Statistics from Altmetric.com
Strengths and limitations of this study
This is the most comprehensive review to date of prescribers’ barriers and enablers to minimising potentially inappropriate medications which are chronically prescribed in adults.
Although database and manual searching was protracted and extensive, it is possible that not all relevant studies were found due to the poor indexing and inconsistent terminology for this topic.
Utilisation of a peer-reviewed, published method for thematic synthesis and a checklist to assess potential bias in studies contributed to the review’s methodological rigour.
The included studies largely explored primary care physicians’ perspectives on managing older, community-based adults in relation to relatively few drug classes and may limit the generalisability of the findings.
Studies in the USA and Australia indicate that at least one in two older people (aged 65 years or greater) living in the community use five or more prescription, over-the-counter or complementary medicines every day, and the number used increases with age.1 ,2 Polypharmacy (the use of multiple medications concurrently) predisposes older people to being prescribed potentially inappropriate medications (PIMs), that is, where the actual or potential harms of therapy outweigh the benefits.3–5 Recent international data suggest that one in five prescriptions for community-dwelling older adults is inappropriate.6 In Australia, approximately 20–50% of individuals in this age group are prescribed one or more PIMs, with higher rates seen in residential aged care facilities (RACFs).3 ,7–10 For adults younger than 65 years of age, rates of prescribing of PIMs have not been quantified beyond single medication classes (eg, benzodiazepines, proton pump inhibitors). The rates and harms of polypharmacy in this population remain uncertain, although they are likely to be considerably less than that seen in older adults. In contrast, the harms of polypharmacy and prescribing of PIMs in older people are well established. Prescribing of PIMs is independently associated with adverse drug events, hospital presentations, poorer health-related quality of life and death.11 ,12 Up to 15% of all hospitalisations involving older people in Australia are medication-related, with one in five potentially preventable.13
These well-documented harms of prescribing PIMs should evoke a response from clinicians to identify and stop, or reduce the dose of, inappropriate medications as a matter of priority. While there is some evidence that PIM exposure has decreased marginally over recent years, its prevalence remains high.3 ,14–16 The process of reducing or discontinuing medications, with the goal of minimising inappropriate use and preventing adverse patient outcomes, is increasingly referred to as ‘deprescribing’.17 Although the term may be new, appropriate cessation or reduction of medication is a long accepted component of competent prescribing.18 ,19
The act of stopping a medication prescribed over months to years, however, is complicated by many factors related to patients and prescribers. These need to be understood if effective deprescribing strategies are to be developed. A recent review by Reeve et al20 identified patient barriers to, and enablers of, deprescribing, but to the best of our knowledge, no comprehensive review of prescribers’ perspectives has been reported, which this paper aims to provide.
In the absence of a universally accepted method to conduct a systematic review of qualitative data, we utilised principles of quantitative systematic review, applied to qualitative research,21 and were guided by the Cochrane endorsed ENTREQ (Enhancing transparency in reporting the synthesis of qualitative research) position statement.22
Search strategy and sources
An initial search was conducted to ensure that no systematic review on the same topic already existed. Two experienced health librarians were independently consulted in developing a comprehensive search strategy, which was informed by extensive prior scoping.23
PubMed, EMBASE, Scopus (limited to Health Sciences), PsycINFO, CINAHL and INFORMIT (Health Collection) electronic databases were searched from inception to March 2014. Filters to identify qualitative research were used and adapted to improve search sensitivity.24 These were combined with terms and text words for: medical and non-medical prescribers and either inappropriate prescribing or reducing, stopping or optimising medications. Terms/text words were searched in all/any fields or restricted to the title, abstract or keyword, depending on the size of the database and sophistication of its indexing. Reference lists and related citations of relevant articles were reviewed for additional studies. The full search strategy is detailed in the online supplementary appendix.
After duplicate citations were excluded, one reviewer (KA) screened titles, abstracts and, where necessary, full text, to create a list of potentially relevant full text articles. Articles were required to meet provisional, intentionally overly inclusive, eligibility criteria to minimise the risk of inappropriate exclusions by the single reviewer. This list was forwarded to three reviewers (CF, DS and IS) who independently assessed the articles for inclusion. Discrepant views were resolved by group discussion to create the final list of included papers based on the refined eligibility criteria.
