marks the first revision in which all ATV-related fatalities are grouped under a single code (V86.X), thereby facilitating more complete collection of these incidents by CPSC staff than was accomplished before 1999. It should be noted that the ICD-10 codes (V86.X) characterizing the external cause of death as “ATVrelated,” include fatalities resulting from all specialty motor vehicles intended primarily for off-road use (World Health Organization, 2007). Thus, other types of off-road vehicles are captured in this set of codes, such as dune buggies, ROVs, UTVs, and dirt bikes. By conducting in-depth investigations (IDIs), CPSC staff attempts to verify that the vehicles involved in these incidents were “ATVs,” as defined by CPSC staff (i.e., an ATV is a motorized vehicle intended for off-road use and having three or four low pressure tires, a straddle seat for the operator, and handlebars for steering control). In cases where the specific type of off-road vehicle cannot be ascertained, CPSC staff counts the death report as an ATVrelated fatality. This assumption may result in an overestimation of ATV-related deaths. Estimation of ATV-Related Fatalities (1999–Present) CPSC staff estimates the number of deaths associated with ATVs by use of a capture-recapture approach. This approach involves examining the numbers of reports of ATV-related fatalities gathered via two different avenues:  The first avenue is the collection of death certificates obtained by CPSC staff,13 where the death is deemed by the medical examiner to be ATV related. These incidents are entered into CPSC staff’s death certificate database (DTHS).  The second avenue involves the collection of reports of fatal ATV-related incidents by any other means available to CPSC staff (denoted non-DTHS). Sources of these types of reports include news clips; reports from the Medical Examiners and Coroners Alert Project (MECAP); reports 13 CPSC staff purchases death certificates from the 50 states, the District of Columbia, Puerto Rico, and New York City for fatalities involving selected consumer products, including ATVs. Determination of the association between a fatality and a consumer product is based on the external cause of death code(s) reported on the death certificate. Since 1999, the external causes of death reported on death certificates have been coded in accordance with ICD-10 (National Center for Health Statistics, 2007). 25 from consumers or their representatives via telephone or the Internet (www.SaferProducts.gov); and hospital reports from the National Electronic Injury Surveillance System (NEISS). It should be noted that the NEISS database primarily includes product-related injuries rather than fatalities. However, all ATV-related NEISS cases are reviewed to identify incidents where an emergency department-treated, ATV-related injury was reported to result in death; and these deaths are included in the ATV-related fatality reports available to CPSC staff. In many cases, CPSC staff receives fatality reports for the same incident from multiple sources. The reports are either about deaths counted in a previous annual report, or deaths reported for the first time in this annual report. For example, CPSC staff may receive a MECAP report for a fatality that previously was reported to CPSC staff via a news clip. Reports from non-DTHS sources are reviewed carefully to match multiple source reports about the same incident for both the current reporting year and for previous years. Incidents that have been reported in multiple sources are counted only once in Table 1. The calculation of the capture-recapture estimate entails the matching of fatality reports from DTHS and non-DTHS sources. Then, for each year of interest, CPSC staff determines the total number of fatalities included in DTHS, the total number of fatalities included in non-DTHS sources, and the total number of deaths included in both sources (i.e., DTHS and non-DTHS). Thereafter, the estimate is calculated using the following equation (Hook and Regal, 1992; Morrison and Stone, 2000; Hook and Regal, 2004): 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒 = (𝑀 + 1)(𝑁 + 1) −1 𝑛+1 Equation 1 where 𝑀 is the number of incidents captured by purchase of death certificates from the states (DTHS); 𝑁 is the number of incidents collected by other means (non-DTHS); and 𝑛 is the number of incidents captured by both death certificate purchase