Article Text

Download PDFPDF

Epidemiology of non-fatal US emergency room visits for road crashes involving pedestrians in wheelchairs
  1. John D Kraemer
  1. Correspondence to Professor John D Kraemer, Department of Health Systems Administration, O'Neill Institute for National & Global Health Law, Georgetown University, 3700 Reservoir Road, NW, 231 St Mary's Hall, Washington, DC 20007, USA; jdk32{at}georgetown.edu

Abstract

This study aims to quantify and describe the risk of non-fatal pedestrian injury among persons who use wheelchairs in the US. Cases of pedestrian injury between 2002 and 2010 among persons using wheelchairs were identified in the National Electronic Injury Surveillance System to generate national injury estimates. Between 2002 and 2010, an estimated 9348 (95% CI 4912 to 13 784) people were treated in emergency departments for non-fatal pedestrian injuries sustained while using wheelchairs. Using wheelchair-use denominators calculated from the Survey of Income and Program Participation, this equates to an incidence rate of 31.3 (95% CI 16.4 to 46.1) per 100 000 person-years. Injury risk was 3.5 times higher for men than women (p<0.001). Contusions, abrasions, and lacerations (42.7%) and fractures (16.4%) were most common. The head and neck (24.7%) and lower extremities (28.4%) were most often injured. A fifth (21.4%) of injuries required hospitalisation, and 89.2% occurred in traffic on public roadways.

View Full Text

Statistics from Altmetric.com

Introduction

Reducing pedestrian injuries is a top road safety priority,1 but little research has been conducted on pedestrian risks among persons who use wheelchairs. In 1997, it was conservatively estimated that there are 360 injuries per year from collisions between vehicles and wheelchair users.2 Later research estimated an average of 890 traffic-related injuries annually from 2004 to 2008.3 During this time, wheelchair use increased by about 5% per year, with use highest among women and those aged 65 years and over.4 However, no research has determined pedestrian injury incidence rates (IRs) among persons using wheelchairs, impeding understanding about risk levels and trends. Beyond direct public health impact, qualitative research has found that pedestrian safety concerns limit community mobility and, therefore, community integration among people who use wheelchairs.5 This paper aims to quantify and describe pedestrian injury risk experienced by wheelchair users.

Methods

Data sources

Injury data came from the National Electronic Injury Surveillance System (NEISS), a probability-sample of US emergency departments (EDs). It includes data on injuries associated with consumer products resulting in ED visits. NEISS included wheelchair-related injuries in community settings through 2010, after which it was removed.6

The NEISS sampling, variables and data validation approach have been published elsewhere.7 ,8 To summarise, NEISS data are extracted by trained coders from records at a 100 nationally representative sampled hospitals, stratified by hospital size, plus a stratum for children's hospitals. Records include demographic data, summary clinical information, and a two-line narrative incident description, as well as the product involved (to a two-product maximum) and incident location.

The narrative descriptions of incidents involving wheelchairs (product code=1707) were examined, and cases were included if there was a clear reference to a person in a wheelchair being injured in a pedestrian collision with a vehicle. Mobility scooters, coded as miscellaneous vehicles and indistinguishable from other scooter types, were not included. Observations were not included as injury if aetiology was unclear (eg, “patient unresponsive in street”) or if the injured person was not the wheelchair user.

Denominator data for rate calculations came from the US Census Bureau's Survey of Income and Program Participation (SIPP), which is fully described elsewhere.9 SIPP is a population-representative survey of non-institutionalised persons with a periodic disability module that includes a current wheelchair use item. SIPP wheelchair use data are available for 2002, 2003, 2005 and 2010; for other years, estimates were calculated using geometric interpolation between the two nearest years, which assumes a constant rate of population change.10 To assess the bias risk from denominator misestimation, three alternative approaches were used for sensitivity analyses: linear interpolation between the two nearest years; fitting a quadratic equation using ordinary least squares; and estimating a log-linear relationship by fitting a generalised linear model with a log link, which avoids log-retransformation bias, using maximum likelihood.

For comparison purposes, the pedestrian injury risk in the general US population came from CDC's WISQARS database,11 which derives data from the NEISS-All Injury Program (NEISS-AIP). NEISS-AIP is a version of the NEISS survey, without sampling restricted to consumer products, at a subset of 66 NEISS hospitals.12 The comparison rates include all intentional and unintentional pedestrian crashes regardless of whether they were traffic-related or non-traffic-related.

