- Email: [email protected]
Drug screening versus history in detection of substance use in ED psychiatric patients
Drug Screening Versus History in Detection of Substance Use in ED Psychiatric Patients JEANMARIE PERRONE, MD, FRANCIS DE ROOS, MD, SUDHA JAYARAMAN, BA, AND JUDD E. HOLLANDER, MD Because self-reporting of substance use may not be reliable, physicians rely on drug screening. We tested the hypothesis that drug screening alone is sufficient to detect substance use in ED psychiatric patients. We prospectively evaluated patients receiving psychiatric consultation over 6 months ending in April 1998 in an urban medical/psychiatric ED with 42,000 annual visits. After informed consent, patients underwent a structured interview by trained research associates who queried regarding substance use in the past 3 days. This self-report was compared with urine drug screen results for 11 substances of abuse. Standard descriptive statistical techniques were used. Kappa statistics were used to assess concordance between history and drug screens. Two hundred eighteen patients participated, 124 had a urine drug screen obtained. Patients with and without urine drug screens were similar with respect to age (34.9 versus 34.9 years, P = .3) and psychiatric diagnosis (P = .24). Overall, there was only fair concordance between history and drug screens (kappa = 0.46). History alone detected substance use in 70 patients (57%); drug screening alone detected substance use in 77 patients (62%). The combination of history and drug screening more often detected substance use than either alone (90 pts (73%); P < .05 for both comparisons). Depending on the particular drug, there was wide variation in concordance between history and drug screen (kappa's varied from 0.07 for ethanol to 0.79 for cocaine). History was better than drug screening for ethanol use (40 versus 10 patients), and TI-IC(28 versus 15 pts). Drug testing alone was never significantly better than history. Although self-reporting of substance use is not reliable, reliance on drug screening alone is also flawed. Optimal identification of drug use in emergency department psychiatric patients requires both history and drug screening. (Am J Emerg Med 2001;19:49-51. Copyright © 2001 by W.B. Saunders Company) Drug screens are commonly obtained in patients with suspected drug use in the emergency department (ED). 1 However, the utility of this practice is sometimes questioned. Drug screening is expensive, resulting in patient charges of $200 to $300 per test, and may be falsely negative at the lower limits of drug detection. Additionally, most simple drug screens are not comprehensive, but rather, limited to the few drugs which are assayed for specifically. 2 Furthermore, drug screening is often qualitative, and a positive screen may reflect use during the past several days, and may not account for the current presentation or symptoms of From the Department of Emergency Medicine, University of Pennsylvania, Philadelphia, PA. Manuscript received March 7, 2000, accepted July 12, 2000. Presented at the Society for Academic Emergency Medicine Annual Meeting, Boston, May 1999. Address reprint requests to Jeanmarie Perrone, MD, Hospital of the University of Pennsylvania, Department of Emergency Medicine, 3400 Spruce Street, Philadelphia, PA 19104. Email: [email protected] Key Words: Drug screen, substance abuse. Copyright © 2001 by W.B. Saunders Company 0735-6757/01/1901-0013510.00/(3 doi:l 0.1053/ajem.2001.20003
the patient? Drug screening in certain groups may be especially misleading, because symptoms may be falsely attributed to the results of a drug screen and halt further diagnostics. 4 For example, a patient who has a seizure for the first time and who has a drug screen revealing the use of cocaine, may have his seizure attributed to cocaine. However, the possibility of another cause of the seizure may be overlooked. Recognizing the limitations of the drug screen is especially important in the clinical evaluation of a patient. 2,3 The actual impact of drug screening has been questioned because it rarely changes patient management, s Kellerman et al report that in their evaluation of 405 adult drug overdose patients, clinical management changed after drug tests in only 4.4% of the patients. Drug screens usually have prolonged turnaround times and it has been observed that the total time for discharged patients in the ED was greater in cases evaluated with drug screens as compared with those who were not. s It is recommended that "drug screening should be considered only when anticipated results are likely to affect patient disposition." Although there is evidence that drug testing may not always be necessary or cost effective, there has been little research on self-reporting of drug use and toxicology screening in those patients. An assessment of the correlation between self-reported drug use and laboratory confirmation of drug use would assist us in determining whether this test is necessary in the future evaluation of these patients.