Inclusion and exclusion criteria
Inclusion criteria comprised: (1) original research articles with a qualitative component (ie, qualitative, mixed or multimethod studies all accepted); and (2) focus on eliciting prescribers’ perspectives of factors that influence their decision to continue or cease chronically prescribed PIMs (as defined by the authors of each study) in adults.
No limits were placed on the care or practice setting of the patient or prescriber, respectively, or whether the article related to single or multiple medications.
Exclusion criteria comprised: (1) reviews, papers not published in English, and those for which the abstract or full text were not available; (2) focus on medication management decisions in the final weeks of life; (3) focus entirely on initiation of PIMs and (4) reported only quantitative data derived from structured questionnaires.
Assessment of the quality of studies
One researcher (KA) assessed the reporting of studies using the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist. This reporting guideline, endorsed by the Cochrane Collaboration, assesses the completeness of reporting and potential for bias in studies of interviews or focus groups.25 Any instances of interpretive uncertainty arising from the checklist were discussed and resolved within the four investigators.
Studies were not excluded or findings weighted on the basis of the COREQ assessment. Rather, we elected to include all studies, ascribing to the theory that the value of insights contained within individual studies may only become apparent at the point of synthesis rather than during the appraisal process.26
Data extraction process
For all included articles, data were extracted about study aims, location, setting, study design, participants, recruitment, PIMs examined and prescribers’ perspectives of factors influencing the chronic prescription of PIMs. Data for thematic analysis were only extracted from the results (not discussion) section of papers, with particular notice taken of quotations from prescriber participants.
Synthesis of results
The method used to synthesise results was based on the technique of thematic synthesis described by Thomas and Harden.27 Following multiple readings of the papers to achieve immersion, KA manually coded and extracted the text, and developed subthemes until no further subthemes could be identified. Two reviewers (DS and IS) independently read all papers and then reviewed the extracted, coded text and subthemes to confirm the comprehensiveness and reliability of the findings.28 Descriptive and draft analytical themes were subsequently developed by KA and then presented to, and discussed with, all investigators in developing and finalising the new analytical construct. The study characteristics and results were analysed for associations between specific themes and studies.
The search yielded 6011 papers, 21 of which met the selection criteria (see figure 1). There were no studies exploring the perspectives of non-medical prescribers.
The characteristics of included studies are presented in table 1. All but one, which collected data by survey, used focus groups and semistructured interviews to collect qualitative data.29 Four papers explored prescribers’ views in relation to multiple medications (ie, polypharmacy)30–33 while the remaining papers investigated prescribers’ views in relation to single PIMs or classes of medications (10 described one or more centrally acting agents such as psychotropics, hypnotics, benzodiazepines, minor opiates and antidepressants34–43; 2 for proton pump inhibitors44 ,45 and 5 for miscellaneous PIMs defined according to prespecified criteria, a preset medication list or clinical judgement.29 ,46–49 Eighteen studies elicited the views of prescribers practising in primary care,29–41 ,44–48 one of the prescribers in secondary care,49 and two of the prescribers servicing RACFs.42 ,43
The completeness of reporting varied across studies, with an average of 17 (range 8–22) of 32 items from the COREQ checklist clearly documented (table 2). The single descriptive survey reported 9 of 24 applicable fields.29 See online supplementary table for the completed COREQ assessment for each study.