Measures

Injury counts and IRs were calculated for the population of people using wheelchairs as well as by available personal characteristics, clinical features and incident-related factors. Variables for personal characteristics include sex (dichotomous) and age (categorised to maintain sample size as <30, 30–49, 50–64 and >65). Injury diagnoses were categorised as contusions/abrasions/lacerations, fractures, strains and sprains, internal organ injury and other). Injury location was categorised as head/neck, upper trunk, lower trunk, arm/hand, leg/foot or other. ED discharge disposition was categorised as treated and released, hospitalised (admitted or transferred to another hospital) or other (held for observation or who left without being seen). The narrative description was examined to determine the wheelchair type used (dichotomised as motorised vs manual or unknown) and striking vehicle (categorised as car, truck/SUV, van, other or unknown). Finally, crashes were classified as traffic-related or non-traffic-related based on location and colliding vehicle using International Classification of Diseases (ICD) coding rules.13

Statistical analyses

Estimated injury counts were calculated using NEISS inverse sampling probability weights and all SE for CI construction were adjusted for the complex survey design using Taylor series linearisation. Population estimates for rate denominators were calculated using SIPP sampling weights. IRs were stratified by age, sex and sex-specific age categories. For the latter, ages below 30 years were not included because of small sample sizes. Differences in injury IRs between wheelchair users and the general US population and between men and women were tested using a z-test for differences with SEs of the difference adjusted for the survey design. Bivariate associations between ED discharge disposition, whether the crash was traffic-related, and age, sex, injury diagnosis, and injury location were calculated using a design-corrected χ2 test. Analyses used Stata V.13.1.

Georgetown University's IRB does not require approval of secondary analyses of public data sets without identifiers.

Results

The NEISS sample included 294 cases, corresponding to a weighted national estimate of 9348 (95% CI 4912 to 13 784) pedestrian injuries among persons using wheelchairs (table 1) during 29.9 million person-years (py) at risk. This is an injury IR of 31.3 per 100 000 py (95% CI 16.4 to 46.1). Annual injury risk did not change meaningfully from 2002 to 2010 (figure 1). Eighty-nine per cent of crashes were traffic-related.

Table 1

National estimates of the number and rate of pedestrian injuries treated in US emergency departments among persons using wheelchairs compared with the general US population, 2002–2010

Figure 1

Incidence rate of pedestrian injury among persons using wheelchairs by year 2000–2010.

Women comprised 29.1% of pedestrian injuries among wheelchair users, representing a significantly lower injury rate (IR=15.4 per 100 000 py; 95% CI 5.8 to 25.0) than men (IR=54.1 per 100 000 py; 95% CI 30.2 to 78.1; p<0.001 comparing sex). While the pedestrian injury rate for women using wheelchairs was lower than in the general population (p<0.001), the difference was not significant for men (p=0.23).

Among those using wheelchairs, injury risk was lower under age 30 years (IR=29.6 per 100 000 py; 95% CI 10.3 to 49.0) and ages 65 years and older (IR=13.9 per 100 000; 95% CI 6.8 to 21.0) and higher between 30–49 year-olds (IR=69.0 per 100 000 py; 95% CI 35.5 to 102.4) and 50–64 year-olds (IR=48.9 per 100 000 py; 95% CI 23.8 to 74.1). Risk in the latter ages was qualitatively similar to the general population with differences statistically insignificant (p=0.62 and 0.80, respectively).

The injury risk age-distribution differed between male and female wheelchair users. The point estimate for female risk was lower among wheelchair users than among the general pedestrian population for all age groups, with the difference statistically significant among women aged 30–49 years (IR=19.6 vs 47.1 per 100 000 py; p=0.003) and over 65 years (IR=4.3 vs 37.3 per 100 000 py; p<0.001). Among men, the risk difference between wheelchair users and the overall US population was not statistically significantly at any age, but point estimates were higher for men aged 30–49 years (IR=123.8 per 100 000 py; 95% CI 65.3 to 182.4) and 50–64 years (IR=68.5 per 100 000 py; 95% CI 35.3 to 101.8).

The most common injury diagnoses were contusions, abrasions and lacerations (42.7%); fractures (16.4%); and strains or sprains (16.3%) (table 2), with lower extremities (28.4%), head and neck (24.7%), and the upper trunk (19.1%) most frequently injured. Of the injuries 21.4% resulted in hospital admission. Injury diagnoses (p<0.001) and injured body part (p=0.02) varied between admitted and non-admitted patients (not shown in tables). Head and neck injuries (37.7% vs 21.4%), fractures (51.4% vs 6.8%) and internal injuries (33.6% vs 4.9%) were more common among admitted patients. Hospitalisation did not vary by age (p=0.61) or sex (p=0.83). Diagnosis, injury site, hospitalisation, age and sex did not vary significantly between traffic-related and non-traffic-related crashes. Of the crashes 25.4% involved a person recorded by ED staff as using a motorised wheelchair. Cars were recorded as the most common striking vehicle (71.5%).

