Forensic Science International 234 (2014) 154–161
Contents lists available at ScienceDirect
Forensic Science International journal homepage: www.elsevier.com/locate/forsciint
Examining the impact of opioid analgesics on crash responsibility in truck drivers involved in fatal crashes Paula Reguly a,b, Sacha Dubois a,b,c,d,*, Michel Be´dard a,b,c,d a
Department of Health Sciences, Lakehead University, 955 Oliver Road, Thunder Bay, ON P7B 5E1, Canada Centre for Research on Safe Driving, Lakehead University, 955 Oliver Road, Thunder Bay, ON P7B 5E1, Canada c Research Department, St. Joseph’s Care Group, 580 North Algoma Street, Thunder Bay, ON P7B 5G4, Canada d Northern Ontario School of Medicine, Human Sciences Division, Lakehead University, 955 Oliver Road, Thunder Bay, ON P7B 5E1, Canada b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 27 June 2013 Received in revised form 28 October 2013 Accepted 6 November 2013 Available online 15 November 2013
Introduction: Commercial motor vehicle (CMV) drivers, particularly drivers of large trucks continue to be a population of concern regarding traffic safety despite the reduction in large truck crash rates over the past decade. Medication and drug use while driving is one important risk factor for large truck crashes. Work-related exposures, such as vibration, manual handling and poor ergonomics contribute to an increased risk for injuries and chronic conditions and are common reasons for opioid analgesic (OA) use by CMV truck drivers. The objectives of this study were to examine the role of OA use in CMV truck drivers involved in fatal crashes by: (a) generating prevalence estimates of OA use; (b) documenting the relationship between OA use and crash responsibility. Methods: Case-control study using logistic regression to compare Fatality Analysis Reporting System (1993–2008) record of one or more crash-related unsafe driver actions (UDAs – a proxy measure of responsibility) between drivers with a positive drug test and drivers with a negative drug test for OA, controlling for age, other drug use, and driving history. Results: The annual prevalence of OA use among all CMV drivers of large trucks involved in fatal crashes did not exceed 0.46% for any year in the study period and mostly ranged between 0.1 and 0.2%. Male truck drivers using OA had greater odds of committing an UDA (OR: 2.80; 95% CI: 1.64; 4.81). Middleaged users had greater odds than younger or older users. Conclusion: The results of our study indicate that the presence of OAs is associated with greater odds of committing an UDA. This association may have implications for the commercial transport industry and traffic safety. However, the limited prevalence of OA use is encouraging and further research is needed to address the limitations of the study. ß 2013 Elsevier Ireland Ltd. All rights reserved.
Keywords: Opioid analgesics Truck drivers Crash culpability Impairment
1. Introduction Commercial motor vehicle (CMV) drivers, particularly drivers of large trucks continue to be a population of concern regarding traffic safety despite the reduction in large truck crash rates over the past decade [1]. Occupational fatality rates for the transport industry are consistently among the highest [2], and in large truck crashes involving multiple vehicles there is a tendency for the occupants of the other vehicle(s) to be severely injured or killed [3– 5]. There are also significant economic costs incurred by large truck
* Corresponding author at: Research Department, St. Joseph’s Care Group, LPH Site, 580 N. Algoma Street, P.O. Box 2930, Thunder Bay, Ontario P7B 5G4, Canada. Tel.: +1 807 343 4300x4480; fax: +1 807 346 5243. E-mail addresses: [email protected] (P. Reguly), [email protected] (S. Dubois), [email protected] (M. Be´dard). 0379-0738/$ – see front matter ß 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.forsciint.2013.11.005
crashes through medical costs and insurance settlements, lost productivity, and property damage [6]. Medication and drug use while driving is one important risk factor for large truck crashes. Because driving is a complex activity that involves a range of cognitive and psychomotor functions, both licit and illicit drugs effect on the central nervous system can impair driving ability [7]. A recent large-scale study on large truck crash causation found that prescription drug use and over the counter drug use was among the top ten factors associated with crashes out of hundreds of factors examined [8]. Nonetheless, the impact of drug use on driving performance has received little research attention [4], despite the potential for impairment. Opioid analgesics (OAs) are one of the drug groups warranting further attention. Opioid analgesics are most commonly used to treat pain but they are also used in substitution therapy to treat substance abuse [9]. The term opioid refers to naturally occurring, synthetic and
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
semi-synthetic compounds that are derived from the opium of the poppy plant, such as codeine, morphine, oxycodone and hydromorphone [9]. According to the American Chronic Pain Association (ACPA), common and anticipated central nervous system (CNS) side effects of OAs are thought and memory impairment, drowsiness and nausea, mild sedation, and impaired judgment and co-ordination. The ACPA warns against driving until tolerance or a baseline is reached [9]. With respect to OA use and driving performance, there are several reasons why commercial large truck drivers are a population of interest. Truck drivers experience work-related exposures such as vibration, manual handling and ergonomic factors that may increase the risk for injuries and chronic conditions (e.g., lower back pain and musculoskeletal disorders) [10–12] that are typically treated with OAs. In fact, treatment for chronic pain has been recognized as the most common reason for opioid use by CMV truck drivers [4]. Also, fatigue has been recognized as a risk-factor for large truck crashes [5,8,13] and the side effects associated with OAs, such as sedation, might compound the risk of crashes. Finally, there is some evidence that drugs are used by many truck drivers to cope with boredom and other aspects of the job [14]. Beyond these job-related factors, there have been dramatic increases in the prescribing and use of OAs in the general population over the last two decades [4,15–17], as well as some indication of an increased prevalence of detection of opioids among drivers in the general population [18,19]. However, studies of opioid use among CMV truck drivers have found a low prevalence of opioid use (generally not exceeding 4.0% of the study sample) compared to other types of drugs such as stimulants, depressants and cannabinoids [13,14,20,21]. Prevalence studies that have relied on the voluntary provision of biological samples [14,21] may have resulted in the under-detection, and hence underestimation of prevalence rates. Conversely, the use of fatally injured drivers as a study sample population provides an overestimate of prevalence rates within the wider truck driver population [13]. Finally, these prevalence studies [13,14,20,21] have not addressed long-term trends and may not accurately reflect current prevalence rates, especially considering the recent dramatic increase in the prescribing and use of OAs in the general population. It should be noted that even a low prevalence of opioid use can translate into a considerable absolute number of opioid users, given that, in 2012, there were 5,700,000 CMV drivers operating in the U.S. alone [22]. Based on this figure, a prevalence rate of 2.0% among U.S. CMV drivers is equivalent to 114,000 opioid users. Experimental studies investigating the effect of OAs on driving ability have used driving simulators, on-road driving, and cognitive and psychomotor tests specifically for driving. The key findings from this research are that therapeutic long-term stable doses of opioids do not negatively impact driving ability [23–28], but a change in dosing of opioid medication (30% increase in dose) results in significant cognitive impairment [23]. Furthermore, administering an OA to healthy individuals not previously exposed to OAs did not significantly affect driving ability, but study participants had significantly reduced pupil size and they reported that significantly more effort was needed to perform the driving test and reported significantly more sedation and reduced alertness [29]. But there are several limitations of the experimental research regarding the effect of OAs on driving ability, such as small sample sizes and insufficient statistical power; the use of healthy and young study participants; insufficient doses to elicit effects; and highly controlled environments that do not always reflect actual driving conditions and experiences. There is scant research from observational studies investigating the association between OA use and crash risk and crash
