Impact of a substance use screening program on racial disparities in urine toxicology testing

Abstract

Objective

To evaluate whether implementing a standardized substance use screening program reduced racial disparities in urine toxicology (Utox) testing among delivery admissions.

Methods

We conducted a retrospective cohort study of delivery admissions at a single academic center from June 2019 to June 2025. To address known disparities in Utox testing, a standardized iPad‐based screening tool using a modified 4Ps screen was implemented February–May 2023. Screening responses triggered protocolized Utox testing based on a predefined algorithm. Clinical and demographic predictors of Utox testing were compared pre‐ and post‐implementation using chi‐square and t‐tests. Logistic regression with predictor × era interaction terms assessed changes in associations with testing; final models were adjusted for age, race/ethnicity, parity, and insurance.

Results

Among 36,656 admissions, 216 (0.59%) underwent Utox testing: 181 (0.80%) pre‐implementation and 32 (0.26%) post‐implementation. Pre‐implementation, Black patients accounted for 26.8% of tests but only 7.0% of the population. Post‐implementation, Black patients represented only 6.3% of tests and 6.5% of the population. The testing rate for Black patients declined from 3.2% to 0.5% (p  <  0.001). In unadjusted models, testing for effect modification by implementation period, race (p = 0.02 for Black vs. White; interaction odds ratio [OR], 0.17; 95% confidence interval [CI], 0.05–0.66), and age (p = 0.001; interaction OR 1.14; 95% CI, 1.06–1.23) showed significant interactions. In adjusted models, the disparity in testing by race significantly decreased post‐implementation (interaction adjusted OR [aOR] for Black vs. White 0.17; 95% CI 0.05–0.66; p = 0.01). Among those tested, positive Utox screens did not differ significantly by period (43.2% pre‐implementation vs. 56.3% post‐implementation, p  =  0.11).

Discussion

Implementation of a standardized substance use screening tool was associated with a significant reduction in racial disparities in Utox testing at delivery.

1. INTRODUCTION

Substance use is reported in 5%–14% of pregnancies, depending on the type of substance involved [1, 2, 3]. Identifying substance use in pregnancy may allow for timely interventions that benefit both the pregnant person and their newborn. Despite these potential benefits, substance use screening in the United States has been inconsistently implemented, and at times has been inequitable and even discriminatory [4, 5, 6, 7].

Professional organizations including the American College of Obstetrics and Gynecologists, Society for Maternal‐Fetal Medicine, World Health Organization, and United States Preventative Services Task Force recommend universal screening early in pregnancy through one of several screening tools including the NIDA Quick Screen, 4Ps Plus, and CRAFFT instruments [8, 9, 10, 11]. Positive screens allow for a brief intervention (also referred to as Screening, Brief Intervention, and Referral to Treatment [SBIRT]) and, when clinically indicated, biologic testing with urine toxicology (Utox).

Screening for substance use on labor and delivery represents an additional opportunity to identify patients who were missed in outpatient screening; who may have an undiagnosed, newly diagnosed, or untreated substance use disorder; or whose newborns may be at risk of withdrawal. Additionally, screening in the outpatient setting is imperfect, with one study showing that more than 20% of obstetricians did not routinely screen for substance use in their offices [12].

However, when screening on labor and delivery is non‐universal and dependent on subjective judgment, bias may be introduced. In the United States, approaches range from universal Utox testing to universal questionnaire‐based screening (either verbal or written) to discretionary use of Utox testing–based individual provider suspicion. These variations in policy, even when created in good faith, can have negative consequences. Despite generally similar estimated rates of substance use across racial groups, Black pregnant patients are more likely to be screened, more likely to have Utox tests sent, and more likely to experience punitive consequences or newborn separation compared with White patients. [4, 5, 6, 7] Screening strategies that incorporate sociodemographic criteria such as limited or late entry to prenatal care risk amplifying racial disparities in testing [7].

We previously found that at our institution, Black and Hispanic patients, publicly insured patients, and patients with a hospital‐assigned provider (vs. a private provider) were more likely to have Utox tests despite similar rates of positive tests across racial groups. This pattern is well documented in the literature and reflects long‐standing inequities in substance use screening approaches [13, 14, 15]. Because of these findings, a multidisciplinary team of providers collaborated to design and implement a standardized, universalized screening process for substance use screening on labor and delivery. The objective of this study was to compare Utox testing before and after the implementation of the substance use screening protocol.

