Stimulant drug misuse in patients presenting to level 1 trauma centres: a retrospective analysis of 80 000 patients in Canada and the United States

Abstract

Background

: Despite increasing global misuse of stimulant drugs, there is a paucity of literature regarding stimulant use among trauma patients. The primary aim of this study was to assess hospital mortality among patients with stimulant intoxication who presented to level 1 trauma centres in Canada and the United States. Secondary outcomes included emergency department mortality, hospital length of stay, and admission to the intensive care unit (ICU).

Methods

: This is a retrospective analysis of data from the Trauma Quality Improvement Program, including patients presenting emergently to participating level 1 trauma centres from January 2017 to December 2018. We included patients aged 19 to 88 years who had positive toxicology to stimulant drugs or negative toxicology to all drugs. We adjusted for patient demographic factors in the analysis.

Results

: We included 89 130 patients, with 10 017 patients (11.2%) having a positive screen for stimulants. The mean age in the stimulants cohort was 42.0 years, with a 75.9% male preponderance and mean Injury Severity Score of 9. The control cohort was older, with a mean age of 55 years, and a smaller proportion of males (63.0%, p < 0.001). We found no significant differences between emergency department mortality and length of stay in the ICU after we adjusted for confounders. Patients with stimulants in their system were predicted to spend 1.11 (95% confidence interval 1.09 to 1.13) times as many days in hospital as controls, after we accounted for confounders.

Conclusion

: Stimulant misuse is a relevant issue in the trauma population, associated with increased hospital length of stay, but no significant difference in mortality. By addressing stimulant intoxication in trauma patients, health care providers can work toward improving patient outcomes and reducing the associated health care burden.

In 2019, about 1.1 million Canadians had used illegal substances within the past year, most under the age of 25 years.¹ The Canadian Alcohol and Drugs Survey has shown a steady increase in the prevalence of illegal drug use (e.g., cocaine, methamphetamines, ecstasy, hallucinogens, heroin, and salvia) since 2013.¹ The Canadian Centre on Substance Use and Addiction has shown an increase in the per-person cost of substance use of 11.8% from 2007 to 2020.² The increase in per-person cost of substance use for stimulants (including amphetamines) was the third largest overall, from Can$46 in 2007 to $80 in 2020.² Given that about 10% of trauma patients are found to test positive for stimulants in their urine or serum toxicology on presentation, there is an increased interest in determining the impact of stimulants on clinical outcomes.^3,4

N-methyl-3,4-methylenedioxy-amphetamine, or 3,4-methylenedioxy-methamphetamine (MDMA), is a relatively common street drug, often referred to as “ecstasy.”⁵ The chemical composition of MDMA differs slightly from that of amphetamine and methamphetamine by 1 methylenedioxy group attached to the aromatic ring of the amphetamine molecule (i.e., it is “ring-substituted”).⁵ Like amphetamine and methamphetamine, MDMA is a completely synthetic substance.^5,6 With the chemical structure of these drugs resembling that of epinephrine and dopamine, they induce a sense of euphoria, restlessness, and increased body temperature. With amphetamines as well as other stimulants, users can experience a “crash” around 48 hours after drug use, which can manifest as difficulty concentrating, depression, anxiety, and fatigue.⁷

Cocaine, another common stimulant drug of abuse, causes tachycardia, hypertension, and vasoconstriction via its action on α and β receptors within the body.⁸ Multiple studies have shown the effect of cocaine on the cardiovascular and vascular systems. Cocaine induces vasoconstriction, endothelial dysfunction, and accelerated atherosclerosis, leading to an increased risk of thromboembolic complications, such as stroke and myocardial infarction.⁹

