Incidence rate ratios for substance use by year of admission for the entire cohort.
Incidence rate ratios for substance use by year of admission for the select cohort.
London JA, Battistella FD. Testing for Substance Use in Trauma PatientsAre We Doing Enough?. Arch Surg. 2007;142(7):633-638. doi:10.1001/archsurg.142.7.633
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
Only a fraction of trauma patients are being tested for substance use, and the proportion of those tested may have decreased over time.
Retrospective review of longitudinal data.
National Trauma Data Bank.
Individuals aged 15 to 50 years admitted with injuries from 1998 to 2003.
Main Outcome Measures
The primary outcomes of interest are the incidence of drug and alcohol testing and the results of these tests. The primary exposure of interest is year of admission.
Half of patients admitted with injuries are being tested for alcohol use, and half of these patients have positive test results. Only 36.3% of patients admitted with injuries are tested for drug use, and 46.5% of these patients have positive test results. There have been no significant trends for either alcohol testing or results in the past 6 years. Compared with 1998, patients are significantly less likely to be tested for drugs, but more likely to have positive test results.
Only a small proportion of patients who are admitted with injuries are tested for substance use. The proportion of patients tested for drugs has decreased significantly during the past 6 years. Routine testing would maximize identification of patients who may benefit from interventions. Several obstacles exist to routine screening, including legal and physician-related barriers. Future efforts to facilitate routine testing of trauma patients for substance use should concentrate on protecting patient confidentiality and educating physicians on the techniques and benefits of brief interventions.
Injuries are the leading cause of death and disability in the United States for persons younger than 35 years.1 The 2004 National Survey on Drug Use and Health found that 7.9% of the US population 12 years or older were current users of illicit drugs, and 23% participated in binge drinking of alcohol.2 Rates of substance use in patients admitted for traumatic injuries exceed those of the general population, with positive drug and alcohol test results ranging from 30% to 40% and from 27% to 47%, respectively.3- 7 In addition to injury and death, psychosocial, health, and economic problems often result from substance use. Estimated national costs of alcohol-related gunshot and stab wounds are $6.3 billion annually; and those for motor vehicle crashes, $15.7 billion annually.8
Despite the strong association between substance use and injuries, drug and alcohol testing among patients admitted for traumatic injuries is not routine. Rates of testing range from 35% to 89% and from 35% to 62% for alcohol and other drugs, respectively.9- 11 Testing patients for substance use is often the first step toward identifying the problem and providing intervention. Several physician-related and legal barriers exist to routine drug and alcohol testing. We hypothesize that only a fraction of trauma patients are being tested for substance use, and because of these barriers, the proportion of those tested may have decreased over time.
The National Trauma Data Bank (NTDB) was established by the American College of Surgeons' Committee on Trauma, and is the largest aggregation of trauma registry data in the United States. The NTDB is not a population-based sample because it includes a disproportionate percentage of larger hospitals with younger and more severely injured patients. The data are voluntarily submitted by contributing hospitals. Therefore, the data may not be an accurate representation of all hospitals. The database includes only patients who were admitted to the hospital and does not include prehospital deaths.12
Patients admitted from 1998 to 2003 with a primary diagnosis of injury (code E800-E999) were included in the study. The years before 1998 represent relatively few centers and only 14.3% of the patients included in the entire database. Patients with primary diagnoses of injuries from poisoning (codes E850-E869), surgical misadventures (codes E870-E879), environmental factors (codes E900-E909), late effects of injury (code E929), or drugs in therapeutic use (codes E930-E949) were excluded. Patients aged 15 to 50 years, the ages of those most at risk for traumatic injuries and substance use, were included in the analysis. The term substance use refers to alcohol and other drugs unless individually stated. The NTDB has fields labeled “alcohol present” and “drugs present.” The potential entries were “yes,” “no,” or a blank field. Alcohol and other drug testing were considered performed when the respective fields in the database were coded as either positive or negative. We assumed that no test was performed if the fields were blank.
Potential confounding variables, such as sex, age, race, Injury Severity Score (ISS), Glasgow Coma Scale (GCS) score, and mechanism of injury, were included in the multivariate analysis. We evaluated age in 4 categories (15-20, 21-30, 31-40, and 41-50 years), ISS in 5 categories (≤8, 9-15, 16-25, 26-45, and >45), and GCS score in 3 categories (<9, 9-13, and 14-15). Race was classified as white or nonwhite.
The primary outcomes of interest were the prevalence of drug and alcohol testing and the results of these tests. The primary exposure of interest is year of admission. Two separate analyses were performed. In the first analysis, the entire cohort (EC) that submitted data to the NTDB any time during the 6-year interval was included. We recognize that bias may be introduced as centers with different testing policies enter or leave the database through the years. For example, if a facility that tests all patients submitted data for the first half of the study period, but did not submit data for the second half, the apparent trend of decreased testing over time may be artificially low because of the absence of 1 facility. In an attempt to minimize selection bias, we performed a second analysis on a select cohort (SC) of institutions that submitted data for the entire 6-year period.
