Adolescent Risk Behavior Screening and Interventions in Hospital Settings: A Scoping Review

We conducted a scoping review given expected heterogeneity of the body of literature on this topic. Scoping reviews map out broad themes and identify knowledge gaps when the published works of focus use a wide variety of study designs.¹⁵  We used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews guidelines to inform our search and synthesis of the literature.¹⁶ 

We conducted a literature search in June 2019. Data sources included PubMed (1965–2019) and Embase (1947–2019). In the Supplemental Information, we outline the details of our search strategy. We pooled results from both queries together and removed duplicates. Inclusion criteria were study population age (adolescents aged 10–25 years), topic (risk behavior screening or risk behavior interventions), and setting (urgent care, ED, or hospital). Given that guidelines recommend universal risk screening of all adolescents, we excluded studies that were focused only on high-risk adolescents, such as patients admitted to adolescent medicine, trauma, or psychiatry services or patients admitted for toxic ingestions, suicide, or eating disorders. We also excluded any studies with interventions taking place outside the urgent care, ED, or hospital because we aimed to identify interventions that could be completed during acute care encounters. In this study, the terms “hospitalized” or “hospital setting” refer to patients admitted to pediatric units under either inpatient or observation status. We only included studies published in English.

Two independent reviewers screened, extracted, and summarized the studies (N.P. and A.D.). We first screened titles and abstracts using Rayyan software (Qatar Computing Research Institute, Doha, Qatar),¹⁷  and we resolved conflicts regarding the title and abstract screen through discussion. Next, the 2 reviewers independently completed a full-text screen. We calculated Cohen’s κ to assess interrater reliability. The 2 reviewers made joint final decisions on inclusion of studies with conflicting initial determinations.

Data extracted from the full texts included the full citation, study type, risk of bias, risk behavior domain, intervention or screening tool, results of the study, and conclusions. We described and summarized major findings, organized by the following risk behavior categories: comprehensive, sexual activity, mood and suicidal ideation (SI), substance use, and abuse and violence. Within each category, we grouped studies by subcategory: screening rates, screening and intervention tools, and attitudes toward screening and intervention. We did not combine and quantitatively analyze study results because of heterogeneity in study design.

This study was determined exempt by the Institutional Review Board at the University of California, San Francisco.

Our initial search yielded 1336 studies in PubMed and 656 studies in Embase. After duplicates were removed, 1867 unique studies were identified. After a title and abstract screen, 75 studies remained. In the full-text screen, both reviewers included 43 studies and excluded 25 studies; 7 studies were in conflict. Cohen’s κ was calculated and determined to be 0.8, correlating with a 90.7% agreement. One study that met inclusion criteria was found post hoc and included in the final review for a total of 46 studies (Fig 1). Included studies were published between 2004 and 2019, and the majority (n = 38) of the studies took place in the ED setting, whereas 7 took place in the hospital setting, and only 1 took place in the urgent care setting. Study design and risk of bias are presented in Table 1. Using methods from a study by Rea et al,¹⁸  we analyzed risk of bias for each of the included studies and found that only 2of 46 studies had a low risk of bias, 33 of 46 had moderate risk of bias, and 11 of 46 had a high risk of bias.

Below we report results by risk behavior domain, with studies organized into subcategories of screening rates, screening tools and interventions, and adolescent, parent, and clinician attitudes toward screening and intervention.

Six of 46 studies that were included in our review were focused on comprehensive risk behavior screening and/or interventions (across all risk behavior domains), as summarized in Table 2. Two of the studies took place in the hospital setting and 4 in the ED setting.

Yeo et al¹³  found that ∼10% of admitted patients at a tertiary children’s hospital had a comprehensive risk behavior assessment documented (defined as ≥5 of 7 domains: home, education, activities, tobacco use, drug and/or alcohol use, sexual activity, suicide and/or depression). An additional 28% had partial or incomplete screening, with less sensitive issues, such as home life, education, and employment, documented significantly more often than sexual activity, depression, or drug use (P = .013). In 75% of cases in which risk behaviors were identified, interventions were provided. Similarly, in a hospital study of surgical adolescent patients by Wilson et al,¹⁹  the authors found that only 16% of patients were offered screening, and of these, 30% required interventions.

