Amusement Ride-Related Injuries and Deaths in the United States: 1987-1999 July 2000 C. Craig Morris, Ph.D. U.S. Consumer Product Safety Commission Directorate for Epidemiology Division of Hazard Analysis 4330 East West Highway Bethesda, MD 20814 Page 1 Executive Summary This report describes U.S. Consumer Product Safety Commission (CPSC) data on amusement ride-related fatalities and hospital emergency room-treated injuries. Fatality data are presented for calendar years 1987 through 1999. Injury data are presented for calendar years 1993 through 1999. Hazard scenario data derived from in-depth investigations and incident reports are presented for the period from 1 Jan 1990 through 30 April 2000. • An estimated 10,400 hospital emergency room-treated injuries occurred in 1999, with about 7,000 involving fixed rides and 3,000 involving mobile rides. An estimated 23.5 injuries per million attendance occurred at fixed-site parks in 1999. Attendance data are not currently available for mobile rides. • There was a marginally significant upward trend in fixed-site and total amusement ride- related injuries from 1993 through 1999 due to a sharp increase in fixed-site injuries beginning in 1997. • The estimated annual average number of non-occupational amusement ride-related fatalities from 1987 through 1998 was 4.3 fatalities each year, and there was a marginally significant upward trend in fatalities from 1993 through 1999. • A hazard sketch summarizing CPSC investigations of amusement ride incidents revealed hazard patterns associated with mechanical failure, operator behavior, consumer behavior, other factors, and combinations of these factors. Page 2 Table of Contents Page Executive Summary 2 Introduction and Definition of Amusement Rides 4 Amusement Ride-Related Injuries 4 Amusement Ride-Related Fatalities 9 Hazard Patterns Associated with Amusement Ride-Related Incidents 12 References 13 Appendix A – Technical Note 14 Appendix B – Reported Fatalities 15 Appendix C – Investigated Cases 26 Page 3 Amusement Ride-Related Injuries and Deaths in the United States: 1987-1999 This report describes U.S. Consumer Product Safety Commission (CPSC) data on amusement ride-related fatalities and hospital emergency room-treated injuries. Fatality data are presented for calendar years 1987 through 1999. Injury data are presented for calendar years 1993 through 1999. A hazard sketch is presented to summarize in-depth investigations and incident reports of amusement ride incidents from 1 Jan 1990 through 30 April 2000. This report replaces previous CPSC reports summarizing amusement ride incidents [1,2,3,4]. Definition of Amusement Rides Section 3(a)(1) of the Consumer Product Safety Act describes an amusement ride as: ...any mechanical device which carries or conveys passengers along, around, or over a fixed or restricted route within a defined area for the purpose of giving its passengers amusement, which is customarily controlled or directed by an individual who is employed for that purpose and who is not a consumer with respect to that device, and which is not permanently fixed to a site. Although fixed-site amusement rides are excluded from CPSC jurisdiction by Section 3 (a) (1), CPSC data collection systems receive data on amusement rides that do not, at the outset, distinguish between fixed-site and mobile rides. Additional analysis must be undertaken in order to determine which incidents involve mobile rides and whether a fixed-site ride that was the subject of an incident is also a mobile ride. Data regarding fixed-site ride incidents are included in this report, along with data on mobile ride incidents, since those data are immediately available to CPSC through the process described above and are useful for comparison purposes. Only non-occupational incidents, which involve non-employee victims injured while on, in, or around an amusement ride, are included in this report. Amusement Ride-Related Injuries Method Data on non-occupational amusement ride-related injuries were obtained from CPSC's National Electronic Injury Surveillance System (NEISS). The NEISS is a stratified probability sample of hospitals with emergency rooms and 6 or more beds in the United States. There are currently 100 participating