Urine toxicology screening in an urban stroke and TIA population

DISCUSSION

In this cohort of patients with stroke and TIA, representing individuals admitted to a tertiary care center in a major metropolitan city, we found that 40% of patients underwent urine toxicology testing and 11% (and 9% of patients 50 years or older) were positive for cocaine. We further identified a disparity in care whereby individuals who were younger and black were more likely to undergo screening. Despite the fact that patients from this profile were more likely to have positive urine toxicologies for cocaine (but not opiates), patients not fitting this profile also had positive toxicologies for cocaine and opiates. The potential impact of profiling in this situation is 2-fold. First, patients with positive cocaine toxicologies appeared less likely to receive thrombolytic therapy despite having similar stroke severities at admission. From the chart review, it was unclear why thrombolytics were not given in these cases. In many instances, the toxicology results came after the time window had passed so there may have been other factors leading to a decision not to treat. It is possible that the same factors that were associated with increased urine toxicology testing were associated with reduced treatment with thrombolytics, e.g., race.¹² Nevertheless, patients with positive toxicologies may indeed benefit from thrombolysis. In a study of 29 patients, none of the patients with positive urine toxicologies for cocaine had an intracerebral hemorrhage and many had a good outcome.¹³ Second, patients who are not screened and are in fact using cocaine or opiates will miss the opportunity for counseling.

How health care to individuals of different race, sex, sexual orientation, and primary language is delivered has been the subject of a number of studies and reviews. In a study published in 1999, physicians viewing different actors describing chest pain symptoms were less likely to recommend cardiac catheterization for women and blacks.¹⁴ Analysis of race-sex interactions showed that black women were 60% less likely to be referred for cardiac catheterization than white men. A study published a decade later found that women were 50% more likely to have delayed ambulance transport for chest pain than men, after multivariate adjustment.¹⁵ Race was not associated with delays in transport time. During the same period, a study in Canada found that physicians were more likely to recommend total knee arthroplasty to a male standardized patient than a female standardized patient.¹⁶ When analyzed by specialty, family physicians were twice as likely to recommend the surgery to the male patient and orthopedic surgeons 22 times as likely. Language barriers may also produce health care disparities. In a cross-sectional study, the percentages of patients receiving all eligible health care services among different groups were as follows: English-speaking whites 57.0%, Hispanics speaking English at home 53.6%, Hispanics not speaking English at home but comfortable speaking English outside the home 44.9%, Hispanics not speaking English at home and uncomfortable speaking English outside the home 35.0%.¹⁷ Physicians as patients have also described their own experiences of disparities in care based on sexual orientation¹⁸ and race.¹⁹

