“…Two RCTs 36 , 37 both had a low risk of bias across five domains; see Figure 2 . For the 15 NRCTs, 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 seven of the studies 21 , 24 , 26 , 29 , 30 , 32 , 35 may have been affected by confounding variables. One of which was patient recruitment into the studies, as this was not reflective of all service users admitted into the hospital setting as some only included elderly patients or those who were prescribed two or more medications.…”