hospitals in the NEISS. NEISS hospital coders identify injury incidents associated with amusement rides by using the NEISS product code for amusement rides (1293). In this analysis, all NEISS records for calendar years 1993 through 1999 containing product code 1293 were reviewed. Based on information in narrative comments in the records, a single experienced coder (the author) classified each case into 1 of 5 mutually exclusive and exhaustive categories: not a ride (out of scope), fixed-site ride, mobile-site ride, unknown-site ride, or unknown if ride. Cases involving coin-operated rides or free-play attractions often found at restaurants or shopping centers, alpine and water slide amusements, wave machines, "moon walks," inflatable slides, mechanical bulls, playground equipment, etc., are examples of cases coded not a ride. Cases involving roller coasters or “whirling” rides are examples of cases coded fixed-site, mobile-site, or unknown-site ride: if the comment stated the name of an amusement park or that the incident occurred at a park or involved an amusement park ride, then the case was coded fixed-site; if the comment stated that the incident occurred at a carnival, fair, Page 4 or festival, then the case was coded mobile-site; if the comment gave no site information, then the case was coded unknown-site. Cases involving a “merry go round,” with no indication of whether it was playground equipment or an amusement ride as defined by the Consumer Product Safety Act, are examples of cases coded unknown if ride. Estimate Adjustments Adjustments of amusement ride-related injury estimates for prior years appear in this report. First, all NEISS cases from 1993 to 1999 were reviewed by an experienced coder to verify consistency in the coding of cases. A few codings were adjusted to improve consistency. Second, the NEISS sample of hospitals was updated in 1997 to reflect changes in the distribution and size of emergency room hospitals in the U.S. since the previous sample update in 1990. Periodic updates are required for the sample of NEISS hospitals to accurately represent the universe of hospitals with emergency rooms in the U.S. Data were collected concurrently from both the old and updated NEISS samples for 9 months in 1997 to provide a statistical basis for adjustments of prior estimates. Third, unknown-site estimates were allocated to fixed-site and mobile-site estimates in the proportions observed in known-site estimates each year. The adjusted estimates in this report differ from estimates given in previous CPSC publications. Results Figure 1 gives NEISS estimates with unknown-site estimates allocated to the fixed-site and mobile-site estimates in the proportions observed in known-site estimates each year. For example, the fixed-site estimate for 1999 is .699 of the sum of the fixed-site and mobile-site estimates, so .699 of the unknown-site estimate for 1999 is added to the fixed-site estimate, and .301 of the unknown-site estimate for 1999 is added to the mobile-site estimate. The proportions of unknown-site estimates allocated to fixed-site versus mobile-site estimates each year depend only on estimates for each year as in the above example for 1999. An estimated total of 10,400 – 5,600 hospital emergency room-treated injuries occurred in 1999. (The margin of error is 5,600 for a 95% confidence level). The increasing trends in fixed-site and total injuries from 1993 through 1999 were marginally significant, both p = .07, 1-tail exact Kendall Tau tests. 12,000 11,000 Total 10,000 9,000 8,000 Fixed-Site 7,000 6,000 5,000 4,000 Mobile Estimated Injuries 3,000 2,000 1,000 0 1993 1994 1995 1996 1997 1998 1999 Year Figure 1. Estimated amusement ride-related injuries treated at US hospital emergency rooms from 1993 through 1999. Source: U.S. Consumer Product Safety Commission. Page 5 Table 1 gives the unallocated NEISS estimates for calendar years 1993 through 1999. Note that the fixed-site and mobile-site injury estimates in Table 1 underestimate actual totals due to the exclusion of unknown-site cases, all of which involve fixed-site or mobile amusement rides. The adjusted estimates in Figure 1 correct for the underestimation of fixed-site and mobile injuries in Table 1 by allocating the unknown-site