The lowest rates of reporting were observed in domain 1, meaning that researcher bias (poor confirmability) cannot be excluded.26 Greater transparency was apparent with domains 2 and 3 allowing comparatively better assessment of the credibility, dependability and transferability of study findings. For example, all studies reported the sample size and method and most reported a description of the sample and interview guide. There was consistency between raw data and interpretive findings in all papers except one in which the interpretation was so brief that its accuracy was considered doubtful.36 For five papers, it was unclear whether ethics approval was obtained.29 ,34 ,43 ,44 ,46
Synthesis of results
Thematic synthesis yielded 42 subthemes, 12 unique descriptive themes and 4 analytical themes (figure 2), with multiple interdependencies and relationships. Barrier and enabler descriptive themes and subthemes tended to mirror each other for each analytical theme of Awareness, Inertia, Self-efficacy and Feasibility. The first three themes reflect factors intrinsic to the prescriber and his/her decision-making process while the fourth deals with extrinsic factors. Tables 3 and 4 provide illustrative quotations from either primary study participants or study authors relating to barrier and enabler subthemes, respectively.
Fewer enablers were reported than barriers and there was variation in the relative contribution of each study to each theme.
Awareness refers to the level of insight a prescriber has into the appropriateness of his/her prescribing. This theme was apparent in the three papers which utilised audit or informal third-party (eg, other health professional) observation and feedback.46 ,47 ,49 Poor insight was an observed rather than reported barrier, with interventions to raise prescriber awareness an enabler to minimising the prescription of PIMs. Prescriber beliefs at a population level did not necessarily translate to prescribing practices at an individual level. For example, agreement among prescribers that benzodiazepines should not be used regularly or in the long term did not necessarily preclude such prescribing in individual patients.34 ,38 ,41
Inertia is defined as the failure to act, despite awareness that prescribing is potentially inappropriate, because ceasing PIMs is perceived to be a lower value proposition than continuing PIMs.
Fear of unknown/negative consequences of change featured in 15 of 22 papers, and related to consequences for: the prescriber (threatened therapeutic relationship, diminished credibility, increased initial and ongoing workload, potential for litigation, conflict with other prescribers/health professionals)29–31 ,34–36 ,38 ,40 ,43–47 ,49; the patient (withdrawal syndrome, symptom relapse or increased risk of the condition/event for which preventive medication was originally prescribed)36 ,38 ,40 ,42–47 and other health professionals (increased workload and safety concerns of staff in RACFs).42 ,43 The prescriber beliefs that facilitate cessation were the converse, that is, fear of unknown/negative consequences of continuation,44 a positive attitude to stopping medicines31 and a belief that this practice can bring benefits.36 ,37 ,48
The barrier belief that drugs appear to work with few adverse effects was apparent in nine papers34 ,35 ,38 ,39 ,41 ,43–45 ,47 of which two studied ‘high-rate’ and ‘low-rate’ benzodiazepine prescribers. ‘High-rate’ prescribers consistently downplayed risks of harm, whereas ‘low/ medium-rate’ prescribers were more conscious of such risks.34 ,41 The futility and potential harm of cessation in patients of advanced age was a subtheme predominantly present in papers considering psychoactive agents.34 ,35 ,38 ,43 ,46 ,47
Another barrier was the devolvement of responsibility to another party for the decision to continue or cease a medication (eg, another prescriber, health professional, society or the patient). One example was continuation of PIMs in patients that prescribers had inherited from colleagues where the former failed to question the rationale used by the latter in prescribing such drugs.29 ,34 ,41 ,49 Another example was the provision of PIMs on the request of RACF nursing staff42 or patients34 ,40 ,43 without a critical prescriber review. Finally, inappropriate prescribing of psychotropics, while viewed as a public health concern, was considered beyond the scope of individual prescribers.35
This analytical theme refers to factors that influence a prescriber's belief and confidence in his or her ability to address PIM use. It involves subthemes relating to knowledge, skill, attitudes, influences, information and decision support.
Knowledge or skill deficits,30–35 ,40 ,45 ,49 including difficulty in balancing the benefits and harms of therapy,30–33 recognising adverse drug effects31 ,32 and establishing clear-cut diagnoses/indications for medicines,34 ,35 ,40 were challenges prescribers faced in identifying and managing PIMs. Balancing the benefits and harms was perceived to be especially difficult when reviewing preventive medications in multimorbid older people with polypharmacy where shorter life expectancy, uncertain future benefits and higher susceptibility to more immediate adverse drug effects must all be considered.30–33 On the other hand, better quantification of the benefits and harms of therapy,30–32 ,48 confidence to deviate from guidelines and stop medications if thought necessary,33 ,45 greater experience,30 ,45 and targeted training, especially in prescribing for older people,49 were seen as enabling factors.
