Pharmacist-led primary care interventions to promote medicines optimisation and reduce overprescribing: a systematic review of UK studies and initiatives

Abstract Objectives To systematically review and synthesise evidence on the effectiveness and implementation barriers/facilitators of pharmacist-led interventions to promote medicines optimisation and reduce overprescribing in UK primary care. Design Systematic review. Setting UK primary care. Methods We searched MEDLINE, Embase, CINAHL PsycINFO and The Cochrane Library for UK-based studies published between January 2013 and February 2023. Targeted searches for grey literature were conducted in May 2023. Quantitative and qualitative studies (including conference abstracts and grey literature) that addressed a relevant intervention and reported a primary outcome related to changes in prescribing were eligible for inclusion. Quality of included studies was assessed using the Multiple Methods Appraisal Tool. We performed a narrative synthesis, grouping studies by publication status, setting and type of data reported (effectiveness or implementation). Results We included 14 peer-reviewed journal articles and 11 conference abstracts, together with 4 case study reports. The journal articles reported 10 different interventions, 5 delivered in general practice, 4 in care homes and 1 in community pharmacy. The quality of evidence was higher in general practice than in care home settings. It was consistently reported that the intervention improved outcomes related to prescribing, although the limited number of studies and wide range of outcomes reported made it difficult to estimate the size of any effect. Implementation was strongly influenced by relationships between pharmacists and other health and care professionals, especially general practitioners. Implementation in care homes appeared to be more complex than in general practice because of differences in systems and ‘culture’ between health and social care. Conclusions Pharmacist-led interventions have been reported to reduce overprescribing in primary care settings in the UK but a shortage of high-quality evidence means that more rigorous studies using high-quality designs are needed. More research is also needed in community pharmacy settings; to assess intervention effects on patient outcomes other than prescribing and to investigate how reducing overprescribing can impact health inequalities. PROSPERO registration number CRD42023396366.https://fundingawards.nihr.ac.uk/award/NIHR135767


INTRODUCTION
This evidence review was performed to support implementation of the National Overprescribing Review for England (NOR; see below)) by examining research on pharmacist-led overprescribing interventions in UK primary care settings.Pharmacists are trained to provide advice and support to patients and other health professionals, pharmacist-independent prescribers (PIPs) have existed since 2006 and patients are increasingly asked to consider the community pharmacy as a first source of support for minor health conditions.Alongside community pharmacies, many general practices have pharmacists as members of the practice team.Pharmacists, working with general practitioners (GPs) and other healthcare professionals, are thus well placed to support interventions directed towards medicines