and at least one other source. Estimates of fatalities that occurred on or after January 1, 1999, and that are associated with ATVs having three, four, or an unknown number of wheels are calculated using equation 1. Estimation of ATV-Related Fatalities (1985–1998) Before 1999, CPSC staff received death certificates for only two types of ATV-related fatality incidents: (1) ATV-related fatalities occurring in places other than public roads, and (2) ATV-related fatalities occurring on public roads that were erroneously reported as nonpublic roads. Because of this, the capture-recapture procedure for estimating pre-1999 ATV-related deaths had two parts:  For public road fatalities, the estimates were based solely on the number of fatalities reported to CPSC staff. Reports of these fatalities were largely contained in CPSC staff’s Injury or Potential Injury Incident file (IPII). As noted, death certificates generally were not received for these fatalities.  For incidents occurring in other places, the capture-recapture approach was applied. Using equation 2, these two parts (i.e., incidents occurring on public roads and incidents occurring in other places) were combined to derive the pre-1999 annual estimates of ATV-related deaths for ATVs having three, four, or an unknown number of wheels. 26 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒 = (𝑀𝑁𝑃 + 1)(𝑁𝑁𝑃 + 1) − 1 + 𝐶𝑃 𝑛𝑁𝑃 + 1 Equation 2 where 𝑀𝑁𝑃 is the number of reports of nonpublic-road fatalities captured by purchase of death certificates from the states; 𝑁𝑁𝑃 is the number of reports of nonpublic-road fatalities collected by other means; 𝑛𝑁𝑃 is the number of reports of nonpublic-road fatalities captured by both death certificate purchase and at least one other source; and 𝐶𝑃 is the count of reports of ATV-related fatalities occurring on public roads from any source. CPSC staff believes that the ATV-related fatality estimates for the years before 1999 were likely to be underestimated because the pre-1999 estimates used only the available counts of public road fatalities and did not account for missing reports in these types of incidents. As noted previously, CPSC staff now receives death certificates for ATV-related incidents occurring on public roads. Consequently, since 1999, the capture-recapture approach has been applied fully to both components (i.e., incidents occurring on public roads and incidents occurring in other locations) of the annual estimates of ATVrelated deaths. For this reason, CPSC staff expects that the annual death estimates for 1999 and later represent better estimates of ATV-related fatalities than were possible in the years before 1999. Estimation of Fatalities Associated with Four-Wheel ATVs A number of incidents reported to CPSC staff involve ATVs for which the number of wheels is unknown. Because some of these likely involve four-wheel ATVs, the unknowns are apportioned in the calculation of the estimated number of deaths associated with four-wheel ATVs. This estimate is calculated by first dividing the reported number of deaths for four-wheel ATVs by the combined reported number of deaths for three- and four-wheel ATVs, and then multiplying this quotient by the estimated number of deaths for all ATVs (three, four, or unknown number of wheels). Thus, the estimate of deaths associated with four-wheel ATVs is given by equation 3. 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒4𝑊 = 𝑟𝑒𝑝4𝑊 𝑒𝑠𝑡3𝑊+4𝑊+𝑈𝑊 𝑟𝑒𝑝3𝑊+4𝑊 Equation 3 where 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒4𝑊 is the estimated number of fatalities associated with four-wheel ATVs; 𝑟𝑒𝑝4𝑊 is the reported number of fatalities associated with four-wheel ATVs; 𝑟𝑒𝑝3𝑊+4𝑊 is the reported number of fatalities associated with three- and four-wheel ATVs; and 𝑒𝑠𝑡3𝑊+4𝑊+𝑈𝑊 is the estimated number of fatalities associated with ATVs having three, four, or an unknown number of wheels. [Note: this is the “estimate” derived in equations 1 and 2]. 