155
responsibility among CMV drivers of large trucks. We found only one study regarding the association between opioid use and crash risk among CMV drivers. Howard et al. [30] collected self-reported information on crash history, drug and alcohol use from a sample of CMV drivers and found that drivers that reported using narcotic analgesics had greater odds of being involved in a crash than unexposed drivers after adjusting for age, hours of driving and alcohol intake (Adjusted OR: 2.40; CI: 1.46; 3.92). There are, however, a number of observational studies, using samples from the general population, that have investigated the effect of OA use on crash risk and crash responsibility. A review by Fishbain et al. [7] concluded that there was consistent evidence that opioids are not associated with crashes; but the methodological shortcomings of many of the studies reviewed, such as the lack of control group, weaken their conclusion. A meta-analysis by Mona´rrez-Espino et al. considered studies from the period between 1991 and 2012 specifically for older drivers (55 years old) [31]. They argued that the evidence fails to provide convincing support that opioids are associated with increase crash risk among drivers 55 and older, due to the small number of studies, the generally inadequate control of confounders (e.g., other medications and illness), and inconsistent results. Conversely, the DRUID Project compared the risk of being seriously injured or killed while driving with psychoactive substances including medicinal opioid use [32,33]. Pooled data from six European countries was used. Cases were obtained from hospitalization data, controls were obtained from roadside surveys. Using odds ratios as an estimate of risk, the DRUID Project reported that drivers had significantly increased estimates of risk of being seriously injured (OR: 9.06; CI: 6.40–12.83) or killed (RR: 4.82; CI: 2.60–8.93) when positive for medicinal opioids. Other studies, using samples from the general population, have demonstrated a small, but positive association between opioid use and increased crash risk [34,35]. Arguably better evidence for the effect of opioids on crash involvement comes from crash culpability studies. Dubois et al. [36] demonstrated a significant positive association between OAs and crash culpability using a sample of passenger vehicle drivers (survivors and fatally injured) who were involved in fatal crashes (controlling for other medications, driving history and other factors). Drummer et al. [18] showed a positive but non-significant association between opioid use and crash culpability study using a sample of fatally injured drivers of all vehicle types (controlling for crash type, drug use and other factors). The DRUID Project considered only illicit opiates in its culpability analysis, and found that the odds of being responsible for a fatal crash did not differ significantly between the drivers that tested positive for illicit opiates and those that tested negative [32]. Overall, factors that may be contributing to the observed inconsistencies for the association between opioids and crash risk and crash culpability include sample type (e.g., fatally vs. no-fatally injured), opioid(s) of interest, determination of opioid exposure (e.g. prescription records vs. blood or urine test), and study design (e.g., cohort or case-control vs. cross-sectional). We designed our study to: (a) generate estimates of the prevalence of opioid use over time in a sample of drivers of large trucks; and (b) document the relationship between OA use (Schedule II opioid analgesics in particular) and crash responsibility among drivers of large trucks involved in fatal crashes using a representative data source with high external validity and standardized toxicological testing, while controlling for other possible contributory factors to crash initiation including age, previous driving history, and substance use (other than OAs) such as alcohol and other medications. We hypothesized that the odds of committing an unsafe driving action (UDA) preceding a fatal crash would be greater for truck drivers who tested positive for an OA compared to truck drivers who tested negative.
156
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
2. Materials and methods
using UDAs as proxy measures of crash responsibility has been demonstrated in several other studies [3,36,38–42].
2.1. Data source 2.4. Opioid analgesic classification and exposure Data were obtained from the Fatality Analysis Reporting System (FARS), available from the U.S. National Center for Statistics and Analysis of the National Highway Traffic Safety Administration. It is a repository of information pertaining to fatal traffic crashes occurring in all 50 states, the District of Columbia, and Puerto Rico, dating back to 1975. A crash qualifies for inclusion if it involves a motor vehicle traveling on a traffic way customarily open to the public, and results in the death of a person (i.e., vehicle occupant or non-motorist) within 30 days of the crash [37]. Crash information is collected, coded, and entered into the FARS database according to standard methods by trained FARS analysts [37]. FARS analysts obtain information from a variety of sources such as police accident reports, hospital medical reports, coroner reports, and state vehicle registration and licensing files. The FARS data are organized into three levels. The crash level includes crash time and location, the number of vehicles and people involved, and weather conditions. The vehicle level includes ‘‘the vehicle type, role in the crash, initial and principal impacts points, the most harmful event, the driver’s record, and license status.’’ The person level includes data on the people involved in the crash (e.g., age, gender, role in the crash [e.g., driver or passenger], alcohol and drug involvement, injury, severity). Information entered into the database is subjected to a rigorous Quality Control Program which includes checks for consistency, timeliness, completeness, and accuracy [37]. 2.2. Sample We used the same sampling strategy as Gates 2013 [38]. In brief, the sample was based on crashes between 1993 and 2008. The prevalence analysis included drivers operating a single or combination-unit heavy truck with a gross vehicle weight rating of greater than 26,000 lbs or a truck-tractor (cab only, or with any number of trailing units; any weight). To ensure a minimal level of driver history could have occurred we excluded drivers below the age of 20. Two additional criteria were applied to those drivers included in the crash responsibility analysis. First, to rule out potential confounding effects of alcohol only drivers with a confirmed BAC of zero were included. Second, all drivers included in the crash responsibility analysis were blood tested for drugs. 2.3. Design Cases and controls were classified according to the presence (case) or absence (controls) of at least one recorded UDA. The UDAs served as a proxy measure of crash responsibility. Drivers with any UDA(s) recorded were deemed to have contributed to the initiation of the collision. Blower has argued that driver-related factors (i.e., UDAs) are preferable to traffic violations as indicators of crash culpability because not all contributing factors are chargeable offenses; traffic violations are not uniformly applied or enforced; and various factors influence whether a violation is charged including the severity of the crash and the degree of supporting evidence [3]. FARS analysts can enter up to four driver-related factors (i.e., UDAs) (three before 1997). The driver-related factors are based on crash narrative in the police report and ‘‘the narrative allows a fuller description of the factors that contributed to the crash’’ [3]. Finally, driver-related factors have been shown to be consistent with the physical evidence and configuration of crashes, which in turn are strongly related to the relative contribution of drivers in most types of large truck crashes [3]. This method of
