Quality of Research
Documents Reviewed
The documents below were reviewed for Quality of Research. The research point of
contact can provide information regarding the studies reviewed and the availability
of additional materials, including those from more recent studies that may have been conducted.
Outcomes
| Outcome 1: Alcohol use |
|
Description of Measures
|
Alcohol use was assessed for the prior 1-month period using four measures from a self-administered, online version of the Brief Drinker Profile (BDP): (1) drinks per week, (2) estimated peak blood alcohol concentration (BAC) in a typical week, (3) average number of drinks during two heavy drinking episodes, and (4) average estimated peak BAC in the two heavy drinking episodes. The BDP, derived from the Comprehensive Drinker Profile, is a 50-minute structured interview that measures the quantity and frequency of current drinking and the severity of risk factors across eight life domains: demographics, family and employment status, history of problem development, alcohol-related problems, severity of dependence, other drug use, additional life problems, and motivation for treatment.
|
|
Key Findings
|
In a 12-month clinical trial, heavy-drinking college students (ages 18-24) from a 4-year public university and a community college were randomly assigned to the intervention group, which received CDCU, or an assessment-only control group. Heavy drinking was defined according to the National Institute on Alcohol Abuse and Alcoholism (NIAAA) criteria for heavy, episodic drinking (i.e., having four or more drinks per occasion for women and five or more drinks per occasion for men at least once in the past 2 weeks with an estimated peak BAC of 0.08 gram-percent or above). Assessments occurred at baseline and at 1 and 12 months after baseline (follow-ups). Findings included the following:
- At the 1-month follow-up, participants in the intervention group had a lower peak BAC during a typical week (p = .017), a lower average number of drinks during two heavy drinking episodes (p = .017), and a lower average BAC during the two heavy drinking episodes (p = .010) compared with participants in the control group, after controlling for baseline assessments. These group differences were associated with small effect sizes (Cohen's d = 0.41, 0.41, and 0.44, respectively).
- At the 12-month follow-up, participants in the intervention group had fewer drinks per week (p = .044), a lower average number of drinks during two heavy drinking episodes (p = .021), and a lower average estimated BAC in the two heavy drinking episodes (p = .024) compared with participants in the control group, after controlling for baseline assessments. These group differences were associated with small effect sizes (Cohen's d = 0.36, 0.41, and 0.40, respectively).
In a 1-month clinical trial, heavy-drinking college students (ages 18-24) from a 4-year public university and a community college were randomly assigned to the intervention group, which received CDCU, or a delayed-assessment control group. Heavy drinking was defined according to the NIAAA criteria for heavy, episodic drinking. Assessments occurred at baseline and at 1 month after baseline (follow-up) for participants in the intervention group and at 1-month follow-up for participants in the delayed-assessment control group, who completed a retrospective baseline assessment. At the 1-month follow-up, participants in the intervention group had fewer drinks per week (p = .008), a lower estimated peak BAC in a typical week (p = .001), a lower average number of drinks during two heavy drinking episodes (p = .001), and a lower average estimated peak BAC in the two heavy drinking episodes (p = .001) compared with participants in the control group, after controlling for baseline assessments. These group differences were associated with medium and large effect sizes (Cohen's d = 0.60, 0.80, 0.91, and 0.97, respectively).
|
|
Studies Measuring Outcome
|
Study 1, Study 2
|
|
Study Designs
|
Experimental
|
|
Quality of Research Rating
|
3.1
(0.0-4.0 scale)
|
Study Populations
The following populations were identified in the studies reviewed for Quality of
Research.
|
Study
|
Age
|
Gender
|
Race/Ethnicity
|
|
Study 1
|
18-25 (Young adult)
|
62.5% Male
37.5% Female
|
55.6% White
28.5% Hispanic or Latino
6.9% Race/ethnicity unspecified
5.6% Black or African American
2.1% American Indian or Alaska Native
0.7% Asian
0.7% Native Hawaiian or other Pacific Islander
|
|
Study 2
|
18-25 (Young adult)
|
56.1% Male
43.9% Female
|
46.3% White
37.8% Hispanic or Latino
9.8% Race/ethnicity unspecified
2.4% American Indian or Alaska Native
2.4% Black or African American
1.2% Asian
|
Quality of Research Ratings by Criteria (0.0-4.0 scale)
External reviewers independently evaluate the Quality of Research for an intervention's
reported results using six criteria:
For more information about these criteria and the meaning of the ratings, see Quality of Research.
|
Outcome
|
Reliability
of Measures
|
Validity
of Measures
|
Fidelity
|
Missing
Data/Attrition
|
Confounding
Variables
|
Data
Analysis
|
Overall
Rating
|
|
1: Alcohol use
|
2.8
|
2.9
|
3.3
|
3.0
|
2.9
|
3.5
|
3.1
|
Study Strengths The Brief Drinker Profile (BDP), a well-known assessment instrument, is supported by acceptable reliability and validity in the field, and its adaptation from an interview to an online self-report for use by a college student population was supported by sample test-retest reliability estimates from the second study. Breathalyzer tests and a collateral informant option improved reliability and validity of the assessments in both studies. Intervention fidelity was strengthened by a stand-alone computer delivery system, feedback norms that were appropriately tailored for college students, and the placement of research assistants in the computer room to answer questions while participants accessed the intervention. Attrition was low in both studies (3% and 10% at the 1- and 12-month follow-ups, respectively, in one study and 2% at the 1-month follow-up in another study). The use of a stratified, randomized study and the implementation of a delayed-assessment control (using a retrospective pretest methodology) in one study minimized potential confounding variables. The analysis of covariance modeling of the data in both studies was appropriate and used an adequate sample size on the basis of a prospective power analysis, resulting in excellent control of type I and type II error rates. The use of Bonferroni corrections, the 95% confidence interval, and effect sizes for condition contrasts in addition to the traditional alpha tests of significance added to the overall strength of data modeling and control of type I and type II errors.
Study Weaknesses Reliability of the peak BAC estimates may have been compromised because participants estimated their weight and amount of time spent drinking, a potential reliability issue also noted in the BDP manual. Psychometrics for the self-report drinking measures were provided by the BDP manual; however, criterion validity correlations between self-reports and collateral reports for estimated peak BAC were substantially lower for steady and per episode drinking (0.39 and 0.36, respectively) than for drinks per week (0.76). The convergent validity provided for the online self-report form of the BDP with the well-established Form 90 was based on an adult community sample, and it is unclear whether generalization to a college student population is justified. There was no direct measurement of implementation fidelity during interactions of participants and research assistants (i.e., when research assistants conducted in-person screenings of participants, when participants accessed the intervention while research assistants were in the computer room). Although attrition at the 12-month follow-up in one study was low (10%), no data comparing noncompleters and completers were reported. As the investigators noted, general maturation, regression to the mean, and a Hawthorne effect on the outcomes cannot be ruled out in both studies. An intent-to-treat approach was not used, and participants with incomplete assessment data were excluded from the statistical analyses in both studies.
|
|