We evaluated trends in the covariates over time using Poisson regression, with the dependent variable being the year of admission. Incidence rate ratios were generated for each covariate, with the dependent variable being drug or alcohol testing or the results of those tests. We recognize that patients admitted to a single facility may not be independent compared with those admitted to another facility by virtue of their geographic location and socioeconomic factors. We accounted for this variability by using the “cluster” option in a computer program (Stata).13 Robust 95% confidence intervals were generated using the Huber-White sandwich estimator.13 Finally, variables potentially confounding the relationship between drug and alcohol testing and the year of admission were evaluated using Poisson regression. Results of the multivariate analysis are graphically depicted. All statistical analyses were performed using a computer program (Stata 8/SE).
There were 996 225 patients from 258 facilities who met the inclusion criteria during the study period. The characteristics of the EC are illustrated in Table 1. The characteristics of the SC are shown for comparison. There were 25 SC facilities, accounting for 21.8% of the patients. When comparing SC with EC, there were similar proportions of mechanism of injury, age distribution, and sex. The SC facilities contained more level I centers when compared with the entire study population. The SC facilities also had more severely injured patients, as evidenced by higher ISSs and lower GCS scores. There were no significant trends in age, sex, race, mechanism of injury, ISS, or GCS score during the 6-year period (data not shown). Overall, only 36.3% of the patients were tested for drug use and 48.7% were tested for alcohol use.
Univariate analysis revealed that patients admitted with penetrating trauma are less likely to be tested for alcohol than those admitted after blunt trauma, but more likely to have positive test results. A similar relationship exists for drug testing and results with respect to mechanism of injury, although this did not reach statistical significance. While patients were more likely to have positive test results for alcohol with increasing age, the same was not true for other drugs. Men were consistently tested, and had positive results for substance use more often than women. Nonwhite patients are tested less often for alcohol. As ISS increases, and GCS score decreases, patients are more likely to be tested for substance use. The results of the univariate analysis for the SC facilities are similar to those of the EC facilities (Table 2 and Table 3).
Multivariate analysis shows that compared with 1998, there were no significant differences in alcohol testing or results for either the EC or the SC facilities. In contrast, drug testing consistently and significantly decreased during the study period. There was a trend toward increasingly positive results in drug testing, but this only reached statistical significance in 2003 for the EC analysis. Table 4 and Table 5 show the results of the multivariate analysis of testing for substance use by year of admission for the EC and the SC, respectively. These results are graphically depicted in Figure 1 and Figure 2, respectively.
During a 6-year period, fewer than half of patients admitted with traumatic injuries were tested for substance use. When examining the trend from 1998 to 2003, we found that while there were no significant changes in alcohol testing and results, there was a statistically significant decrease in the proportion of the population being tested for drugs, and a trend toward increasingly positive results.
The results of this study are not unprecedented. The National Survey on Drug Use and Health found that from 2002 to 2003 there has been an increase in the nonmedical use of pain relievers, from 29.6 to 31.2 million.2 Soderstrom et al9 examined the trends of alcohol and other drug use in a single institution for trauma patients between 1984 and 2000. They found that while there has been a decrease in those with positive test results for alcohol, there has been a significant increase in those with positive test results for cocaine or opiates.9
The finding of decreased drug testing and the trend toward increased use has several possible explanations. Drug testing may be performed more selectively in patients in recent years. It is feasible that clinicians have become more astute in recognizing drug intoxication and test those most likely to be under the influence. It is also possible that drug testing has decreased nonselectively and that there is a true increase in the proportion of patients using drugs. It is unclear why similar trends are not found with alcohol testing. Comparable results between the EC and the SC analysis decrease the possibility that the findings are a result of facilities entering and leaving the database. In either case, more than half of the population is not being tested for substance use and, therefore, potentially not identifying individuals who may benefit from interventions.
Testing for substance use, in addition to screening questionnaires, is the first step in the identification of patients with potential substance abuse problems. Many trauma patients do not have medical insurance or a primary care physician. Therefore, their injury and contact with the medical profession represents potentially the only window of opportunity for intervention.14 Brief interventions are time-limited counseling for patients that focus on changing a patient's behavior and increasing the success of therapy. Individuals are given feedback on their level of substance consumption, the relation of consumption to injury, and the negative consequences related to consumption. Emphasis is placed on personal responsibility, and options are presented to the patient for limiting or ceasing intake.
The widespread social, economic, and personal impact of traumatic injury make it one of the nation's primary public health concerns. When routinely screened, more than 40% of trauma patients show evidence of chronic alcohol abuse or dependence, making it the most common chronic condition among these patients.4 A public health paradigm for prevention identifies risk factors and addresses them in such a way to effect a change. Substance use is a risk factor for traumatic injuries in a similar way that smoking is a risk factor for lung cancer, hypertension is for cardiac disease, and obesity is for diabetes mellitus.15 One could arguably classify all of these diseases resulting from lifestyle choices. Smokers often receive periodic screening chest radiographs to identify pulmonary nodules, patients with hypertension are screened for cardiac disease, and obese patients are screened for diabetes mellitus. In addition to routine screening, these patients receive information regarding cessation of smoking, controlling serum glucose level, and weight loss. Similarly, patients presenting with traumatic injuries should be routinely tested for substance use and, if present, interventions should be initiated. This does not happen for several reasons.