Opioid analgesics were defined as Schedule II opioid analgesics in accordance with section 1308.12 of the U.S. Code of Federal Regulations [43]. All Schedule II substances are licit drugs with legitimate medical purposes, but they also have high potential for physical and psychological dependence. Up to three drugs can be reported in the FARS database, however to reduce the potential bias inherent with multiple OAs only drivers testing positive for one OA were included in the analyses; those drivers positive for two or more OAs were excluded from the responsibility analysis. 2.5. Potential confounders The FARS database provided the information used to control for potential confounders: age, exposure to other drugs, and risky or poor driving history. To isolate the effects of exposure to OAs we controlled for exposure to drugs other than OAs. For this study, drugs other than OAs were categorized as depressants, stimulants, cannabinoids, narcotics, and other drugs. The drugs included in the depressants, stimulants and cannabinoids categories are the same drugs included in the corresponding FARS categories [44]. The category of other drugs combined the hallucinogens, PCP, inhalants, steroids and other drugs categories in FARS due to the small number of drivers testing positive for these categories in our sample. The narcotics category includes all of the drugs in the FARS category of narcotics except Schedule II opioid analgesics. Thus, the narcotics category included illicit, Schedule I substances such as heroin, as well as licit Schedule III, IV and V controlled substances. We also controlled for poor driving skills and risky driving tendencies using past driving history. Specifically, FARS records include the following infraction variables over the previous three years: collisions, DWI convictions, other convictions, speeding, and license suspensions. Each of these variables were dichotomized as present or absent. 2.6. Statistical analyses The annual prevalence rates of positive OA drug tests among drivers of large trucks were determined for each year of FARS data included in the study, and the entire study period (1993–2008) by calculating first, the overall percentage of all truck drivers 20 years of age and older (males, females and unknown sex, regardless of BAC level or alcohol test status) that tested positive for an OA. We further refined our overall prevalence estimates by examining prevalence only for those truck drivers that were drug-tested. Finally, we also calculated prevalence for drivers included in the responsibility analysis. Given that the observed increase in prevalence over time may be an artifact of increased annual drug screening, we also present adjusted prevalence estimates that assume the proportion of drivers tested remains constant at the proportion observed in the initial year of data collection (1993). The responsibility analyses were based on the cases remaining after the inclusion and exclusion criteria were applied (e.g., male truck drivers, 20 and over, drug-tested, BAC = 0, if OA positive only for one OA). Driver age, other drug use, previous driving history, and UDAs committed were compared by OA exposure. An Independent Samples T-Test was used to compare mean age; Pearson Chi-square test, Fisher’s Exact test, and N 1 Chi-square test were used as appropriate [45] to compare other drug use, past driving history, and UDAs. Logistic regression was used to evaluate the hypothesis that truck drivers exposed to OAs had greater odds of committing an
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
UDA compared to non OA exposed drivers. Initially, we evaluated the impact of OAs on the odds of committing an UDA without adjusting for other factors. Next we examined the impact of OAs adjusting for driver age, sex, and potential confounders (previous driving history, poly-drug use) and their interactions with OA status. Age was transformed into decades and centered at 45 years. Given the possibility of a curvilinear relationship between age and UDAs (i.e., the youngest and oldest drivers might have the worst driving performance) we also included the quadratic age term. The final model included OA status, driver age, sex, and poly-drug use, previous driving history and significant two-way interactions identified using a step-down hierarchical process [46]. All statistical analyses were conducted using SPSS version 19. 3. Results 3.1. Prevalence of opioid analgesics As described in Gates 2013 [38] between 1993 and 2008 there were 65,867 driver records for trucks with a GVWR greater than 26,000 lbs or truck-tractors. Of these, 174 drivers were below the age of 20 and 332 drivers had no age recorded. These were excluded, leaving 65,361 truck drivers for the prevalence analysis (99%) of which 10,190 (15.6%) were tested for drugs. The prevalence rates based on all truck drivers fluctuated between 0.1 and 0.2% from 1993 until 2004, after which time the prevalence rates appeared to increase, with the highest rate of 0.5% observed in 2007. For drug-tested and those drivers included in the analysis sample the prevalence rates typically ranged between 1.0 and 1.5% (see Fig. 1(panel A)). The proportion of all truck drivers tested increased from 9.3% in 1993 to 25.5% in 2008. Fig. 1(panel B) presents the adjusted prevalence which assumes the proportion of drivers tested remains constant at the proportion observed in 1993. After adjustment, prevalence rates for all truck drivers remained between 0.1 and 0.2% for the entire study period. Most annual prevalence rates for both drug-tested and analysis sample drivers ranged between 0.5 and 1.0%. Eighty-four percent (N = 8531) of the 10,190 drug-tested drivers had a confirmed BAC of zero. Although research has suggested that the association between OAs and UDAs is mediated by sex [36], female drivers represented only 2.4% (N = 205) of the analysis sample and were removed along with one driver without sex identified. Therefore the analysis sample of 8325 truck drivers included males 20 and over, blood-tested for drugs, with a confirmed BAC of zero either driving trucks with a GVWR greater than 26,000 lbs (21%) or truck-tractors (79%). One-hundred and two truck drivers from the analysis sample tested positive for a single OA, 14 drivers tested positive for two OAs and one driver for three. The majority of drivers testing positive for a single OA tested positive for either: morphine (18.6%), hydrocodone (17.6%), methadone (12.7%), codeine (11.8%), and propoxyphene (10.8%). The complete findings are presented in Table 1. The mean age of drivers testing positive for OAs (M = 45.5, SD =11.3) did not differ from drivers who tested negative (M = 43.5, SD = 11.8, t(8308) = 1.7, p = .083). For those 15 drivers testing positive for multiple OAs, combinations included codeine and hydrocodone (N = 2), codeine and Morphine (N = 2), Opium and morphine (N = 2); codeine (N = 3) with propoxyphene, oxycodone, or opium; hydrocodone (N = 3) with hydromorphone, oxycodone, methadone and oxycodone; propoxyphene (N = 2) with methadone or opium; oxycodone (N = 1) with oxymorphone. Poly-drug use was more common among the drivers that tested positive for OAs. A significantly higher proportion of drivers testing positive for OAs also tested positive for depressants, stimulants, cannabinoids, and other drugs (see Table 2). Overall, previous driving history by OA use
157