2. METHODS

This is a retrospective cohort study evaluating the impact of a standardized substance use screening program on Utox testing among pregnant individuals admitted for delivery at a single labor and delivery unit serving a large, urban metropolitan area. The study period spanned from June 1, 2019 to June 30, 2025.

2.1. Intervention

Before 2023, Utox at the time of labor and delivery admission was ordered at individual clinician's discretion. To reduce inequities in discretionary testing, our unit implemented a standardized iPad‐based screening process using a modified version of the 4Ps Plus tool, administered at the time of delivery admission. We selected the 4Ps Plus tool because it is brief, validated for use in pregnancy, and feasible for integration into our unit's workflow [16, 17]. Although the 4Ps Plus has not been independently validated for administration on labor and delivery, it was used in this protocol as a standardized screening mechanism to promote consistency and reduce clinician discretion in the assessment of substance use risk.

The original 4Ps Plus tool asks patients whether their parents or partner have problems with alcohol or drugs, and whether they have used these substances in the past or during pregnancy [16]. For our screening protocol, we adapted the tool to comprehensively identify substances with implications for maternal and neonatal care in the peripartum setting, and to incorporate the timing of use. Specifically, we expanded the questionnaire to ask about specific drug categories, including prescribed and unprescribed opioids and medications used in opioid use disorder (e.g., buprenorphine or methadone), which are not explicitly emphasized in the original 4Ps Plus (Supporting Information). We also incorporated questions about timing of use (within the last week, within the last month, in the first trimester, or never), given the relevance of recent use for peripartum management and neonatal monitoring.

Screening responses automatically triggered protocolized Utox testing based on a predefined algorithm that was developed iteratively over 12 months by a multidisciplinary team of maternal‐fetal medicine physicians, pediatricians, resident physicians, nursing leadership, and quality improvement staff (Figure 1). Algorithm development occurred through regular multidisciplinary meetings and was informed by review of relevant literature, existing institutional practices, and expert consensus regarding screening and testing practices related to substance use in pregnancy, with refinement through stakeholder feedback prior to implementation. Patients who reported recent (within the last month) drug or opioid use, were on maintenance medications such as methadone or buprenorphine, or exhibited signs or symptoms of intoxication or withdrawal were considered high‐risk; these responses prompted the clinician to obtain a focused history, request and document informed consent for Utox testing, and notify the primary obstetrical provider and social work team for SBIRT evaluation.

FIGURE 1.

Standardized substance use screening and urine toxicology testing workflow at delivery admission. DOA, drugs of abuse; MD/NP, physician or nurse practitioner; OB, obstetric provider; RN, registered nurse; SBIRT, Screening, Brief Intervention, and Referral to Treatment; UDS, urine drug screen.

Additionally, as part of the intervention, we intentionally removed tetrahydrocannabinol (THC/cannabinoids) from the routine Utox panel. Marijuana is legal for adult use in our state, and its detection in peripartum toxicology screens may contribute to disproportionate Child Protective Services (CPS) referrals and other punitive actions that disproportionally affect Black and Hispanic birthing people [18, 19]. To mitigate these potential downstream consequences, cannabinoids were excluded while retaining all the remaining components of the hospital's standard drug screen, including opioids, stimulants (e.g., cocaine, amphetamines), ethanol, benzodiazepines, methadone, buprenorphine, and dissociative agents such as phencyclidine (PCP).

Prior to implementation, the screening protocol and associated workflow were introduced to labor and delivery providers through unit‐based education sessions, written guidance, and integration into existing admission processes. The iPad‐based screening tool was embedded into routine delivery admission workflows. Formal monitoring of provider adherence to the screening tool or testing algorithm, including individual‐level screening completion, was not assessed as part of the study.

2.2. Analysis

The primary objective was to assess whether implementation of the standardized substance use screening program reduced racial disparities in Utox testing at delivery admission. Secondary outcomes included Utox positivity among those tested and changes in the association between other patient demographics and Utox testing before and after implementation.

We defined three non‐overlapping periods: (1) pre‐implementation period (June 1, 2019 to January 31, 2023), (2) washout period (February 1, 2023 to May 31, 2023), and (3) post‐implementation period (June 1, 2023 to June 30, 2025). The washout period was excluded to allow for the rollout of the screening program, during which staff underwent education and the workflow processes were refined.

