Methodology and Demographics of a Brief Adolescent Alcohol Screen Validation Study

Can accurate demographic information about people who use prescription medications nonmedically be derived from Twitter?

Traditional substance use (SU) surveillance methods, such as surveys, incur substantial lags. Due to the continuously evolving trends in SU, insights obtained via such methods are often outdated. Social media-based sources have been proposed for obtaining timely insights, but methods leveraging such data cannot typically provide fine-grained statistics about subpopulations, unlike traditional approaches. We address this gap by developing methods for automatically characterizing a large Twitter nonmedical prescription medication use (NPMU) cohort (n = 288,562) in terms of age-group, race, and gender. Our natural language processing and machine learning methods for automated cohort characterization achieved 0.88 precision (95% CI:0.84 to 0.92) for age-group, 0.90 (95% CI: 0.85 to 0.95) for race, and 94% accuracy (95% CI: 92 to 97) for gender, when evaluated against manually annotated gold-standard data. We compared automatically derived statistics for NPMU of tranquilizers, stimulants, and opioids from Twitter with statistics reported in the National Survey on Drug Use and Health (NSDUH) and the National Emergency Department Sample (NEDS). Distributions automatically estimated from Twitter were mostly consistent with the NSDUH [Spearman r: race: 0.98 (P < 0.005); age-group: 0.67 (P < 0.005); gender: 0.66 (P = 0.27)] and NEDS, with 34/65 (52.3%) of the Twitter-based estimates lying within 95% CIs of estimates from the traditional sources. Explainable differences (e.g., overrepresentation of younger people) were found for age-group-related statistics. Our study demonstrates that accurate subpopulation-specific estimates about SU, particularly NPMU, may be automatically derived from Twitter to obtain earlier insights about targeted subpopulations compared to traditional surveillance approaches.

Association of Physical Activity, Sports, and Screen Time With Adolescent Behaviors in Youth Who Visit the Pediatric Emergency Department

Moderate to vigorous physical activity (MVPA), sports, and reduced screen time are associated with favorable youth risk profiles. We evaluated the association of MVPA, sports, and screen time with adolescent behaviors among pediatric emergency department youth. Adolescents were assessed for alcohol/drug use, risky behavior, conduct disorder, and depressive mood. MVPA was activity for ≥5 days/week and ≥60 minutes/day. Increased screen time was ≥3 hours/day computer/TV use for non-schoolwork. Multivariable regression studied association between MVPA, sports, and increased screen time and outcomes adjusting for demographics and academic achievement. Older age and lower academic achievement were significantly associated with risky behaviors, conduct disorder, and depression. Youth who endorsed MVPA and sports participation had less depression (odds ratio [OR] = 0.76; confidence interval [CI] = 0.66-0.87). Increased screen time was associated with conduct disorder (OR = 1.6; CI = 1.3-2.1), depression (OR = 1.2; CI = 1.0-1.4), and drug use (OR = 1.8; CI = 1.1-2.8). In pediatric emergency department youth, MVPA and sports participation is associated with less depression. Increased screen time is associated with conduct disorders, depression, and drug use.

Adolescent alcohol use predicts cannabis use over a three year follow-up period

Background: Alcohol and cannabis use frequently co-occur, which can result in problems from social and academic impairment to dependence (i.e., alcohol use disorder [AUD] and/or cannabis use disorder [CUD]). The Emergency Department (ED) is an excellent site to identify adolescents with alcohol misuse, conduct a brief intervention, and refer to treatment; however, given time constraints, alcohol use may be the only substance assessed due to its common role in unintentional injury. The current study, a secondary data analysis, assessed the relationship between adolescent alcohol and cannabis use by examining the National Institute of Alcohol Abuse and Alcoholism (NIAAA) two question screen's (2QS) ability to predict future CUD at one, two, and three years post-ED visit. Methods: At baseline, data was collected via tablet self-report surveys from medically and behaviorally stable adolescents 12-17 years old (n = 1,689) treated in 16 pediatric EDs for non-life-threatening injury, illness, or mental health condition. Follow-up surveys were completed via telephone or web-based survey. Logistic regression compared CUD diagnosis odds at one, two, or three-year follow-up between levels constituting a single-level change in baseline risk categorization on the NIAAA 2QS (nondrinker versus low-risk, low- versus moderate-risk, moderate- versus high-risk). Receiver operating characteristic curve methods examined the predictive ability of the baseline NIAAA 2QS cut points for CUD at one, two, or three-year follow-up. Results: Adolescents with low alcohol risk had significantly higher rates of CUD versus nondrinkers (OR range: 1.94-2.76, p < .0001). For low and moderate alcohol risk, there was no difference in CUD rates (OR range: 1.00-1.08). CUD rates were higher in adolescents with high alcohol risk versus moderate risk (OR range: 2.39-4.81, p < .05). Conclusions: Even low levels of baseline alcohol use are associated with risk for a later CUD. The NIAAA 2QS is an appropriate assessment measure to gauge risk for future cannabis use.