was quite similar. Approximately one in four drivers, regardless of OA status had at least one prior speeding or other driving conviction in the past three years. The top five most frequently reported UDAs for truck drivers in the final sample are presented in Table 3. Truck drivers using OAs had significantly greater reported frequencies than non-users of OAs for any UDA (56.9% vs. 41.9%). This difference was mostly accounted for by truck drivers testing positive failing to keep in their proper lane compared to drivers testing negative (35.3% vs. 17.8%). There were no statistically significant differences by OA status for the other UDAs. The crude odds ratio computed from the first logistic regression model indicated that drivers testing positive for an OA had 83% greater odds of committing at least one UDA than drivers testing negative for an OA (Unadjusted OR: 1.83; 95% CI: 1.23; 2.71). For the adjusted model, age-squared (Wald = 8.13, p = .004) interacted significantly with OA use. None of the other drugs detected interacted significantly with OA use. Thus, the final model included: age, age-squared, age*OA status, age-squared*OA status, poly-drug status (depressants, stimulants, cannabinoids, other medications), and previous driving history (crash, DWI, other driving infraction, speeding infraction, suspended license). After this adjustment, drivers testing positive for an OA had 180% greater odds of committing an UDA compared to a truck driver testing negative (OR: 2.80; 95% CI:1.64; 4.81). In addition, truck drivers: (a) testing positive for either stimulants or other medications; or (b) having a previous crash, other driving infraction, or speeding conviction had greater odds of an UDA. See Table 4 for the full results. The interaction between age-squared and OA exposure was examined further given the level of significance for this term in the final model. Predicted odds and odds ratios (and 95% CIs) were generated for select ages (every 10 years, 25 through 75) [36,46,47]. The resulting odds ratios exhibited an inverse curvilinear relationship, with the greatest odds of committing an UDA observed for age centered at 45 and 55 (see Table 5). Given that some truck drivers had multiple OAs detected we re-ran both the unadjusted and final models substituting any presence of OAs (e.g., 0, 1 or more) for just one OA present (e.g., 0 or 1).1 Including the 15 drivers who had 2 or 3 OAs detected had very little impact on the results, unadjusted OR for OAs = 1.86 (95% CI: 1.29; 2.69) with those 15 drivers included versus 1.83 (95% CI: 1.23; 2.71) when they were excluded. For the final model the ORs were 2.65 (95% CI: 1.62; 4.33) when those 15 drivers were included and 2.80 (95% CI: 1.64; 4.81) when they were excluded. 4. Discussion Between 1993 and 2008, the prevalence rate of OA use among all drivers of large trucks, 20 years and older, who were involved in a fatal USA crash fluctuated between 0.1 and 0.2% up until 2004 after which point the prevalence appeared to increase, with the highest rate of 0.46% observed for 2007. Similar, amplified, patterns in OA prevalence rates were seen when examining the more restrictive samples (drug-tested truck drivers; drivers included in the responsibility analysis) which typically ranged between 1.0 and 1.5% annually and in 2007 approached 3.0%. However, the proportion of total drivers tested for drugs increased by almost 150% between 1993 and 2008. Assuming the proportion of drivers tested remained constant at the proportion observed in 1993 resulted in lower, and more consistent, prevalence estimates across the study period. After accounting for age, use of substances 1 We also considered examining the polynomial contrasts (i.e., linear and quadratic effects), similar to the approach we took with a related study [38], but decided against this approach given the sparseness of the data.
158
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
Fig. 1. Percentage of truck drivers testing positive for opioid-analgesics by year (1993–2008) for all truck drivers, drug-tested truck drivers, and truck drivers included in the analysis sample.
other than OAs, and past driving record, truck drivers who tested positive for an OA had 2.80 times greater odds of being responsible for the crash (i.e., committed at least one UDA directly preceding the crash) than truck drivers who tested negative. Middle aged
drivers using OAs had the highest odds of being deemed responsible for the crash; the youngest and oldest drivers had the lowest odds. Exposure to stimulants and other drugs were associated with significantly greater odds of committing an UDA.
P. Reguly et al. / Forensic Science International 234 (2014) 154–161 Table 1 Breakdown of opioid analgesics reported for single opioid-positive drivers (n = 102).
Morphine Hydrocodone Methadone Codeine Propoxyphene Opium Meperidine (Pethidine) Oxycodone Dihydrocodeine Fentanyl Oxymorphone Hydromorphone Metopon
n
%
19 18 13 12 11 8 7 5 3 3 1 1 1
18.6 17.6 12.7 11.8 10.8 7.8 6.9 4.9 2.9 2.9 1.0 1.0 1.0
Note: Codeine includes codeine (6 cases) and acetominophen + codeine (6 cases).
One or more collisions, speeding infractions, and other convictions in the past three years were also associated with significantly greater odds of committing an UDA. It is important to note that because this study only used data from truck drivers with a confirmed BAC of zero and confirmed blood test for drugs, the findings can only be generalized to drivers with these characteristics. In our study, the annual prevalence of opioid use among all CMV drivers of large trucks involved in fatal crashes did not exceed 0.46% for any year in the study period and mostly ranged between 0.1 and 0.2%. Prevalence rates were lower after adjustment. The low prevalence rates for all CMV truck drivers observed in this study were somewhat lower than other prevalence rates in CMV truck drivers likely due to the fact that the other studies either included fatally-injured truck drivers [13] or used urine to test for presence of OAs [14,20]. The low prevalence rates of opioid use
159
observed among CMV drivers of large trucks may be due in part to the policies and regulations for drug use (licit and illicit) in the CMV industry. Further, our prevalence rates among all CMV truck drivers (ranging between 0.1 and 0.5%) would have underestimated the actual OA prevalence rates because blood drug test results were available for only 15.5% of all truck drivers, yet the truck drivers without blood drug tests results were classified as negative (for this particular prevalence analysis). If we presume that at least some of the truck drivers without drug test results would have tested positive for an OA, the prevalence rates would have been higher. Conversely, our prevalence estimates using the more restrictive samples (drug-tested truck drivers; drivers included in the responsibility analysis) which were typically 1.0–1.5% annually may provide an overestimation for the general population. The positive association between OA use by drivers of large trucks and crash responsibility supported the study hypothesis. The association between OA use and crash culpability was the same direction, but of greater magnitude than other studies investigating the issue for passenger vehicle drivers (OR: 1.72; 95% CI: 1.45; 2.03) [36] and all vehicle drivers who were fatally injured in crashes (OR: 1.41; 95% CI: 0.7; 2.9) [18]. One possible explanation for the greater odds ratio for truck driver culpability is that the sedative effects or other side effects of OAs are compounding the fatigue experienced by truck drivers, but further investigation into the synergistic effects of OA side effects and fatigue is needed. Another potential explanation is that the various OAs recorded may have different effects on drivers. The most common OAs detected in this study were morphine, hydrocodone, methadone, and codeine which accounted for 60.8% of the OAs detected. They form a heterogeneous group, with a range of potencies and pharmacological properties [48,49]. However, due to the small
Table 2 Chi-square results for drug use and driving records. Opioid analgesic positive n (%) (N = 102)a
Opioid analgesic negative n (%) (N = 8208)
Drugs Depressants Stimulants Cannabinoids Narcotics Other drugs
17 13 6 1 22
(16.7%) (12.7%) (5.9%) (1.0%) (21.6%)
66 288 177 6 283
Driving record Collisions DWI convictions Other Convictions Speeding Suspensions
15 1 26 27 12
(14.7%) (1.0%) (25.5%) (26.5%) (8.3%)
1444 77 2081 2297 661
a b c d
(0.8%) (3.5%) (2.2%) (0.1%) (3.4%)
(17.6%) (0.9%) (25.4%) (28.0%) (8.1%)
x2 (Pearson or N 1)
p-Value
