• There were large differences in inclusion criteria and types of treatment between studies.
• Given that studies show that assessment can positively change alcohol use, trials of alcohol use
interventions in people with psychosis require non-assessment control groups to more accurately
assess treatment effects.
Baker et al.
244
reported, a conservative estimate of 0.5 was used
for the pre- ⁄ post effect size calculations. As there
was substantial diversity in clinical presentation of
the participants, type of treatment, and alcohol use
measures between the studies, calculating an overall
effect size, which would occur during a metaanalysis,
was inappropriate and therefore not
completed.
To estimate the level of bias within the identified
studies, an assessment of their methodological
quality was conducted using the validated Physiotherapy
Evidence Database (PEDro) scale (24).
PEDro scores are calculated by assessing whether
a study has i) specified participant eligibility criteria;
ii) allocated participants randomly to groups; iii)
concealed allocation; iv) used groups similar at
baseline on main prognostic signs; v) employed
blinding of assessors who measured at least one key
outcome; vi) had adequate follow-ups; vii) used
intent-to-treat analysis; viii) employed between
group statistical comparison of outcomes; and ix)
given both point estimates and variability for
outcomes. Consistent with the recent study by
Spring et al. (25), two items regarding blinding of
subjects and therapists were not scored in the present
review, as these criteria were not appropriate for the
studies under review. Two raters (AB and SH)
independently rated the seven RCTs on the PEDro
scale and reached consensus on the ratings.