Compounding generic knowledge and skill gaps were information deficits specific to individual prescribing decisions, resulting from poor communication with multiple prescribers and specialists involved in patient care, inadequate transfer of information at care interfaces, fragmented and difficult-to-access patient medical records, and failure of patients to know/disclose their full medical history/medication lists to prescribers.30–33 ,40 ,41 ,46 ,47 ,49 This subtheme linked strongly with time and effort demands on prescribers, and in two papers was associated with low motivation arising from a perceived inability to efficiently access all information required for optimal prescribing.40 ,49
Eight papers discussed the influence of care recommendations from guidelines and specialists.30–33 ,38 ,44 ,46 ,49 Guidelines were often viewed negatively, with prescribers feeling pressured to comply with recommendations at odds with the complexities of clinical practice.30–32 ,44 ,46 Pressure from staff to continue prescribing PIMs, often to maintain facility routines, was presented as a barrier unique to RACFs.42 ,43 Offsetting this were enablers centred on greater dialogue with patients to increase understanding and facilitate shared decision-making,29 ,30 ,31 ,44 ,46 as well as timely access to, and decision support from, specialists, particularly geriatricians and psychiatrists.37 ,40 ,41 ,44 ,46 ,49
Feasibility refers to factors, external to the prescriber, which determine the ease or likelihood of change. They relate to patient characteristics, resource availability, work practices, medical and societal health beliefs and culture, and regulations.
The most frequently expressed barrier concerning patients was their ambivalence or resistance to change29–32 ,35 ,37 ,38 ,40 ,43 ,44 ,46 ,48 ,49 and their poor acceptance of alternative therapies.37 ,38 ,42–44 In contrast, receptivity and capacity to change were identified as enablers in three studies,33 ,37 ,46 as was a poor prognosis which helped crystallise care goals and prompt review of the appropriateness of existing drug regimens.49
The limited time and effort to review and discontinue medications30 ,33 ,34 ,37 ,38 ,40–42 ,46 ,48 ,49 was the most common resource constraint followed by the limited availability of effective non-drug treatment options.35 ,37 ,38 ,41–43 Adequate reimbursement38 and access to support services such as mental health workers and pharmacists for medication review31 ,37 ,41 ,46 emerged as enablers.
Certain work practices were raised as barriers to deprescribing, such as provision of repeats for a prescriber's own or colleague's patients,34 ,46 ,47 and the absence of explicit treatment plans or a formal or scheduled medication review.34 ,43 The mirroring enablers were opportunities to review medication regimens (eg, hospital admission,29 ,49 change of prescriber,31 specialist40 or scheduled review).44 ,48
The remaining descriptive themes related to medical and societal health beliefs and cultural and regulatory factors. The most frequently mentioned barrier was discomfort and reluctance to question a colleagues’ prescribing decisions29 ,30 ,34 ,37 ,45 ,46 ,49 associated with respect for professional autonomy or the medical hierarchy when specialist prescribers were involved.
Externally imposed guideline-based quality measures were presented as a barrier to minimising the prescription of PIMs.33 Raising the prescribing threshold for medications (eg, through increased cost or restricted access)44 ,45 and monitoring by authorities34 were seen by prescribers as unwelcome, perverse enablers.
This systematic review comprehensively investigates prescriber barriers and enablers to minimising the prevalence of chronically prescribed PIMs in adults. The thematic construct which was developed from published literature centres on Awareness, Inertia, Self-efficacy and Feasibility. It principally reflects the perspectives of primary care physicians managing older, community-based adults. Although the themes and subthemes have been presented separately, the reasons doctors continue to prescribe, or do not cease, PIMs are multifactorial, highly interdependent and impacted by considerable clinical complexity.
Many subthemes were common to papers regardless of interstudy differences in the PIMs discussed, patient age and clinical setting (eg, primary, secondary or residential aged care).