Using REDCap and Apple ResearchKit to integrate patient questionnaires and clinical decision support into the electronic health record to improve sexually transmitted infection testing in the emergency department

OBJECTIVE

Audio-enhanced computer-assisted self-interviews (ACASIs) are useful adjuncts for clinical care but are rarely integrated into the electronic health record (EHR). We created a flexible framework for integrating an ACASIs with clinical decision support (CDS) into the EHR. We used this program to identify adolescents at risk for sexually transmitted infections (STIs) in the emergency department (ED). We provide an overview of the software platform and qualitative user acceptance.

MATERIALS AND METHODS

We created an ACASI with a CDS algorithm to identify adolescents in need of STI testing. We offered it to 15- to 21-year-old patients in our ED, regardless of ED complaint. We collected user feedback via the ACASI. These were programmed into REDCap (Research Electronic Data Capture), and an iOS application utilizing Apple ResearchKit generated a tablet compatible representation of the ACASI for patients. A custom software program created an HL7 (Health Level Seven) message containing a summary of responses, CDS recommendations, and STI test orders, which were transmitted to the EHR.

RESULTS

In the first year, 1788 of 6227 (28.7%) eligible adolescents completed the survey. Technical issues led to decreased use for several months. Patients rated the system favorably, with 1583 of 1787 (88.9%) indicating that they were "somewhat" or "very comfortable" answering questions electronically and 1291 of 1787 (72.2%) preferring this format over face-to-face interviews or paper questionnaires.

CONCLUSIONS

We present a novel use for REDCap to combine patient-answered questionnaires and CDS to improve care for adolescents at risk for STIs. Our program was well received and the platform can be used across disparate patients, topics, and information technology infrastructures.

Sex Without Contraceptives in a Multicenter Study of Adolescent Emergency Department Patients

OBJECTIVES

In the United States, rates of teenage pregnancy and sexually transmitted infections (STIs) remain exceptionally high, and racial and ethnic disparities persist. Emergency departments (EDs) care for over 19 million adolescents each year, the majority being minority and low socioeconomic status. Single-center studies demonstrate infrequent use of contraceptives among adolescent ED patients and an association between risky sex and behaviors such as alcohol and drug use; however, no multicenter ED data exist. The objectives of this study were to 1) determine the prevalence of sex without contraceptives in a large multicenter adolescent ED study and 2) assess patient demographic and risky behaviors associated with sex without contraceptives.

METHODS

Participants aged 14 to 17 years (n = 3,247) in 16 pediatric EDs across the United States completed an electronic survey. Questions focused on validated measures of risky sex; use of alcohol, tobacco, marijuana, and other drugs; and depression and violence. In this secondary analysis, we constructed univariable and multivariable models to identify demographic and behavioral factors associated with sex without contraceptives (our primary outcome), separately for adolescent males and females.

RESULTS

In the prior year, 17.4% (236/1,356) of males and 15.8% (299/1,891) of females had sex without contraceptives. In the multivariable model, sex without contraceptives for both genders was more likely among teens who were black, with conduct problems and participated in casual sex, binge drinking, or cannabis use. Sex without contraceptives was also more likely among Hispanic and cigarette-smoking males, as well as depressed females.