STRENGTHS AND LIMITATIONS OF THIS STUDY
⇒ We included evidence often excluded from systematic reviews to get as full a picture as possible of how pharmacist-led interventions are implemented and sustained in practice as well as their characteristics and effectiveness.⇒ Many of the studies lacked a control group and the research took place in a highly complex and evolving system, meaning that results could have been influenced by confounding factors such as other interventions in the health and social care system.⇒ Some review processes were performed by a single reviewer and meta-analysis was not feasible.
Open access optimisation and the reduction of overprescribing.Such interventions include carrying out structured medication reviews (SMRs) directly with patients and carers and/or reviewing data from patient records.The aims and objectives of the review are outlined below, following a brief clarification of terminology.
Overprescribing has been defined as 'the use of a medicine where there is a better non-medicine alternative, or the use is inappropriate for that patients' circumstances and wishes'. 1 Overprescribing is often related to the concept of problematic polypharmacy, where harmful effects result from the prescription of multiple medications.However, there is no agreed definition of polypharmacy and patients with complex health conditions may require multiple medications.
Medicines optimisation is an umbrella term for interventions designed to ensure that medicines are used safely and effectively, producing the best possible outcomes for patients.In this context, deprescribing refers to the process of stopping medications that are no longer appropriate to a patient's needs.Deprescribing is a response to overprescribing and problematic polypharmacy and involves collaboration between health professionals and patients and/or carers to ensure shared decision-making.Shared decision-making with patients and/or carers is fundamental to successful medicines optimisation 2 but the need for time and resources to ensure that this takes place can create barriers to service delivery.Another related term, medicines reconciliation, is a more technical process to ensure consistency between prescription records and the medications the patient is actually receiving and taking.The terminology around overprescribing and other forms of medicines misuse was recently reviewed by Singier et al. 3 Medication review involves examining a patient's prescriptions as a whole and is separate from measures to reduce inappropriate prescribing of specific medications or types of medication such as antibiotics or proton pump inhibitors.
Overprescribing can cause direct harm to patients in a variety of ways.It has been estimated that about 6.5% of hospital admissions are caused by harmful effects of medication, rising to 20% for people aged over 65. 1 In addition to physiological harms, long-term use of some medications can lead to dependency and problems when attempting to withdraw the medication.
Issues relating to prescribed medication can arise from a whole range of causes, including patients requiring treatment for multiple conditions, lack of coordination between different health professionals or organisations and failures of communication between health professionals and patients (eg, failing to gather information because of time constraints on appointments).Availability of new medications and increasing numbers of people living with long-term conditions such as arthritis and diabetes have resulted in patients being prescribed more medications and continuing to take them for long periods of time, often for life.The average number of prescription items per head of population doubled between 1996 and 2016, and over 75% of prescriptions are repeat prescriptions. 1 Pharmacists are thus well placed to support processes of medicines optimisation, which involve them working closely with medical professionals (particularly GPs), commissioners of healthcare and patients.The report of the NOR for England, published in 2021, provides numerous examples and case studies. 1 The NOR for England was set up in 2018 to evaluate the extent of overprescribing in the National Health Service (NHS) and recommend measures to reduce it, particularly in primary care.A review of existing research (overview of systematic reviews) was commissioned to support the national review. 4The NOR identified a need for a more consistent and effective approach to medication review, which requires both the identification of effective interventions and an understanding of the factors that need to be addressed in terms of organisational and cultural barriers to implementation.The national review's recommendations included changes to systems (patient records, transfers of care and clinical guidance) and culture (reduced dependence on medication and support for shared decision-making), as well as the appointment of a National Clinical Director for Prescribing. 1 This evidence review was commissioned to support implementation of the NOR recommendations by examining research on pharmacist-led overprescribing interventions in UK primary care settings.Our focus on pharmacist-led interventions complements recent research on deprescribing in the UK context.The TAILOR evidence synthesis sought to identify how best to support deprescribing in older people living with multimorbidity and polypharmacy.The authors concluded that effective deprescribing requires 'attention to providing an enabling infrastructure, access to data, tailored explanations and trust'. 5More recently, Radcliffe et al conducted a realist review and synthesis examining multidisciplinary medication review and deprescribing interventions for older people in primary care. 6This study identified a number of key mechanisms that could contribute to the design of effective interventions, including integration of pharmacists into the multidisciplinary team delivering the intervention.Pharmacist-led interventions could fall within the scope of both of these studies, but characterisation of the evidence base is required to support the application of insights derived from these more general, theory-based reviews.
We aimed to assess the effects of relevant interventions on outcomes related to prescribing, identify key characteristics of the interventions and examine barriers and facilitators to implementation in routine practice.A further aim was to assess the quality of the evidence base and identify priorities for further research.

Review aims and objectives
We aimed to perform a systematic review of published literature and published or informally published evaluations We excluded interventions aimed at reducing overprescribing of specific medications or types of medication, for example, antibiotics or proton pump inhibitors.Studies of children and young people were also excluded.

Search methods
The literature search harnessed economies of scale by identifying primary studies for inclusion in this review and reviews for inclusion in a scoping review for internal use to inform the wider project.Searches were conducted by an information specialist (MC) in order to identify published and unpublished evidence on primary care interventions to reduce overprescribing.
Phase 1: peer-reviewed literature A first phase of database searches was run in February 2023 to retrieve relevant peer-reviewed literature.Searches were designed around the concepts mentioned in table 1.
While we are aware of the Morel filter (2022) for identifying studies of deprescribing, 7 our focus was specifically on a primary care setting.Search strategies are provided in online supplemental file 1.
Searches covered the databases MEDLINE, Embase, CINAHL, PsycINFO and The Cochrane Library and were limited to studies published since 2013 and in OECD countries with healthcare systems similar to the UK.
Phase 2: grey literature A further phase of targeted searches was conducted in May 2023 to identify unpublished or 'grey' literature.This involved searching for the case studies identified by the NOR (to identify any which had produced a report or evaluation), and then searching the Overton.io platform for pharmacist-led deprescribing/overprescribing and medicines optimisation.
Searches were complemented by input from stakeholders (internal and external topic advisers) to minimise the risk of missing any other relevant evidence.

