Significance

Screening tools for substance use are mostly used once in pregnancy and miss assessments of substance use throughout pregnancy. This is the first study to show that prenatal screening starting in the first trimester and performed periodically throughout pregnancy resulted in decreased substance use by the end of the pregnancy and improved maternal and infant outcomes

Introduction

Opioid use disorder (OUD) among women of reproductive age is a national epidemic (Kennedy-Hendricks et al., 2016), leading to adverse health outcomes in pregnancy, including increased risk of maternal and fetal mortality, obstetric complications, co-occurring mental health disorders, and comorbid medical conditions ((WHO), 2017; Benningfield et al., 2012; Costello & Thompson, 2015; Jones et al., 2008; Lisonkova et al., 2019; McCarthy et al., 2017; Mittal, 2014). In turn, adverse health outcomes increase the risk of relapse and worsening OUD (Clark et al., 2015; Marchuk, 2014).

In the most recent national study, the incidence of illicit drug use among pregnant women age 15–44 years old is 30%, SAMHSA (2017) increasing more than four-fold over the past decade (Haight et al., 2018). As OUD rates in pregnancy have increased, so have rates of neonatal opioid withdrawal syndrome (NOWS) (Patrick et al., 2019). Given the rapidly rising cases of OUD, the Centers for Disease Control and Prevention (CDC), and the World Health Organization (WHO) have named the opioid epidemic a research priority ((ANA), 2016; (WHO), 2017; (ODPHP), 2014; (ACOG), 2012).

In 2012, the American College of Obstetrics and Gynecology ((ACOG), 2012) and CDC recommended substance use screening for all pregnant women regardless of social status, educational level, race, or ethnicity (ACOG, 2017; Wright et al., 2016). Screening, brief intervention, and referral to treatment (SBIRT) is an evidence-based approach to identify and initiate OUD management in pregnancy (Babor et al., 2007). Examples of validated screening tools to identify substance use include the NIDA Quick Screen-ASSIST (Modified Alcohol, Smoking, and Substance Involvement Screening Test), ((NIDA), 2017), the 4P's Plus, (Chasnoff et al., 2007) and the Substance Use Risk Profile-Pregnancy (SURP-P) scale (Yonkers et al., 2010).

A single screening during pregnancy is inadequate to assess the variability of substance use throughout pregnancy. A validated screening tool used instead at regular intervals throughout pregnancy may be more sensitive. Interval screening throughout pregnancy can provide the opportunity for education, assistance, and referral to treatment services early in pregnancy and throughout pregnancy to improve maternal and infant outcomes and potentially to reduce the burden of maternal substance use (ACOG, 2015; Coleman-Cowger et al., 2019; Gopman, 2014; Wong et al., 2011; Wright et al., 2016).

To address the inadequacy of screening only once per pregnancy, we developed a standardized substance use prenatal screening tool given once during each trimester and upon any hospitalization during pregnancy. The study’s primary aim was to determine if serial, early self-reported prenatal substance screening leads to decreased substance use by the end of the pregnancy and to improved perinatal outcomes. Our long-term clinical goal is to provide evidence that early administration of a serial standardized prenatal substance screening tool is necessary to identify both illicit and prescribed substance use throughout pregnancy. This tool should ideally be easily adopted into diverse clinical practice models.

Materials and Methods

We conducted a retrospective cohort study of women and their infants using the prenatal clinic and hospital data from an academically affiliated community hospital from January 2015 to December 2017. This study was approved by the Medical College of Wisconsin’s Institutional Review Board with a waiver of informed consent.

A substance use screening tool (Boden screening tool, Electronic Supplementary material A) was developed through our clinical program by self-reporting substance use, prescribed and nonprescribed. It was administered at the first prenatal visit and subsequent visits once per trimester until delivery. Using various questions within other research-based alcohol and drug use screening questionnaires, including the modified CAGE and TWEAK questions (Morse et al., 1997), the Boden screening tool was created to identify both prescribed and illicit substances used by pregnant women: marijuana, morphine, methadone, meperidine, oxycodone, propoxyphene, hydromorphone, fentanyl, diazepam, cocaine, heroin, codeine, phenobarbital, clomipramine, hydroxyzine, theophylline, lithium, chlorpromazine, clonidine, diphenhydramine, hydrocodone, lysergic acid diethylamide (LSD), and solvents/aerosols (Electronic Supplementary material A). This tool was completed in the clinic waiting area at routine obstetrical prenatal visits and took approximately 5 min to complete.