METHODS This is a prospective, cross sectional study of a convenience sample of patients who request or require psychiatric consultation presenting to the ED. This study was approved by the Committee on Studies in Human Subjects of the Hospital of the University of Pennsylvania. Patients who requested or received psychiatric consultation in our ED and in whom a urine drug screen was requested were eligible for inclusion. Our ED serves an urban indigent adult population as well as a tertiary care referral center. Patients were enrolled if they agreed to be interviewed regarding substance use in the past 3 days and had a urine drug screen ordered and obtained by their treating physician. Written informed consent was obtained in all subjects. Each patient who receives psychiatric evaluation is initially evaluated with a screening history and physical examination by a house officer and attending physician from the ED. After this screening examination, patients were eligible for the study if they agreed to participate. After informed consent, patients underwent a structured interview by trained research associates regarding their substance use in the past 3 days of any of the substances listed in Table 1. This self-report of recent substance use was compared with urine drug screen results for the same 11 substances of abuse. This urine drug screen was performed in our hospital laboratory by 49
50
AMERICAN JOURNAL OF EMERGENCY MEDICINE • Volume 19, Number 1 • January 2001
TABLE1. Substances of Abuse Amphetamines Barbiturates Benzodiazepines Cocaine Ethanol Methadone Methaqualone Opioids Phencyclidine Propoxyphene THC
enzyme immunoassay technique (EMIT; Behring Diagnostics, San Jose, CA). The threshold for detection of the principal drugs of abuse (amphetamines, tetrahydracannibinol (THC), cocaine metabolite, opiates, and phencyclidine) were consistent with the National Institute on Drug Abuse thresholds for these agents. The research associate also specifically addressed whether any substances had been used in the past 3 days (the threshold for most drugs to be detected in the urine). Trained research assistants assisted in data collection and in follow-up of drug screen results. Although it was not possible to assign these research assistants to all shifts, many day, evening, and night shifts throughout the week and on weekends were covered to get a wide representation of the patient population. Presence of the research assistant was not required for enrollment of the patient. When research associates were not present, housestaff treating the patients in the ED were expected to enroll the patients. Data were analyzed by standard methods for cross sectional studies. Statistical significance (P < .05) was determined by the Fisher's Exact test. Kappa statistics were used to assess concordance between history and drug screens.
RESULTS Two hundred eighteen patients participated; 124 had a urine drug screen obtained. Patients with and without urine drug screens were similar with respect to age (34.9 versus 34.9 years, P = .3) and psychiatric diagnosis (P = .24). There was only fair concordance between history and drug screens kappa = 0.46 (see Table 2). Depending on the particular drug, there was wide variation in concordance between history and drug screen. The combination of history and drug screening more often detected substance use than either alone (90 patients; 73%; P < .05). History alone detected substance use in 70 patients (57%). Drug screening alone detected substance use in 77 patients (62%). History was best for ethanol use and THC. Drug screening was never significantly better than history.
DISCUSSION Although drug screening is widely used in ED patients to detect and document substance use, often, the patient may admit to substance use if carefully addressed in the history. Certain patients may even be seeking help for their substance use, and thus, documentation of substance use may be unnecessary. We sought to compare the results of urine drug screening with a careful patient history in ED patients who requested or required psychiatric consultation. Not surprisingly, drug screen results infrequently correlated
with patient history and was very dependent on the substance being analyzed. There are several explanations for this. Depending on the substance, certain drugs may actually be present in the urine for more than 3 days depending on the chronicity of use, fat solubility of the drug, patient nutritional factors, other coprescribed medications which may affect drug clearance and urine retention times in the assays used. Cocaine, the most prevalent drug used in this group and which has a urine retention time of approximately 72 hours in most persons, revealed the highest kappa value for correlation between self reported history (of cocaine use in the past 3 days) and drug screen results. This may be because patients admitting to cocaine use most often still had urine drug screens which were positive. Alternatively, this may reflect a patient population requesting psychiatric consultation secondary to depression associated with cocaine use. Kappa values were poorest for ethanol use, which may reflect a lack of ability of the drug screen to detect ethanol use greater than 24 hours before ED evaluation in most patients. Certain drugs, such as barbiturates, opioids and THC had kappa values between 0.4 and 0.48 which may reflect their greater urine retention time and relative prevalence in this patient population. Additionally, poor kappa values for benzodiazepines may be secondary to administration of some benzodiazepines while the patient was in the department to treat agitation or anxiety. This study specifically addressed the correlation between self-reported substance use and drug screening of the same substances in ED psychiatric patients, these results may not be generalizable to all El) patients. Some of these patients were actually seeking help for their substance use and may be more willing to reveal their substance use in their history. However, this study does address the practice of psychiatrists and health maintenance organizations in our region which "require" a urine drug screen for documentation before evaluation of these patients in dual diagnosis units where their substance abuse and depression or psychosis is jointly addressed. As shown here, drug screening was rarely better than history of substance use. Patients rarely claim drug use when they are not really using the drugs. 