Study selection
Records retrieved by the literature search were stored in a shared EndNote library and deduplicated.Screening for inclusion at the title level was performed by single reviewers after piloting a test set.Reviewers could refer records to another team member in the event of uncertainty and a 20% sample of records was screened by a second reviewer to validate title-level inclusion decisions.
Screening for inclusion at the abstract and full-text level was performed by pairs of reviewers acting independently.Disagreements were resolved by discussion among the reviewers involved (AJC, DC and LP).A good level of agreement was achieved, values of kappa between pairs of reviewers ranging from 0.67 to 0.96.Reasons for exclusion at the full-text stage were recorded.

Data extraction
Data extraction tables and summary tables were developed in Microsoft Word.Extraction was performed by a single reviewer, with a 10% sample being checked for consistency and accuracy.In addition to standard data extraction fields (study design/sample size, setting, intervention, key findings and strengths/limitations), we used the TIDieR Lite framework to collect information on the features of interventions reported as 'successful' to determine whether service commissioners and providers should consider specific factors when commissioning/ delivering services.TIDieR Lite is a simplified version of the TIDieR checklist. 8ality assessment Methodological quality of peer-reviewed journal articles was assessed using the Mixed Methods Appraisal Tool (MMAT) version 2018. 9The tool includes screening questions and methodological quality questions for Open access different study designs (qualitative, randomised trials, non-randomised quantitative studies, descriptive studies and mixed methods).Quality assessment results were combined with identified strengths and limitations (including those reported by study authors) to characterise the contribution of individual studies and groups of studies to the overall evidence base.

Data synthesis
We performed a narrative synthesis of the included studies using text and tables to describe study and intervention characteristics in line with methodological and reporting guidelines. 10 11We initially grouped studies by publication status, considering peer-reviewed journal articles (regardless of study design and quality) separately from conference abstracts and case studies.Within these three categories, we grouped studies by setting (general practice, care homes or community settings).We also distinguished between studies reporting effectiveness of interventions and those reporting implementation of interventions (eg, qualitative studies and process evaluations).In view of study heterogeneity and reporting limitations, effectively implemented interventions were defined as those where the study authors' conclusions indicated that the service was regarded as a success and was planned to continue or be expanded.Studies reported a wide variety of outcomes using diverse effect measures.For this reason, we did not attempt to calculate a standardised metric to compare effect sizes across outcomes.The synthesis used a 'votecounting' method (number and proportion of studies reporting positive, negative or neutral outcomes), prioritising prescribing-related outcomes over patient and other outcomes.Reported effect measures and associated 95% CIs were recorded in the text and tables.Tables of study characteristics and findings were presented alphabetically by author for consistency.While reporting results from all study designs, we prioritised stronger study designs (experimental and quasi-experimental) over those of uncontrolled observational studies.In terms of exploring heterogeneity, the structure of the synthesis allowed consideration of potential modifiers including study design, study quality and setting.Intervention components and aspects of implementation were examined using modifications of existing frameworks, the component analysis was prespecified in the review protocol.
We did not use the Grading of Recommendations Assessment, Development and Evaluation approach to assess certainty of evidence because of its emphasis on randomised trials and downgrading of other study designs.Instead, we distinguished between controlled and uncontrolled studies, identified areas of consistency and inconsistency and highlighted areas of particularly limited evidence (eg, settings or outcomes represented by single studies).A similar approach has been used by team members in previous reviews. 12

Public involvement
The review was supported by a public panel that provided feedback on public perceptions that informed the review and are reflected in the Discussion.

Variations from protocol
We used Tidier Lite instead of the full TIDieR framework.This was because the full framework is designed to allow the replication of interventions and therefore goes beyond the degree of detail required for evidence synthesis.The scoping review of reviews referred to in the protocol was not completed (see 'Search methods' above).

Results of literature search
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram (figure 1) summarises the study selection process.After screening 1774 records at the title and abstract stage and 215 full-text articles, we included 14 published articles, 11 conference abstracts and 4 case study reports.The majority of exclusions were of studies conducted outside the UK, with a smaller number excluded because the intervention was not pharmacist led or the article did not report empirical data.Characteristics of the included studies are reported in the following sections.