256.37 24.62 6.49 – 93.56
Contents lists available at ScienceDirect
Forensic Science International journal homepage: www.elsevier.com/locate/forsciint
Examining the impact of opioid analgesics on crash responsibility in truck drivers involved in fatal crashes Paula Reguly a,b, Sacha Dubois a,b,c,d,*, Michel Be´dard a,b,c,d a
Department of Health Sciences, Lakehead University, 955 Oliver Road, Thunder Bay, ON P7B 5E1, Canada Centre for Research on Safe Driving, Lakehead University, 955 Oliver Road, Thunder Bay, ON P7B 5E1, Canada c Research Department, St. Joseph’s Care Group, 580 North Algoma Street, Thunder Bay, ON P7B 5G4, Canada d Northern Ontario School of Medicine, Human Sciences Division, Lakehead University, 955 Oliver Road, Thunder Bay, ON P7B 5E1, Canada b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 27 June 2013 Received in revised form 28 October 2013 Accepted 6 November 2013 Available online 15 November 2013
Introduction: Commercial motor vehicle (CMV) drivers, particularly drivers of large trucks continue to be a population of concern regarding traffic safety despite the reduction in large truck crash rates over the past decade. Medication and drug use while driving is one important risk factor for large truck crashes. Work-related exposures, such as vibration, manual handling and poor ergonomics contribute to an increased risk for injuries and chronic conditions and are common reasons for opioid analgesic (OA) use by CMV truck drivers. The objectives of this study were to examine the role of OA use in CMV truck drivers involved in fatal crashes by: (a) generating prevalence estimates of OA use; (b) documenting the relationship between OA use and crash responsibility. Methods: Case-control study using logistic regression to compare Fatality Analysis Reporting System (1993–2008) record of one or more crash-related unsafe driver actions (UDAs – a proxy measure of responsibility) between drivers with a positive drug test and drivers with a negative drug test for OA, controlling for age, other drug use, and driving history. Results: The annual prevalence of OA use among all CMV drivers of large trucks involved in fatal crashes did not exceed 0.46% for any year in the study period and mostly ranged between 0.1 and 0.2%. Male truck drivers using OA had greater odds of committing an UDA (OR: 2.80; 95% CI: 1.64; 4.81). Middleaged users had greater odds than younger or older users. Conclusion: The results of our study indicate that the presence of OAs is associated with greater odds of committing an UDA. This association may have implications for the commercial transport industry and traffic safety. However, the limited prevalence of OA use is encouraging and further research is needed to address the limitations of the study. ß 2013 Elsevier Ireland Ltd. All rights reserved.
Keywords: Opioid analgesics Truck drivers Crash culpability Impairment
1. Introduction Commercial motor vehicle (CMV) drivers, particularly drivers of large trucks continue to be a population of concern regarding traffic safety despite the reduction in large truck crash rates over the past decade [1]. Occupational fatality rates for the transport industry are consistently among the highest [2], and in large truck crashes involving multiple vehicles there is a tendency for the occupants of the other vehicle(s) to be severely injured or killed [3– 5]. There are also significant economic costs incurred by large truck
* Corresponding author at: Research Department, St. Joseph’s Care Group, LPH Site, 580 N. Algoma Street, P.O. Box 2930, Thunder Bay, Ontario P7B 5G4, Canada. Tel.: +1 807 343 4300x4480; fax: +1 807 346 5243. E-mail addresses: [email protected] (P. Reguly), [email protected] (S. Dubois), [email protected] (M. Be´dard). 0379-0738/$ – see front matter ß 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.forsciint.2013.11.005
crashes through medical costs and insurance settlements, lost productivity, and property damage [6]. Medication and drug use while driving is one important risk factor for large truck crashes. Because driving is a complex activity that involves a range of cognitive and psychomotor functions, both licit and illicit drugs effect on the central nervous system can impair driving ability [7]. A recent large-scale study on large truck crash causation found that prescription drug use and over the counter drug use was among the top ten factors associated with crashes out of hundreds of factors examined [8]. Nonetheless, the impact of drug use on driving performance has received little research attention [4], despite the potential for impairment. Opioid analgesics (OAs) are one of the drug groups warranting further attention. Opioid analgesics are most commonly used to treat pain but they are also used in substitution therapy to treat substance abuse [9]. The term opioid refers to naturally occurring, synthetic and
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
semi-synthetic compounds that are derived from the opium of the poppy plant, such as codeine, morphine, oxycodone and hydromorphone [9]. According to the American Chronic Pain Association (ACPA), common and anticipated central nervous system (CNS) side effects of OAs are thought and memory impairment, drowsiness and nausea, mild sedation, and impaired judgment and co-ordination. The ACPA warns against driving until tolerance or a baseline is reached [9]. With respect to OA use and driving performance, there are several reasons why commercial large truck drivers are a population of interest. Truck drivers experience work-related exposures such as vibration, manual handling and ergonomic factors that may increase the risk for injuries and chronic conditions (e.g., lower back pain and musculoskeletal disorders) [10–12] that are typically treated with OAs. In fact, treatment for chronic pain has been recognized as the most common reason for opioid use by CMV truck drivers [4]. Also, fatigue has been recognized as a risk-factor for large truck crashes [5,8,13] and the side effects associated with OAs, such as sedation, might compound the risk of crashes. Finally, there is some evidence that drugs are used by many truck drivers to cope with boredom and other aspects of the job [14]. Beyond these job-related factors, there have been dramatic increases in the prescribing and use of OAs in the general population over the last two decades [4,15–17], as well as some indication of an increased prevalence of detection of opioids among drivers in the general population [18,19]. However, studies of opioid use among CMV truck drivers have found a low prevalence of opioid use (generally not exceeding 4.0% of the study sample) compared to other types of drugs such as stimulants, depressants and cannabinoids [13,14,20,21]. Prevalence studies that have relied on the voluntary provision of biological samples [14,21] may have resulted in the under-detection, and hence underestimation of prevalence rates. Conversely, the use of fatally injured drivers as a study sample population provides an overestimate of prevalence rates within the wider truck driver population [13]. Finally, these prevalence studies [13,14,20,21] have not addressed long-term trends and may not accurately reflect current prevalence rates, especially considering the recent dramatic increase in the prescribing and use of OAs in the general population. It should be noted that even a low prevalence of opioid use can translate into a considerable absolute number of opioid users, given that, in 2012, there were 5,700,000 CMV drivers operating in the U.S. alone [22]. Based on this figure, a prevalence rate of 2.0% among U.S. CMV drivers is equivalent to 114,000 opioid users. Experimental studies investigating the effect of OAs on driving ability have used driving simulators, on-road driving, and cognitive and psychomotor tests specifically for driving. The key findings from this research are that therapeutic long-term stable doses of opioids do not negatively impact driving ability [23–28], but a change in dosing of opioid medication (30% increase in dose) results in significant cognitive impairment [23]. Furthermore, administering an OA to healthy individuals not previously exposed to OAs did not significantly affect driving ability, but study participants had significantly reduced pupil size and they reported that significantly more effort was needed to perform the driving test and reported significantly more sedation and reduced alertness [29]. But there are several limitations of the experimental research regarding the effect of OAs on driving ability, such as small sample sizes and insufficient statistical power; the use of healthy and young study participants; insufficient doses to elicit effects; and highly controlled environments that do not always reflect actual driving conditions and experiences. There is scant research from observational studies investigating the association between OA use and crash risk and crash