There are multiple well-documented acute medical sequelae of stimulant use, including hypertension, cardiac arrythmias, electrolyte imbalance, and psychosis.^10,11 Although vital signs have not been shown to be significantly different in patients with substance ingestion compared with controls, the effects of acute stimulant intoxication can increase the complexity of the clinical presentation in trauma patients.^4,12 These patients can also present with increased agitation secondary to amphetamines, or cocaine-induced chest pain in addition to injuries from a traumatic mechanism, and therefore, early recognition is important.^3,14 The American College of Surgeons Committee on Trauma has advocated for routine toxicology testing in all trauma patients on trauma team activation, and multiple subsequent studies have supported this as well.^15–17 However, guidelines on trauma team activation vary by institution, and therefore this testing is not always carried out.^16,18

Substance abuse is an important risk factor for poorer health care outcomes, which can lead to increased hospital stays and health care costs.¹⁹ Studies have supported an increased risk of traumatic injury among patients with stimulant intoxication.^4,20 Several studies have also linked stimulant use with increased use of health care resources, compounding the impact stimulant use has on public health.^4,21,22 With higher rates of substance abuse and known adverse effects, there is a need for further study to better characterize the impact of stimulant drug misuse in the trauma patient.

Despite the increasing prevalence of stimulant drug misuse and associated health care costs, there is a paucity of literature regarding the prevalence of stimulant drugs among trauma patients. This study aims to better define the effect of stimulant intoxication on mortality, hospital length of stay, and intensive care unit (ICU) admission among trauma patients. By understanding the effects of stimulant intoxication in this population, health care providers can implement strategies for improved care and resource allocation.

Methods

Data source and cohort collection

This retrospective analysis used data from the Trauma Quality Improvement Program (TQIP), encompassing all patients aged 19 to 88 years admitted to participating trauma centres between January 2017 and December 2018. The program is a comprehensive database that prospectively collects information (including demographic and clinical outcome data) on all patients presenting to level 1 trauma centres throughout Canada and the United States. All adults aged 19 to 88 years who were tested by the hospital’s drug panel and who presented to the emergency department following a trauma team activation were included. Patients who tested positive for 1 or more non-stimulant illicit drugs on urine or serum toxicology were excluded from the study. Our comparator group included patients who tested negative for all drugs on urine and serum toxicology and were admitted following a trauma team activation. Allocation into intervention and comparator groups was based on first recorded positive drug screen results within 24 hours of first hospital encounter. Different toxicology testing with varying sensitivity and specificity may have been used at different centres.

We extracted data from the TQIP for all patients meeting the inclusion criteria, including demographic variables, Injury Severity Score (ISS), urine or serum toxicology results (alcohol, stimulants, or both), hospital length of stay, length of stay in the ICU, and death. We counted each partial or full day as 1 calendar day when measuring length of stay in the ICU.

The primary outcome was overall mortality. Secondary outcomes included emergency department mortality, hospital length of stay, and ICU length of stay. Deaths that occurred while patients were in the emergency department were excluded in the calculation of hospital mortality.

Statistical analysis

We conducted descriptive statistics and univariate analyses for all variables. We expressed frequencies as percentages and means (medians with interquartile range [IQR] if the variable was not normally distributed) with standard deviations (SDs) to describe our overall cohort. We used the χ² test to compare categorical variables, and we used the Student t test or Mann–Whitney U test for continuous variables. A review of the literature was performed to identify independent predictors of mortality.^23–25 We performed multivariable logistic regression models to evaluate the effect of the presence of stimulants among trauma patients on hospital mortality, adjusting for covariates, including age, sex, smoking, body mass index (BMI), and comorbidities (i.e., advanced directive limiting care, anticoagulant therapy, cirrhosis, chronic obstructive pulmonary disease, cerebrovascular accident, diabetes, disseminated cancer, congestive heart failure, myocardial infarction, mental or personality disorder, and chronic renal failure). We reported the adjusted risk for death with odds ratios (ORs) and 95% confidence intervals (CIs). Similarly, we examined the effect of stimulant use among trauma patients on emergency department mortality. We used generalized linear models with a negative binomial distribution and zero truncation to examine the effect of the presence of stimulants on hospital length of stay and length of stay in the ICU. Owing to the over-dispersion present, the use of Poisson models was not possible. We considered p values less than 0.05 significant. We performed all analyses using IBM SPSS Statistics (Version 28; IBM Corp).