Study characteristics
Study characteristics are summarised in table 2, with full data extraction tables in online supplemental file 2. The 14 publications reported on 10 interventions, of which 5 were delivered in general practice (7 publications [13][14][15][16][17][18][19] ), 3 in care homes for older people (5e publications [20][21][22][23][24] ), 1 in care homes for people with intellectual disabilities (IDs) 25 and 1 in community pharmacies. 26ll the interventions involved medication review in some form.Distinctive features of interventions included use of IT to identify patients for review 13-15 17 18 ; a key role for PIPs in medication management in care homes 23 24 ; and employment of pharmacists by groups of general practices (primary care networks, PCNs) to provide a holistic patient-centred service specified by NHS England. 16Intervention characteristics are considered in more detail below.
Study designs used included one individual randomised controlled tril (RCT) 19 and two cluster RCTs (CHIPPS 20 23 and PINCER 13 ), although the primary publications of the latter two trials fell outside the time period covered by this review.Two studies used an interrupted time series (ITS) design 17 18 and five used qualitative approaches. 14-16 20 24ne study was a mixed-methods process evaluation. 23The remaining studies were described as service evaluations or quality improvement reports with an uncontrolled before versus after design. 21 22 25 26pen access Included studies reported a wide range of outcomes (table 2).For further analysis, see below under 'effects of interventions' and 'Implementation/system issues, respectively.None of the studies reported details of participants other than age and sex, making it difficult to assess equity, diversity and inclusion across the evidence base.

Intervention characteristics
Table 3 in online supplemental file 2 summarises the characteristics of the included interventions using the TIDieR Lite checklist.][29] The pharmacists involved in delivering the interventions were variously described as PIPs 23 ; trained pharmacists and pharmacy technicians 13 18 ; primary care pharmacists 21 ; clinical pharmacists working in general practice [15][16][17] ; GP practice-based pharmacists working as part of a wider primary care team 19 ; community and specialist mental health pharmacists 25 ; and community pharmacists and pharmacy team members. 26One study simply referred to 'pharmacists'. 15our interventions were explicitly stated to require training of pharmacists to deliver them 13 19 23 26 ; the extent of training was described for three of these. 19 23 26Training pharmacists to deliver the PINCER intervention was described in a separate paper. 13Interventions were delivered with other primary care team members depending on the setting of the study and in some cases with staff employed by clinical commissioning groups (CCGs).In particular, only the CHIPPS study involved pharmacists with the power to prescribe medication independently; in other studies recommendations were passed to the patient's GP or another medically qualified professional for implementation.Shared decision-making with patients and/or families was specifically reported for three interventions. 16 19 22eporting of interventions varied between studies.Most studies reported the process of medication review including patient selection for review and the review itself in more detail than resulting follow-up actions.Two qualitative studies reported limited details of the review process, 14 16 although a service specification was available for the NHS England SMR investigated by Madden et al. 16 For studies where the intervention was primarily directed at improving medication review processes using general practice data, [13][14][15] it was unclear whether there was a

Open access
standard process to discuss findings with the patient and make changes to their prescriptions.All studies reporting on effectiveness of medication reviews stated that the person undertaking the review had access to relevant patient records. 17-19 21 22 25 26ntensity of interventions was also variably reported.In the CHIPPS study, PIPs committed a minimum of 16 hours/month to deliver care to approximately 20 care home residents. 27Madden et al reported that SMR appointments were recommended to allow at least 30 min for review and shared decision-making. 16The medicines optimisation intervention evaluated by Syafhan et al involved up to three meetings between patient and pharmacist 19 while the FOMM study in community pharmacies estimated times of 25 min for initial consultation, 10 min for monthly review and 11 min for quarterly review. 26ther studies reported that time and level of support allocated to interventions varied between and within CCG areas depending on local resources and priorities. 18 21Another measure of intervention intensity was the number of recommended actions, averaging 3.3/resident in care home residents with ID. 25 Most included studies reported on a single round of medication reviews with variable periods of follow-up.As noted above, some interventions required multiple interactions between pharmacists and patients.