Demographic and clinical characteristics were compared between individuals who did and did not receive Utox testing in the pre‐ and post‐implementation periods using chi‐square tests for categorical variables and t‐tests for continuous variables. Race and ethnicity were self‐reported by patients at the time of hospital registration and abstracted from the electronic medical record. Logistic regression was performed in both periods to assess the relationship between demographic predictors and Utox testing. To evaluate whether associations between patient characteristics and Utox testing differed before versus after implementation, we performed unadjusted logistic regression models including interaction terms between each predictor of interest and implementation period (i.e., race x implementation period). Multivariable logistic regression was performed to adjust for potential confounders. Lastly, as a sensitivity analysis, we performed segmented regression to account for temporal trends in Utox testing. Models included terms for time (continuously by month), implementation period, and post‐implementation time to assess pre‐intervention trends and changes after implementation, including changes in the association between race and testing. A p value of < 0.05 was considered statistically significant. Stata MP Version 19.5 (StataCorp LLC) was used for all analyses. The study was reviewed by the Cedars‐Sinai IRB and was granted an exemption as the study involved analysis of existing de‐identified clinical data.

3. RESULTS

Over the 6‐year study period, there were 36,656 delivery admissions. Excluding deliveries during the washout period, there were 22,618 deliveries in the pre‐implementation period and 12,148 deliveries in the post‐implementation period. A total of 216 individuals underwent Utox testing, including 181 (0.8%) of pre‐implementation period deliveries and 32 (0.26%) of post‐implementation period deliveries (p < 0.001). During the pre‐implementation period, Utox testing rates declined over time, from approximately 2.4% of deliveries in 2019 to 0.4% in 2022. Baseline characteristics of the pre‐ and post‐implementation groups are described in Table 1. Absolute differences in baseline characteristics were minimal, although there was a reduction in proportion of hospital‐affiliated delivery provider in the post‐implementation period.

TABLE 1.

Baseline characteristics of delivery admissions in pre‐ and post‐implementation periods.

Characteristic	Pre‐implementation (n = 22,618)	Post‐implementation (n = 12,148)	p
Maternal age (years)	34.1 ± 4.8	34.4 ± 4.8	<0.001
Parity			0.455
Nulliparous	11,802 (52.9%)	6375 (53.3%)
Multiparous	10,507 (47.1%)	5580 (46.7%)
Race/ethnicity			0.001
Non‐Hispanic White	12,980 (57.4%)	6836 (56.3%)
Non‐Hispanic Black	1681 (7.4%)	790 (6.5%)
Hispanic	3497 (15.5%)	1971 (16.2%)
Asian/Pacific Islander	2856 (12.6%)	1650 (13.6%)
American Indian/Alaska Native	29 (0.1%)	17 (0.1%)
Other/Unknown	1575 (7%)	882 (7.3%)
Hospital‐affiliated OB			<0.001
Private OB provider	16,907 (74.7%)	10,866 (89.4%)
Hospital OB provider	5711 (25.3%)	1282 (10.6%)
Insurance Type			0.006
Private	20,845 (93.1%)	11,091 (92.3%)
Public	1000 (4.5%)	567 (4.7%)
Other/Unknown	546 (2.4%)	357 (3%)

Abbreviation: OB, obstetrical.

Rates of Utox testing by demographic characteristics are shown in Table 2. Positive Utox tests did not differ by race/ethnicity, insurance, parity, or age in either period. In the pre‐implementation period, there was a higher rate of Utox testing among Black patients and publicly insured patients. Black patients accounted for 29.3% of those tested, but only 7.2% of the population overall (p < 0.001). This pattern persisted in the years immediately preceding implementation, with Black patients remaining overrepresented among those undergoing Utox testing during 2021–2022. Post‐implementation, Black patients represented 6.3% of those tested compared with 6.5% of the population (p = 0.17). In the pre‐implementation era, 3.2% of Black patients had Utox performed compared to 0.5% in the post‐implementation era (p < 0.001).

TABLE 2.

Comparison of demographic and clinical characteristics among patients with and without Utox testing across implementation periods.

