Summary of the Evidence: Behavioral Counseling Interventions in Primary Care to Reduce Risky/Harmful Alcohol Use by Adults. Appendix Table 2

You Are Here: AHRQ Home > Clinical Information > U.S. Preventive Services Task Force (USPSTF) > Guide to Clinical Preventive Services, 2001-2004 > Screening: Alcohol Misuse > Summary of the Evidence

Appendix Table 2. Excluded Studies and Reasons for Exclusion^a

Study, Year (Reference)	Population/Setting	Reason for Exclusion
Cordoba, et al., 1998⁸³	Males, 14-50 y in Spain	Poor Quality: 270/546 excluded from analysis (49%) due to loss to followup or non-adherence to protocol (60/270). Therefore, not intent-to-treat analysis. Unit of randomization was the primary care practice; unit of analysis was the patient. Other problems not listed as analytic approach represents a fatal flaw.
Dimeff, 1998 (PhD dissertation)⁸⁴ Dimeff & McNeely, 2000⁸⁵	College student health center patients in United States	Poor Quality: Post randomization attrition from both groups. Early dropouts from intervention group excluded from further analysis (did not use intent-to-treat design). Small sample (n = 36), short followup (30 days), very low intensity computer-based intervention.
Blair, 2000 (PhD dissertation)⁸⁶	Undergraduate college students in United States	Poor Quality: Non-randomized allocation to treatment and control; subjects were "assigned." Unequal assignment to treatment (n = 74) vs. control (n = 103) from using pilot study participants to increase sample size. Assembly of comparable groups at baseline was not reported. Loss to followup was significant at 4-wk followup (20/74 in treatment and 29/103 in controls). Outcomes in pre- and post-test design could not be matched between individuals at baseline and followup.
McIntosh et al., 1997⁸⁷	United States	Poor Quality: Unclear allocation concealment with baseline non-comparability between comparison group numbers and probably differences in alcohol quantity frequency measures, particularly among women. Inadequate power for analyses, including all participants at baseline; thus results reported by gender subgroups must be underpowered.
Burton et al., 1995⁸⁸	United States	Poor Quality: Single alcohol outcome measure not clearly defined; presumed measure (change in lifetime CAGE responses) is insensitive and lacks content, concurrent, and predictive validity for improvement among problem drinkers addressed as part of this population-based multi-factorial risk factor intervention.
Aalto et al., 2000⁸⁹	Finland	Poor Quality: Inadequate allocation concealment with intervening physician "drawing a card" to assign randomization condition to patients during intervention. 34% overall loss to followup with large differences between groups that could affect results, even though not statistically significant, possibly due to small sample sizes. No replacement of missing values in analyses. Unclear blinding of participants or outcome assessors and unclear intervention delivery.
Aalto et al., 2001⁹⁰	Finland	Poor Quality: Inadequate allocation concealment with intervening physician "drawing a card" to assign randomization condition to patients during intervention. Unequal number of participants in comparison groups at baseline. 32% overall loss to followup with large differences between groups that could affect results, even though not statistically significant, possibly due to small sample sizes. No replacement of missing values in analyses. Unclear blinding of participants or outcome assessors and unclear intervention delivery.
Logsdon et al., 1989⁹¹	United States	Poor Quality: Use of single alcohol outcome measure without definition of how measured at baseline and without definition of how change was quantified. Otherwise well-conducted feasibility controlled clinical trial of multi-factorial preventive intervention in primary care.
Persson & Magnusson, 1989⁹²	Sweden	Poor Quality: Alcohol consumption measures reported for intervention groups but not controls. Unclear allocation concealment and 31% overall attrition rate. Unclear blinding of participants or outcome assessors.
Heather et al., 1987⁹³	Scotland	Poor Quality: Less than half of intended intervention participants received the full intervention due to implementation design difficulties. Post-randomization exclusions of participants with numbers not reported.
Israel et al., 1996⁹⁴	Canada	Poor Quality: Loss to followup 30%, with no adjustment for missing data at followup. Baseline comparison of study group composition unclear.
Waterson et al., 1990⁹⁵	England	Poor Quality: Concealment of allocation a concern because clinics were assigned to conditions non-randomly. High and differential attrition between groups (41% and 50% to first followup assessment, 26% and 66% at second followup assessment), which analyses do not address, reduce interpretability of findings.
Kristenson et al., 1983⁹⁶	Sweden	Poor Quality: High attrition at first followup (2 y): 41% in intervention group, 27% in control group, unclear blinding at followup assessment.
Heather et al., 1987⁹⁷	Scotland	Excluded Setting and Poor Quality: Media-recruited problem drinkers received 2 levels of self-help intervention. Attrition rate 55% with differences between groups and no replacement of missing values in analysis.