Effects of interventions
Seven studies reported on effects of pharmacist-led interventions in some form (table 3): three in general practice, [17][18][19] three in care homes 21 22 25 (including one in a care home for people with ID 25 ) and one in community pharmacies. 26he strongest evidence for the effectiveness of interventionscame from the studies in general practice.The ITS studies of Peek et al 17 and Rodgers et al, 18 which used indicators of inappropriate prescribing to identify patients for intervention, reported significant decreases in inappropriate prescribing at 6 and 12 months after intervention (table 3).Estimated reductions were larger in Peek et al (27.9% and 40.7%) compared with Rodgers et al (16.7% and 15.3%). 17 18The 95% CIs of the two studies at 12 months did not overlap, suggesting some uncertainty about the magnitude of the effect.The randomised trial by Syafhan et al 19 preferentially recruited patients based on prescription of six or more medications and a history of recent unplanned hospital admission.The intervention was associated with a reduction in medication-related problems in those who completed the full programme (up to three appointments) and an improvement in Medicines Appropriateness Index (MAI) scores.
Of the three studies set in care homes, only Baqir et al reported a direct effect on prescribing associated with medication review, a 19.5% reduction in number of prescribed medicines. 22Alves et al 30 reported on pharmacist interventions and potential financial savings over 5 years.In the 1-year reported in detail, 24.5% of interventions involved deprescribing.Potential drug cost savings were estimated at £812 441 annually, of which £431, 493 (55%) was attributed to deprescribing.The study of Thayer et al 25 differed from the others in involving care home residents with IDs.There was a high level of polypharmacy at baseline and pharmacists made an average of 3.3 interventions/recommendations per resident, of which 12.8% involved deprescribing.A large majority of pharmacist recommendations were accepted by GPs/ psychiatrists caring for the residents.
One study in a community pharmacy setting recruited patients aged 65 or older who were prescribed four or more medications. 26Of 620 patients recruited, 441 (71.1%) completed the 6-month study.Pharmacists made 142 recommendations related to 110 patients, largely dealing with potentially inappropriate prescribing of nonsteroidal anti-inflammatory drugs (NSAIDs) and PPIs or duplication of therapy.The study also reported a significant decrease in falls and improvements in medication adherence and quality of life at follow-up.
The review included two publications from the CHIPPS Care Homes Independent Pharmacist Prescriber Study) trial 20 23 but the paper reporting effectiveness and safety results from this cluster RCT 31 was published too late for formal consideration for inclusion in our review.The primary outcome was rate of falls, with Drug Burden Index (DBI) being one of the secondary outcomes.Fall rate at 6 months did not differ significantly between intervention and control groups but DBI was lower in the intervention group (mean 0.66 vs 0.73; adjusted rate ratio 0.83, 95% CI 0.74 to 0.92).

Implementation/system issues
Seven studies provided quantitative and/or qualitative evidence on factors affecting implementation of pharmacist-led interventions, of which four were performed in general practice [13][14][15][16] and three in care homes. 20 23 24he general practice studies focused on different parts of the implementation pathway.Two dealt with implementation of IT systems to support detection of potentially hazardous prescribing 14 15 ; one was a process evaluation of the PINCER trial 13 ; and one focused on implementation of SMRs as recommended by NHS England in routine practice. 16The studies of IT-supported interventions were broadly positive about the potential for implementation and sustainability, but the study of NHS England's SMR programme concluded that its early implementation failed to deliver the planned holistic and patient-centred approach.

Conference abstracts
We included 11 conference abstracts (table 4), of which 2 were earlier reports of studies subsequently published as full papers. 30 32All of the included abstracts focused on intervention effects on prescribing and related outcomes.
4][35] As a group, Open access these three abstracts provided weak evidence of associations between pharmacist-led medication reviews and changes in medication and cost savings together with high levels of patient satisfaction (table 5).
Two abstracts reported on selected general practice populations.The only comparative study in this group reported that patients living with frailty who were reviewed by a pharmacist as part of a multidisciplinary team review had a reduction in total medications compared with a control cohort. 36When patients recently discharged from hospital were reviewed by a pharmacist working in their general practice, 16 out of 35 had changes made to their medication, with 74% of changes involving deprescribing. 37urning to studies performed in care homes, two abstracts by Doherty et al 38 39 evaluated an intervention entitled Medicines Optimisation in Older People which involved case management by pharmacists.The authors reported that inappropriate prescribing (based on the MAI) was highly prevalent at baseline×84%) but declined significantly following the intervention.Swift reported that a team comprising pharmacists and pharmacy technicians who both performed medication reviews and supported care home staff significantly reduced inappropriate polypharmacy (measured by prescribing quality indicators) between 2024 and 2017. 40For care home residents receiving palliative care, SMRs involving shared decision-making were associated with high rates of changes to medication (1787 suggested changes from 574 reviews, 76% of which were implemented) and associated cost savings. 41

Grey literature case studies
We included reports of four case studies reporting on local initiatives in three areas of England (see table 4).Details of all case studies may be found in Annex C of the NOR report. 1 Case studies were submitted by NHS organisations (mainly CCGs) and included varying amounts of data on intervention characteristics, support for implementation and outcome measures.Three interventions were delivered in general practice and one in care homes.The initiative developed by Swale CCG was distinctive in using pharmacy technicians to review less complex cases, although the initiative was targeted at patients considered high risk for ADRs.Although not classified as research, such case studies can provide useful data on implementation of interventions and outcomes achieved in routine practice