Results

We included a total of 89 130 patients in the study, with 10 017 (11.2%) testing positive for stimulants only (i.e., cocaine, amphetamines, or methamphetamine) and 79 113 (88.8%) with negative urine or serum toxicology (Table 1). In the stimulant cohort, the average age was 42.0 (IQR 32.0 to 53.0) years, 7599 (75.9%) were males, and the mean BMI was 25.7 (IQR 22.8 to 29.7). In the control group, the median age was 55.0 (IQR 35.0 to 71.0) years, 49 821 (63.0%) were males, and the median BMI was 27.2 (IQR 23.7 to 31.7). The stimulant cohort had a greater proportion of males, smokers, and comorbidities (with the exception of mental or personality disorders) (p < 0.001). In addition, the control group was older, had a higher BMI, and had a significantly smaller proportion of males (p < 0.001). The median ISS was similar between both groups: 9.0 (IQR 5.0 to 17.0) in the stimulant and control groups.

Table 1.

Patient demographic characteristics and comparison between groups

Characteristic	No. (%)of patients*		p value	Missing, %
	Control group n = 79 113	Stimulant group n = 10017
Age, yr, median (IQR)	55.0 (35.0–71.0)	42.0 (32.0–53.0)	< 0.001	0.0
Sex, male	49 821 (63.0)	7599 (75.9)	< 0.001	0.0
BMI, median (IQR)	27.2 (23.7–31.7)	25.7 (22.8–29.7)	< 0.001	11.1
ISS, median (IQR)	9.0 (5.0–17.0)	9.0 (5.0–17.0)	0.02	0.1
Race
Asian	2795 (3.5)	225 (2.2)	< 0.001	0.0
Black	11 159 (14.1)	2157 (21.5)	< 0.001	0.0
Indigenous	321 (0.4)	90 (0.9)	< 0.001	0.0
Pacific Islander	392 (0.5)	125 (1.2)	< 0.001	0.0
White	23 141 (29.3)	3820 (38.1)	< 0.001	0.0
Not applicable or unknown	1150 (1.5)	195 (1.9)	< 0.001	0.0
Ethnicity, Hispanic or Latino	11 833 (15.4)	1903 (19.5)	< 0.001	2.7
Comorbidities
Smoker	13 657 (17.3)	4636 (46.3)	< 0.001	0.0
Advanced directive limiting care	2216 (2.8)	42 (0.4)	< 0.001	0.0
Anticoagulant therapy	8114 (10.3)	171 (1.7)	< 0.001	0.0
Cirrhosis	930 (1.2)	134 (1.3)	0.2	0.0
Chronic obstructive pulmonary disease	4682 (5.9)	420 (4.2)	< 0.001	0.0
Cerebrovascular accident	2399 (3.0)	135 (1.3)	< 0.001	0.0
Diabetes	12 870 (16.3)	760 (7.6)	< 0.001	0.0
Disseminated cancer	459 (0.6)	27 (0.3)	< 0.001	0.0
Congestive heart failure	3089 (3.9)	163 (1.6)	< 0.001	0.0
Myocardial infarction	730 (0.9)	27 (0.3)	< 0.001	0.0
Mental or personality disorder	9977 (12.6)	1896 (18.9)	< 0.001	0.0
Renal disease	1082 (1.4)	52 (0.5)	< 0.001	0.0
Hospital characteristics
Teaching status			< 0.001	0.0
Community	18 959 (24.0)	2263 (22.6)
Non-teaching	469 (0.6)	62 (0.6)
University	59 685 (75.4)	7692 (76.8)
Hospital type			< 0.001	0.0
For profit	2395 (3.0)	180 (1.8)
Non-profit	76 491 (96.7)	9809 (97.9)
Government	227 (0.3)	28 (0.3)
Bed size			< 0.001	0.0
≤ 200	2881 (3.6)	289 (2.9)
201–400	10 684 (13.5)	1648 (16.5)
401–600	26 842 (33.9)	3980 (39.7)
> 600	38 706 (48.9)	4100 (40.9)
Outcomes
Hospital mortality	3841 (5.2)	339 (3.6)	< 0.001	7.0
Emergency department mortality	204 (0.3)	35 (0.4)	0.1	1.2
Hospital length of stay, d, mean (IQR)	5.0 (2.0–9.0)	5.0 (3.0–9.0)	0.002	1.0
ICU length of stay, d, mean (IQR)	3.0 (2.0–7.0)	3.0 (2.0–7.0)	0.3	0.0

