consistent changes in summated MAI with variable effects on heterogeneity (table 2). Second, postintervention pooled data from five studies27–31 (488 intervention participants, 477 control participants) showed a lower summated MAI score (mean difference −3.88, 95% CI −5.40 to −2.35) in the intervention group compared with the control group (table 2). There was little evidence of heterogeneity between these estimates (I2=0%). This was consistent with the findings of Gallagher et al,25 which were not included in the meta-analysis because the data were skewed. Third, one study32 expressed the MAI score as the number of inappropriate prescriptions. The percentage of inappropriate prescriptions decreased in all MAI domains (n=10) in the intervention group and increased in five domains in the control group. These data could not be included in a meta-analysis.Beers’ criteria Pooled data from two studies30 31 (298 intervention participants, 288 control participants) showed that intervention group participants were prescribed fewer Beers’ drugs than control group participants postintervention (mean difference −0.1, 95% CI −0.28 to 0.09; I2=89%; table 2). Spinewine et al31 also reported the proportion of patients taking one or more Beers’ drugs preintervention and postintervention. Similar improvements were reported in the proportion of intervention and control group patients receiving one or more Beers’ drugs between hospital admission and discharge (OR 0.6, 95% CI 0.3 to 1.1). As this was the only study to report the results in this format, meta-analysis was not possible.McLeod criteria One study used the McLeod criteria38 to identify the initiation and discontinuation rates of 159 prescriptionrelated problems.34 The reported relative rate of initiation of inappropriate prescriptions for the intervention group was 0.82 (95% CI 0.69 to 0.98). However, the intervention did not appear to have an effect on the relative rate of discontinuation of pre-existing prescription-related problems (1.06, 95% CI 0.89 to 1.26). Meta-analysis was not possible as these criteria were not used in other studies.STOPP and START criteria Two studies25 26 used the Screening Tool of Older Person’s Prescriptions (STOPP) criteria to screen for PIP in older patients admitted to hospital. Gallagher et al25 reported lower (p