estimates to the known fixed-site and mobile estimates in the same proportions observed in the known-site estimates (see Table 2). Table 1 Unallocated Estimated Non-Occupational Amusement Ride Injuries by Year and Site of Ride Year Fixed Mobile Unknown-Site Total ME (Total) 1999 5,980 2,580 1,820 10,380 5,560 1998 4,760 2,110 2,500 9,370 3,950 1997 4,590 2,170 1,280 8,050 4,600 1996 3,220 2,530 900 6,650 4,010 1995 3,530 2,680 1,330 7,540 4,340 1994 3,080 2,420 1,260 6,760 3,930 1993 3,750 2,240 1,710 7,700 3,980 Note: Details may not sum to totals due to rounding. Margin of error (ME) for 95% confidence level. Source: U.S. Consumer Product Safety Commission. Table 2 gives the allocated NEISS estimates for calendar years 1993 through 1999 presented graphically in Figure 1. As explained above, the adjusted estimates in Table 2 correct for the underestimation of fixed-site and mobile injuries in Table 1 by allocating the unknown- site estimates to the known fixed-site and mobile estimates in the same proportions observed in the known-site estimates. Table 2 Allocated Estimated Non-Occupational Amusement Ride Injuries by Year and Site of Ride Year Fixed Mobile Total ME (Total) 1999 7,260 3,120 10,380 5,560 1998 6,500 2,870 9,370 3,950 1997 5,460 2,580 8,050 4,600 1996 3,720 2,930 6,650 4,010 1995 4,290 3,260 7,540 4,340 1994 3,790 2,970 6,760 3,930 1993 4,830 2,880 7,700 3,980 Note: Details may not sum to totals due to rounding. Margin of error (ME) for 95% confidence level. Source: U.S. Consumer Product Safety Commission. Page 6 Table 3 gives estimated fixed-site injuries divided by estimated attendance at amusement parks [7] each year. The upward trend in risk is marginally significant, p < .07, 1-tail exact Kendall Tau test. Unknown-site estimates were allocated to fixed-site and mobile estimates as described above. Attendance data for mobile rides are not available. Table 3 Risk of Non-Occupational Fixed-Site Amusement Ride Injury by Year Year Attendance (millions) Risk 1999 309 23.5 1998 300 21.6 1997 300 18.2 1996 290 12.8 1995 280 15.3 1994 267 14.2 1993 275 17.5 Note: Risk is number of injuries per million attendance. Source: U.S. Consumer Product Safety Commission and International Association of Amusement Parks and Attractions. Because an increasing trend in amusement ride-related injuries began in 1997, the same year the NEISS sample of hospitals was updated, it is important to determine whether the increase might be attributable to revision of the NEISS sample, i.e., replacement of about 30 hospitals in the sample of about 100 hospitals. Hospitals continuously present in NEISS from 1995 through 1999 were identified to determine the trend in amusement ride-related injuries at those hospitals. Table 4 gives sample counts, not estimates, from these hospitals. Although the total count dropped from 121 in 1995 to 95 in 1996, it steadily increased to 117 in 1997, 124 in 1998, and 167 in 1999. The increasing trend from 1996 to 1999 was statistically significant by an exact Kendall Tau 1-tailed test, p = .04. Thus, the trend in fixed-site and total amusement ride-related injuries does not appear to be an artifact of the updated NEISS sample in 1997. Table 4 Non-Occupational Amusement Ride-Related Injury Counts for Hospitals in NEISS from 1995 through 1999 by Year and Site of Ride Year Fixed Mobile Unknown Total 1999 87 44 36 167 1998 62 29 33 124 1997 63 29 25 117 1996 49 33 13 95 1995 60 43 18 121 Source: U.S. Consumer Product Safety Commission. Page 7 Analysis of individual NEISS hospital data for calendar years 1995 through 1999 revealed one hospital (#58) that both records more amusement ride-related cases each year and exhibits a larger increasing trend over the period than all other NEISS hospitals. However, as discussed in Appendix A, reporting amusement ride-related injury estimates with hospital 58 excluded is inappropriate because the NEISS sample with hospital 58 removed is not statistically representative of all emergency room hospitals in the U.S. In conclusion, current NEISS data indicate a total in 1999 of about 10,400 – 5,600 emergency room-treated, amusement ride-related injuries, with about 7,000 involving fixed-site rides and about 3,000 involving mobile rides (Figure 