6 In a small number of patients in this study, drug screening was positive with a negative history (see Table 3). This information could be clinically relevant in patients who get admitted to the hospital for evaluation and treatment of one TABLE 2, Correlation Between History of Substance Use and Toxicology Screen Results Drug Cocaine Barbiturates THC Opioids Benzodiazepines Ethanol Amphetamines Methadone Phencyclidine Propoxyphene Methaqualone
History+ (%) Tox Screen-}- (%) Kappa Value 37 3.4 24 8.3 8.4 34 4 0.9 1,7 0 0
38 3.4 12.8 10 7.6 8.6 0 0 0 1.7 0
0.787 0.482 0.414 0.4 0.257 0.07 NA NA NA NA NA
PERRONE ET AL • DRUG SCREENING IN ED PSYCHIATRIC PATIENTS
TABLE 3. 2 × 2 Analysis of Discrepancies Between History and Drug Screen Drug
HX+ DS+
HX+ DS-
HXDS+
HXDS-
Kappa
Cocaine Barbiturates THC Opioids Benzodiazepines Ethanol
39 2 11 5 3 5
5 2 17 5 7 35
7 2 4 7 6 5
69 112 85 103 103 71
0.787 0.482 0.414 0.400 0.257 0.072
Abbreviations: HX+ DS+, history+, drug screen_+; HX+ DS-, history+, drug screen-; HX- DS+, history-, drug screen+; HXDS-, history-, drug screen-.
substance disorder (ie, cocaine abuse) and who subsequently develop signs and symptoms of withdrawal from another substance such as benzodiazepines. Prophylaxis may be initiated at the time of admission to the hospital (after results of the drug screen) to prevent deterioration secondary to unanticipated drug withdrawal. The practice of obtaining toxicology screens in all patients who admit substance use should be done with more discretion because it increases health care costs while having inconsistent impact on patient disposition or treatment. This study did have several limitations. We enrolled a convenience sample of patients presenting to the ED for psychiatric evaluation. It is possible that some patients were missed on the overnight shifts when staffing was lower and enrollment was more time consuming for the housestaff. We may have missed some patients with substance use seeking help who may have been more willing to describe their substance use accurately. This would have been expected to
51
reinforce our results and increase our kappa values of correlation between history of substance use and drug screen results. There may have also been selection bias in that patients who agreed to consent for the study may be more willing to admit to their substance use. This would be expected to bias our results towards finding a higher correlation between history of substance use and drug screen results. Because the retention times of drugs in the urine will vary by individual and each drug, this may impact our results. Although self-reporting of substance abuse is not reliable, reliance on drug screening alone is also flawed. Optimal identification of substance use in ED psychiatric patients requires both history and drug screening. Patients who selfreport cocaine abuse had the highest correlation between history and drug screen in this population. Documentation of cocaine abuse in patients admitting to cocaine use may be unnecessary. REFERENCES 1. Epstein FB, Hassan M: Therapeutic drug levels and toxicology screen. Emerg Med Clin North Am 1986;4:367-376 2. Wiley JF: Difficult diagnoses in toxicology: Poisons not detected by the comprehensive drug screen. Pediatr Clin North Am 1991 ;38:725-737 3. Schwartz RH: Urine testing in the detection of drugs of abuse. Arch Intern Med 1988;148:2407-2412 4. Gold MS, Dackis CA: Role of the laboratory in the evaluation of suspected drug abuse. J Clin Psych 1986;47:17-23 (suppl) 5. Kellermann AL, Fihn SD, LoGerfo JP, et al: Impact of drug screening in suspected overdose. Ann Emerg Med 1987;16:12061216 6. Hollander JE, Todd KH, Green G, et al: Chest pain associated with cocaine: An assessment of prevalence in suburban and urban emergency departments. Ann Emerg Med 1995;26:671-676
the patient? Drug screening in certain groups may be especially misleading, because symptoms may be falsely attributed to the results of a drug screen and halt further diagnostics. 4 For example, a patient who has a seizure for the first time and who has a drug screen revealing the use of cocaine, may have his seizure attributed to cocaine. However, the possibility of another cause of the seizure may be overlooked. Recognizing the limitations of the drug screen is especially important in the clinical evaluation of a patient. 2,3 The actual impact of drug screening has been questioned because it rarely changes patient management, s Kellerman et al report that in their evaluation of 405 adult drug overdose patients, clinical management changed after drug tests in only 4.4% of the patients. Drug screens usually have prolonged turnaround times and it has been observed that the total time for discharged patients in the ED was greater in cases evaluated with drug screens as compared with those who were not. s It is recommended that "drug screening should be considered only when anticipated results are likely to affect patient disposition." Although there is evidence that drug testing may not always be necessary or cost effective, there has been little research on self-reporting of drug use and toxicology screening in those patients. An assessment of the correlation between self-reported drug use and laboratory confirmation of drug use would assist us in determining whether this test is necessary in the future evaluation of these patients.