Table 2

Characteristics of injuries among pedestrians using wheelchairs treated in US emergency departments, 2002–2010

In the sensitivity analyses, injury rates were not meaningfully different when calculated using alternative denominator interpolation approaches (see online supplementary appendices 1–3).

Discussion

Key findings and limitations

This study finds that an average of 1040 persons using wheelchairs were treated in US EDs for injuries from non-fatal pedestrian collisions with vehicles annually between 2002 and 2010, approximately 90% on public trafficways. This is approximately consistent with a study of traffic-related crashes using slightly older data. This is the first study to calculate pedestrian injury IRs for persons using wheelchairs and finds a significant disparity between women and men, who have about 3.5-fold increased risk. Increased risk among men using wheelchairs is consistent with previous research on pedestrians more generally.14

Rates calculated in this study are likely conservative. There is likely incomplete ascertainment of wheelchair use in ED records, a problem that may be exacerbated if a patient was transported by ambulance following a crash or if there is less complete investigation of minor injuries.2 These pose a risk of differential bias resulting in undercounting of injured wheelchair users, and may also result in overestimation or underestimation of the proportion of injuries that result in hospitalisation. Additionally, it is possible—though no data exist—that wheelchair users may more often die from crashes. If so, this would not bias non-fatal injury estimates per se, but it would bias inferences using non-fatal crashes as a proxy for all crashes. Minor injuries also may have been treated at facilities without an ED and not captured in NEISS.15 This issue applies to NEISS and the NEISS-AIP sample used for comparisons, so this bias is likely non-differential.

Calculating injury rates based on risk per person-kilometer-travelled would be preferable but there is scant data on distances travelled by persons who use wheelchairs. Limited evidence suggests that persons using motorised wheelchairs travel shorter average daily distances than non-wheelchair users.16 ,17 (Data on manual wheelchair users are not available.) This may artificially reduce the observed risks and would be doubly problematic if persons in wheelchairs limit incommunity travel because of pedestrian safety concerns.5

For several variables, there is a risk of inaccurate ascertainment. It is likely that motorised wheelchairs were under-reported by ED personnel—who are not required to report wheelchair type—and cars may be over-recorded as a generic term for ‘motor vehicle’.

Policy and research implications

Understanding causes of pedestrian crashes involving persons using wheelchairs requires more research. Prior studies and anecdotes have suggested difficulty seeing wheelchair users, behavioural risk factors, and unsafe pedestrian facilities as possible factors disproportionately affecting wheelchair users.3 ,18 As with most injuries, optimal prevention likely requires built environment improvements and safety promotion interventions among drivers and persons using wheelchairs.19 ,20 Based on their increased risk, behavioural interventions may be best targeted to relatively young men.

Additionally, little research exists on the risk of pedestrian deaths among wheelchair users. One study, using NEISS data, found one fatal crash per 40 non-fatal crashes—a likely underestimate because of deaths prior to ED transport.2 An analysis of media reports, which likely overinclude deaths, found that approximately half of reported crashes were deadly.18

Crashes involving pedestrians using wheelchairs will likely increase if current trends towards more wheelchair use continue. Against this backdrop, a focus on safe pedestrian infrastructure—including through better enforcement of the Americans with Disabilities Act—should remain prioritised.

What is already known on the subject

  • Few studies have investigated the risk of pedestrian injury among persons who use wheelchairs, though there is a trend towards more frequent pedestrian injuries and more wheelchair use in the USA.

  • Existing research has found a doubling of the number of pedestrian injuries among persons who use wheelchairs in the USA, but it is unclear whether this represents an increase in per-person risk.

What this study adds

  • Approximately 1040 persons who use wheelchairs are treated in US emergency departments for pedestrian injuries each year, a risk equal to 31.3 injuries per 100 000 person-years.

  • Men are at greater risk than women, and relatively young men are at the highest risk.

  • The annual per capita risk of injury did not change meaningfully during the study period.

Most embarrassing

In its monthly round-up of intriguing news, Retraction Watch posted this item judged to be the most embarrassing (and unsafe) field research experience. “Accidentally glued myself to a crocodile while attaching a radio transmitter.” Editor’s note: Not a problem for most epidemiologists. But one asked the logical follow-up: What happened next, and another asked about the glue. (Noted by IBP)

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors JDK designed the study, acquired and analysed its data, and drafted the manuscript.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All raw data used in this study are publicly available. Cleaned data will be made available to other researchers upon request.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.