1). From 1993 through 1999, there was a marginally significant upward trend in fixed-site and total amusement ride-related injuries and in the risk of fixed-site injury defined as the estimated number of injuries per million attendance at amusement parks. It should be noted that, although there are fewer mobile as compared to fixed- site ride injuries, the amount of exposure is probably different for mobile versus fixed-site rides, so the relative risk of injury attributable to one versus the other cannot be determined without comparable exposure information for both types of rides. Page 8 Amusement Ride-Related Fatalities CPSC's files were searched for records of fatalities involving amusement rides during calendar years 1987 through 1999. These files were the death certificate file (DTHS), the injury or potential injury incident file (IPII), and the National Electronic Injury Surveillance System file (NEISS). Information in the narrative field of the records was used to classify cases using the criteria defined above in the Method section for the NEISS injury analyses. DTHS and IPII files recorded 49 non-occupational fatalities from 1987 through 1999; NEISS files recorded no amusement ride-related fatalities during this period. Appendix B lists the 49 documented non-occupational fatality cases. Due to the logistical difficulty of capturing all fatalities related to any product or event, these counts do not account for all amusement ride- related fatalities. Methods for estimating true totals given 2 or more independent data sources are known as “capture-recapture” or “multiple record systems” methods [8,9]. Capture- recapture analyses were conducted on 43 documented cases during 1987 through 1998 to estimate the total and annual average number of non-occupational amusement ride-related fatalities during that period. Lags in fatality reporting preclude estimation of fatalities for the last full calendar year. The capture-recapture analyses treated the DTHS and IPII files as independent samples of all non-occupational amusement ride-related deaths during the study period. The estimated number of non-occupational amusement ride-related fatalities from 1987 through 1998 was 52 (95% confidence interval = 43 to 62), for an estimated annual average of 4.3 fatalities each year during that period. Documented non-occupational fatalities from 1987 through 1999 were classified by ride type, location (state), and ride site. Site refers to fixed-site rides as in amusement parks, mobile rides as in carnivals or fairs, or rides of unknown site. From 1987 through 1999, there were 49 documented fatalities, including 30 fixed ride-related fatalities, 8 mobile ride-related fatalities, and 11 fatalities involving unknown ride mobility status. The following tables give documented fatalities by year and ride mobility (Table 4), ride type and mobility (Table 5), and state and ride mobility (Table 6). Page 9 Table 4 gives fatalities by year and site. The fatalities observed in 1998 (7) and 1999 (6) were the most since 1988. From 1993 through 1999, there was a marginally significant upward trend in both fixed-site and total fatalities, both p = .08, exact Kendall Tau 1-tailed tests. Table 4 Non-Occupational Amusement Ride-Related Fatalities by Year and Ride Site Year Fixed Mobile Unknown Total 1999 6 0 0 6* 1998 3 2 2 7* 1997 1 0 3 4 1996 2 1 0 3 1995 3 1 0 4 1994 2 0 0 2 1993 1 1 2 4 1992 0 2 0 2 1991 3 0 0 3 1990 0 0 0 0 1989 3 0 0 3 1988 2 1 4 7 1987 4 0 0 4 Total 30 8 11 49 * Data for 1998-1999 are incomplete; counts will increase if additional reports are received. Source: U.S. Consumer Product Safety Commission. Table 5 gives documented non-occupational amusement ride fatalities by type and mobility. The majority of deaths were associated with roller coasters and “whirling” rides. Table 5 Non-Occupational Amusement Ride-Related Fatalities from 1987 through 1999 by Ride Type and Mobility Type of Ride Fixed Mobile Unknown Total Roller Coaster 12 0 3 15 Whirling 2 4 4 10 Water 5 0 0 5 Train 2 1 0 3 Ferris Wheel 2 0 0 2 Sleigh 1 0 0 1 Unknown 6 3 4 13 Total 30 8 11 49 Note: Data for 1998-1999 are incomplete; counts will increase if additional reports are received. Source: U.S. Consumer Product Safety Commission. Page 10