Sources of Error in Substance Use Prevalence Surveys

Johnson, Timothy P.

doi:10.1155/2014/923290

Cited by 155 publications

(138 citation statements)

References 230 publications

(229 reference statements)

Supporting

Mentioning

130

Contrasting

Unclassified

Order By: Relevance

“…Our final models, which include statistical adjustment for demographic and socioeconomic characteristics as well as substance use prior to age 18, found no associations between the policy and substance use; the estimated odds ratios were tightly clustered around 1 and none reached statistical significance. These findings confirm and extend those of a prior study using data from the BRFSS, by examining additional outcomes and using more robust data on substance use outcomes; in contrast to the BRFSS, the NSDUH uses in-home computer assisted interview methods rather than telephone interviews resulting in higher and likely more accurate prevalence estimates (Johnson, 2014; Miller et al, 2004). Contrary to the theory of ex ante moral hazard, risky substance use does not appear to be responsive to health insurance coverage in this age group.…”

Section: Discussionsupporting

confidence: 80%

Did the dependent coverage expansion increase risky substance use among young adults?

Breslau

Han

et al. 2017

Drug and Alcohol Dependence

View full text Add to dashboard Cite

Background The dependent coverage expansion (DCE) enacted through the Affordable Care Act increased health insurance coverage among young adults. Increasing insurance coverage in this age group has the potential for unintended consequences on risky substance use. Methods Repeated cross-sectional surveys were used to compare change in substance use during the period the DCE was implemented in the 19–25 year old target age group (Pre-DCE n=15,772, Post-DCE n=22,719) with contemporaneous change in a slightly older age group that was not targeted by the policy (Pre-DCE=19,851, Post-DCE n=28,157). Outcomes include 11 measures of alcohol, illicit drug and cigarette use. Statistical controls were included for demographic and socioeconomic factors and for early initiation of substance use to adjust for historical trends in developmental trajectories. Results Risky substance use decreased in young adults relative to the older age group over the period that the DCE was implemented. However, statistical adjustment for initiation of substance use prior to age 18, which is prior to exposure to the DCE, accounted for the differences between the age groups. In adjusted models, associations between the DCE and substance use outcomes range from 0.96 to 1.08 with p-values ranging from .330 to .963. Conclusions Historical trends in initiation of substance use prior to age 18, not the DCE, account for change in risky substance use among 19 to 25 year olds relative to 26 to 34 year olds. The evidence does not support the suggestion that health insurance coverage would increase risky substance use among young adults.

show abstract

Section: Discussionsupporting

confidence: 80%

Did the dependent coverage expansion increase risky substance use among young adults?

Breslau

Han

et al. 2017

Drug and Alcohol Dependence

View full text Add to dashboard Cite

show abstract

“…Because of how survey items were nested (e.g., someone who reported use in the past 30 days would have also reported use in the past 6 months), sensitivities across recall periods for a substance should have been equal. Instead, the monotonic increase with longer recall period, as observed in other studies [5, 46, 47], suggests participants may be more willing to disclose substance use if there is perceived distance from the behavior, possibly due to impression management [2, 3, 48]. Alternatively, there may be temporal recall errors, such as a telescoping effect [48].…”

Section: Discussionmentioning

confidence: 84%

“…Framed differently, self-reported substance use data appeared as valid at baseline as they were after multiple waves of follow-up and multiple years of research participation, respectively, though a marginal finding toward greater disclosure of non-marijuana drug use among those recruited in 2015 may hint at possible generational differences. Survey methodologists have long been concerned about threats to construct validity due to the research process itself, such as altered behavior in response to self-evaluation, reduced social desirability bias due to familiarity with the researchers and no observed consequences from reporting illegal or stigmatized behaviors, and practice effects [2, 3, 48, 60]. Our study, however, demonstrated that impression management, the main component of social desirability bias, was sustained despite participants’ becoming more comfortable with study procedures and staff.…”

Section: Discussionmentioning

confidence: 99%

“…However, most research continues to rely on self-reported substance use behaviors. Sources of bias in such data are extensive and encompass many issues of measurement error [2]. Certain cognitive (e.g., social desirability) and situational factors (e.g., perceptions of confidentiality within the interview setting) have been linked to underreporting behaviors deemed unfavorable [3, 4], including substance use [5].…”

Section: Introductionmentioning

confidence: 99%

“…Consistent across studies is lower reliability among racial/ethnic minorities [1, 2, 8–10]. Concerns about confidentiality, cultural distrust due to histories of mistreatment by researchers, and fear of legal consequences have been posited as possible explanations [8, 9].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Predictors of correspondence between self-reported substance use and urinalysis screening among a racially diverse cohort of young men who have sex with men and transgender women