155
responsibility among CMV drivers of large trucks. We found only one study regarding the association between opioid use and crash risk among CMV drivers. Howard et al. [30] collected self-reported information on crash history, drug and alcohol use from a sample of CMV drivers and found that drivers that reported using narcotic analgesics had greater odds of being involved in a crash than unexposed drivers after adjusting for age, hours of driving and alcohol intake (Adjusted OR: 2.40; CI: 1.46; 3.92). There are, however, a number of observational studies, using samples from the general population, that have investigated the effect of OA use on crash risk and crash responsibility. A review by Fishbain et al. [7] concluded that there was consistent evidence that opioids are not associated with crashes; but the methodological shortcomings of many of the studies reviewed, such as the lack of control group, weaken their conclusion. A meta-analysis by Mona´rrez-Espino et al. considered studies from the period between 1991 and 2012 specifically for older drivers (55 years old) [31]. They argued that the evidence fails to provide convincing support that opioids are associated with increase crash risk among drivers 55 and older, due to the small number of studies, the generally inadequate control of confounders (e.g., other medications and illness), and inconsistent results. Conversely, the DRUID Project compared the risk of being seriously injured or killed while driving with psychoactive substances including medicinal opioid use [32,33]. Pooled data from six European countries was used. Cases were obtained from hospitalization data, controls were obtained from roadside surveys. Using odds ratios as an estimate of risk, the DRUID Project reported that drivers had significantly increased estimates of risk of being seriously injured (OR: 9.06; CI: 6.40–12.83) or killed (RR: 4.82; CI: 2.60–8.93) when positive for medicinal opioids. Other studies, using samples from the general population, have demonstrated a small, but positive association between opioid use and increased crash risk [34,35]. Arguably better evidence for the effect of opioids on crash involvement comes from crash culpability studies. Dubois et al. [36] demonstrated a significant positive association between OAs and crash culpability using a sample of passenger vehicle drivers (survivors and fatally injured) who were involved in fatal crashes (controlling for other medications, driving history and other factors). Drummer et al. [18] showed a positive but non-significant association between opioid use and crash culpability study using a sample of fatally injured drivers of all vehicle types (controlling for crash type, drug use and other factors). The DRUID Project considered only illicit opiates in its culpability analysis, and found that the odds of being responsible for a fatal crash did not differ significantly between the drivers that tested positive for illicit opiates and those that tested negative [32]. Overall, factors that may be contributing to the observed inconsistencies for the association between opioids and crash risk and crash culpability include sample type (e.g., fatally vs. no-fatally injured), opioid(s) of interest, determination of opioid exposure (e.g. prescription records vs. blood or urine test), and study design (e.g., cohort or case-control vs. cross-sectional). We designed our study to: (a) generate estimates of the prevalence of opioid use over time in a sample of drivers of large trucks; and (b) document the relationship between OA use (Schedule II opioid analgesics in particular) and crash responsibility among drivers of large trucks involved in fatal crashes using a representative data source with high external validity and standardized toxicological testing, while controlling for other possible contributory factors to crash initiation including age, previous driving history, and substance use (other than OAs) such as alcohol and other medications. We hypothesized that the odds of committing an unsafe driving action (UDA) preceding a fatal crash would be greater for truck drivers who tested positive for an OA compared to truck drivers who tested negative.
156
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
2. Materials and methods
using UDAs as proxy measures of crash responsibility has been demonstrated in several other studies [3,36,38–42].
2.1. Data source 2.4. Opioid analgesic classification and exposure Data were obtained from the Fatality Analysis Reporting System (FARS), available from the U.S. National Center for Statistics and Analysis of the National Highway Traffic Safety Administration. It is a repository of information pertaining to fatal traffic crashes occurring in all 50 states, the District of Columbia, and Puerto Rico, dating back to 1975. A crash qualifies for inclusion if it involves a motor vehicle traveling on a traffic way customarily open to the public, and results in the death of a person (i.e., vehicle occupant or non-motorist) within 30 days of the crash [37]. Crash information is collected, coded, and entered into the FARS database according to standard methods by trained FARS analysts [37]. FARS analysts obtain information from a variety of sources such as police accident reports, hospital medical reports, coroner reports, and state vehicle registration and licensing files. The FARS data are organized into three levels. The crash level includes crash time and location, the number of vehicles and people involved, and weather conditions. The vehicle level includes ‘‘the vehicle type, role in the crash, initial and principal impacts points, the most harmful event, the driver’s record, and license status.’’ The person level includes data on the people involved in the crash (e.g., age, gender, role in the crash [e.g., driver or passenger], alcohol and drug involvement, injury, severity). Information entered into the database is subjected to a rigorous Quality Control Program which includes checks for consistency, timeliness, completeness, and accuracy [37]. 2.2. Sample We used the same sampling strategy as Gates 2013 [38]. In brief, the sample was based on crashes between 1993 and 2008. The prevalence analysis included drivers operating a single or combination-unit heavy truck with a gross vehicle weight rating of greater than 26,000 lbs or a truck-tractor (cab only, or with any number of trailing units; any weight). To ensure a minimal level of driver history could have occurred we excluded drivers below the age of 20. Two additional criteria were applied to those drivers included in the crash responsibility analysis. First, to rule out potential confounding effects of alcohol only drivers with a confirmed BAC of zero were included. Second, all drivers included in the crash responsibility analysis were blood tested for drugs. 2.3. Design Cases and controls were classified according to the presence (case) or absence (controls) of at least one recorded UDA. The UDAs served as a proxy measure of crash responsibility. Drivers with any UDA(s) recorded were deemed to have contributed to the initiation of the collision. Blower has argued that driver-related factors (i.e., UDAs) are preferable to traffic violations as indicators of crash culpability because not all contributing factors are chargeable offenses; traffic violations are not uniformly applied or enforced; and various factors influence whether a violation is charged including the severity of the crash and the degree of supporting evidence [3]. FARS analysts can enter up to four driver-related factors (i.e., UDAs) (three before 1997). The driver-related factors are based on crash narrative in the police report and ‘‘the narrative allows a fuller description of the factors that contributed to the crash’’ [3]. Finally, driver-related factors have been shown to be consistent with the physical evidence and configuration of crashes, which in turn are strongly related to the relative contribution of drivers in most types of large truck crashes [3]. This method of