Obstetric providers used a positive screen to identify candidates for referral to community resources and drug treatment programs. Pregnant women using substances, either prescribed or illicit, were referred to a pediatric provider for education regarding the need for observation and possible NOWS treatment, if indicated, after birth. The data gathered from the screening tool was part of the pregnant woman’s electronic medical record.

We used electronic medical records of patient-infant dyads to collect the following: demographics and urine drug screen results; tobacco and alcohol use, family history of substance use, and the number of prenatal substance use screenings. Urine drug screens were voluntarily collected only if self-reported prenatal screening were positive. Infant data included gestational age at delivery, diagnosis of NOWS, pharmacologic treatment of NOWS with morphine, hospital length of stay (LOS), and results of infant urine and meconium drug screenings.

For analysis, women were divided into three groups based on the timing of their first screen: first trimester (0–13 weeks), second trimester (14–26 weeks), and third trimester (27–41 weeks). We performed a McNemar test by doing before and after substance use measurements (matched pairs) to determine if early prenatal screening affected prenatal substance use by the end of pregnancy. We performed Poisson regression to find differences in LOS between the three groups. We also performed bivariate and multivariate logistic regression to examine the relationship between the three groups regarding pregnancy and infant outcomes.

Analyses were conducted using Stata15/SE (Stata Corp, College Station, TX). Categorical variables are presented as percentages. Logistic and ordinal logistic regression results are presented as Odds Ratios (OR) and 95% confidence intervals with the third trimester (27–41 weeks) as the reference group. Regressions were adjusted for maternal demographics and risk factors.

Results

Two hundred fifty-one women were identified to have substance use on their first prenatal screen based on the positive substance use screening tool (Boden screening tool, Electronic Supplementary material A). Significant differences were noted in demographics and risk factors between the three trimesters (Table 1). Older maternal age (p < 0.05), nulliparity (p < 0.05), marital status (p < 0.001), higher income (p < 0.05), private insurance (p < 0.01), and current employment (p < 0.05) were associated with early prenatal screening. Alcohol use was associated with early prenatal screening, but not tobacco use. In the first trimester, screening resulted in 52% of pregnancies having ≥ 3 screens before birth compared to 6% of women with ≥ 3 screens in their third trimester (p < 0.001). Positive screening in the second and third trimesters resulted in 56–60% of obtaining urine drug screens compared to a 26% rate when positively screened in the first trimester.

Table 1 Significant maternal demographics and risk factors between the three groups of the earliest prenatal substance screena
Full size table

For women with early prenatal screening in their first trimester, there was a decrease of any self-reported substance use by 72%, marijuana by 57%, all opiates by 40%, and benzodiazepine use by 76% when rescreened at the end of their pregnancy (Fig. 1). For those with initial screening in their second trimester, there was a smaller decrease in any self-reported substance use by 44%, marijuana by 36%, all opiates by 20%, and benzodiazepine use by 50% when rescreened at the end of their pregnancy. We even found a much smaller decrease when screened in their third trimester in any substance by 6%, marijuana by 9%, all opiates by 0%, and benzodiazepine use by 15% when rescreened at the end of their pregnancy. Please see Electronic Supplementary material B for the list of other substances analyzed.

Fig. 1
figure 1

Significant decrease of any substance, marijuana and opiate use between previous vs. end of pregnancy screenings†

Full size image

At the end of pregnancy, 15% of women who were initially screened in the first trimester had a positive urine drug screen compared to 44–45% of women screened in their second and third trimesters (Fig. 2). Also, infants of women screened in the first trimester had an 8% positive infant urine drug screen and a 10% positive meconium drug screen. We found a much higher positive urine and meconium drug screen result when women were initially screened later in pregnancy (Fig. 3). Please see Electronic Supplementary material C for the substances found in the combined maternal and infant urine drug screens and infant meconium drug screen.