	Pre‐implementation			Post‐implementation
Characteristic	No Utox (n = 22,437)	Utox (n = 181)	p	No Utox (n = 12,116)	Utox (n = 32)	p
Maternal age, mean (SD)	34.1 (4.8)	28.8 (6.4)	<0.001	34.4 (4.8)	32.8 (5.2)	0.068
Parity			0.009			0.27
Nulliparous	11,729 (53.0)	73 (42.9)		6362 (53.4)	13 (43.3)
Multiparous	10,410 (47.0)	97 (57.1)		5563 (46.6)	17 (56.7)
Race/ethnicity			<0.001			0.029
Non‐Hispanic White	12,940 (57.7)	40 (22.1)		6823 (56.3)	13 (40.6)
Non‐Hispanic Black	1628 (7.3)	53 (29.3)		788 (6.5)	2 (6.3)
Hispanic	3430 (15.3)	67 (37.0)		1960 (16.2)	11 (34.4)
Asian/PI	2849 (12.7)	7 (3.9)		1649 (13.6)	1 (3.1)
AI/AN	29 (0.1)	0 (0.0)		17 (0.1)	2 (6.3)
Other/Unknown	1561 (7.0)	14 (7.7)		879 (7.3)	3 (9.4)
Insurance type			<0.001			<0.001
Private	20,798 (93.6)	47 (28.1)		11,081 (92.5)	10 (31.3)
Public	925 (4.2)	75 (44.9)		551 (4.6)	16 (50)
Other/Unknown	501 (2.3)	45 (27)		351 (2.9)	6 (18.8)
Hospital‐affiliated OB			<0.001			<0.001
Private OB provider	16,880 (75.2)	27 (14.9)		10,856 (89.6)	10 (31.3)
Hospital OB provider	5557 (24.8)	154 (85.1)		1260 (10.4)	22 (68.8)

Note: Data are presented as mean (SD) and n(%).

Abbreviations: AI, American Indian; AN, American Native; OB, obstetrical; PI, Pacific Islander.

In analysis adjusting for age, parity, race, and insurance, Black race (adjusted odds ratio [aOR], 2.3; 95% confidence interval [CI], 1.4–3.7) and Hispanic race (aOR, 1.7; 95% CI, 1.1–2.7) were both predictors of Utox testing in the pre‐implementation period. Advanced maternal age was associated with lower odds of Utox testing (aOR, 0.43; 95% CI, 0.3–0.7) and public insurance was associated with increased odds of Utox testing (aOR, 22.9; 95% CI, 14.9–35.1) compared with private insurance in the pre‐implementation period. In the post‐implementation period, Black race was no longer associated with increased Utox testing (aOR, 0.4; 95% CI, 0.1–1.5) and neither was Hispanic ethnicity (aOR, 0.72; 95% CI, 0.3–1.8) when compared to White race.

Logistic regression with interaction terms was performed to assess the impact of the screening program on the pre‐implementation period associations (Table 3). In both adjusted and unadjusted interaction models, the association between Black race and Utox testing significantly changed after implementation, indicating attenuation of the pre‐existing disparity. Maternal age demonstrated a significant interaction with implementation period, such that advanced maternal age remained associated with Utox testing in both periods and this association strengthened in the post‐implementation period. The remainder of the interactions were non‐significant. Though public insurance remained associated with Utox testing in the post‐implementation period, this association was not significant in adjusted analysis by period.

TABLE 3.

Unadjusted and adjusted odds ratios for interaction between implementation period and predictors of urine toxicology testing. Covariates for each adjusted analysis included race/ethnicity, advanced maternal age (AMA), parity, and insurance type.

Characteristic [× Implementation period]	OR (95% CI)	p	aOR (95% CI)	p
Age	1.14 (1.05–1.22)	0.001	1.11 (1.05–1.18)	<0.001
Advanced maternal age	2.98 (1.35–6.60)	0.007	3.3 (1.4–7.5)	0.005
Multiparous × Era	1.00 (0.46–2.19)	0.998	1.09 (0.49–2.42)	0.83
Non‐Hispanic Black (White ref)	0.25 (0.08–0.84)	0.024	0.17 (0.05–0.66)	0.01
Hispanic (White ref)	0.47 (0.19–1.14)	0.095	0.44 (0.18–1.09)	0.074
Public Insurance × Era (Private ref)	0.9 (0.37–2.2)	0.81	0.95 (0.39–2.4)	0.93
Hospital OB Provider (Private OB ref)	1.1 (4.66–2.57)	0.84	0.82 (0.34–1.99)	0.66

Note: OR and aOR are presented for the interaction term for predictor × implementation period.

Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; OB, obstetrical; OR, odds ratio.

In a segmented regression sensitivity analysis accounting for temporal trends, Utox testing declined prior to implementation, while the Black–White disparity in testing decreased after implementation, indicating that the observed reduction in disparity was not solely attributable to secular trends. Lastly, among those who underwent Utox testing, positive Utox screens did not differ significantly by period (43.2% vs. 56.3%, p  =  0.11).