CONCLUSIONS

Adolescent ED patients across the United States are participating in risky sexual behaviors that increase their likelihood of pregnancy and STI acquisition. These adolescents report a number of problem behaviors, including substance use, which are strongly correlated with unprotected sex. The ED visit may be an opportunity to identify at-risk adolescent patients, address risky behaviors, and intervene to improve adolescent health.

Predictive Validity of the CRAFFT for Substance Use Disorder

BACKGROUND

The utility of CRAFFT (Car, Relax, Alone, Forget, Friends, Trouble) in identifying current and future problematic substance use and substance use disorders (SUDs) in pediatric emergency department (PED) patients is unknown. We conducted a secondary analysis of a study in 16 PEDs to determine the concurrent and predictive validity of CRAFFT with respect to SUD.

METHODS

At baseline, 4753 participants aged 12 to 17 years completed an assessment battery (CRAFFT and other measures of alcohol, drug use, and risk behaviors). A subsample was readministered the battery at 1-, 2-, and 3-year follow-up to investigate future SUDs.

RESULTS

Of 2175 participants assigned to follow-up, 1493 (68.6%) completed 1-year, 1451 (66.7%) completed 2-year, and 1265 (58.1%) completed the 3-year follow-up. A baseline CRAFFT value of ≥2 was significantly associated with problematic substance use or mild or moderate to severe SUD diagnosis on the Diagnostic Interview Schedule for Children at baseline (P < .001). The results persisted after 1, 2, and 3 years (P < .001). The best combined sensitivity and specificity was achieved with a baseline CRAFFT value of ≥1 as a cutoff for predicting problematic substance use and a Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition diagnosis of mild SUD at 1, 2, and 3 years. The baseline CRAFFT score that best predicted a moderate to severe SUD at 1 year was ≥2; but at 2 and 3 years, the cutoff score was ≥1.

CONCLUSIONS

CRAFFT has good concurrent validity for problematic substance use and SUD in PED patients and is useful in predicting SUDs at up to 3 years follow-up but with limited sensitivity.

Reliability and Validity of the Newton Screen for Alcohol and Cannabis Misuse in a Pediatric Emergency Department Sample

OBJECTIVES

To determine the test-retest reliability, concurrent, convergent, and discriminant validity of a recently devised screen (the Newton screen) for alcohol and cannabis use/misuse, and its predictive validity at follow-up.

STUDY DESIGN

Adolescents, 12-17 years old (n = 4898), treated in 1 of 16 participating pediatric emergency departments across the US were enrolled in a study as part of a larger study within the Pediatric Emergency Care Applied Research Network. Concurrent and predictive validity (at 1, 2, and 3 years of follow-up) were assessed in a random subsample with a structured Diagnostic and Statistical Manual of Mental Disorders-based interview. Convergent validity was assessed with the Alcohol Use Disorders Identification, a widely used alcohol screening measure.

RESULTS

The sensitivity of the Newton screen for alcohol use disorder at baseline was 78.3% with a specificity of 93.0%. The cannabis use question had a baseline sensitivity of 93.1% and specificity of 93.5% for cannabis use disorder. Predictive validity analyses at 1, 2, and 3 years revealed high specificity but low sensitivity for alcohol and high specificity and moderate sensitivity for cannabis.

CONCLUSIONS

The Newton screening instrument may be an appropriate brief screening tool for use in the busy clinical environment. Specificity was high for both alcohol and cannabis, but sensitivity was higher for cannabis than alcohol. Like other brief screens, more detailed follow-up questions may be necessary to definitively assess substance misuse risk and the need for referral to treatment.

Predictive Validity of a 2-Question Alcohol Screen at 1-, 2-, and 3-Year Follow-up

BACKGROUND

The National Institute on Alcohol Abuse and Alcoholism (NIAAA) 2-question screen is a valid adolescent alcohol screening tool. No studies have examined if this tool predicts future alcohol problems. We conducted a study at 16 pediatric emergency departments to determine the tool's predictive validity for alcohol misuse and alcohol use disorders (AUDs).

METHODS

Participants (N = 4834) completed a baseline assessment battery. A subsample of participants completed the battery at 1, 2, and 3 years follow up.

