External validation of a tool to assess medication-related admissions in four Swedish hospitals

Background

The MedBridge trial is a large ongoing clinical trial on medication reviews in elderly patients in Sweden with participating hospitals from Uppsala, Enköping, Västerås and Gävle. A tool for assessing medication-related admissions (MRAs) has been developed and validated in Uppsala: Assessment Tool for identifying Hospital Admissions Related to Medications (AT-HARM10). MRAs is often used as an outcome for medication reviews and similar interventions. Validation of AT-HARM10 outside of Uppsala is therefore warranted.

Aim

The aim was to externally validate AT-HARM10. This was needed since AT-HARM10 was for the first time used outside Uppsala.

Materials and methods

The readmissions of elderly patients within the MedBridge trial were collected from all participating hospitals. The assessors in this study were two final-year pharmacy students. First, the assessors assessed the readmissions independently of each other and then a consensus meeting was held where the readmissions were discussed, and a consensus was reached. The outcomes in this study were the assessment time, inter-rater reliability and content validity between the hospitals from Uppsala, Enköping, Västerås, and Gävle. The time was measured every day during the assessment period. The inter-rater reliability was calculated for every consensus meeting. The content validity was based on the questions from AT-HARM10 and was collected for each of the assessed readmissions and compared between the four participating hospitals. The content validity of the distribution of questions U1-P10 between the hospitals was tested with a χ²-test.

Results

In this study, 1687 unplanned readmissions from 893 patients were assessed. Uppsala University Hospital had 6.6 minutes as total mean assessment time including consensus meeting. The total mean time from Enköping, Västerås and Gävle were 6.48, 6.32 and 6.16 minutes respectively. The strength of agreement between the two assessors was substantial in Uppsala, Västerås and Gävle (ĸ 0.80, 0.79 and 0.76 respectively) and perfect in Enköping (ĸ 0.89). For the content validity were there no differences between the hospitals in the distribution of question U1-P10.

Conclusion

AT-HARM10 was for the first time used outside Uppsala and has been externally validated with data from the hospitals in Enköping, Västerås, and Gävle. The time needed per admission for the two assessors was low which confirms that it is possible to have final-year pharmacy students as assessors. This information is important both within the MedBridge trial and if the tool would be used in outer clinical studies.