BMI = body mass index; ICU = intensive care unit; IQR = interquartile range; ISS = Injury Severity Score.

^*Unless stated otherwise.

Hospital characteristics were significantly different between the control and stimulant groups, with a greater proportion of patients testing positive for stimulants presenting to academic hospitals, non-profit centres, and hospitals with between 201 and 600 beds (Table 1).

Our primary outcome, overall hospital mortality, was not significantly different between the stimulant and control groups, with an adjusted OR of 1.09 (95% CI 0.95 to 1.26). However, the unadjusted odds ratio for overall hospital mortality was significant, with the stimulant group having lower odds of dying in hospital than the control group (OR 0.69, 95% CI 0.61 to 0.77). In the control group, 5.2% (n = 3841) of patients died, which was a greater proportion than in the stimulant group (3.6%, n = 339) (Table 2).

Table 2.

Unadjusted and adjusted association between presence of stimulants and hospital mortality in adult trauma patients

Variable	Unadjusted odds ratio (95% CI)	p value	Adjusted odds ratio (95% CI)*	p value
Control	Ref. (1.00)	–	Ref. (1.00)	–
Stimulants present	0.69 (0.61–0.77)	< 0.001	1.09 (0.95–1.26)	0.2

CI = confidence interval; Ref. = reference category.

^*Adjusted model with covariates.

Secondary outcomes included emergency department mortality, hospital length of stay, and length of stay in ICU. There were no significant differences between the groups for emergency department mortality, with an unadjusted OR of 1.36 (95% CI 0.95 to 1.94) and an adjusted OR of 1.66 (95% CI 0.93 to 2.95). The median length of stay was 5.0 (IQR 2.0 to 9.0) days in the stimulant group and 5.0 (IQR 3.0 to 9.0) days in the control group (p = 0.002), with an incidence rate ratio (IRR) of 1.11 (95% CI 1.09 to 1.13). This indicates that individuals with stimulants in their system were predicted to spend 1.11 times as many days in hospital as those without stimulants in their system, while controlling for confounding factors. However, patients who ingested stimulants had no significant difference in length of stay in the ICU, with an IRR of 1.03 (95% CI 0.998 to 1.06). These results are summarized in Table 2 and Table 3.

Table 3.

Unadjusted and adjusted association between presence of stimulants and emergency department mortality in adult trauma patients

Variable	Unadjusted odds ratio (95% CI)	p value	Adjusted odds ratio (95% CI)*	p value
Control	Ref. (1.00)	–	Ref. (1.00)	–
Stimulants present	1.36 (0.95–1.94)	0.1	1.66 (0.93–2.95)	0.09

CI = confidence interval; Ref. = reference category.

^*Adjusted model with covariates.

Discussion

The prevalence of stimulant drug use in the trauma patient population is varied depending on geographical location and health care institution. In level 1 trauma centres within the US, there is a reported 10% rate of positive toxicology reports for amphetamines in all trauma patients.^3,4 Hadjizacharia and colleagues reported an 18.4% rate of positive toxicology results for cocaine after exclusion of polysubstance users, at their level 1 trauma centre in Los Angeles.²⁶ Similar rates were reported by Soderstrom and colleagues, who observed 13.4% of patients testing positive for cocaine only at their level 1 trauma centre in Baltimore, Maryland.¹⁷ London and colleagues reported a proportion of 9.0% (n = 557) of trauma patients with an ISS less than 9 who tested positive for MDMA, with a higher prevalence of males in this population (67.0% in the MDMA group v. 62.6% in the no-drug group, p = 0.04).⁴ Our results showed an 11.2% prevalence of trauma patients testing positive for cocaine, amphetamines, or methamphetamines, as well as a larger proportion of males (n = 7599, 75.9%), which is similar to findings described for patients presenting to trauma centres and testing positive for stimulants in the US.

