Characteristics of Emerging Road and Trail Users and Their Safety

SAFETY/CRASH DATA AVAILABILITY

As part of this study, potential sources of information regarding the safety of emerging road and trail users were identified. The primary data sources examined were:

• Data from the NEISS.

• Other hospital discharge, trauma registry, or emergency department (ED) data.

• State motor vehicle crash databases.

• State narrative crash data.

This section presents relevant information on each of these sources and discusses their advantages and disadvantages.

The Fatality Analysis Reporting System (FARS) contains information on motor vehicle crashes that result in a fatality to either a vehicle occupant or a nonoccupant (such as a pedestrian) within 30 days of the crash. However, the majority of crashes are not fatal and thus are not included in FARS. Moreover, FARS does not indicate whether a person was using a wheelchair, inline skates, or another emerging user device. Therefore, FARS was not considered to be a potential source of information and is not discussed in this section.

NEISS Data

NEISS was felt to be by far the most useful data source available for studying the safety of the identified emerging road and trail users. NEISS is an injury surveillance data collection system that is operated by the U.S. CPSC. It is currently based on a statistically valid sample of 100 hospital EDs nationwide. NEISS has been operational for 30 years, and recently (in 2000) expanded its scope to collect data on all injuries, rather than just those related to consumer products. Reported cases (generally between 200,000 and 300,000 a year before the recent expansion) are weighted to provide overall national estimates of injuries serious enough to require ED treatment.

Data are collected electronically at participating hospitals and immediately forwarded to CPSC. The data collection protocol includes information on the injury victim's age, gender, race, injury diagnosis, ED disposition (treated and released, or admitted to the hospital), and the locale of the accident (home, farm/ranch, street or highway, school, place of recreation or sports, etc.). While the latter does not specifically include off-road trails, communication with CPSC revealed that this information might be captured in the 144-character narrative descriptions accompanying the reports.

What makes the NEISS data especially valuable is the level of detail captured with regard to involved products. Currently, more than 900 different products can be identified. From the coding manual, it is not clear how some of the newer assistive vehicle types such as powered scooters, hand cycles, and powered wheelchairs would be identified, or how bicycle trailers are coded. Also, jogging strollers are not differentiated from conventional strollers. As in the case of location information, further detail may be available in the report narratives. Bicycle trailers have been studied retrospectively using NEISS data.⁽²⁰⁾ Additionally, if a subject was not using a device, but rather was injured by someone else using the device, that information might not be coded in the database; a review of narratives would be required to obtain this data.

NEISS data are publicly available, and CPSC has a long history of cooperation with the Centers for Disease Control and Prevention (CDC), National Highway Traffic Safety Administration (NHTSA), and other Government agencies in using the data for research purposes other than identifying potential product hazards. In addition to reports taken from the computerized database, the CPSC regularly conducts special "follow-back" studies in which it contacts the victim, the victim's parent, or a witness (usually by telephone, but sometimes involving on-site investigations) for more detailed information. Generally, these contacts can be made within a few weeks of the occurrence of the injury. This approach was used for an in-depth study of bicycle-related injuries occurring in 1991.⁽²¹⁾

In reviewing the literature on the safety of the various emerging road and trail users being examined in the current study, NEISS data were frequently cited.^(10,22,23)

Wheelchair-related injuries and deaths may also be reported in greater detail to the Food and Drug Administration (FDA), as required under its Medical Device Reporting program (data available at http://www.fda.gov/cdrh/mdrfile.html), or as part of its voluntary MedWatch program. For example, FDA data served as the basis for a study on wheelchair safety.⁽²³⁾ Devices examined in this study included manual wheelchairs, powered wheelchairs, and assistive scooters.

In summary, the NEISS data appear to be an especially rich source of information on the safety of many emerging vehicle types targeted in this study. Because its basis is in hospital EDs, it incorporates data on injury events occurring on and off public roadways, and regardless of whether or not a motor vehicle was involved. The data constitute a statistically valid national sample, and CPSC has a long tradition of working cooperatively with researchers and other Government agencies interested in accessing the data. In addition, opportunities exist for further expanding the available data by incorporating follow-back telephone interviews into the data collection process.