Subthemes varied according to whether studies focused on polypharmacy or single PIMs or classes of PIMs, which were also associated with differing levels of prescriber insight and certainty. In the four studies focused on polypharmacy, prescribers were aware of polypharmacy-related harm but could not easily identify which medications were inappropriate, as reflected by the subthemes ‘difficulty/inability to balance benefits and harms of therapy’,30–33 ‘inability to recognise adverse drug effects’,31 ,32 ‘lack of evidence’30 ,31 ,33 and ‘incomplete clinical picture’.30–33 In other studies focusing on specific classes of overprescribed medications, prescribers were aware of this inappropriateness, but in response voiced various rationalisations for continued prescribing such as ‘drugs work, few adverse effects’,34 ,35 ,38 ,39 ,41 ,43–45 ,47 ‘prescribing is kind and meets needs’,34 ,37–41 ,43 ,44 ‘stopping is difficult, futile, has or will fail’,34 ,36–38 ,42 ,43 ,47 ‘poor (patient) acceptance of alternatives’,37 ,38 ,42–44 and ‘difficult and intractable adverse (patient) circumstance’.34 ,35 ,37 ,39 ,40
However, in other studies focusing on miscellaneous PIMs, prescribers were generally not aware of their inappropriate prescribing until this was revealed to them (eg, through audit and feedback).46 ,47 ,49
No definite thematic pattern was observed from the subthemes of six studies which did not specifically focus on the care of older people29 ,37 ,39 ,41 ,44 ,45 compared with the remaining 15 which did. Compared with studies in primary care, unique themes emerged from papers set in RACFs and acute care settings. For example, pressure on prescribers to continue prescribing PIMs at the request of RACF nursing staff was unique to this setting.42 ,43 The one study set in acute care highlighted inexperience and training deficiencies of junior prescribers, as viewed by three geriatricians.49
The finding that poor insight into potentially inappropriate prescribing (PIP) practices was only apparent in studies where prescribers were made aware of this is unsurprising given that prescribers do not intentionally prescribe medications inappropriately. It demonstrates the importance of awareness-raising strategies for prescribers. Inertia, as in failure to deprescribe when appropriate, sits at odds with the more traditional use of the word as symbolising failure to intensify therapy when indicated.50 Inertia has been linked to ‘omission bias’ where individuals deem harm resulting from an act of commission to be worse than that resulting from an act of omission.51 ,52 In the case of deprescribing as an act of commission, it becomes more a matter of reconciling a level of expected utility (accrual of benefits) with a level of acceptable regret (potential to cause some harm).53 Fear of negative consequences resulting from deprescribing contributes to inertia and is not easily allayed by the current limited evidence base regarding the safety and efficacy of deprescribing.54 In the same papers in which prescribers rationalised continuation of therapy with the belief that drugs work and have few adverse effects,34 ,35 ,38 ,39 ,41 ,43–45 ,47 prescribers also identified different thresholds for initiating versus continuing the same therapy. This anomaly suggests a lack of prescriber insight, clear differences in prescribers’ attitudes towards initiation versus continuation, or a social response bias towards a false belief induced by the methodology used by interviewers.
Relevance to previous literature
One meta-synthesis of seven papers has recently been published online exploring prescribers’ perspectives of why PIP occurs in older people.55 Compared with our review, this study had a generic focus on PIP, including underprescribing, and its search strategy retrieved fewer articles (n=7). Scanning their reference list did not reveal any additional papers which would have met our selection criteria and their results yielded no additional themes.
Our findings are consistent with those in the literature (largely focused on initiation of therapy), suggesting that pharmacological considerations are not the only factors impacting doctors' prescribing decisions.56 Interacting clinical, social and cultural factors relating to both the patient and prescriber influence prescribing decisions.56–58
Reeve et al20 recently published a review of patient barriers and enablers to deprescribing and have emphasised the importance of a patient-centred deprescribing process.59 When comparing their results with ours, we find that prescribers’ barriers are concordant with those of patients with respect to resistance to change, poor acceptance of non-drug alternatives, and fear of negative consequences of discontinuation. However, prescribers also underestimate enabling factors including patients’ experiences/concerns of adverse effects, dislike of multiple medicines, and being assured that a ceased medication can be recommenced if necessary. Patients also reported that their primary care physician could be highly influential in encouraging them to discontinue therapy, a perception not echoed among prescribers in this review.20 Prescribers need to discuss, rather than assume, patient attitudes towards their medicines and to deprescribing, in the context of their current care goals.
