In a consecutive series of patients admitted to a stroke unit at a single tertiary care medical center, we found that the yield of urine toxicology screening for a substance that has been associated with an increased risk of stroke was only 4.6%. Cannabinoids and cocaine were the principal drugs detected. Current smokers under 60 years of age had the highest likelihood of a positive toxicology result (14.1%). Our study suggests that universal screening in patients with stroke or TIA is likely to be associated with low diagnostic yield and selected approaches may be more reasonable.

Multiple mechanisms of stroke have been related to illicit drug use. Cocaine ingestion may cause ischemic stroke through vasospasm, by preventing sympathomimetic neurotransmitters and cerebral vasoconstrictors reuptake [3, 15]. It may also result in acute myocardial infarction and arrhythmia, or cardiomyopathy with subsequent cerebral embolism [16]. Cannabinoids, particularly marijuana may induce reversible cerebral vasoconstriction syndrome and multifocal arterial stenosis, which may cause cerebral ischemia [17, 18]. Proposed mechanisms of ischemic stroke with amphetamine exposure include intracranial cerebral necrotizing vasculitis, non-arteritic vasculopathy, as well as possible cardiomyopathy with subsequent cerebral embolism [5, 19, 20]. The mechanism of ischemic stroke in relation to phencyclidine use is unclear. It should be noted that stroke in the setting of illicit drug use may not be secondary to the effects of acute exposure and traditional mechanisms should not be overlooked.

Several differences may explain our lower yield (1.2% vs. 11% cocaine-positive screens) compared to a prior study [10] and compared to the population prevalence’s of drug use. The racial/ethnic background of the population evaluated in the previous report differed in comparison to the cohort in our study, which may contribute to risk of drug use. In addition, the highest rate of cocaine positive toxicology in the prior study was in patients with hemorrhagic stroke as opposed to those with ischemic stroke or TIA [10]. However, selection bias such that only 40% of the total cohort completed toxicology screening may have resulted in the higher prevalence compared to our study where 86% of all eligible patients underwent screening. While population lifetime prevalence of any illicit drug abuse is approximately 9%, the rate of drug use restricted by type to those associated with stroke and ingested within 48 of stroke is likely much lower. Similar to their findings and consistent with prior reports, however, we also observed a higher risk of illicit drug use in younger populations with stroke [21, 22]. No prior study has demonstrated an association of tobacco use and detection of illicit drug use in patients with cerebrovascular events. Based on these risk factors and contrary to the prior paper suggesting universal testing, our data support a targeted or selective approach to urine toxicology screen in stroke patients.

Given the concerning trends in healthcare spending and costly diagnostic evaluation of patients with stroke, testing should be limited when yield is low [23]. Though it may vary between institutions, a single urine toxicology screen at some centers may exceed 300 US Dollars. An approach based on pretest probability and clinical context seems to be the most appropriate strategy for allocating health care resources. It is also not advisable to screen at time periods after which illicit drugs cannot be detected by current techniques. Unlike cocaine and cannabis use, testing for amphetamines and phencyclidine is likely to be of very low yield in cerebrovascular ischemia, though could be considered if their use is more common in the local practice setting or in patients with hemorrhagic stroke.

There are some limitations to our study. First, despite a protocol of performing urine toxicology screening universally on patient admission to the stroke unit or NSICU, 14.3% did not complete testing, introducing the potential for some selection bias. Reasons for non-adherence to protocol included: failure to obtain an adequate urine sample from the patient, testing not appropriately ordered or cancelled, or the sample being lost. Second, our toxicology screening does not differentiate distinct types of substances within each tested drug class (e.g., dextroamphetamine vs. methylphenidate). Third, since illicit drug use can result in systemic organ involvement (i.e., myocardial ischemia) in addition to cerebrovascular complications, some patients with concomitant cardiac symptoms may have been admitted to non-stroke units and not included in our registry. Lastly, it is theoretically possible that obtaining a urine sample immediately after substance use and stroke symptom onset could have led to false negative urine toxicology screens in certain cases.

In conclusion, our study supports the current recommendation that selective urine toxicology screening be considered in stroke and TIA [9]. The highest diagnostic yield is likely to be for cannabinoids and cocaine screening, in younger patients with a history of concurrent tobacco use. Local epidemiology, particularly patient demographics, should be considered along with these predictive factors in the decision for testing. Determining exposure to drugs of abuse, by history or diagnostic screening, would allow for early detection and opportunity for counseling and patient education regarding the association of substance abuse and stroke. Further studies are warranted to evaluate overall diagnostic yield of toxicology screening in different populations, for newly emerging illicit drugs, and for cost effectiveness of testing.