^10.  Rutherford, Jr. G. W. and R. Ingle. Unpowered Scooter-Related Injuries-United States, 1998-2000. CDC MMWR Weekly, Vol. 49, No. 49, December 15, 2000, pp. 1108-1110. www.cdc.gov/mmwr/preview/mmwrhtml/mm4949a2.htm. Accessed August 10, 2001.

^20.  Powell E.C. and R.R. Stanz. Tykes and bikes: injuries associated with bicycle-towed child trailers and bicycle-mounted child seats. Archives of Pediatric and Adolescent Medicine Vol. 155, No. 4, 2001, p. 525.

^21.  Rodgers, G.E. Bicycle Use and Hazard Patterns in the United States, and Options for Injury Reduction. U.S. Consumer Product Safety Commission, Washington, DC, 1993.

^22.  Schieber, R.A. and C.M. Branche-Dorsey. In-line skating injuries: epidemiology and recommendations for prevention. Journal of Sports Medicine, Vol. 19, No. 6, 1995, pp. 427-432.

^23.  Kirby, R.L. and S.A. Ackroyd-Stolarz. Wheelchair safety-adverse reports to the United States Food and Drug Administration. American Journal of Physical Medicine and Rehabilitation, Vol. 74, No. 4, 1995, pp. 308-312.

Other Hospital-Based Sources of Data

Literally hundreds of studies in the published literature deal with skateboard and skating injuries, and a rapidly growing body of literature addresses nonmotorized and motorized scooter injuries; a few specialized studies consider bike trailers, golf carts, and racing wheelchairs. The primary source of data for these studies has been hospital ED data-either individual case series from a single hospital, or local or regional trauma registry data. Studies have involved both retrospective examination of ED records and prospective case identification, and have most often targeted a single user category (e.g., inline skate injuries in children). In addition to studies conducted here in the United States, a significant number of studies have been conducted in Australia, Great Britain, and the Scandinavian countries.

While much can be learned from these studies, they generally do not represent viable data sources for further study. The data are typically not publicly available, and often involve some level of specialized data collection. They are also relatively small-scale studies that may not be generalizable to larger populations.

An exception would be studies based on data from a population-based trauma registry, such as a State, regional (multihospital), or national registry. Trauma registries most often capture information on hospitalized patients, but in some cases collect information on those treated in the ED, as well. An example of a large, publicly available trauma registry database is the National Trauma Data Bank (NTDB), established by the American College of Surgeons in 1997 (with data extending back to 1994). As of 2001, the NTDB contained data from four State registries and from 67 hospitals in 29 States, representing all regions of the United States. Still, overall case numbers were relatively small, at slightly more than 181,000 cases (1994-1999 data).⁽²⁴⁾  Another example of a national trauma registry is the National Pediatric Trauma Registry, a voluntary system of reporting of pediatric trauma patients ages 0 to 19 years that was operational until February 2002. Although no new data are being added to the system, researchers can still access the available data, including a total of more than 43,000 submitted cases from 1994-2001.⁽²⁵⁾

A major disadvantage of trauma registry and most other routinely collected hospital-based data is that it is not sufficiently detailed with respect to the cause of injury, and in particular the involvement of specific equipment such as inline skates, scooters, and motorized wheelchairs. Most rely on standardized E-Codes for coding the cause of injury. But while E-Codes differentiate among pedestrians and bicyclists being struck by motor vehicles on and off public roadways, they do not identify the specific equipment being used at the time, or the specific location of the user if not in the roadway. Neither is it possible to obtain this information through special follow-up studies because the identity of subjects is generally stripped from the files. Thus, existing hospital-based data sources would appear to be of limited usefulness for studying the safety characteristics of emerging road and trail users.