4. DISCUSSION

In this retrospective cohort of over 36,000 delivery admissions, implementation of a standardized substance use screening program was associated with a marked reduction in racial disparities in Utox testing. Prior to implementation, patients who were Black, publicly insured, or cared for by a hospital‐affiliated obstetrical provider were disproportionately tested despite similar rates of positive screens across racial groups. After introducing a universal, structured questionnaire and a clearly defined algorithm for when testing was clinically indicated, these disparities narrowed substantially. Among Black patients, in particular, the absolute rate of Utox testing declined from 3.15% to 0.51%, and the difference in testing by race was no longer statistically significant.

Data on whether there are racial disparities in the prevalence of substance use in pregnancy are mixed. Nonetheless, screening based on risk factors or subjective concern may ultimately lead to disparate consequences for pregnant people of color [7, 20, 21]. Chasnoff et al. found that despite similar alcohol and drug use among White and Black pregnant women at the first prenatal appointment, Black women were 10 times more likely than White women to be reported to child protective services at delivery [16]. Even risk‐based screening protocols that include scant or no prenatal care as an indication for Utox testing may lead to disproportion urine toxicology testing in Black and Latinx patients given the higher prevalence of obstacles to care in these communities.

Our approach incorporated evidence suggesting that a history of recent use is a strong predictor of a positive Utox test, and knowledge of this may inform both maternal and pediatric follow‐up [13]. Importantly, this intervention did not involve universal Utox testing, but rather universal iPad‐based screening using a validated questionnaire paired with standardized, criteria‐based consented testing. In previously reported institutional data from a subset of delivery admissions, cannabinoids were the most commonly detected analyte on Utox screening [15]. However, because marijuana is legal for adult use in California and detection at delivery admission has limited implications for intrapartum management, cannabinoids were removed from the routine panel to reduce the potential for disproportionate downstream consequences.

Utox testing itself is imperfect, and providers should be thoughtful about the appropriate use of this test given the potential ripple effects to the family unit. Besides report of recent substance use, our protocol included current medication‐assisted treatment and signs/symptoms of intoxication or withdrawal as criteria to consent for Utox testing. Martinez and authors proposed a framework for decision‐making when considering sending an Utox test, including specifying the clinical question the provider is trying to answer, considering whether the Utox test will answer that question, how testing compares with the risks of testing, whether there are more effective tests to address the question, and whether there are biases affecting the decision [6].

Though universal screening may help reduce the contribution to disparate punitive consequence for positive Utox screenings, this is not the final solution. Roberts et al. suggested in their work that though universal screening may address some of the downstream implications of biased Utox testing, addressing the disparities in response to positive screens is where a larger impact may lie. Currently, over 50% of physicians including obstetricians and pediatricians support removing newborns from parental custody in the setting of substance use according to a survey‐based study in Michigan, despite studies showing higher likelihood of remission and reduction in neonatal abstinence syndrome when dyads are kept together [22, 23, 24, 25].

This study is not without limitations. It reflects practices within a single academic center and may not generalize to settings with different patient populations or workflows, particularly in areas with higher rates of substance use, greater proportion of uninsured patients, or different provider practice models. Although the intervention was designed to standardize testing, some provider variability likely remained, and we could not measure adherence to the screening tool or algorithm. In addition, individual‐level data completion or refusal of the iPad‐based screening questionnaire was not systemically captured, precluding assessment of potential non‐response bias.

Screening at delivery admission also provides a limited snapshot and may not capture substance use patterns earlier in pregnancy, including frequency, quantity, or mode of use. The relatively small number of Utox tests, particularly during the post‐implementation period, may have limited the power to detect subtle differences by period. We also were unable to collect information on CPS follow‐up on positive screens, and thus the downstream implications of reduced testing in this study were not described. Lastly, though we observed that our protocol did not appear to affect the association of Utox with public insurance, this finding should be interpreted cautiously, as publicly insured patients represented < 5% of the cohort and the number of Utox tests in the post‐implementation period was small.

Despite these limitations, our study demonstrates that a universal substance use screening process on labor and delivery can meaningfully reduce inequitable testing practices as it pertains to race. As institutions grapple with the potential harms of punitive perinatal drug policies, this approach offers a feasible model to screen for substance use while minimizing bias. Future work should evaluate the impact of such protocols on long‐term maternal outcomes, neonatal outcomes, consequences following positive Utox screens, and patient experience.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

FUNDING INFORMATION

The authors received no specific funding for this work.

ETHICS STATEMENT

This study was reviewed and approved by the Cedars‐Sinai Medical Center Institutional Review Board and was conducted in accordance with institutional policies and internationally accepted ethical standards. Due to de‐identified retrospective design, the study qualified for a waiver of informed consent.

Supporting information

Supporting Information

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.