METHODS This is a prospective, cross sectional study of a convenience sample of patients who request or require psychiatric consultation presenting to the ED. This study was approved by the Committee on Studies in Human Subjects of the Hospital of the University of Pennsylvania. Patients who requested or received psychiatric consultation in our ED and in whom a urine drug screen was requested were eligible for inclusion. Our ED serves an urban indigent adult population as well as a tertiary care referral center. Patients were enrolled if they agreed to be interviewed regarding substance use in the past 3 days and had a urine drug screen ordered and obtained by their treating physician. Written informed consent was obtained in all subjects. Each patient who receives psychiatric evaluation is initially evaluated with a screening history and physical examination by a house officer and attending physician from the ED. After this screening examination, patients were eligible for the study if they agreed to participate. After informed consent, patients underwent a structured interview by trained research associates regarding their substance use in the past 3 days of any of the substances listed in Table 1. This self-report of recent substance use was compared with urine drug screen results for the same 11 substances of abuse. This urine drug screen was performed in our hospital laboratory by 49
50
AMERICAN JOURNAL OF EMERGENCY MEDICINE • Volume 19, Number 1 • January 2001
TABLE1. Substances of Abuse Amphetamines Barbiturates Benzodiazepines Cocaine Ethanol Methadone Methaqualone Opioids Phencyclidine Propoxyphene THC
enzyme immunoassay technique (EMIT; Behring Diagnostics, San Jose, CA). The threshold for detection of the principal drugs of abuse (amphetamines, tetrahydracannibinol (THC), cocaine metabolite, opiates, and phencyclidine) were consistent with the National Institute on Drug Abuse thresholds for these agents. The research associate also specifically addressed whether any substances had been used in the past 3 days (the threshold for most drugs to be detected in the urine). Trained research assistants assisted in data collection and in follow-up of drug screen results. Although it was not possible to assign these research assistants to all shifts, many day, evening, and night shifts throughout the week and on weekends were covered to get a wide representation of the patient population. Presence of the research assistant was not required for enrollment of the patient. When research associates were not present, housestaff treating the patients in the ED were expected to enroll the patients. Data were analyzed by standard methods for cross sectional studies. Statistical significance (P < .05) was determined by the Fisher's Exact test. Kappa statistics were used to assess concordance between history and drug screens.
RESULTS Two hundred eighteen patients participated; 124 had a urine drug screen obtained. Patients with and without urine drug screens were similar with respect to age (34.9 versus 34.9 years, P = .3) and psychiatric diagnosis (P = .24). There was only fair concordance between history and drug screens kappa = 0.46 (see Table 2). Depending on the particular drug, there was wide variation in concordance between history and drug screen. The combination of history and drug screening more often detected substance use than either alone (90 patients; 73%; P < .05). History alone detected substance use in 70 patients (57%). Drug screening alone detected substance use in 77 patients (62%). History was best for ethanol use and THC. Drug screening was never significantly better than history.