Antti-Poika et al., 1988⁹⁸	Finland	Excluded Health Care Setting: Nurse and physician counseling of hospitalized injured male patients who screened as heavy drinkers or alcoholics was evaluated in randomly assigned intervention vs. controls.
Blondell et al., 2001⁹⁹	United States	Excluded Health Care Setting: Brief physician intervention with and without peer intervention was compared to usual care among non-randomly assigned patients hospitalized for alcohol-related injuries.
Elvy et al., 1988¹⁰⁰	New Zealand	Excluded Health Care Setting: Evaluation of inpatient referral of hospitalized problem drinkers.
Forsberg et al., 2000¹⁰¹	Sweden	Excluded Health Care Setting: Randomized comparison of brief vs extensive alcohol intervention in an emergency surgical ward by surgical staff.
Gentilello, 1997¹⁰²	United States	Excluded Health Care Setting: Randomized comparison of subsequent alcohol consumption and emergency department visits among alcohol-affected patients receiving an onsite intervention in a trauma center vs. controls.
Gentilello et al., 1999¹⁰³	United States	Excluded Health Care Setting: Randomized comparison of reinjury rates among alcohol-affected patients in a level 1 trauma center receiving brief intervention vs. controls.
Heather et al., 1996¹⁰⁴	Australia	Excluded Health Care Setting: Controlled trial of brief motivational interviewing, skills-based counseling, or usual care on alcohol consumption after discharge among hospitalized male heavy drinkers.
Monti et al., 1999¹⁰⁵	United States	Excluded Health Care Setting: Randomized comparison of brief motivational interviewing or usual care on alcohol-related consequences among adolescents seen in the emergency department.
Watson, 1999¹⁰⁶	Scotland	Excluded Health Care Setting: Comparison of 3 brief nursing interventions to reduce alcohol consumption on potential problem drinkers in general hospital wards.
Welte et al., 1998¹⁰⁷	United States	Excluded Health Care Setting: Comparison of risk reduction intervention with treatment referral or usual care among general hospital patients at risk for (or with) alcohol dependence.
Romelsjo et al., 1989¹⁰⁸	Adults, age 18-64 y in Sweden	Poor Quality: Randomization process was not simple, but rather a quota sample, stratified on general practitioner and then on practice. Masking of general practitioner not assured. Significant post-randomization exclusion (151/258 participants). Inclusion criteria not adequately applied, resulting in missing of the most eligible persons based on drinking (and not laboratory levels). Non-comparable groups assembled at baseline with respect to alcohol consumption and problems and no adjustment for differences. Attrition rate relatively low (11/83) and non-differential. Does not appear to be intent-to-treat analysis, as some cases followed up were not included in reported analyses. No statistical testing of results reported.
Oliansky, et al., 1997¹⁰⁹	Adults age 18-55 y Adolescents age 12-18 y Women age 18-55 y in the United States	Poor Quality: Three different populations in 3 different clinics were "randomly" assigned to intervention vs. control conditions. In 2 clinics, random assignment was based on odd/even medical record numbers. In the third, the randomization method was not described. Comparability of intervention and control groups at baseline was not reported. The intervention was not clearly defined. The measures used to determine eligibility for the study and to measure outcomes (Substance Use Screening Instrument [SUSI]) is a novel instrument developed for this project. It is reported as being based on AUDIT and CAGE, but the actual items included in SUSI are not provided. Loss to followup was up to 39% in 1 clinic and was greater in intervention conditions if there were equal numbers in intervention and control groups initially (cannot be sure from report). Maintenance of comparable groups not reported. Outcomes were reported for all substances combined (alcohol, tobacco, and other drugs of abuse).
Tomson et al., 1998¹¹⁰	Adults, age 25-54 y in Sweden	Poor Quality: Unequal randomization results (intervention n = 100, control n = 122) without rationale. Comparability of intervention and control groups at baseline not assured because control group not assessed for CAGE or alcohol consumption at baseline. Change in CAGE and alcohol consumption not measured in control group, therefore measures not equal. Possible contamination of control condition by receipt of general practitioner intervention. High attrition rates (50-62%), with loss to followup greater in the intervention condition. Analyses do not account for baseline differences and no test of between-group differences for primary outcomes (except gamma-glutmyltranasferase).

^a AUDIT = Alcohol Use Disorders Identification Test (10-item instrument for risky/harmful use); CAGE = 4-item screening questionnaire to detect alcoholism.

Return to Document

Appendix Table 2. Excluded Studies and Reasons for Exclusiona

Appendix Table 2. Excluded Studies and Reasons for Exclusion^a