Opioid analgesics were defined as Schedule II opioid analgesics in accordance with section 1308.12 of the U.S. Code of Federal Regulations [43]. All Schedule II substances are licit drugs with legitimate medical purposes, but they also have high potential for physical and psychological dependence. Up to three drugs can be reported in the FARS database, however to reduce the potential bias inherent with multiple OAs only drivers testing positive for one OA were included in the analyses; those drivers positive for two or more OAs were excluded from the responsibility analysis. 2.5. Potential confounders The FARS database provided the information used to control for potential confounders: age, exposure to other drugs, and risky or poor driving history. To isolate the effects of exposure to OAs we controlled for exposure to drugs other than OAs. For this study, drugs other than OAs were categorized as depressants, stimulants, cannabinoids, narcotics, and other drugs. The drugs included in the depressants, stimulants and cannabinoids categories are the same drugs included in the corresponding FARS categories [44]. The category of other drugs combined the hallucinogens, PCP, inhalants, steroids and other drugs categories in FARS due to the small number of drivers testing positive for these categories in our sample. The narcotics category includes all of the drugs in the FARS category of narcotics except Schedule II opioid analgesics. Thus, the narcotics category included illicit, Schedule I substances such as heroin, as well as licit Schedule III, IV and V controlled substances. We also controlled for poor driving skills and risky driving tendencies using past driving history. Specifically, FARS records include the following infraction variables over the previous three years: collisions, DWI convictions, other convictions, speeding, and license suspensions. Each of these variables were dichotomized as present or absent. 2.6. Statistical analyses The annual prevalence rates of positive OA drug tests among drivers of large trucks were determined for each year of FARS data included in the study, and the entire study period (1993–2008) by calculating first, the overall percentage of all truck drivers 20 years of age and older (males, females and unknown sex, regardless of BAC level or alcohol test status) that tested positive for an OA. We further refined our overall prevalence estimates by examining prevalence only for those truck drivers that were drug-tested. Finally, we also calculated prevalence for drivers included in the responsibility analysis. Given that the observed increase in prevalence over time may be an artifact of increased annual drug screening, we also present adjusted prevalence estimates that assume the proportion of drivers tested remains constant at the proportion observed in the initial year of data collection (1993). The responsibility analyses were based on the cases remaining after the inclusion and exclusion criteria were applied (e.g., male truck drivers, 20 and over, drug-tested, BAC = 0, if OA positive only for one OA). Driver age, other drug use, previous driving history, and UDAs committed were compared by OA exposure. An Independent Samples T-Test was used to compare mean age; Pearson Chi-square test, Fisher’s Exact test, and N 1 Chi-square test were used as appropriate [45] to compare other drug use, past driving history, and UDAs. Logistic regression was used to evaluate the hypothesis that truck drivers exposed to OAs had greater odds of committing an
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
UDA compared to non OA exposed drivers. Initially, we evaluated the impact of OAs on the odds of committing an UDA without adjusting for other factors. Next we examined the impact of OAs adjusting for driver age, sex, and potential confounders (previous driving history, poly-drug use) and their interactions with OA status. Age was transformed into decades and centered at 45 years. Given the possibility of a curvilinear relationship between age and UDAs (i.e., the youngest and oldest drivers might have the worst driving performance) we also included the quadratic age term. The final model included OA status, driver age, sex, and poly-drug use, previous driving history and significant two-way interactions identified using a step-down hierarchical process [46]. All statistical analyses were conducted using SPSS version 19. 3. Results 3.1. Prevalence of opioid analgesics As described in Gates 2013 [38] between 1993 and 2008 there were 65,867 driver records for trucks with a GVWR greater than 26,000 lbs or truck-tractors. Of these, 174 drivers were below the age of 20 and 332 drivers had no age recorded. These were excluded, leaving 65,361 truck drivers for the prevalence analysis (99%) of which 10,190 (15.6%) were tested for drugs. The prevalence rates based on all truck drivers fluctuated between 0.1 and 0.2% from 1993 until 2004, after which time the prevalence rates appeared to increase, with the highest rate of 0.5% observed in 2007. For drug-tested and those drivers included in the analysis sample the prevalence rates typically ranged between 1.0 and 1.5% (see Fig. 1(panel A)). The proportion of all truck drivers tested increased from 9.3% in 1993 to 25.5% in 2008. Fig. 1(panel B) presents the adjusted prevalence which assumes the proportion of drivers tested remains constant at the proportion observed in 1993. After adjustment, prevalence rates for all truck drivers remained between 0.1 and 0.2% for the entire study period. Most annual prevalence rates for both drug-tested and analysis sample drivers ranged between 0.5 and 1.0%. Eighty-four percent (N = 8531) of the 10,190 drug-tested drivers had a confirmed BAC of zero. Although research has suggested that the association between OAs and UDAs is mediated by sex [36], female drivers represented only 2.4% (N = 205) of the analysis sample and were removed along with one driver without sex identified. Therefore the analysis sample of 8325 truck drivers included males 20 and over, blood-tested for drugs, with a confirmed BAC of zero either driving trucks with a GVWR greater than 26,000 lbs (21%) or truck-tractors (79%). One-hundred and two truck drivers from the analysis sample tested positive for a single OA, 14 drivers tested positive for two OAs and one driver for three. The majority of drivers testing positive for a single OA tested positive for either: morphine (18.6%), hydrocodone (17.6%), methadone (12.7%), codeine (11.8%), and propoxyphene (10.8%). The complete findings are presented in Table 1. The mean age of drivers testing positive for OAs (M = 45.5, SD =11.3) did not differ from drivers who tested negative (M = 43.5, SD = 11.8, t(8308) = 1.7, p = .083). For those 15 drivers testing positive for multiple OAs, combinations included codeine and hydrocodone (N = 2), codeine and Morphine (N = 2), Opium and morphine (N = 2); codeine (N = 3) with propoxyphene, oxycodone, or opium; hydrocodone (N = 3) with hydromorphone, oxycodone, methadone and oxycodone; propoxyphene (N = 2) with methadone or opium; oxycodone (N = 1) with oxymorphone. Poly-drug use was more common among the drivers that tested positive for OAs. A significantly higher proportion of drivers testing positive for OAs also tested positive for depressants, stimulants, cannabinoids, and other drugs (see Table 2). Overall, previous driving history by OA use
157