27 ATV-Related Injuries Estimation of Emergency Department-Treated Injuries Associated with ATVs All injury estimates in this report have been derived from data collected through the CPSC’s NEISS, a probability sample of U.S. hospitals with 24-hour emergency departments with more than six beds (Schroeder and Ault, 2001a and 2001b). Thus, ATV-related injury estimates in this report represent hospital emergency department-treated injuries only. ATV-related injuries that were not treated in hospital emergency departments are not included in these estimates. Injury estimates have been adjusted to reflect revisions in the NEISS Coding Manual in 1985, as well as to account for NEISS sampling frame updates (Marker et. al., 1988; Marker and Lo, 1996). Estimates for 1982 through 1985 also were adjusted, based on a review of NEISS comments, to exclude dune buggies and include ATVs that had been misclassified as mini or trail bikes. Injury estimates for 1985, 1989, 1997, and 2001 are based on injury surveys using NEISS cases. Injury estimates for 2010 are based on a partial-year study of surveys of NEISS cases. Injury estimates for other years have been adjusted by factors to account for out-of-scope (occupational, intentional, and/or non-ATV) cases, based on injury studies in these years (Garland, 2011; Levenson, 2003b; Levenson 2005). An “in-scope injury case” is defined to be any nonoccupational, unintentional case involving an ATV, whether or not the victim was operating the ATV at the time of the incident, i.e., the victim could have been a passenger or a bystander. Note that NEISS does not collect occupational injuries; and, thus, the definition of “in-scope, ATV-related injuries,” differs slightly from the definition of “in-scope, ATV-related fatalities.” The applied adjustment factors were as follows: 0.93 for 1986 through 1988, 0.95 for 1990 through 1996, 0.903 for 1998 through 2000 (amended from 0.935), 0.922 for 2001 through 2009, and 0.899 for 2010 through present. As the market and ridership changes for off-road vehicles, including the substantial increase of ROVs in use, the adjustment factors may not reflect the changes in injuries due to different types of vehicles as time progresses. Each adjustment factor is calculated based on a special study performed in the NEISS (Levenson, M. (2003b), Garland, S. (2011)). Each adjustment factor reflects the year in which the special study was performed, and each is applied in subsequent years until another special study is performed. Thus, if there have been substantial changes to the records that would be considered out of scope, this cannot be reflected on a year-to-year basis, nor can it be determined if this is occurring. For example, if the increase in the number of ROVs in use is impacting the injury estimates, specifically if records coded in the NEISS as ATVs are increasingly actually related to ROVs, then ATV-related injuries are being overestimated. Again, it is unknown if this is occurring in the data; however, this is a possibility with using an adjustment factor that cannot be updated yearly, but only periodically. Coefficients of Variation A coefficient of variation (CV) is an expression of the standard deviation relative to the estimate itself. In this report, CVs for injury estimates are given as percentages. The adjustment factors discussed above are also estimated and have associated variability. This variability (along with the variability of the injury estimates) affects significance tests and tests for trends. Calculation of NEISS estimates and their variances is discussed in Schroeder and Ault (2001a) and Schroeder and Ault (2001b). Adjustment factors and other concepts specific to variability associated with ATV estimates are discussed more fully in Levenson (2003b, 2005) and Garland (2011). An alternative ante-dependence model for the variance-covariance matrix was used in modelling the trend for ATVs having three, four, or an unknown number of wheels for all ages, collectively during the years 2007 through 2014. The unstructured variance-covariance matrix was used in modelling the trend for ATVs having four wheels for all ages, collectively during the years 2007 through 2014. 28 Estimation of Emergency Department-Treated Injuries Associated with Four-Wheel ATVs NEISS includes injuries that are associated with ATVs for which the number of wheels is unknown. Because of this, the estimated injuries associated with ATVs having an unknown number of wheels are apportioned in the calculation of the estimated injuries associated with four-wheel ATVs, using equation 4. 𝑡𝑜𝑡𝑎𝑙 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒4𝑊 = 𝑒𝑠𝑡4𝑊 (𝑒𝑠𝑡3𝑊 𝑒𝑠𝑡3𝑊 +𝑒𝑠𝑡4𝑊 + 𝑒𝑠𝑡4𝑊 + 𝑒𝑠𝑡𝑈𝑊 ) Equation 4 where 𝑡𝑜𝑡𝑎𝑙 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒4𝑊 is the total estimated injuries associated with four-wheel ATVs with unknowns apportioned; 𝑒𝑠𝑡4𝑊 is the injury estimate associated with four-wheel ATVs (excluding unknowns); 𝑒𝑠𝑡3𝑊 is the injury estimate associated with three-wheel ATVs (excluding unknowns); and 𝑒𝑠𝑡𝑈𝑊 is the injury estimate associated with ATVs with an unknown number of wheels. 