The database search retrieved 36,570 records (see ESM) of which 8708 were duplicates identified through Endnote (5405) and Rayyan (3303). A total of 27,862 records underwent screening followed by eligibility checking to identify 21 studies that met the inclusion criteria. Screening references of included studies identified one additional study, resulting in a final number of 22 studies included in the review. Details of the search results are demonstrated in a PRISMA chart (Fig. 1).

Flow chart of the screening process for included studies. DRP drug-related problems, PRFs potential risk factors

Of the included studies, eight identified PRFs of PIP [43,44,45,46,47,48,49,50], seven of MEs [20, 21, 51,52,53,54,55], three of ADRs [17, 50, 56], two of MDs [57, 58], one of ADEs [4], and two included a wider range of DRPs [59, 60]. Sixteen studies were conducted in specialised mental health institutions/hospital trusts [4, 20, 21, 45, 47,48,49, 51,52,53,54, 56,57,58,59,60] whereas six studies were completed in psychiatry units from general hospitals [17, 43, 44, 46, 50, 55]. The majority of the studies originated in Europe (19/22, 86.4%) [4, 17, 20, 21, 43,44,45,46,47,48,49,50,51,52,53,54,55, 58, 60], particularly in the UK (n = 7/21, 33.3%) [4, 20, 21, 52,53,54, 58]. The remaining studies were conducted in India (n = 2/21, 9.5%%) [56, 59] and the USA (n = 1/21, 4.8%) [57]. All studies were observational except for one that was interventional [44]. Thirteen studies used a multivariable logistic regression alone [47, 48, 54, 55, 58] or a multivariable logistic regression preceded by a univariate regression [4, 43,44,45,46, 50, 53] or bivariate regression [17] to test PRFs. Three studies were published between 2000 and 2010 [20, 51, 52] with 13 published between 2010 and 2020 [17, 21, 43,44,45,46, 53,54,55, 57,58,59,60] and six in 2021 [4, 47,48,49,50, 56]. Two non-English studies were included in the review [50, 51] and each was translated by a native speaker fluent in English. A summary of the study characteristics is presented in Table 3.

Only two studies reported all necessary research sections described by Allan and Barker [41] in question one [17, 54], including a clear and detailed introduction, methodology, results, interpretation and conclusion. None of the studies reported checking the assumptions for the statistical tests they performed. An approach to confirm validity of the DRP identification method was reported by the majority of studies (n = 15/22, 68%) while reliability was reported by eight studies (n = 8/22, 36.4%). The full quality assessment is presented in Table