Work

The burden of cannabis‐attributed pediatric and adult emergency department visits

Year: 2021

Type: article

Abstract: States that have legalized cannabis for medicinal or retail use have noted increases in emergency department (ED) visits associated with cannabis-related harms.1-3 However, most studies are retrospective and identify cannabis-related visits using diagnosis codes or THC-positive urine drug immunoassays.2-5 These identification methods are limited; ICD codes are infrequently used in adult ED cases of exposure to drugs of abuse6, 7 and clinicians may not explicitly document diagnosis of drug use, even when highly suspected or confirmed by patients. A positive urine screen implies drug exposure but does not necessarily provide evidence of intoxication or a relationship between exposure and the ED visit.8 We sought to determine a more accurate burden of cannabis on ED visits by examining use of an electronic medical record (EMR) query designed to capture treating clinicians' real-time impressions about whether visits were attributable to cannabis use and to describe those visits in terms of triage acuity, length of stay, ultimate disposition, final diagnoses, and hospital charges. The study was performed in an adult and pediatric ED at an urban, academic Level I trauma center and was approved by the institutional review board of the Oregon Health and Science University. As part of a quality improvement project, we embedded a single question related to cannabis use ("Do you think that cannabis use was a significant factor leading to today's ED visit?" Answers: yes, no) into the EMR. The question was linked to a folder icon on the ED tracking board that changed color from red to green when the question was completed, similar to required fields, to drive participation. The question was additionally added to the disposition page in the EMR. Attending physicians, residents, and advanced practice providers were introduced to the field via presentations—including guidance about using the field for cases in which cannabis was suspected to be directly attributable for the reason for the ED visit and concrete examples of clinical presentations that might be considered cannabis attributable (e.g., cannabis intoxication, cannabis hyperemesis)—at faculty meeting and teaching conference and via mass email with periodic reminder emails. Participants were sent monthly feedback on the percentage of records in which they answered the query as well as percentile rank among their peers. Coffee cards were distributed to those with > 90% field completion rates. The cannabis field was activated on June 1, 2018. We excluded data from the four months after the field was introduced as a "wash-in period" during which clinicians became accustomed to completing the field. Data were extracted for all ED visits between October 1, 2018, and September 30, 2019. Demographic information and patient visit characteristics were abstracted by standardized query from the EMR. Two of the authors (RGH and EKC) performed chart review on a convenience sample of 10% of the adult and 10% of the pediatric visits to validate the performance of the cannabis question field. Visits were selected to comprise 50% "yes" and 50% "no" answers in the cannabis field. A visit was considered confirmed as cannabis-related if there was documentation of cannabis in the history, differential diagnosis, medical decision making, biological testing, and/or ICD-10 codes. We examined distributions of variables of interest among visits, stratified according to the response to the cannabis item. Visit characteristics analyzed included patient demographics, insurance type, mode of arrival to the ED, time of arrival, triage category, ED length of stay, hospital length of stay, ICD-10-CM ED diagnosis codes (first 20 for each visit), and disposition. ICD-10-CM codes considered "cannabis attributed" were 40.7x (poisoning by narcotics and psychodysleptics, cannabis) and F.12.xx (use of cannabinoids). ICD-10-CM codes were aggregated into 18 clinically relevant medical condition groups. Hospital charges include all charges (not costs) billed from the ED stay and hospitalization (if applicable) stemming from an ED visit in which clinicians identified cannabis as a reason for the visit. Data were extracted from the EMR (Epic, Information Technology Health Informatics, Madison, WI) into a study database (Microsoft Excel, Microsoft Office Professional Plus 2016). Results are presented using descriptive statistics including 95% confidence intervals (CIs). Response rates are described as median with interquartile range. Percentages for characteristics of noncannabis and unclassified visits were compared to those of cannabis-attributed visits and tested for significance using chi-square at the 95% confidence level. Fisher's exact test was used if any data field contained variables that were less than or equal to five. We examined a total of 47,888 ED visits in the study period. Seventy-four percent (35,287) were adult and 26% (12,555) were pediatric (age < 18 years). Clinician completion of the EMR cannabis question was 76% (35,287/47,888) of visits, 73% in adults and 84% in children. Clinicians completed the question in a median of 80.9% (IQR = 62.2%–90.8%) of their cases. The median percentage of cannabis-attributed ED visits per total visits by individual providers was 1.18% (IQR = 0.74%–1.50%). Among adult patient visits, 412 visits were cannabis attributed, representing 1.6% of classified visits