Of note, 1 in 10 tested patients in this study were positive for cocaine (including those aged 50 and older). When only considering patients with intracerebral hemorrhage, 1 in 4 were positive. Positive results were identified up to age 71. Outside of this study, we have seen positive results in patients older than 80 years. The prevalence of stroke among individuals with drug use and the prevalence of drug use among stroke patients of all ages are not known nationally but several studies have examined this issue at a local level. In a study of 3,712 drug users, 13 (0.4%) were identified as having stroke.²⁰ Seven (54%) events were ischemic stroke, 3 (23%) were subarachnoid hemorrhage, and 3 (23%) were intracerebral hemorrhage. Of 214 stroke patients aged 15 to 44 years admitted to San Francisco General Hospital, 63 (34%) were using drugs compared with 18 (8%) of control patients admitted for other diagnoses and matched for age, sex, and year of hospitalization.²¹ In this case-control study, the adjusted relative risk for stroke compared with non–drug-using patients was 6.5. A population-based study of 3,148,165 hospitalized patients in Texas found that amphetamine use was associated with hemorrhagic but not ischemic stroke whereas cocaine was associated with both stroke subtypes.²² Cocaine use was associated with an approximately 2-fold increased risk of both stroke subtypes whereas amphetamine use increased the risk of hemorrhagic stroke 5-fold. A case-control study of 347 stroke patients in northern and southern California among women aged 15 to 44 years found an adjusted odds ratio of 7.0 among women reporting use of cocaine and/or amphetamines.²³ The study was done in an insured urban population as compared with prior studies suggesting that the potential mechanism of injury may occur independently of sociodemographic factors. In a study in Maryland, 167 stroke patients were asked about drug use. Information could not be obtained in 51 patients (31%) because of the stroke or failure to obtain a drug history.²⁴ Of the remaining 116 patients, 11 (9.5%) were historically associated with drug use. The age range of patients was 25 to 56 years. Drugs reported included cocaine in 5 (45%), over-the-counter sympathomimetics in 3 (27%), phencyclidine in 2 (18%), and heroin in 1 (9%). Finally, in distinction to prior studies, a case-control study from Atlanta, Georgia did not find a relationship between crack cocaine use and stroke among 144 stroke patients aged 20 to 39 years.²⁵ However, in that study, 43% of controls and 32% of stroke patients had used crack cocaine at any time whereas 21% of stroke patients and 16% of controls had used crack cocaine within 4 hours. Moreover, data regarding crack cocaine use at any time were missing in 54% of stroke patients and 33% of controls while date regarding use within 4 hours was missing in 61% of stroke patients and 54% of controls. The gaps in information in this study underscore the selectivity and potential bias of ad hoc screening.

Current data support the notion that drug-related stroke may be a consequence of the pharmacodynamic effect rather than a vasculopathy. An autopsy study of 12 patients who had intracerebral or subarachnoid hemorrhage in the setting of cocaine use found no evidence of vasculitic changes within the vessel walls.²⁶ In fact, arterioles were either normal or showed nonspecific changes. A separate autopsy study of 17 patients who had parenchymal or subarachnoid hemorrhage, of whom 10 were positive for cocaine, also found no evidence of vasculitis or vasculopathy.²⁷ In the San Francisco General Hospital study, the relative risk of stroke increased to 49.4 for patients who had symptoms within 6 hours of drug consumption.²¹ Although the absolute number of patients was relatively small, we observed a higher rate of lacunar mechanism of ischemic stroke (43%) among cocaine-positive patients than all other subgroups (i.e., cocaine-negative patients and opiate-positive or -negative patients [26%–28%]). Mechanisms of stroke among patients with substance abuse merit further study.

There are several limitations of this study. First, the study was conducted in a major metropolitan city and may not be representative of other areas of the country. Second, despite the fact that cases of stroke and TIA were collected prospectively, the analysis of cases was done retrospectively through chart review. Therefore, relevant information such as NIHSS score was not available for all patients and may have led to inaccurate estimates of stroke severity at admission for either group. Third, the clinical evaluation plan of the treating physician was not explored in this study and may have provided additional clues as to differences observed. Fourth, there is no long-term follow-up beyond the period of discharge, which may have affected the findings in patient outcome. Fifth, the identification of race was by an administrative clerk at admission rather than patient self-report. In many cases, patients were cognitively impaired and would not have been able to respond to that question. However, this limitation may actually be beneficial within the context of this study. How another individual assigns race of a patient may influence the likelihood of urine toxicology screening. Therefore, the assignment of category by an outside observer rather than the patient themselves may be more relevant in this circumstance. Sixth, the toxicology screening described in this study does not distinguish between different types of opiates or systemic levels. Such information would be useful to not only identify abuse of illicit substances but also to identify overuse/abuse of prescription opiates. In addition, the toxicology screening did not include newer drugs of abuse such as MDMA, methamphetamine, and phencyclidine. Finally, information on patients with subarachnoid hemorrhage was not collected. This subgroup of patients could be a subject of a future study on urine toxicology screening.

Given the findings in this study, and despite its limitations, we submit that current guidelines for stroke and TIA should be updated to recommend urine toxicology screening for all patients. Additional studies evaluating the health effects and cost implications of routine screening in patients with stroke and TIA would be informative on this topic.