Previous reviews of interventions to reduce inappropriate prescribing/polypharmacy in older patients have not been able to conclude with certainty that multifaceted interventions are more effective than single strategies.60 ,61 Although our findings suggest that the former are likely to be more successful, further research is required to identify the barriers and enablers with the greatest potential for impact in designing targeted deprescribing interventions.
Strengths and limitations
Inconsistent terminology and poor indexing of search terms relating to deprescribing and inappropriate therapy greatly hampered our ability to identify relevant studies. Our mitigation efforts comprised a comprehensive prescoping exercise, a highly iterative search strategy tailored to each database, and snowballing from reference lists and related citations.
Despite no search restrictions on patient age, clinical setting or type of PIM, most study participants were experienced primary care physicians caring for older, community-based adults. Caution is therefore needed when transferring our results to other settings or patient groups. However, two recent cross-sectional studies looking at barriers to discontinuation of benzodiazepines and antipsychotics in nursing homes reflected subthemes identified in our review—fear of negative consequences of discontinuation such as poorer quality of life, symptom recurrence, greater workload and a lack of available, effective, non-drug alternatives.62 ,63
Many of the papers focused on relatively few drug classes (psychotropics and PPIs) and only four focused on polypharmacy. Although some subthemes were common to all types of studies (single and miscellaneous PIMs and polypharmacy papers), others were not. It is possible that, had more medication classes been studied, some of our results may have been different.
The strengths of our review include adherence to a peer-reviewed, documented methodology for thematic synthesis, COREQ assessment of studies allowing the assessment of potential for bias, compliance with ENTREQ reporting requirements and a multidisciplinary team of investigators to validate theme identification and synthesis.
Implications for clinicians and policymakers and future research
The results of this review disclose prescriber perceptions of their own cognitive processes as well as patient, work setting and other health system factors which shape their behaviour towards continuing or discontinuing chronically prescribed PIMs. The thematic synthesis provides a clear conceptual framework to understand this behaviour. Rendering these issues visible for both clinicians and policymakers is the first stage in minimising inappropriate prescribing in routine clinical practice. It facilitates what has been lacking in deprescribing intervention studies to date—a pragmatic approach towards identifying and accounting for local barriers and enablers which will determine the overall effectiveness of targeted interventions.
Further high-quality prospective clinical trials are urgently needed in demonstrating the safety, benefits and optimal modes of deprescribing, especially in relation to multimorbid older people.61 ,64 The fog of polypharmacy clouds a prescriber's capacity and confidence to identify PIMs which, to be overcome, requires complete and accurate clinical information and decision support.
Professional organisations and colleges have an important role in encouraging the necessary cultural and attitudinal shifts towards ‘less can be more’ in appropriate patients. The push for guideline adherence and intensification of therapy needs to be counterbalanced by the view that judicious reduction, discontinuation or non-initiation of medication, in the context of shared decision-making and agreed care goals, is an affirmation of highest quality, individualised care.65 This view needs to be embraced in the education and training of all health professionals, not just doctors, who influence the prescribing process.
Prescribers are making decisions in the face of immense clinical and health system complexity. Appropriate deprescribing needs to be regarded as equally important and achievable as appropriate initiation of new medications. Understanding how prescribers perceive and react to prescribing and deprescribing contexts is the first step to designing policy initiatives and health system reforms that will minimise inappropriate overprescribing.
The authors thank the University of Queensland librarians Mr Lars Eriksson and Ms Jill McTaggart for their assistance in developing the search strategy and Ms Debra Rowett for her invaluable insights when scoping the search and developing the manuscript.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.