was quite similar. Approximately one in four drivers, regardless of OA status had at least one prior speeding or other driving conviction in the past three years. The top five most frequently reported UDAs for truck drivers in the final sample are presented in Table 3. Truck drivers using OAs had significantly greater reported frequencies than non-users of OAs for any UDA (56.9% vs. 41.9%). This difference was mostly accounted for by truck drivers testing positive failing to keep in their proper lane compared to drivers testing negative (35.3% vs. 17.8%). There were no statistically significant differences by OA status for the other UDAs. The crude odds ratio computed from the first logistic regression model indicated that drivers testing positive for an OA had 83% greater odds of committing at least one UDA than drivers testing negative for an OA (Unadjusted OR: 1.83; 95% CI: 1.23; 2.71). For the adjusted model, age-squared (Wald = 8.13, p = .004) interacted significantly with OA use. None of the other drugs detected interacted significantly with OA use. Thus, the final model included: age, age-squared, age*OA status, age-squared*OA status, poly-drug status (depressants, stimulants, cannabinoids, other medications), and previous driving history (crash, DWI, other driving infraction, speeding infraction, suspended license). After this adjustment, drivers testing positive for an OA had 180% greater odds of committing an UDA compared to a truck driver testing negative (OR: 2.80; 95% CI:1.64; 4.81). In addition, truck drivers: (a) testing positive for either stimulants or other medications; or (b) having a previous crash, other driving infraction, or speeding conviction had greater odds of an UDA. See Table 4 for the full results. The interaction between age-squared and OA exposure was examined further given the level of significance for this term in the final model. Predicted odds and odds ratios (and 95% CIs) were generated for select ages (every 10 years, 25 through 75) [36,46,47]. The resulting odds ratios exhibited an inverse curvilinear relationship, with the greatest odds of committing an UDA observed for age centered at 45 and 55 (see Table 5). Given that some truck drivers had multiple OAs detected we re-ran both the unadjusted and final models substituting any presence of OAs (e.g., 0, 1 or more) for just one OA present (e.g., 0 or 1).1 Including the 15 drivers who had 2 or 3 OAs detected had very little impact on the results, unadjusted OR for OAs = 1.86 (95% CI: 1.29; 2.69) with those 15 drivers included versus 1.83 (95% CI: 1.23; 2.71) when they were excluded. For the final model the ORs were 2.65 (95% CI: 1.62; 4.33) when those 15 drivers were included and 2.80 (95% CI: 1.64; 4.81) when they were excluded. 4. Discussion Between 1993 and 2008, the prevalence rate of OA use among all drivers of large trucks, 20 years and older, who were involved in a fatal USA crash fluctuated between 0.1 and 0.2% up until 2004 after which point the prevalence appeared to increase, with the highest rate of 0.46% observed for 2007. Similar, amplified, patterns in OA prevalence rates were seen when examining the more restrictive samples (drug-tested truck drivers; drivers included in the responsibility analysis) which typically ranged between 1.0 and 1.5% annually and in 2007 approached 3.0%. However, the proportion of total drivers tested for drugs increased by almost 150% between 1993 and 2008. Assuming the proportion of drivers tested remained constant at the proportion observed in 1993 resulted in lower, and more consistent, prevalence estimates across the study period. After accounting for age, use of substances 1 We also considered examining the polynomial contrasts (i.e., linear and quadratic effects), similar to the approach we took with a related study [38], but decided against this approach given the sparseness of the data.
158
P. Reguly et al. / Forensic Science International 234 (2014) 154–161
Fig. 1. Percentage of truck drivers testing positive for opioid-analgesics by year (1993–2008) for all truck drivers, drug-tested truck drivers, and truck drivers included in the analysis sample.
other than OAs, and past driving record, truck drivers who tested positive for an OA had 2.80 times greater odds of being responsible for the crash (i.e., committed at least one UDA directly preceding the crash) than truck drivers who tested negative. Middle aged
drivers using OAs had the highest odds of being deemed responsible for the crash; the youngest and oldest drivers had the lowest odds. Exposure to stimulants and other drugs were associated with significantly greater odds of committing an UDA.
P. Reguly et al. / Forensic Science International 234 (2014) 154–161 Table 1 Breakdown of opioid analgesics reported for single opioid-positive drivers (n = 102).
Morphine Hydrocodone Methadone Codeine Propoxyphene Opium Meperidine (Pethidine) Oxycodone Dihydrocodeine Fentanyl Oxymorphone Hydromorphone Metopon
n
%
19 18 13 12 11 8 7 5 3 3 1 1 1
18.6 17.6 12.7 11.8 10.8 7.8 6.9 4.9 2.9 2.9 1.0 1.0 1.0
Note: Codeine includes codeine (6 cases) and acetominophen + codeine (6 cases).
One or more collisions, speeding infractions, and other convictions in the past three years were also associated with significantly greater odds of committing an UDA. It is important to note that because this study only used data from truck drivers with a confirmed BAC of zero and confirmed blood test for drugs, the findings can only be generalized to drivers with these characteristics. In our study, the annual prevalence of opioid use among all CMV drivers of large trucks involved in fatal crashes did not exceed 0.46% for any year in the study period and mostly ranged between 0.1 and 0.2%. Prevalence rates were lower after adjustment. The low prevalence rates for all CMV truck drivers observed in this study were somewhat lower than other prevalence rates in CMV truck drivers likely due to the fact that the other studies either included fatally-injured truck drivers [13] or used urine to test for presence of OAs [14,20]. The low prevalence rates of opioid use
159
observed among CMV drivers of large trucks may be due in part to the policies and regulations for drug use (licit and illicit) in the CMV industry. Further, our prevalence rates among all CMV truck drivers (ranging between 0.1 and 0.5%) would have underestimated the actual OA prevalence rates because blood drug test results were available for only 15.5% of all truck drivers, yet the truck drivers without blood drug tests results were classified as negative (for this particular prevalence analysis). If we presume that at least some of the truck drivers without drug test results would have tested positive for an OA, the prevalence rates would have been higher. Conversely, our prevalence estimates using the more restrictive samples (drug-tested truck drivers; drivers included in the responsibility analysis) which were typically 1.0–1.5% annually may provide an overestimation for the general population. The positive association between OA use by drivers of large trucks and crash responsibility supported the study hypothesis. The association between OA use and crash culpability was the same direction, but of greater magnitude than other studies investigating the issue for passenger vehicle drivers (OR: 1.72; 95% CI: 1.45; 2.03) [36] and all vehicle drivers who were fatally injured in crashes (OR: 1.41; 95% CI: 0.7; 2.9) [18]. One possible explanation for the greater odds ratio for truck driver culpability is that the sedative effects or other side effects of OAs are compounding the fatigue experienced by truck drivers, but further investigation into the synergistic effects of OA side effects and fatigue is needed. Another potential explanation is that the various OAs recorded may have different effects on drivers. The most common OAs detected in this study were morphine, hydrocodone, methadone, and codeine which accounted for 60.8% of the OAs detected. They form a heterogeneous group, with a range of potencies and pharmacological properties [48,49]. However, due to the small
Table 2 Chi-square results for drug use and driving records. Opioid analgesic positive n (%) (N = 102)a
Opioid analgesic negative n (%) (N = 8208)
Drugs Depressants Stimulants Cannabinoids Narcotics Other drugs
17 13 6 1 22
(16.7%) (12.7%) (5.9%) (1.0%) (21.6%)
66 288 177 6 283
Driving record Collisions DWI convictions Other Convictions Speeding Suspensions
15 1 26 27 12
(14.7%) (1.0%) (25.5%) (26.5%) (8.3%)
1444 77 2081 2297 661
a b c d
(0.8%) (3.5%) (2.2%) (0.1%) (3.4%)
(17.6%) (0.9%) (25.4%) (28.0%) (8.1%)
x2 (Pearson or N 1)
p-Value
256.37 24.62 6.49 – 93.56