RESULTS

Of the 2209 participants assigned to follow-up, 1611 (73%) completed a 1-year follow-up, 1591 (72%) completed a 2-year follow-up, and 1377 (62%) completed a 3-year follow-up. The differences in AUDs between baseline NIAAA screen nondrinkers and lower-risk drinkers were statistically significant at 1 year (P = .0002), 2 years (P <.0001), and 3 years (P = .0005), as were the differences between moderate- and highest-risk drinkers at 1 and 2 years (P < .0001 and P = .0088, respectively) but not at 3 years (P = .0758). The best combined score for sensitivity (86.2% at 1 year, 75.6% at 2 years, and 60.0% at 3 years) and specificity (78.1% at 1 year, 79.2% at 2 years, and 80.0% at 3 years) was achieved by using "lower risk" and higher as a cutoff for the prediction of a Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition diagnosis.

CONCLUSIONS

The NIAAA 2-question screen can accurately characterize adolescent risk for future AUDs. Future studies are needed to determine optimaluse of the screen.

Screening for Adolescent Alcohol Use in the Emergency Department: What Does It Tell Us About Cannabis, Tobacco, and Other Drug Use?

BACKGROUND

The pediatric emergency department (PED) represents an opportune time for alcohol and drug screening. The National Institute of Alcohol Abuse and Alcoholism (NIAAA) recommends a two-question alcohol screen for adolescents as a predictor of alcohol and drug misuse.

OBJECTIVE

A multi-site PED study was conducted to determine the association between the NIAAA two-question alcohol screen and adolescent cannabis use disorders (CUD), cigarette smoking, and lifetime use of other drugs.

METHODS

Participants included 12-17-year olds (n = 4834) treated in one of 16 participating PEDs. An assessment battery, including the NIAAA two-question screen and other measures of alcohol, tobacco and drug use, was self-administered on a tablet computer.

RESULTS

A diagnosis of CUD, lifetime tobacco use or lifetime drug use was predicted by any self-reported alcohol use in the past year, which indicates a classification of moderate risk for middle school ages and low risk for high school ages on the NIAAA two-question screen. Drinking was most strongly predictive of a CUD, somewhat weaker for lifetime tobacco use, and weakest for lifetime drug use. This same pattern held for high school and middle school students and was stronger for high school students over middle school students for all three categories. This association was also found across gender, ethnicity and race. The association was strongest for CUD for high school students, sensitivity 81.7% (95% CI, 77.0, 86.5) and specificity 70.4% (95% CI, 68.6, 72.1). Conclusions/Importance: A single question about past year alcohol use can provide valuable information about other substance use, particularly marijuana.

Problem Behaviors and Psychological Distress Among Teens Seen in a National Sample of Emergency Departments

BACKGROUND

Problem behaviors, such as substance use and peer aggression, frequently coexist and are common among youth seen in emergency departments (EDs). EDs are increasingly urged to screen for both psychological distress and problem behaviors.

OBJECTIVE

To inform screening and intervention efforts, we aimed to identify classes of problematic substance use and peer aggression in a sample of adolescents from 16 pediatric EDs, and to examine the relative prevalence of psychological distress in identified classes.

METHODS

We completed a cross-sectional survey of youth (n = 5001) presenting for any reason to 16 pediatric EDs across the United States, with the use of validated measures of demographics, alcohol and substance use, and peer aggression. We used standard latent class analysis techniques to create behavioral risk classes of adolescents based on violence and substance use variables; then we conducted logistic regression to examine the relationship between psychological distress and the latent classes.

RESULTS

Three classes of problem behaviors were identified: low-risk (few problem behaviors, 91.2% of sample), medium risk (high cigarette smoking; moderate violence, alcohol/substance use; 5.2%), and high risk (high levels of all problem behaviors, 3.5%). A significant directional association (P < .001) between worse psychological distress and higher-risk behavior classes was noted, even after adjusting for demographics.

CONCLUSIONS

Youth seen in the ED for any reason who report higher levels of past-year substance use and peer aggression are significantly more likely to report negative mood symptoms. Targeted screening and interventions for this population may be indicated.