Fig. 2
figure 2

Decreased positive maternal urine drug screens at end of pregnancy*

Full size image
Fig. 3
figure 3

Decreased positive infant urine and meconium drug screens with first trimester screening*

Full size image

Table 2 shows differences in infant outcomes. We noted lower rates of positive opioid urine and meconium drug screen results, diagnosis of NOWS, and NOWS requiring morphine treatment when pregnancies were screened early in the first trimester, compared to later screenings in the second and third trimesters. The mean infant hospital LOS with initial screening in the second trimester was twice as long as with early screening in the first trimester. Using Poisson regression, we identified a significant increase in LOS in those with initial second-trimester screening, but not for initial third-trimester screening, using the first trimester as reference.

Table 2 Significant differences in infant outcomes between the three groups of the earliest prenatal substance screenb
Full size table

Table 3 shows the adjusted OR of outcomes when analyzed by the trimester, where the earliest prenatal substance screening was administered. There was a five-fold decrease of any substance use with the initial screen in the second trimester for maternal outcomes, and a seven-fold decrease with the initial screen in the first trimester. Compared to initial screening in the third trimester, there was also a nine-fold decrease in marijuana use with screening in the first trimester but no significant reduction with screening in the second trimester. We found no significant differences in decreased opioid use and benzodiazepine use between the three trimesters. We also found a significant five-fold increase in a negative maternal urine drug screen when self-reported screening was first done in the first trimester compared to in the third trimester. For infant outcomes, we found more than a three-fold increase of no newborn diagnosis of NOWS, and a five-fold increase of no infant morphine treatment for NOWS (when identified) when patients were screened early in the first trimester compared to initial screening in the third trimester.

Table 3 Significant Adjusted Odds Ratio [OR (95%CI)] for maternal and infant outcomes between the three groups of the earliest prenatal substance screena
Full size table

Discussion

In a cohort of women at an academic-affiliated community hospital, this study demonstrates that prenatal screening starting in the first trimester and performed periodically throughout pregnancy resulted in decreased reported substance use by the end of the pregnancy and improved outcomes. This approach to periodic screening during pregnancy has several advantages. Periodic screening provides additional information to the health care team about the timing of substance use and indications of ongoing use. Given the legal and social barriers which often discourage pregnant women from reporting substance use to their healthcare providers, periodic screening may further identify ongoing substance use. In this study, early self-reported serial prenatal screening throughout pregnancy was associated with earlier identification of substance use and prompt referral for counseling and community services. We speculate that this may contribute to the study’s improved outcomes.

Substance use and substance use disorders continue to be frequently missed diagnoses in pregnancy. In order to decrease substance use and increase substance use disorder detection, a more patient-centered approach to screening in pregnancy may be required (Price et al., 2018). Current recommendations advise early universal screening in pregnancy by using a screening tool at the first prenatal visit. Those with a substance use disorder in pregnancy can be identified and referred for treatment to improve pregnancy outcomes ((ACOG), 2017).

When examining individual screening tools, the 4P’s Plus is the only validated tool for assessing drug and alcohol use in pregnancy. The SURP-P screen asks about marijuana and alcohol use only but is not validated in pregnancy. Both 4P’s Plus and SURP-P have high sensitivity as screening tests but do not assess behavioral substance use patterns, cravings, or substance use consequences. The NIDA Quick Screen-ASSIST assesses cravings and functional consequences as well as behavioral substance use patterns. However, this screen has a lower sensitivity of 85.4%, making it a lesser desirable screening test compared to the 4P’s Plus and the SURP-P (Coleman-Cowger et al., 2019).

Compared to the 4P’s Plus, SURP-P, and the NIDA Quick Screen-ASSIST, the Boden screening tool asks what prescribed and illicit substances were used in the past 3 months. Like the NIDA Quick Screen-ASSIST, we modified the CAGE and TWEAK alcohol screening questionnaires for adults to better assess craving and behavioral substance use patterns (Armstrong et al., 2001; Morse et al., 1997).