Multiple studies have analyzed the effect of stimulants on types of traumatic injury, mainly associated with violent and nonviolent crimes. The association between substance use and violent crime has been well described in the literature.²⁷ In their case series, Tominaga and colleagues observed a higher percentage of trauma patients presenting with intentional self-inflicted injuries or injuries from intentional assaults who tested positive for methamphetamine on admission than patients with negative toxicology screening.²¹ Abdalla and colleagues determined that cocaine use was a predictor of violent crimes with their population-based study in Brazil. Their results indicated that 19.7% of cocaine users were victims of violent crime, and cocaine users were 4 times as likely to be an aggressor in a violent crime.²⁸ Fujii and McCague similarly described a 2.6-times higher rate of having an ISS greater than 15, or “major trauma,” among those who had used amphetamines, as described in their cross-sectional study involving 731 trauma patients.²⁹ Owing to limitations in data from the TQIP database, we were unable to report the nature of trauma sustained or whether patients were victims or aggressors in violent crimes. However, given that the proportion of stimulant drug use in the TQIP population (11.2%) is higher than that of the general North American population (1.9% for cocaine and 2.0% for amphetamines), this finding would likely have been replicated if data had been available.³⁰

The results of this study did not show a significant difference in mortality between trauma patients testing positive for stimulants and our control group after we adjusted for covariates. However, the current literature has presented conflicting data regarding rates of mortality in this population. Gemma and colleagues found a higher proportion of in-hospital deaths following trauma in patients who tested positive for amphetamines (4.4% v. 2.9% drug-free group, p = 0.015).²² In contrast, Hadjizacharia and colleagues found no association between methamphetamine or cocaine use and increased mortality in their retrospective cohort study involving trauma patients who underwent toxicology testing at their level 1 trauma centre in Brazil (11.1% v. 10.9%, p = 0.87).^3,11 Several other studies found similar results, showing no significant correlation between mortality and stimulant use.^{23,29,31–33} To the best of our knowledge, there are no current studies reporting on the association between positive toxicology to stimulants and emergency department mortality. However, Demetriades and colleagues found an increased likelihood of patients with positive screens (for any illicit drug) to be dead on arrival at the emergency department.²⁴ Therefore, our results of no association between stimulants and hospital mortality are in keeping with the current literature, but it is unclear whether a trend of increased emergency department mortality is seen in these patients in trauma centres not included in the TQIP data set.

Correlation between length of stay and stimulant use also has conflicting evidence. Fujii and McCague found that toxicology positive for amphetamines was associated with an increased length of stay in hospital.²⁹ Gemma and colleagues found that illicit amphetamine use, compared with no drug use, was associated with a 7.4% increase in length of stay (p = 0.043) for the least severely injured patients (ISS < 9), but no association was found in patients with an ISS of 9 or greater.²² Tominaga and colleagues reported a significantly longer length of stay among patients who tested positive for methamphetamine, compared with controls, in 544 trauma patients who presented to their tertiary trauma centre in Hawaii (2.7 ± 0.4 v. 1.7 ± 0.1 d; p = 0.003).²¹ Fujii and McCague also found that amphetamine use was a predictor of increased length of stay in hospital (p = 0.01) but found no association between cocaine use and length of stay.²⁹ Hadjizacharia and colleagues and Bhullar and colleagues found no association between hospital length of stay or length of stay in the ICU and cocaine use, but did find a higher rate of emergency surgery among patients who tested positive for cocaine (14.1% v. 9.4%, 0.001).^3,20 Consistent with some of the current literature, our study found that overall hospital length of stay, but not ICU length of stay, was significantly increased in patients who tested positive for stimulants.

