Bladh 2011.

Study characteristics
Methods	Randomised controlled trial
Participants	400 participants randomised (199 to medication review and 201 to control) Patients admitted to 2 internal medicine wards at a university hospital in Sweden Median (IQR) age: medication review group 81 (72 to 87) years, control group 82 (75 to 86) years 39% male Median number of drugs: 7 (+1 drug on demand)
Interventions	Medication reviews were performed with a computer support system (MiniQ) that identified potentially inappropriate prescribing (PIP) according to 3 drug‐specific quality indicators, and included oral feedback to prescribing physicians. PIPs were (1) drugs that should be avoided in the elderly, for example long‐acting benzodiazepines and drugs with anticholinergic action, (2) 3 or more psychotropic drugs (i.e. antipsychotics, anxiolytics, hypnotic‐sedatives and antidepressants) and (3) potentially serious drug‐drug interactions: Category D interactions according to the Pharmaceutical Specialties in Sweden (FASS) specifying drug combinations that should be avoided. Participants in the control group received normal care.
Outcomes	Primary outcome: EQ‐5D index Secondary outcomes: 'self‐rated global health' (registered as an integer from 1 (very poor) to 5 (very good)) and 'attitudes towards the medication report' (evaluated by questionnaires sent to participants' GPs after discharge from hospital) PIP items: (1) drugs that should be avoided in the elderly, (2) 3 or more psychotropics and (3) potentially serious interactions Potential drug‐related problems (DRPs) identified only in the intervention group (mortality not reported as an outcome per se) All outcomes had 6 months of follow‐up
Notes	Funding: Swedish National Board of Health and Welfare
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomised (1:1) to intervention or control group. Two persons without knowledge about the study protocol performed the randomisation using sequentially numbered envelopes. Authors confirm that the allocation sequence was generated using random generator software.
Allocation concealment (selection bias)	Low risk	A ward physician or nurse judged whether the medical condition of the patient allowed inclusion in the study. Sequentially numbered, sealed envelopes were opened after participant details were written and transferred to the assignment card via a carbon paper inside the envelope.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not described, but probably not blinded.
Blinding of outcome assessment (detection bias) Mortality (all‐cause)	Low risk	Mortality data were likely gathered from a non‐biased national register (through unique patient‐specific social security number).
Blinding of outcome assessment (detection bias) Hospital readmissions (all‐cause)	Low risk	Readmission data should be unbiased, as they were gathered from a non‐biased register.
Blinding of outcome assessment (detection bias) Health‐related quality of life	High risk	It was not stated whether the person who called the participants and asked about EQ‐5D was blinded to group allocation. The trial authors confirmed that the person who called the participants was not blinded.
Incomplete outcome data (attrition bias) Mortality (all‐cause)	Low risk	No missing data were described; all data should be available from a non‐biased national register.
Incomplete outcome data (attrition bias) Hospital readmissions (all‐cause)	Low risk	No missing data were described; all data should be available from a non‐biased national register.
Incomplete outcome data (attrition bias) Health‐related quality of life	High risk	The HRQL follow‐up was completed for 204 participants (59%). Lost to follow‐up: intervention group 69 (20 died, 7 declined, 42 not reached despite repeated attempts), control group 72 (15 died, 6 declined, 51 not reached despite repeated attempts).
Selective reporting (reporting bias)	Low risk	Relevant outcomes were collected and were similar to information provided on www.clinicaltrials.gov.
Contamination bias	High risk	No cluster‐randomisation.
Other bias	Low risk	No evidence of other types of bias.