and 1.1% of overall visits. Among 12,555 pediatric patients, 69 visits were cannabis attributed, representing 0.66% of classified visits and 0.55% of overall visits. Among visits that were cannabis attributed by clinicians, 82% had chart documentation of cannabis use. Among visits determined to be non–cannabis attributed by clinicians, none had documentation of cannabis in the chart. Assuming that the chart review was the criterion standard, the single-item question had a sensitivity of 100% and specificity of 84% (PPV = 82%, NPV = 100%). Of visits identified by clinicians as cannabis attributed, a minority had a cannabis-related ICD-10-CM code: only 22% in adults and 17% in children (Table 1). A cannabis-related ICD-10-CM code was rare in patients who were determined by clinicians to have a non–cannabis-attributed visit (0.01% in adults, 0.30% in children) or were not classified (0.01% in adults, 0.20% in children). ICD-10-CM codes were specific (99.1%) for determining a cannabis-related ED visit, but were poorly sensitive (5.4%) and would have missed 80% of all cannabis-attributed visits. Adults with cannabis-attributed ED visits in our study were more likely to be young; male; and Black, Alaska Native or American Indian, or multiracial (Table 1). Adults with cannabis-attributed ED visits were more likely to use an ambulance but less likely to be admitted to the hospital and had shorter hospital lengths of stay once admitted. When compared to children with non–cannabis-attributed ED visits, children with cannabis-attributed ED visits were older (≥12 years old) and more likely to arrive by ambulance and have much longer ED lengths of stay (10 hours longer on average), but were no more likely to be admitted to the hospital (Table 1). Overall, the most common diagnoses in cannabis-attributed ED visits were poisoning (18%), neurologic (14%), trauma (12%), gastrointestinal (10%), cardiovascular (9%), and mental health (8%). Poisoning, neurologic, and mental health diagnoses were more common in cannabis-attributed than non–cannabis-attributed ED visits. Total hospital charges for cannabis-attributed ED visits in this single ED were $5,573,423 per year. This included $583,749 in pediatric ED visits and $4,989,674 in adult ED visits. Cannabis-attributed visits accounted for 0.74% of all hospital charges related to ED visits during the study time period. ICD-10-CM codes specific for cannabis were present in only a small percentage of cases where the treating clinician attributed the ED visit to cannabis. Our methodology has been used previously to identify methamphetamine-associated ED visits8, 9 and suggests a potential solution to the limits of ICD codes in identifying health system burden of drug-related visits.6, 7 Adult patients with cannabis-attributed ED visits used an ambulance more often but measures of acuity were lower compared to non–cannabis-attributed visits. This may reflect the severity of CNS symptoms and intoxication in adult patients upon presentation—particularly with high potency forms of cannabis now readily available—in combination with the relatively short-lived toxicity of cannabis.10 We report a lower admission rate for cannabis-associated visits than previous studies.3 It is possible that our sensitive identification approach captured more clinically mild presentations. Children with cannabis-attributed ED visits were typically adolescents, likely with intoxication/overuse as opposed to the severe toxicity related to exploratory behavior in younger children.10 Cases in young children were relatively rare but tended to consume a significant amount of ED resources: pediatric cannabis-attributed visits in our study had a length of stay that was 10 hours longer than the general pediatric ED population. Both adults and adolescents had higher rates of traumatic injuries in their cannabis-attributed visits, suggesting that cannabis' role in traumatic injury should be a continued area for public health awareness and safety work. We performed this study in a single, urban, tertiary care academic ED in a state where cannabis is legal for retail as well as medical use. The results may not be generalizable to other institutions or states. We did not objectively confirm clinician gestalt because there is no criterion standard for determining whether a visit is truly attributable to cannabis. Furthermore, although we validated the cannabis attribution field against chart documentation, it is likely that the field captures cases in which chart documentation failed to mention cannabis. It is important to note there are likely biases when relying on providers to determine causation. For example, providers may have been biased in assigning cannabis use to certain racial or ethnic groups, which might account for the differences in these factors between cannabis-attributed and non–cannabis-attributed visits noted. Finally, we report total hospital charges as an estimate of the financial burden of cannabis on a single ED, rather than actual cost to the patient or to the health care system. In this study in an ED within a cannabis-legal state, cannabis-attributed visits accounted for a small percentage of all adult and pediatric ED visits but imposed a significant burden on the health care system's resources. As cannabis legalization expands to more states, analyses of health impacts from cannabis use that depend on ICD-based case definitions may underestimate the true burden.

Source: Academic Emergency Medicine

DOI: 10.1111/acem.14275