Three ED studies described interventions to increase comprehensive risk behavior screening. Bernstein et al²⁰  used nonphysician providers, or health promotion advocates (HPAs), to perform risk behavior screening and were successful in standardizing comprehensive screening and intervention for adolescents in a busy ED setting by having a dedicated role for the task. However, lack of initial physician buy-in and administrative hurdles, such as funding for HPAs, training, and competition with other medical professionals (ie, social workers), made it difficult to transition this intervention into sustainable clinical practice.²⁰  In 2 studies, researchers evaluated physician reminders to screen, including a home, education, activities, drugs, sexual activity, suicide and/or mood (HEADSS) stamp on paper medical charts and a distress response survey in the electronic health record (EHR). The HEADSS stamp resulted in a significant increase in postintervention screening rates (from <1% to 9%; P = .003).²¹  The EHR distress response survey by Nager et al²²  was found to be feasible to integrate into the busy ED physician workflow, but the study offered limited insight into effects on screening or utility of the tool (assessed by using only yes or no questions).

In an ED survey study by Ranney et al,²³  for all risk behavior categories assessed, 73% to 94% of adolescent patients (n = 234) were interested in interventions, even when screen results were negative.

There were no studies on parent or clinician attitudes toward comprehensive risk behavior screening.

Nineteen studies on sexual activity screening and/or intervention were included in our review: 5 in the hospital setting (Table 3) and 14 in the ED (Table 4).

Documentation of sexual activity screening of adolescents was low in both ED and hospital settings. In retrospective cohort studies by Riese et al,²⁴  McFadden et al,²⁵  and Stowers and Teelin,²⁶  sexual activity screening rates in the hospital setting are described. The authors reported screening rates of 55% to 62%.^24,25  For patients who had documented sexual or reproductive history, screening for more specific risk behaviors (such as condom use, birth control use, and number of sexual partners) was often omitted.²⁴ 

Similarly, in the ED, a retrospective study by Beckmann and Melzer-Lange²⁷  reported that even in charts of patients diagnosed with an STI, documentation of sexual activity was incomplete and inconsistent. The authors noted that although 94% of patients in the study were documented as sexually active, only 48% of charts documented condom use, only 38% of charts documented STI history, and only 19% of charts documented the number of partners. No charts contained documentation on other important risk-stratifying details, such as contraception use other than condoms, the sex of partners, partners’ risk of STIs, anal sex practice, or partners’ drug use.²⁷  None of these studies reported on whether privacy was ensured in sexual history taking, although they did mention the need for confidentiality as a possible barrier to higher rates of screening.^23–26 

McFadden et al²⁵  described sexual health services provided in the hospital setting and reported that STI testing was conducted in 12% of patients, that pregnancy testing was done in 60% of female patients, and that contraception was provided for 2% of patients. Forty-six percent of patients were due for the human papillomavirus (HPV) vaccine, and 19% of these received it during admission.²⁵ 

In the ED, researchers of a single-blind randomized controlled trial tested a computerized self-administered screening tool to identify adolescent patients who were at risk for STIs. In the intervention arm, the results of the screen provided decision support for ED physicians. Adolescents in the intervention group were more likely to receive STI testing compared with those in the control arm (52.3% vs 42%; odds ratio [OR] 2.0 [95% confidence interval (CI) 1.1–3.8]). These findings were more pronounced in adolescents without symptoms of STI (28.6% vs 8.2%; OR 4.7 [95% CI 1.4–15.5]).²⁸  In a study by Miller et al²⁹  done in the ED setting, MI was found to be a feasible, timely, and effective technique in promoting sexual health in adolescents.

In several of the included studies in the sexual activity domain, researchers looked at attitudes of adolescent patients, parents, and clinicians toward adolescents being screened in acute care settings. Many adolescents felt the ED should universally provide education on sexual and reproductive health practices and provide contraceptive services, especially for patients who may not have access to a primary provider.^25,30–32  Chernick et al³³  found that one-fourth of the adolescent patients in their study were interested in receiving contraception in the ED. In several studies, researchers found that computerized self-disclosure tools were preferred by adolescent patients, regardless of the presenting chief complaint.^34,35  Regarding counseling and interventions, adolescent patients generally valued clinician-patient interactions. For example, Shamash et al³⁶  found that the majority of adolescents did not support provision of expedited partner therapy and partner notification if an STI was identified, citing reasons such as the importance of interaction between the partner and his or her own clinician. The value of such interaction was echoed in another study in which patients preferred in-person counseling.³⁷  However, in a cross-sectional hospital study, Guss et al³⁸  found that patients who were interested in more information preferred learning about contraceptive options from a brochure rather than from a clinician.

Parents were overall supportive of sexual activity screening and care provision in the ED and hospital setting. In fact, in a study by Miller et al,³⁹  parents were more accepting of sexual activity screening and STI testing than surveyed clinicians.