Such limitations could, of course, be overcome by implementing a hospital-based data collection system specifically designed to gather information on the safety of the various user groups of interest. One could also attempt to modify a data collection system or trauma registry already in place. Both are likely to be costly undertakings, especially in light of the large number of hospitals that would need to participate to generate a sufficient number of cases with respect to the "rarer" user groups, including adult tricycles, tandem bicycles, and racing wheelchairs.

^24.  American College of Surgeons. National Trauma Data Bank Report 2001. Chicago, IL: American College of Surgeons, 2001. www.facs.org/ntdbreport2001/summary.html.

^25.  National Pediatric Trauma Registry. Characteristics and Technical Highlights. Boston, MA: Tufts-New England Medical Center. www.nptr.org.

State Motor Vehicle Crash Data

State motor vehicle crash files are another potential source of information on the safety of emerging road and trail users. To determine whether States collect information on the various user groups identified, the researchers sent a brief e-mail survey (included in the appendix) to the crash form coordinator in each State. The names of the crash form coordinators and their e-mail addresses were obtained from NHTSA's Traffic Records System Inventory Web site (http://www.nhtsa.dot.gov/people/perform/trafrecords/crash/Pages/coordinators.htm). The site also provides telephone numbers and downloadable copies of State crash report forms.

Surveys were initially sent to the 40 States with available e-mail addresses. Of these 40, however, only 22 proved to be viable (i.e., deliverable) addresses, and only 15 responded. Researchers continued to follow up with nonrespondents, and also began trying to contact persons by telephone to obtain updated e-mail addresses. Eventually, researchers were able to identify e-mail addresses for all but a few States, and obtained completed surveys for 35.

The results are not especially encouraging. In response to the question about whether specific user groups could be identified on their computerized crash database, the number of "yes" responses (out of 35) is shown in table 3.

Table 3. User types that can be identified in State motor vehicle crash files.

Four States indicated that they could identify inline skaters (or rollerbladers), but in two of these States (Nevada and North Carolina), inline skaters were coded together with skateboards. Seven States indicated that they could identify motorized scooters; however, we believe that in most if not all of these cases, the motorized scooter that the respondent was referring to was the larger powered motorbike similar to a motorcycle, and not the small, motorized stand-on scooter that we had intended by our question. The exception may be Nevada, which indicated that it had begun collecting data on motorized scooters in November 2001. The only other special use groups identified by any of the States were bicycle trailers (two States), wheelchairs (two States), and racing wheelchairs (one State). In the case of wheelchairs, Oregon indicated that although they had been coded separately since 1998, they were not differentiated by type. This may also be the case in Nevada (i.e., Nevada may code racing wheelchairs along with regular wheelchairs). Thus, there appears to be few data available on emerging road and trail users from computerized State motor vehicle crash databases.

Anticipating this might be the case, we also asked the coordinators whether their State computerized the narrative descriptions of crashes, either full descriptions or key words, and if they did, whether this data could be electronically searched. Five States (Arkansas, Delaware (under development), North Carolina, Oregon, and Rhode Island) indicated that they did computerize some or all of their crash narratives, and three of these (North Carolina, Oregon, and Rhode Island) indicated that these data could be searched using key words to identify potentially relevant crash reports for the various identified user groups.

A final question asked respondents whether there had been any efforts in their State to use available crash data to study the safety of any of the identified emerging user groups (not including standard bicycles). Florida indicated that its department of transportation had studied inline skaters, and Vermont that pedestrians had been studied (although without reference to any of the identified user groups). In North Carolina, the crash data have been queried in response to requests from the public, but no formal studies have been carried out on any of the identified user groups. Several respondents indicated that these user groups were not felt to constitute a safety problem in their State.

As a side task, we searched the North Carolina crash data to identify all crashes in which "Person Type" was coded as "roller skater, roller blader, etc." Over a period of approximately 2.5 years (2000, 2001, and through about August 2002), only 17 such cases were coded.

However, it was clear that the coding was not entirely accurate, since when we examined the hard copies of the crash reports for these cases, 9 of the 17 did not appear to involve skaters. In addition, there were cases that we had identified from our narrative search (see following section) that should have been coded in this category but were not. This admittedly limited pilot effort suggests that routinely coded crash data may not be the most complete or reliable source of information on emerging road users.