29 Appendix B Table 8 Historical ATV-Related Emergency Department Treated Injury Estimates for ATVs with 3, 4, or Unknown Number of Wheels and for 4-wheel ATVs from 1985 to 2015 Estimated Number Estimated number Estimated of Injuries: of Injuries: Number of All Ages Under 16 years Injuries: (3, 4, and unknown (3, 4, and unknown All Ages Year number of wheels) number of wheels) (4-wheel ATVs) 2015 97,200 26,700 95,500 2014 93,700 24,800 92,200 2013 99,600 25,000 97,600 2012 107,900 26,500 105,000 2011 107,500 29,000 105,500 2010 115,000 28,300 111,900 2009 131,900 32,400 128,600 2008 135,100 37,700 131,700 2007 150,900 40,000 146,500 2006 146,600 39,300 140,900 2005 136,700 40,400 130,000 2004 136,100 44,700 129,500 2003 125,500 38,600 116,600 2002 113,900 37,100 104,800 2001 110,100 34,300 98,200 2000 92,200 32,000 82,300 1999 82,000 27,700 68,900 1998 67,800 25,100 57,100 1997 52,800 20,600 39,700 1996 53,600 20,200 40,700 1995 52,200 19,300 36,200 1994 50,800 21,400 33,300 1993 49,800 17,900 32,000 1992 58,200 22,000 33,000 1991 58,100 22,500 34,400 1990 59,500 22,400 30,800 1989 70,300 25,700 35,700 1988 74,600 28,500 39,400 1987 93,600 38,600 33,900 1986 106,000 47,600 23,400 1985 105,700 42,700 14,700 Sources: U.S. Consumer Product Safety Commission: National Electronic Injury Surveillance System, the Directorate for Epidemiology/Division of Hazard Analysis. 30 References Garland, S. (2011) 2010 ATV Injury Adjustment Factor, 2010 Adjustment Factor Standard Error, and 2010 Adjustment Factor Implementation in ATV-related Injury Analyses, memorandum. June 16, 2011. U.S. Consumer Product Safety Commission. Hook, E.B. and Regal, R.R. The Value of Capture-Recapture Methods Even for Apparent Exhaustive Surveys. American Journal of Epidemiology. 1992; 135(9):1060–1067. Hook, E.B. and Regal, R.R. Completeness of Reporting: Capture-Recapture Methods in Public Health Surveillance. From: Monitoring the Health of Populations: Statistical Principles and Methods for Public Health Surveillance, Brookmeyer, R. and Stroup, D. F., eds. 2004. Oxford University Press, New York. Levenson, M. (2003a) All-Terrain Vehicle 2001 Injury and Exposure Studies. January 2003. U.S. Consumer Product Safety Commission. Levenson, M. (2003b) ATV Injury Adjustment Factors for 1997 and 2001, memorandum. September 12, 2003. U.S. Consumer Product Safety Commission. Levenson, M. (2005) Uncertainty Estimates of ATV Adjusted Injury Estimates, memorandum. August 22, 2005. U.S. Consumer Product Safety Commission. Marker, D., Waksberg, J., and Braden, J. NEISS Sample Update. June 3, 1988. Westat, Inc. Marker, D. and Lo, A. Update of the NEISS Sampling Frame and Sample. October 11, 1996. Westat, Inc. Morrison, A. and Stone, D.H. Capture-Recapture: A Useful Methodological Tool for Counting TrafficRelated Injuries? Injury Prevention. 2000; 6:299–304. National Center for Health Statistics (NCHS) “About the International Classification of Diseases/Clinical Modification (ICD-10-CM).” August 2007. http://www.cdc.gov/nchs/icd.htm. Schroeder, T. Trend Analysis of NEISS Data. February 2000. U.S. Consumer Product Safety Commission. Schroeder, T. and Ault, K. (2001a) The NEISS Sample (Design and Implementation) from 1979 to 1996. June 2001. U.S. Consumer Product Safety Commission. Schroeder, T. and Ault, K. (2001b) The NEISS Sample (Design and Implementation) from 1997 to the Present. June 2001. U.S. Consumer Product Safety Commission. U.S. Consumer Product Safety Commission. Standards for All-Terrain Vehicles and Ban of ThreeWheeled All-Terrain Vehicles; Notice of Proposed Rulemaking. Federal Register, Volume 71, Number 154, Pages 45904–45962. August 10, 2006. U.S. Consumer Product Safety Commission. 2014 Annual Report of ATV Deaths and Injuries. November 2015. U.S. Consumer Product Safety Commission, Office of Information and Public Affairs (OIPA) (September 21, 2006). CPSC Launches Major Campaign to Drive Down Deaths and Serious Injuries Associated with All-Terrain Vehicles. Press Release. Retrieved September 16, 2009. 31 World Health Organization (WHO), “International Classification of Diseases.” Retrieved December 4, 2007 from: http://www.who.int/classifications/icd/en/. 32