In the hospital setting, the top 3 barriers to sexual activity screening among clinicians included concerns about follow-up (63%), lack of knowledge regarding contraception (59%), and time constraints (53%). The majority of respondents reported they would be more likely to increase delivery of sexual health services if provided with further education.⁴⁰  Clinicians expressed concerns about the acute nature of illness and injury in the ED and the sensitive nature of sexual activity screening. In several ED studies, authors cited concerns from clinicians that the ED was not the appropriate setting to address sexual activity, particularly if it was not related to the patient’s presenting problem.^39,41  Clinicians in the ED setting had a preference for computerized screening tools as well.⁴² 

Twelve studies on mood and SI screening and intervention were included in our review; 11 took place in the ED setting, and 1 took place in the urgent care setting (Table 4).

No studies were found.

In our review, we found several reports on various SI screening tools in acute care settings, including the Ask Suicide-Screening Questionnaire (ASQ), the Risk of Suicide Questionnaire (RSQ), and the Behavioral Health Screening–Emergency Department (BHS-ED); these studies indicate the potential promise of these tools and also reveal significant SI risk in adolescents presenting for nonpsychiatric issues. The ASQ has been widely referenced in literature as a brief and feasible tool to assess suicide risk in pediatric patients in the ED.⁴³  The ASQ 4-question screen has a sensitivity of 96.9%, a specificity of 87.6%, and a negative predictive value of 99.7%.⁴⁴  In their review, King et al⁴⁵  found that universal screening for mood and SI in the ED setting can identify a clinically significant number of patients who have active SI but are presenting for unrelated medical reasons. To help identify such patients, a cross-sectional study done to validate the RSQ in patients presenting to the ED revealed a clinically significant prevalence (5.7%) of SI in patients with nonpsychiatric chief complaints.⁴⁶  However, another validation study revealed that in a low-risk, nonsymptomatic patient population, the RSQ had high false-positive rates. The authors concluded that a more general psychosocial risk screen, such as the HEADSS, should be implemented instead.⁴⁷  Ambrose and Prager⁴⁸  described potential screening tools for SI (eg, ASQ and RSQ) and concluded that these tools need further prospective study and validation in a general population of adolescents without mental health complaints.

Fein et al⁴⁹  describe successful implementation of a more broad behavioral health screen: the BHS-ED, which is used to assess for mood and behavioral health issues as well as associated risks, such as substance use. Fein et al⁴⁹  found that with the BHS-ED, mental health problem identification increased from 2.5% to 4.2% (OR 1.70; 95% CI 1.38–2.10), with higher rates of social work or psychiatry evaluation in the ED (2.5% vs 1.7%; OR 1.47 [95% CI 1.13–1.90]).

A 2-question SI screen was piloted by Patel et al⁵⁰  in an urgent care setting to identify adolescents at risk for SI. Most adolescents who screened positive did not have mental health–related chief complaints, and positive screening results led to interventions in the form of referrals (82% of positive screen results) or urgent admission to an inpatient psychiatric facility (10% of positive screen results). Specifically, 5 of 10 patients who met criteria for inpatient psychiatric facility admission did not have an initial mental health–related chief complaint.⁵⁰ 

In a cross-sectional survey, O’Mara et al⁵¹  found that after a positive screen result, the majority of adolescent patients and their parents valued the chance for immediate intervention and resources in the ED. Similarly, in 2 qualitative studies by Ballard et al,^52,53  90% to 96% of interviewed adolescents responded positively to SI screening in the ED. Positive themes included detection of youth who may be at risk and have a lack of social support as well as possible prevention of suicide attempts. The biggest concerns from adolescent patients included worries about privacy issues.⁵¹ 

Parental reservations regarding screening were focused on the patient being in too much pain or distress for screening.⁴⁶  Other identified hesitations were fear of a lack of focus on nonpsychiatric chief complaints and possible iatrogenic harm secondary to screening.⁵³ 

Clinicians felt that a computerized depression screen would overcome many of the identified barriers (lack of rapport, time constraints, high patient acuity, lack of training or comfort, privacy concerns, and uncertainty with next steps), but they endorsed a need for support to facilitate connecting patients with mental health resources and interventions.⁵⁴ 

Seven studies on substance use screening and intervention were included in our review; all took place in the ED setting (Table 4).

No studies were found.