Study quality
Quality assessment results using the MMAT are presented in online supplemental file 3. The results should be read in conjunction with the study strengths and limitations (see table 2 in online supplemental file 1).
Five different checklists within the MMAT were used to assess the 14 studies.The sample included one RCT 19 ; six studies were classified as quantitative nonrandomised 17 18 21 22 25 26 ; one as quantitative descriptive 13 ; one as mixed methods 23 and five as qualitative. 14-16 20 24ll studies passed the screening questions (are there clear research questions? and do the collected data allow to address the research questions?) The RCT by Syafhan et al was described as a pragmatic trial and was at relatively high risk of bias for this type of Open access

Open access
design.The trial did not achieve the planned number of participants and there was a high rate of attrition (about 30%), meaning that many participants did not receive the full intervention or provide outcome data.The trial also suffered from unclear reporting: method of randomisation and whether outcome assessors were blinded was not reported, making it difficult to assess overall risk of bias.The quantitative non-randomised studies comprised four observational studies at high risk of bias because of the absence of a control 21 22 25 26 and two large ITS studies. 17 18The MMAT tool identified some limitations of these studies, including some risk of confounding and incomplete outcome data in one study. 18However, these were large studies conducted in routine practice and providing evidence of a statistically significant effect at 12 months postintervention.The process evaluations of the CHIPPS 23 and PINCER 13 studies both scored highly on the MMAT assessment.
The qualitative studies were generally of good quality, with sufficient data presented in support of conclusions and appropriate use of frameworks and thematic analysis to organise presentation of the findings.The study by Alharthi et al 20 was a secondary analysis of data collected for another purpose, making it unclear whether qualitative data collection methods were adequate.
Using the system applied by the authors in previous studies of complex health service interventions, 12 the overall strength of evidence was classified as borderline 'stronger' (generally consistent findings in multiple studies with a comparator group) for general practice, 'weaker' (generally consistent findings in one study with a comparator group design and several non-comparator studies or multiple non-comparator studies) for care homes and 'very limited' (single study) for community pharmacies.

Effectively implemented interventions
Three research studies met the criteria for 'effectively implemented' interventions: the closely related PINCER 18 and SMASH 17 interventions in general practice and the Somerset model of medication review in care homes. 219][40][41] Case studies from Brighton and Hove and Swale CCGs appeared to report effectively implemented interventions targeted at highrisk patients in general practice (table 4).An evaluation of the early implementation of SMRs in PCNs indicated that the service as provided did not match the vision of a patient-centred holistic review with an emphasis on shared decision-making. 16

Summary of findings
In spite of its broad inclusion criteria, this review identified a relatively small number of studies of pharmacist-led interventions in UK primary care (14 peer-reviewed journal articles, 11 conference abstracts and 4 case studies).Overall, the bulk of evidence came from the care home sector but most of the better quality evidence was derived from studies conducted in general practice.The majority (8/14) of peer-reviewed papers were published in 2020 or later, suggesting that this is a developing area of research and practice in the context of encouraging patients to consult pharmacists initially for minor conditions and to increase pharmacists' prescribing rights.It was encouraging that we identified a number of effectively implemented interventions and initiatives in both care homes and general practice.

Outcomes of effective interventions
This systematic review suggests that pharmacist-led interventions may reduce overprescribing in primary care settings in the UK, although more controlled studies are needed.The evidence is strongest for interventions implemented in general practice, where we identified a small randomised trial 19 as well as two large quasi-experimental studies (ITS) 17 18 and various uncontrolled studies and service evaluations.Evidence from care home settings was of lower quality with the exception of the CHIPPS study involving PIPs working in care homes. 23We located only one uncontrolled study based in UK community pharmacies. 26lthough the direction of reported effects was clear, the limited number of controlled studies combined with the wide range of outcomes reported makes it difficult to estimate the size of any effect.For example, the two ITS studies using similar interventions reported markedly different reductions in measures of inappropriate prescribing at 6 and 12 months after implementation of the intervention. 17 18Uncertainty about effect sizes is increased because many of the studies lacked a control group and the results could have been influenced by other interventions in the health and social care system, for example, the Enhanced Health in Care Homes programme implemented in England.While our review focused primarily on outcomes related to prescribing, data on cost savings were also widely reported but the evidence was generally of low quality.We also found limited evidence of a link between reductions in measures of overprescribing and clinical outcomes, mainly because of lack of reporting.The CHIPPS study found no significant difference in its primary outcome of fall rate, although there was a reduction in DBI (a secondary outcome) in the intervention group at 6 months. 31

Characteristics of effective interventions
The TIDieR Lite checklist provided a suitable structure for describing intervention characteristics for evidence synthesis purposes and this discussion follows its structure.Lack of reporting (especially of intervention intensity/frequency) was a limiting factor, as was reporting of varying intervention information across multiple publications.