Although our study did not include a cost–consequence analysis for patients testing positive for stimulants, our finding of increased length of stay could lead to increased health care–associated costs for this subset of patients. Previous studies have supported the hypothesis that patients using stimulants have increased health care costs. Tominaga and colleagues found significantly higher hospital charges (US$15 617 ± $1866 v. $11 600 ± $648; p = 0.01), and increased likelihood of hospital admission from the emergency department (91% v. 70%; p = 0.001) with an ISS of 5 or lower and positive toxicology for stimulants.²¹ Gemma and colleagues reported a mean increase of 4.6% in cost for trauma patients with an ISS lower than 9 and who tested positive for stimulants on admission (p = 0.019).²² London and colleagues found that total direct costs were higher for trauma patients with an ISS lower than 9 who tested positive to methamphetamine than for those who tested negative (US$2998 v. $2667, p < 0.001), and these patients consumed more health care resources in all 10 resource categories (e.g., radiology, laboratory, surgery, nursing, and pharmacy).⁴

The strengths of our study include a substantial population from a large North American database including data from multiple level 1 trauma centres. This is one of the most sizable retrospective cohort studies that has evaluated the effect of stimulants on mortality among trauma patients. This study also investigated both hospital mortality (i.e., death following admission from the emergency department) and emergency department mortality, which few studies have described.²⁵ This finding calls attention to the effects of acute intoxication on trauma patients and the importance of toxicology as part of the initial investigations ordered on trauma team activation.^34,35

Given the results presented in our study, we advocate for routine toxicology testing on all patients admitted following trauma team activation.¹⁸ Early recognition of stimulant intoxication could be useful in identifying trauma patients at risk of death in the emergency department or patients who may be at higher risk of having an increased length of stay and higher consumption of health care resources. Toxicology testing can potentially be used as an early point-of-care test that can give additional information on trauma patients’ inpatient stay and overall outcome. This could enable early engagement with addictions or psychiatry services, and social work for complex social circumstances, to improve outcomes and decrease length of stay.^36,37 Further research should investigate whether mortality (both hospital and emergency department) correlates with chronicity of stimulant use among trauma patients. This would allow risk stratification within this higher-risk trauma population.

Limitations

Limitations of this study include those inherent within a retrospective observational study. Although our analyses controlled for multiple demographic variables, there may be confounders that were unaccounted for that were not represented in the TQIP data. Additionally, the TQIP database may have had selection or reporting bias during the data collection. Different toxicology testing with varying sensitivity and specificity may have been used at different centres, which may result in underestimation of the number of participants who were intoxicated with stimulants at the time of presentation to the trauma bay. Although comorbidity count was included as a covariate in our multivariable linear and logistic regression models, there were likely errors in detailed comorbidity reporting owing to often-rushed history taking in the acute trauma setting and the quality-improvement nature of the TQIP database. There is no toxicology type reported in the TQIP database; therefore, differences in sensitivity and specificity of the different toxicology investigations used are likely present (i.e., urine and serum).³⁸ Additionally, the frequency and chronicity of stimulant use is not reported in the TQIP database, both of which could have a substantial impact on our outcome variables.^39,40 Lastly, the mean age in our stimulant group was younger (42.0 yr) than that of the control group (55.0 yr), which may have led to difference in mortality and length of stay between both groups. However, the adjusted ORs presented did control for age and comorbidities.

Conclusion

This retrospective analysis underscores the importance of recognizing stimulant misuse as an important factor in trauma outcomes and health care utilization. Patients testing positive for stimulants exhibited an elevated risk for longer hospital stays, but no significant difference in mortality. The routine screening of trauma patients for stimulant drug use is a valuable practice, allowing trauma centres to identify at-risk individuals and implement measures to enhance patient care and resource allocation. By addressing stimulant intoxication in trauma patients, health care providers can work toward improving patient outcomes and reducing the associated health care burden.