In a 2011 systematic review of substance use screening tools in the ED, the authors concluded that for alcohol screening of adolescent patients, the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) 2-item scale was best, with a sensitivity of 88% and a specificity of 90% (likelihood ratio of 8.8).⁵⁵  For marijuana screening, they recommended using the Diagnostic Interview Schedule for Children (DISC) Cannabis Symptoms, which is reported to have a sensitivity of 96% and a specificity of 86% (likelihood ratio of 6.83) and is composed of 1 question. More recently, researchers evaluated a self-administered 3-item screening tool based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, the Newton Screen, concluding that it was a brief and effective tool for screening both alcohol (sensitivity of 78.3% and specificity of 93%) and cannabis use (sensitivity of 93.1% and specificity of 93.5%).⁵⁶ 

In a study on the use of the Alcohol Use Disorder Identification Test (AUDIT) tool, researchers observed differences in sensitivity based on the age group of adolescents, noting lower utility in younger adolescents.⁵⁷  The National Institute of Alcohol Abuse and Alcoholism 2-question screen, a self-administered tool via tablet that features 2 different questions for middle school–aged versus high school–aged adolescents, was found to be a valid and brief way to screen for alcohol use in the ED.⁵⁸ 

For positive screen results, MI and brief intervention tools, such as the FRAMES acronym (feedback, responsibility, advice, menu, empathy, self-efficacy) have been found to be effective in addressing high-risk behaviors, particularly in adolescent patients. MI avoids confrontation, and the authors note that both of these evidence-based tools work with a patient’s readiness to change and build awareness of the problem, resulting in increased self-efficacy for the adolescent.⁵⁹ 

In a qualitative study, researchers assessed ED physician use of screening, brief intervention, and referral to treatment (SBIRT) and found that <50% of respondents used a validated tool when screening for alcohol use.⁶⁰  Common perceived barriers were time constraints, inadequate staffing, lack of knowledge of screens, and concerns about parents’ reactions to screening. Falcón et al⁶¹  found that, during implementation of a standardized screening program, it was important to minimize workflow disruption and provide adequate education to achieve participant buy-in.

There were no studies on patient or parent attitudes toward substance use screening or interventions.

Two studies on abuse and violence screening and intervention were included in our review; both took place in the ED setting (Table 4).

No studies were found.

In their study, Erickson et al⁶²  described screening and intervention regarding abuse or violence, specifically focusing on evaluating risk of intimate partner violence with an 8-item screening tool (the Conflict Tactics Survey). They found that the risk of intimate partner violence in female adolescents who presented to the ED was high (37%) and that 4 screening questions had 99% sensitivity.⁶² 

In a narrative review by Jackson et al⁶³  on adolescent relationship abuse screening and interventions in the ED, the authors described successful outpatient interventions that could be easily adapted for the ED setting. They described targeted computer modules as interventions for adolescents who screen positive or, alternatively, use of a universal education intervention, such as a wallet-sized informational card.

No studies were found.

The studies in our review reveal ubiquitously low rates of risk behavior screening in the ED and hospital setting across all risk behavior domains. Our study also highlights the general dearth of studies on the topic (only 7 studies in the hospital setting, only 2 studies with low risk of bias based on our analysis). We outline potential tools and approaches for improving adherence to guideline-recommended comprehensive screening and adolescent health outcomes.

Although comprehensive risk behavior screens (eg, the American Academy of Pediatrics Bright Futures⁶⁴  and HEADSS^3,65 ) remain the gold standard, they have not been validated in the ED or hospital setting. We report on a number of successful domain-specific screening tools validated in ED and hospital settings. The Sexual Health Screen reported on by Goyal et al³⁵  presents a feasible and valid way to screen for sexual and reproductive health. For mood and SI screening, validated tools include the ASQ and RSQ.^48,53  For substance use screening, potential tools include the Newton Screen, the National Institute of Alcohol Abuse and Alcoholism 2-question screen, and SBIRT.^56,58,66  For intimate partner violence screening, Erickson et al⁶²  validated the 8-item Conflict Tactics Survey. These brief validated tools within single risk behavior domains could potentially be combined into a single comprehensive screen (with consideration that these screening tools may have been validated for specific populations and plans to assess feasibility and time burdens).

When patients screen positive for risky behaviors, it is imperative to have strategies and resources in place to address these behaviors. MI has been demonstrated to be feasible, effective, and a preferred method to change risky behavior across all risk behavior domains in ED and hospital settings.^29,59,67  Specifically, the FRAMES acronym provides a promising framework for MI for adolescent substance use but can be applied to any high-risk behavior change.⁵⁹  However, some adolescents may instead prefer paper materials or brochures over face-to-face counseling, so this presents an alternative option.³⁸  As demonstrated in the McFadden et al²⁵  study, other interventions to consider implementing in the ED and hospital settings include STI testing and treatment, contraceptive provision, HPV vaccination, and referral to subspecialty resources (both inpatient and outpatient). For intimate partner violence and adolescent relationship abuse, Jackson et al⁶³  outline successful outpatient interventions (eg, universal wallet-sized educational cards and targeted computerized interventions) that could be feasible in the ED setting but would require further investigation.