Janulis

Mustanski

2019

Addictive Behaviors

View full text Add to dashboard Cite

It is unknown if estimates of illicit drug use among young men who have sex with men and transgender women (YMSM/TW) may be biased due to historical distrust of research or reliable due to more accepting norms for use. Research is needed to examine the validity of drug use self-reports among YMSM/TW. Data came from an ongoing longitudinal study of YMSM/TW aged 16-29 living in Chicago (analytic N = 1029). Baseline urinalysis screens for marijuana, ecstasy, amphetamine, methamphetamine, cocaine, benzodiazepine, and opiate metabolites were compared to self-reported use within different recall periods using measures of concordance. Generalized estimating equations logistic regressions were conducted on three waves of data to identify predictors of disclosing past-6-month use of marijuana and non-marijuana drugs. Past-6-month self-reported use of all non-marijuana substances was <15%. There was excellent agreement between self-reported and drug-tested marijuana use. For other substances, sensitivities within the urinalysis detection window were <0.5 but increased with longer recall periods. Black participants had lower odds of disclosing non-marijuana drug use. Gender minority participants had lower odds of disclosing marijuana use. Participants with a history of arrest had higher odds of disclosing both marijuana and non-marijuana drug use. Wave and year of first research participation were non-significant, suggesting no systematic bias or increasing honesty associated with longer research participation. Programs that rely on self-identification of non-marijuana illicit substance use may be missing a substantial portion of drug-using YMSM/TW. Future epidemiological studies should work to reduce social desirability biases and include biomarker-based drug screenings to increase validity.

show abstract

Association of Constellations of Parental Risk With Children’s Subsequent Anxiety and Depression

et al. 2019

View full text Add to dashboard Cite

IMPORTANCEThe research focus on children of parents with alcohol use disorder has eclipsed the potentially wider-reaching detrimental effects of subclinical parental drinking, both alone and in combination with other parental risk factors.OBJECTIVE To identify constellations of early parental risk characterized by variations in drinking, mental health, and education in both parents and examine their prospective associations with children's contact with the health care system for anxiety and/or depression (ie, diagnoses or treatment). DESIGN, SETTING, AND PARTICIPANTS This prospective study was based on linked survey and health registries data. The sample included 8773 children from 6696 two-parent families in Norway who participated in the Nord-Trøndelag Health Study (HUNT) survey in 1995 to 1997 or 2006 to 2008, when the children were aged 13 to 19 years. Data were analyzed from January to September 2018.EXPOSURES Five constellations of early parental risks, characterized by variations in drinking frequencies and amounts, mental health, and education for both parents, as identified through latent profile analysis. MAIN OUTCOMES AND MEASURESChildren's diagnoses or treatment of anxiety and/or depression from 2008 to 2016 were recorded in 3 health registries. The primary outcome was the total number of registries where participants presented (ranging from 0 to 3). RESULTSOf the 8773 included children, 4404 (50.2%) were boys, and the mean (SD) age at the time of participation in the Nord-Trøndelag Health Study was 16.1 (1.8) years. Prevalence of anxiety and/or depression, as evidenced in at least 1 registry record, was 24.3% (2132 of 8773). Early parental risk profiles risks marked by (1) the lowest parental education (adjusted relative risk, 1.13; 95% CI, 1.01-1.25) and (2) elevated drinking in both parents and elevated mental health symptoms in fathers (adjusted relative risk, 1.52; 95% CI, 1.03-2.22) were associated with a significant increase in risk of anxiety and/or depression in children from those families compared with children from no-risk families.CONCLUSIONS AND RELEVANCE Studies seeking to understand prospective associations of parental drinking with children's mental health need to consider additional risk factors in combination with one another as well as parental behaviors and characteristics below clinically defined levels. When accumulated at a family level, even seemingly innocuous characteristics contributed to meaningful increases in risk of anxiety and/or depression among children, potentially translating into poorer mental health outcomes for many young people.

show abstract

Sources of Error in Substance Use Prevalence Surveys

Cited by 155 publications

References 230 publications

Did the dependent coverage expansion increase risky substance use among young adults?

Did the dependent coverage expansion increase risky substance use among young adults?

Predictors of correspondence between self-reported substance use and urinalysis screening among a racially diverse cohort of young men who have sex with men and transgender women

Association of Constellations of Parental Risk With Children’s Subsequent Anxiety and Depression

Contact Info

Product

Resources

About