Based on these results, it does not appear that State motor vehicle crash databases are especially useful sources of information on the safety of the emerging road and trail users that are the focus of the current study. Although this is primarily because only a few States are collecting data on only a few of the user categories, it is also because the data that are collected are limited almost entirely to crashes involving motor vehicles and occurring on public roadways. Recognizing these limitations, it may still be useful to pool available information from selected States to study the characteristics of, for example, crashes involving inline skaters or skateboard users. While sample sizes are likely to be too small for a statistical analysis of the data, the identified crash reports could be accessed and reviewed as a case series to identify potentially important characteristics, such as involved age groups and factors contributing to the crash.

State Narrative Crash Data

As noted above, at least three States-Oregon, Rhode Island, and North Carolina-computerize some or all of the narrative descriptions contained on their crash reports, and are able to conduct key word searches of these data. Theoretically, at least, this should make it possible to identify crashes involving any road user group desired, regardless of whether they are coded elsewhere on the form. In practice, narrative searches always underestimate the occurrence of a particular crash type, because (1) officers do not always report the information, (2) if they do report the information, they may use words, or spellings of words, that do not coincide with any of the key search terms chosen, and/or (3) there may be omissions or other errors in entering the narrative data into the computerized database. For example, an officer may fail to notice that a child struck by a car was riding a scooter at the time, or may refer to the scooter in his report as a "push bike" or "skooter."

In North Carolina, narrative searches have been used to study a wide array of topics that otherwise could not have been studied from the computerized crash data alone. Examples include post-crash fires, deer crashes, cellular phone use and crashes, driver distractions, billboards, hydroplaning in wet weather crashes, and road debris as a causative factor in crashes. To explore the potential usefulness of narrative searches for identifying crashes involving emerging road and trail users, we used the following search words to identify crashes potentially involving inline skates, roller blades, or skateboards: inline, skate, roller, blade. This list generated 43 narratives for crashes occurring between January and some time in August 2002. However, upon reading the printed narratives, many of these were found to be false hits: for example, reference to a roller tape used in taking measurements, an asphalt roller used in road repair, or a snow plow blade. Eight narratives were true "hits." Examples of these appear below:

"Vehicle 1 made a right turn onto Hillburn Street from Chapel Road. Pedestrian child was rollerblading across Hillburn Street. Driver 1 stated that child skated from his blind spot directly in front of vehicle. Driver 1 stated he did not see the child until the collision."

"After interviewing witnesses and subjects involved it was determined that ped was roller blading in roadway and attempted to cross roadway making contact with the vehicle 2 on the front left quarter panel and then rolling up onto the windshield and was thrown off the front into the roadway."

"Driver Vehicle 1 stated he had just about passed the child on a skateboard when the child moved back into the street and hit [his truck]."

"Driver of Vehicle 1 stated she was going straight ahead and the child was riding a skateboard and came out of side street and struck her vehicle. Pedestrian's mother stated that her son told her he couldn't stop in time and struck vehicle 1 in the side."

While these are only a few example cases, they illustrate how narrative searches can be used to identify crashes of interest that otherwise might go undetected on the computerized crash database.

In addition to underestimating the occurrence of a particular type crash, data resulting from computerized narrative searches will suffer from the same limitations as the computerized report data, namely, a restriction to crashes involving motor vehicles and occurring on public roadways. This data source is again likely to be most useful when used as a basis for a case series study to provide guidance to a larger data collection and analysis activity.

Summary

In summary, a variety of data sources were examined for their potential for providing information on the safety of the emerging road and trail users that are the focus of this FHWA study. While there are advantages and disadvantages to each of the sources, the data available through the NEISS system operated by the CPSC appears to present the greatest advantages (see table 4). It is recommended that FHWA, NHTSA, and the CDC jointly sponsor a study using NEISS data and incorporating follow-back telephone interviews to gather comparative data on the crashes and injuries associated with these various user groups.

Table 4. Summary of advantages and disadvantages to the data sources reviewed.