Reducing Bias in Substance Use Detection in Adolescents Via Standardized Substance Use Screening

Many pediatricians use question-based substance use screening to detect substance use and prevent substance use disorder (SUD).^1,2 Despite availability,³ implementing standardized, universal, question-based substance use screening with adolescents is low,⁴ hindering prevention in health care^2,3 and increasing risk for inequity and bias because of gender, age, and race.⁵ To understand the impact of universal, standardized substance use screening, we compared substance use screening occurrences and results from clinical records before and after implementing a standardized tablet-based substance use screening to determine whether standardization was associated with equitable screening practices.

This 2-cohort comparison study used clinical data for adolescent patients before (cohort 1; n = 485) and after (cohort 2; n = 485) implementing standardized, tablet-based screening. Data was collected from a single freestanding pediatric health care system in a large metropolitan Midwest city, which includes 15 urban and suburban primary care clinics and 60 000 annual visits. Patients seen for primary care encounters between 2017 and 2021 were sampled to examine substance use documentation before standardized screening, and patients seen between 2022 and 2023 were sampled to examine substance use documentation after standardized screening. If patients experienced health care encounters before and after standardized screening, then only encounters after standardized screening occurred were included. Patients who received standardized screening were 1:1 matched to patients who did not receive standardized screening, such that each case was matched to a control of the same gender, race, ethnicity, and date of birth +/− 6 months (N = 970). When more than one individual demographically matched to a patient experiencing standardized screening, the comparison sample was selected at random. An a priori power analysis indicated that with this sample size, there was sufficient power to detect group differences in binomial outcomes ≥9.5% as statistically significant with a Bonferroni-corrected α = .008 and 80% power.

Prior to standardized screening (cohort 1), substance use screening occurrences and results could be captured in the patient record via (1) clinical staff–documented patient responses in a flowsheet, (2) lab results to substance use urine screens, (3) encounter diagnoses related to intoxication or SUD, (4) adding substance use to the problem list, or (5) in the clinic note. Any mention of substance use (or nonuse) in reference to the patient in any clinical note during an encounter was considered evidence of screening (0 = not screened, 1 = screened), and whether substance use was indicated (0 = no, 1 = yes) was captured using previously published methods.^6,7 After implementing tablet-based screening (cohort 2), patients self-administered the CRAFFT 2.1+N substance use screening tool, which was integrated into Epic.⁴ CRAFFT administration (0 = not screened, 1 = screened) and past year substance use based on CRAFFT responses (0 = no, 1 = yes) were extracted from the medical record along with patient demographic characteristics for both cohorts. Differences in screening occurrences and results between cohorts were compared, stratified by sex assigned at birth (male, female), race (white, Black), and age in years. The local institutional review board approved the study; data are available upon request.

By design, the unstandardized and standardized screening cohorts were identical with respect to age, sex assigned at birth (55% female), race (54% Black), and ethnicity (1% Hispanic). With unstandardized screening, 48% of adolescents were screened, and 17% of screenings were positive. With standardized screening, more adolescents (99%) were screened, and more substance use was detected (29% positive screens).

With unstandardized screening, adolescents who were screened were more than 2 years older, on average, than adolescents who were not screened and more than 1 year older than those who received standardized screening (Table 1).

Screening occurrences did not significantly differ between males and females when screening was unstandardized or standardized (Table 1). With unstandardized screening, Black adolescents were more likely to be screened than white adolescents; no difference was detected with standardized screening.

Substance use was detected approximately 6 months earlier when screening was standardized (Table 1). Standardized screening resulted in a higher detection rate of substance use among females (29% positive with standardized; 13% positive with unstandardized screening).

Standardized substance use screening⁸ resulted in more completed screening, less bias, and better substance use detection. Adolescents may inaccurately report substance use, and screening occurrences may be undocumented, which are noted limitations. Investing in standardized universal screening may promote health equity in screening practices, contributing to reduced SUD-related morbidity and mortality.^1–3 

The following study team members contributed to coding the unstandardized clinical notes: Michelle Kalu, Myles Parks-Tiller, Elizabeth Hamik, Houley Sall, Meera Patel, SJ Jacobs, Luke Greiner, Maya Subedi, and Edward Owsley-Longino.

SUD

substance use disorder