Despite a growing body of literature showing the effectiveness of brief interventions, many physicians still do not believe that these programs are effective. Others cite that the information gained from testing for substance use is not clinically important and that screening is not part of the physician's job. Those who want to intervene often lack the confidence to do so because they believe that they do not possess the necessary skill set. Finally, many physicians believe that positive test results will negatively impact patient care.16- 18
Several studies19- 22 have suggested that injury is a potent motivator for changing behavior, potentially making timely interventions more likely to illicit a response. Recent surveys14,23 suggest that trauma patients are amenable to alcohol screening and brief interventions. Contrary to the perceived ineffectiveness of brief interventions for substance use, multiple studies have shown that they are effective in decreasing consumption, arrests and recidivism, and injuries. Gentilello et al24 showed that brief interventions reduced weekly alcohol consumption, and reduced injuries and readmissions by almost 50% at 3 years. Four other randomized trials25- 28 have shown that brief interventions in the setting of alcohol-related injuries decrease consumption, traffic violations, injuries, and other alcohol-related problems.
A survey29 of trauma surgeons found that only 18.7% routinely screened patients, and more than half screened less than 25% of patients. One major disincentive to routine testing and screening is the threat by insurance companies to deny payments for medical care if an injury is sustained while under the influence of a substance of abuse.30,31 A recent survey31 has shown that 38 states and the District of Columbia have provisions in their insurance codes that allow companies to deny coverage to patients injured while under the influence of drugs or alcohol. The problems associated with failure to screen and intervene can be highlighted by the experience in Tennessee, one of the states with such provisions. A statewide study32 found that less than 10% of patients needing substance abuse treatment actually receive any treatment. This unmet need was estimated to cost the state $777 million in 2000 in hospital admissions and emergency department visits alone.32 Nationally, estimated costs of alcohol-related injuries are several billion dollars.8
Several methods have been proposed to help achieve more universal testing and screening.31 One approach is to lobby state policy makers to adopt laws that would prevent insurance companies from denying payments if injuries were sustained under a substance of abuse. In addition to legislative changes, altering the manner in which substance abuse information is obtained may help protect the patient's confidentiality in terms of release of information to third parties. Special federal regulations exist that protect the confidentiality of patients receiving substance abuse therapy. In the hospital setting, these regulations apply only to programs that have either a unit specializing in substance abuse treatment and referral or staff whose primary function is to provide such services. Most information regarding drug or alcohol use is collected during the routine medical care of patients and, therefore, is not protected under federal regulations.
The validity of any observational study is limited by the database used. Although the NTDB is not a population-based sample and the findings may not be an accurate representation of all hospitals in the country, it is the largest aggregation of trauma registry data in the United States.12 Submitting hospitals may use different registry software that may account, in part, for some of the difference between facilities. The manner in which the NTDB collects data precludes the determination of specific drugs for which the facilities were testing, and which drugs produced positive results. Furthermore, there is no method in which to determine if certain drugs, particularly opiates or benzodiazepines, were given to patients by medical personnel in either the prehospital setting or the emergency department.
There are several sources of bias in this study. During the study period, many centers have entered and left the database. If some centers entered the database that routinely did or did not test patients, this may bias the results. The results would then represent the addition or deletion of a selected number of institutions. Restricting the analysis to only centers submitting data for the entire study period reduces this problem. Submitting facilities have the option of leaving the fields of drug and alcohol testing blank. Institutions that did not submit data into these fields were assumed to not have tested the patients. This may not be true, and some institutions may not have submitted complete information.
Substance use and injury in individuals younger than 50 years are leading public health concerns. Only a portion of these patients who are admitted with injuries are tested for substance use. Routine testing of patients is an important first step toward identification of patients in need of interventions. Unfortunately, several obstacles exist to routine testing, including legal and physician-related barriers. Future efforts to facilitate routine testing of trauma patients for substance use should concentrate on legislating changes in insurance policies and on protecting patient confidentiality. In addition, efforts need to be made to educate physicians on the techniques and benefits of brief interventions.
Correspondence: Jason A. London, MD, MPH, Division of Trauma and Emergency Surgery, Department of Surgery, University of California, Davis Medical Center, 2315 Stockton Blvd, Suite 4207, Sacramento, CA 95817 (firstname.lastname@example.org).
Accepted for Publication: March 23, 2006.
Author Contributions: Dr London has full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: London and Battistella. Acquisition of data: London. Analysis and interpretation of data: London and Battistella. Drafting of the manuscript: London. Critical revision of the manuscript for important intellectual content: London and Battistella. Statistical analysis: London. Study supervision: Battistella.
Financial Disclosure: None reported.