We found that although clinicians and patients are receptive to risk behavior screening and interventions in these settings, they also report several barriers.⁵⁴  Clinicians are concerned that parents may object to screening; however, parents favor screening and intervention as long as their child is not in too much pain or distress.⁴⁶  Clinicians additionally identify obstacles such as time constraints, lack of education or knowledge on the topic, and concerns about adolescent patients’ reactions.^40,60,61  Additionally, adolescent patients report concerns around privacy and confidentiality of disclosed information.⁵¹ 

To overcome these collective barriers, future researchers should investigate (1) feasible, efficient risk behavior screening tools with guidance for clinicians on providing risk behavior interventions and (2) tools that increase privacy and comfort for patients (likely through the use of electronic formats). Promising methods to increase screening rates include self-disclosure electronic screening tools coupled with reminders for clinicians (paper or within the EHR). Self-disclosure screening tools have been shown to increase privacy and disclosure of sensitive information by adolescent patients when compared with face-to-face screening by a clinician.⁶⁸  The use of technology and creation of electronic self-disclosure screens may further provide means to maintain comfort and patient privacy while streamlining workflow and maximizing efficiency for clinicians, particularly when a reminder to screen is integrated.^21,22  Special consideration should be given to the interplay between documentation of sensitive information in the EHR and the privacy and confidentiality crucial in screening for adolescent risk behaviors.⁶⁹  One strategy to mitigate possible breaches of confidentiality with EHR documentation is to mark risk behavior screening notes as sensitive or confidential, thus preventing parents or guardians from access to the note (an option that is available on most EHR software). Another option is creating labeling functions within the EHR for children aged 13 to 18 so clinicians can label whether each problem, medication, or diagnostic test result can be accessed by the patient, parents, or both.⁶⁹  In a recently published scoping review, Wong et al⁷⁰  further explore possible systemic solutions in designing digital health technology that captures and delivers preventive services to adolescents while maximizing safety and privacy.

A limitation of this scoping review is heterogeneity in the design and quality of the included studies, with only 1 randomized controlled trial in our area of focus. Additionally, most studies of screens or interventions have thus far been limited to a single study done in 1 center, thus limiting generalizability. With the heterogeneity of studies included, we could only summarize findings but could not perform a meta-analysis. Also, most studies had limited durations of follow-up, so we cannot comment on long-term effects. We excluded studies that involved outpatient follow-up of patients to evaluate interventions that could be completed in the ED or hospital setting, but this may have limited our review of more longitudinal effects.

ED and hospital encounters present a missed opportunity for increasing risk behavior screening and care provision for adolescent patients; current rates of screening and intervention are low. Patients and clinicians are generally receptive to screening in these settings, with barriers including adolescents’ privacy concerns, clinicians’ time constraints, and clinicians’ comfort and knowledge with risk behavior screening and risk behavior interventions. Promising solutions include self-disclosure via electronic screening tools, educational sessions for clinicians, and clinician reminders to complete screening. More prospective controlled studies are needed to evaluate such interventions in ED and hospital settings.

We acknowledge Evans Whitaker, MD, MLIS, for his assistance with the literature search.

 
• ASQ

  Ask Suicide-Screening Questionnaire

 
• AUDIT

  Alcohol Use Disorder Identification Test

 
• BHS-ED

  Behavioral Health Screening–Emergency Department

 
• CI

  confidence interval

 
• DISC

  Diagnostic Interview Schedule for Children

 
• DSM-IV

  Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition

 
• ED

  emergency department

 
• EHR

  electronic health record

 
• FRAMES

  feedback, responsibility, advice, menu, empathy, self-efficacy

 
• HEADSS

  home, education, activities, drugs, sexual activity, suicide and/or mood

 
• HPA

  health promotion advocate

 
• HPV

  human papillomavirus

 
• MI

  motivational interviewing

 
• OR

  odds ratio

 
• RSQ

  Risk of Suicide Questionnaire

 
• SBIRT

  screening, brief intervention, and referral to treatment

 
• SI

  suicidal ideation

 
• STI

  sexually transmitted infection