The Boden tool is an example of a patient-centered care approach. The tool allows for informed consent before screening, thus fulfilling the ethical principle of respect for persons, and fulfills the ethical principle of autonomy by allowing patients to make an informed decision about their care. Compared to universal urine drug screening, where the identification of drug use may be a stressor, this patient-centered approach allows for more collaboration with the health care team, improving the identification and treatment of substance use during pregnancy. One study on the patient-centered care approach identified decreased substance use leading to improved pregnancy and infant outcomes (Wright et al., 2012).

Using our tool early and throughout pregnancy, we were able to identify pregnant women at high risk for substance use. This led to a revision of our clinic workflow based on Kaiser Permanente’s Early Start program (Armstrong et al., 2001). Our workflow used various strategies, including the Boden screening tool, urine drug screening, early intervention, ongoing counseling, and case management by a licensed clinical social worker and pediatric nurse practitioner with expertise in substance abuse (Armstrong et al., 2003). This change in workflow helped the clinic and hospital improve provider and staff awareness of increasing substance use. The workflow change also allowed for earlier identification of substance use, allowing more time for education on the adverse effects of substance use in pregnancy. This expanded the ability to recommend treatment programs when needed, provide women with information on various community resources, and increase collaborative efforts with community programs.

We realize that pregnant women with substance use and substance use disorders face significant obstacles and barriers when seeking prenatal care. Research shows pregnant women fear they may lose custody of their children, as more than 50% of children of women with OUD are not living with their biological parents by the time they are 5 years of age (Burns et al., 2006). They may also face significant stigma from their families, social networks, and healthcare system (Jones et al., 2008). Unfortunately, the use of substances in pregnancy has been identified in some state statutes as evidence of child abuse or neglect. Many patients have been prosecuted, which can be a barrier to care and may play a role in suboptimal outcomes (McCarthy et al., 2017). In particular, rural areas create challenges for accessing community services and substance use disorder treatment (Jackson & Shannon, 2012). Such obstacles may result in adverse health outcomes and increase morbidity and mortality in this population.

Of note, the implementation of the Boden screening tool has had other notable benefits to perinatal care. Expanded education of providers and staff has improved awareness of substance use disorders in pregnancy. As more pregnancies are identified with the risk of substance use disorder complications, this tool has prompted the implementation of a county-wide task force, which has enhanced awareness of substance use disorders in pregnancy and has led to further collaboration within the public health community.

This study identified an overall increased incidence of marijuana use in pregnancy, which has been supported in other studies (Brown et al., 2017). Marijuana is the most common illicit drug identified in pregnancy, followed by amphetamine-type stimulants and opioids (Forray, 2016). Surprisingly in our study, we found a significant decrease in marijuana and opioid use for women screened early in pregnancy but did not find a similar decrease for amphetamine use.

This study identified higher positive urine and meconium drug screen rates in infants when pregnancies were screened in their second and third trimesters. This is consistent with the meconium drug screen detection window for substance use, usually identified as the second and third trimesters of pregnancy (Lozano et al., 2007). Besides, the meconium drug screen provides more complete information on drug exposure during pregnancy compared to the infant urine drug screen.

Weaknesses of the study include the use of self-reporting of substance use. However, we had a very high response rate for a positive self-report, compared to one study that showed only 5% of women admit to using illicit drugs while pregnant ((CBHSQ), 2015). There could also be recall bias when filling out the prenatal screening tool. Besides, we could not confirm the patient report of substance use definitively as urine drug screens were not collected after every screening, and urine drug screens also have significant limitations in confirming substance use. The Boden screening tool has not yet been validated in a more diverse population, and this is a retrospective study from a single obstetric community practice. Despite these limitations, we were able to show improved maternal and infant outcomes with the early administration of a serial standardized prenatal substance screening tool.

In conclusion, using the Boden screening tool throughout pregnancy is associated with decreased reported substance use during pregnancy and improved perinatal outcomes. Further research is needed on newer approaches to substance use screening, such as the patient-centered care approach, and comparing this tool to other validated pregnancy screening tools.