Open access
Medication reviews were undertaken by pharmacists acting independently or in conjunction with GPs or care home staff.In a study in care homes for people with IDs, psychiatrists were also involved in review where appropriate. 25Pharmacy technicians were also involved in the PINCER study and could potentially have a greater role in relatively straightforward medication reviews. 13 18The included studies reported a variety of models of employment of pharmacists, including direct employment by GP practices, CCG Medicines Optimisation Teams, PIPs and community pharmacists.PCNs support employment of pharmacists by general practices and are the route chosen by NHS England to implement its model of SMR.
A major difference between settings is the need to identify patients requiring medication review in general practice, whereas most care home residents take multiple medications and could be considered candidates for review as part of their routine healthcare.A key element of the PINCER 13 18 and SMASH 17 interventions is the use of information technology to search electronic patient records efficiently across large numbers of general practices.Effective interventions were also characterised by attention to training and tools to support and sustain change in practice, for example, an 'audit and feedback' dashboard. 17raining of pharmacists and other staff to deliver interventions was reported to varying degrees, reflecting in part the publication channel of the research.For example, in the CHIPPS study PIPS had comprised 2 days of face-toface instruction plus time in practice to develop relationships with the GP and care home staff. 23Specification and provision of appropriate training will be important for future development of pharmacist-led interventions, as also highlighted by the evaluation of NHS England's SMR programme. 16ntervention intensity is another important factor in developing and delivering interventions.For the CHIPPS study, participating PIPs committed a minimum of 16 hours/month to the service. 23In general practice settings, NHS England recommended allowing 30 min for an SMR to give time for shared decision-making; this was interpreted to include time for preparation and writing up. 16his level of time requirement was also reported in the one study from a community setting, which estimated pharmacist time at 25 min for an initial consultation. 26n terms of intensity more generally, resourcing of interventions was reported to vary between commissioning groups (CCGs) depending on staff availability and other priorities. 13 18 21General practices varied in their use of a medication safety dashboard. 28Frequency of intervention was rarely reported, reflecting the short time frame of most included studies but it seems possible that there could be an ongoing need for review as patients get older and/or their health state changes.

Quality and risk of bias
The MMAT provided a good alternative to the use of multiple tools to assess risk of bias across diverse study designs.The only randomised trial assessed was designed as a pragmatic trial 19 and the assessment confirmed a relatively high risk of bias.Publications from the CHIPPS study were included but the trial per se was not assessed for risk of bias because of the publication date of the main study report.Similarly, the PINCER intervention was supported by a randomised trial published in 2012, before the cut-off date for our review 42 ).Well-conducted studies included in the review included large ITS studies, 17 18 process evaluations 13 14 23 and qualitative studies. 15 16Service evaluations and other lower-quality evidence tended to support higher quality studies by highlighting implementation and results achieved in routine practice, although a causal relationship between intervention and outcome remains uncertain in studies without a parallel control group.

Implementation barriers and facilitators
Implementation of pharmacist-led interventions was strongly influenced by factors affecting relationships between pharmacists and other health and care professionals, especially GPs.Given that most pharmacists are not prescribers, their recommendations around (de) prescribing need to be seen as 'legitimate' by GPs who are generally responsible for acting on the recommendations.This is facilitated by continuity at the system level, including existing links between pharmacists and GPs 23 and good access to data. 14Jeffries et al reported that pharmacists took the lead in developing relationships with GPs, enabling a 'learning health system'. 14The benefits of continuity at the system level could help to explain why early implementation of the SMR programme through the relatively new medium of PCNs was reported to be less successful than initially hoped. 16mplementation in care homes may be more complex than in general practice because of differences in systems and 'culture' between health and social care. 24Patients and their families may be supportive of medication review or oppose it based on real or perceived benefits of medication. 20he main message regarding implementation of pharmacist-led interventions across all settings is the need for involvement of all relevant stakeholders, preferably before starting the process of implementation, to understand the context and anticipate possible barriers. 24

Identification of effectively implemented interventions/initiatives
Our simple criteria for 'effectively implemented' interventions/initiatives identified a number of examples published as research papers, conference abstracts or case studies (see 'Effectively implemented interventions' above).9][40][41] In other studies, despite promising results, it was unclear whether the intervention would be implemented more widely. 19