DISCUSSION Although drug screening is widely used in ED patients to detect and document substance use, often, the patient may admit to substance use if carefully addressed in the history. Certain patients may even be seeking help for their substance use, and thus, documentation of substance use may be unnecessary. We sought to compare the results of urine drug screening with a careful patient history in ED patients who requested or required psychiatric consultation. Not surprisingly, drug screen results infrequently correlated
with patient history and was very dependent on the substance being analyzed. There are several explanations for this. Depending on the substance, certain drugs may actually be present in the urine for more than 3 days depending on the chronicity of use, fat solubility of the drug, patient nutritional factors, other coprescribed medications which may affect drug clearance and urine retention times in the assays used. Cocaine, the most prevalent drug used in this group and which has a urine retention time of approximately 72 hours in most persons, revealed the highest kappa value for correlation between self reported history (of cocaine use in the past 3 days) and drug screen results. This may be because patients admitting to cocaine use most often still had urine drug screens which were positive. Alternatively, this may reflect a patient population requesting psychiatric consultation secondary to depression associated with cocaine use. Kappa values were poorest for ethanol use, which may reflect a lack of ability of the drug screen to detect ethanol use greater than 24 hours before ED evaluation in most patients. Certain drugs, such as barbiturates, opioids and THC had kappa values between 0.4 and 0.48 which may reflect their greater urine retention time and relative prevalence in this patient population. Additionally, poor kappa values for benzodiazepines may be secondary to administration of some benzodiazepines while the patient was in the department to treat agitation or anxiety. This study specifically addressed the correlation between self-reported substance use and drug screening of the same substances in ED psychiatric patients, these results may not be generalizable to all El) patients. Some of these patients were actually seeking help for their substance use and may be more willing to reveal their substance use in their history. However, this study does address the practice of psychiatrists and health maintenance organizations in our region which "require" a urine drug screen for documentation before evaluation of these patients in dual diagnosis units where their substance abuse and depression or psychosis is jointly addressed. As shown here, drug screening was rarely better than history of substance use. Patients rarely claim drug use when they are not really using the drugs. 6 In a small number of patients in this study, drug screening was positive with a negative history (see Table 3). This information could be clinically relevant in patients who get admitted to the hospital for evaluation and treatment of one TABLE 2, Correlation Between History of Substance Use and Toxicology Screen Results Drug Cocaine Barbiturates THC Opioids Benzodiazepines Ethanol Amphetamines Methadone Phencyclidine Propoxyphene Methaqualone
History+ (%) Tox Screen-}- (%) Kappa Value 37 3.4 24 8.3 8.4 34 4 0.9 1,7 0 0
38 3.4 12.8 10 7.6 8.6 0 0 0 1.7 0
0.787 0.482 0.414 0.4 0.257 0.07 NA NA NA NA NA
PERRONE ET AL • DRUG SCREENING IN ED PSYCHIATRIC PATIENTS
TABLE 3. 2 × 2 Analysis of Discrepancies Between History and Drug Screen Drug
HX+ DS+
HX+ DS-
HXDS+
HXDS-
Kappa
Cocaine Barbiturates THC Opioids Benzodiazepines Ethanol
39 2 11 5 3 5
5 2 17 5 7 35
7 2 4 7 6 5
69 112 85 103 103 71
0.787 0.482 0.414 0.400 0.257 0.072
Abbreviations: HX+ DS+, history+, drug screen_+; HX+ DS-, history+, drug screen-; HX- DS+, history-, drug screen+; HXDS-, history-, drug screen-.
substance disorder (ie, cocaine abuse) and who subsequently develop signs and symptoms of withdrawal from another substance such as benzodiazepines. Prophylaxis may be initiated at the time of admission to the hospital (after results of the drug screen) to prevent deterioration secondary to unanticipated drug withdrawal. The practice of obtaining toxicology screens in all patients who admit substance use should be done with more discretion because it increases health care costs while having inconsistent impact on patient disposition or treatment. This study did have several limitations. We enrolled a convenience sample of patients presenting to the ED for psychiatric evaluation. It is possible that some patients were missed on the overnight shifts when staffing was lower and enrollment was more time consuming for the housestaff. We may have missed some patients with substance use seeking help who may have been more willing to describe their substance use accurately. This would have been expected to
51
reinforce our results and increase our kappa values of correlation between history of substance use and drug screen results. There may have also been selection bias in that patients who agreed to consent for the study may be more willing to admit to their substance use. This would be expected to bias our results towards finding a higher correlation between history of substance use and drug screen results. Because the retention times of drugs in the urine will vary by individual and each drug, this may impact our results. Although self-reporting of substance abuse is not reliable, reliance on drug screening alone is also flawed. Optimal identification of substance use in ED psychiatric patients requires both history and drug screening. Patients who selfreport cocaine abuse had the highest correlation between history and drug screen in this population. Documentation of cocaine abuse in patients admitting to cocaine use may be unnecessary. REFERENCES 1. Epstein FB, Hassan M: Therapeutic drug levels and toxicology screen. Emerg Med Clin North Am 1986;4:367-376 2. Wiley JF: Difficult diagnoses in toxicology: Poisons not detected by the comprehensive drug screen. Pediatr Clin North Am 1991 ;38:725-737 3. Schwartz RH: Urine testing in the detection of drugs of abuse. Arch Intern Med 1988;148:2407-2412 4. Gold MS, Dackis CA: Role of the laboratory in the evaluation of suspected drug abuse. J Clin Psych 1986;47:17-23 (suppl) 5. Kellermann AL, Fihn SD, LoGerfo JP, et al: Impact of drug screening in suspected overdose. Ann Emerg Med 1987;16:12061216 6. Hollander JE, Todd KH, Green G, et al: Chest pain associated with cocaine: An assessment of prevalence in suburban and urban emergency departments. Ann Emerg Med 1995;26:671-676