Open access
Relationship to previous research To our knowledge, this is the first systematic review of pharmacist-led interventions and initiatives specifically in UK settings.A scoping literature search identified 20 systematic reviews published between 2014 and 2023.The most recent review covered pharmacist integration into general practice to optimise prescribing and outcomes for patients with polypharmacy. 43The review included 23 studies, of which just three were from the UK.The conclusion that pharmacist integration probably reduced PIP and number of medicines (moderate certainty evidence) was in line with the findings of the present review.A 2016 systematic review by Riordan et al focused on pharmacist-led interventions to optimise prescribing in older community-dwelling adults in primary care. 44The authors concluded that pharmacist-led interventions may improve appropriateness of prescribing but the quality of evidence was low.The review included randomised and quasi-randomised studies published before December 2015, giving it limited overlap with our review.

Strengths and limitations
The UK focus is both a strength and limitation of this review.We included evidence often excluded from systematic reviews to get as full a picture as possible of how pharmacist-led interventions are implemented and sustained in practice as well as their characteristics and effectiveness.The dual focus reflects the fact that pharmacist-led medicines optimisation and deprescribing in primary care is both an area of active research and of implementation within the healthcare system.Nevertheless, some of the evidence is not of high quality and we have tried to be appropriately cautious in our conclusions and identified implications.
Our broad review questions and UK focus resulted in a heterogeneous group of included studies.Meta-analysis was not possible so we performed a narrative synthesis in line with appropriate guidelines. 10 11The review was undertaken by a small but experienced team with expertise in systematic review methods and prescribing.

Implications for service delivery
Several studies indicate that barriers to successful service delivery often arise from 'system' issues and differences in 'culture'. 16 24Commissioners and providers engaged in developing new pharmacist-led services should ensure equitable access to data and information to avoid perceptions of 'ownership' by certain groups at the expense of others. 15In care homes, where medication review is an important component of healthcare for residents, 21 implementation requires health and social care professionals to work together and 'understand each other's systems'. 24The holistic patient-centred SMR envisaged by NHS England may require culture change/training to foster an emphasis on direct patient contact and shared decision-making.Removal of financial incentives for PCNs to carry out SMRs as reported recently (https://pharmaceutical-journal.com/article/news/nhs-england-removes-financial-incentives-for-structured-medication-reviews-in-2023-2024) may complicate delivery, although the service remains a contractual requirement.
Services have been delivered successfully through CCGs Medicines Optimisation Teams with suitable training. 13 18he review also found evidence that services provided by PIPs appear to be a valid alternative to approaches requiring action by GPs or other medical professionals. 23plications for research A major priority for research is to further evaluate the effectiveness of medication review in community pharmacy settings and how pharmacies might be best supported to deliver the service.A related need is for research to better understand public perceptions of community pharmacies as a setting for medication review and their pros and cons compared with alternative settings such as GP surgeries.
Research is needed to support the development of the PIP role and how PIPs might best be used in combination with GPs and other professionals to support optimal prescribing across the health and care system.
Shared decision-making is key to the success of pharmacist-led interventions.Research is needed to better understand patient and family attitudes to shared decision-making in the context of deprescribing and the barriers and facilitators operating in different settings and with different professionals.
The present review focused on outcomes related to prescribing and a review of effects on patient and health system outcomes would be a logical follow-up.Finally, further research is needed to understand the effects of implementing pharmacist-led medication review in general practice on health inequalities and how to reduce unwarranted variations in service delivery between different practices or regions.

Conclusions
The evidence base for pharmacist-led interventions varies widely in terms of quality but studies have consistently reported improvements relative to a comparator group or baseline.The diversity of interventions and outcomes reported makes it difficult to generalise about effect sizes but given the reported extent of the problem, even small relative reductions could be beneficial for patients and the health and care system.
The existing evidence base requires cautious interpretation because of a shortage of controlled studies and this is particularly the case for studies in community pharmacy settings.Further rigorous evaluation of interventions, particularly those delivered in community pharmacies, is required.Although not a focus of this review, there appears to be a shortage of high-quality economic evidence to guide decision-making by service commissioners and providers.
The problems encountered in the early implementation of NHS England's SMR programme 16 suggest a need for further research on the implementation of

Figure 1
Figure 1 PRISMA flow diagram.PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Table 1
Key search concepts

Table 2
Summary of research study characteristics

Table 2 ContinuedTable 3
Summary of studies reporting effects of interventions

Table 5
Summary of studies published as conference abstracts