Falling on stony ground? A qualitative study of implementation of clinical guidelines’ prescribing recommendations in primary care

Abstract

Objectives

We aimed to explore key themes for the implementation of guidelines’ prescribing recommendations.

Methods

We interviewed a purposeful sample of 25 participants in British primary care in late 2000 and early 2001. Thirteen were academics in primary care and 12 were non-academic GPs. We asked about implementation of guidelines for five conditions (asthma, coronary heart disease prevention, depression, epilepsy, menorrhagia) ensuring variation in complexity, role of prescribing in patient management, GP role in prescribing and GP awareness of guidelines. We used the Theory of Planned Behaviour to design the study and the framework method for the analysis.

Results

Seven themes explain implementation of prescribing recommendations in primary care: credibility of content, credibility of source, presentation, influential people, organisational factors, disease characteristics, and dissemination strategy. Change in recommendations may hinder implementation. This is important since the development of evidence-based guidelines requires change in recommendations. Practitioners do not have a universal view or a common understanding of valid ‘evidence’. Credibility is improved if national bodies develop primary care guidelines with less input from secondary care and industry, and with simple and systematic presentation. Dissemination should target GPs’ perceived needs, improve ownership and get things right in the first implementation attempt. Enforcement strategies should not be used routinely.

Conclusions

GPs were critical of guidelines’ development, relevance and implementation. Guidelines should be clear about changes they propose. Future studies should quantify the relationship between evidence base of recommendations and implementation, and between change in recommendations and implementation. Small but important costs and side effects of implementing guidelines should be measured in evaluative studies.

Introduction

It took 50 years for the British navy and 120 years for the merchant fleet to use lemon juice for the prevention of scurvy [1]. Nowadays many innovations take much less to be implemented. Health professionals are keen to implement ‘new’ advances, but this often means using more expensive interventions with limited advantage or questionable effectiveness. The dilemma of evidence-based health care starts here: What are the best ways of helping health professionals keep abreast of new innovations, while doing it in a ‘conscientious, explicit, and judicious’ manner [2] i.e. avoiding innovations with questionable effectiveness and efficiency?

The WHO refers to the ideal state of prescribing and use of medicines as ‘rational use of drugs’:

The rational use of drugs requires that patients receive medications appropriate to their clinical needs, in doses that meet their own individual requirements for an adequate period of time, and at the lowest cost to them and their community (WHO conference of experts, Nairobi 1985, cited in [3]).

Inappropriate use of drugs can result from a variety of situations. Prescribed drugs may not be appropriate for the patients’ needs; they may be expensive; or may not be acceptable for personal, cultural or social reasons. Clinicians may prescribe medicines of no value because of perceived patient pressure or placebo effects [4]. They may also prescribe where medication does not provide any benefit over ‘wait and see’ approach. Irrational prescribing may also be the result of under prescribing of required medicines.

In the last two decades improving quality of prescribing has attracted a lot of interests, not least because an important fraction of health care resources were (and increasingly are) spent on it [5]. Prescribing costs have been growing 6–8% per year in the global context [3]. Most prescribing costs happen in primary care [6]. Soumerai et al. estimated that about 75% of visits to office-based doctors end up in prescribing [7]. In 1995, GP prescribing amounted to 11% of total NHS spending [8]. Focusing on quality and cost of prescribing in primary care is important and vital. In particular if one considers that inappropriate prescribing often results in significant morbidities and costs for patients, societies and health systems.

Clinical guidelines are sought as tools for reducing variation in health care and cost [9], [10] and improving quality of patient care [11] including prescribing behaviour. Increasingly the research findings are summarised in guidelines and a new industry has appeared concerned with guideline development and implementation [12]. The majority of guidelines have not been through rigorous production processes, making it more difficult for clinicians to follow their recommendations [13], [14]. It is even claimed that the quality of guidelines is declining [15]. Even guidelines developed by the WHO may not have been rigorously produced [16]. Also where there are no effective intervention plans, even good quality guidelines are most likely to be ‘words without action’ [17]. In reality, many guidelines have no clear implementation plans [14]. Evaluation of guideline implementation programmes involves careful planning and requires dedicated resources. Multi-stage studies using qualitative and quantitative methods have been recommended [18].

Despite the phenomenal works conducted on understanding effective ways of changing provider behaviour [19] and use of clinical guidelines in behaviour change [20], [21], our understanding of how to improve provider behaviour “is still in its infancy” [22]. For some generic questions, there may never be a concrete answer, e.g. what is the best method of implementing guidelines? Although it is argued for long that single strategies are less likely to change clinical practice [23], this is not much comfort. Multi-faceted interventions are costly and more difficult to implement. Why are some guidelines implemented more successfully than others? Is it because of the difference in quality of guidelines, settings, clinical conditions or dissemination strategies? Previous studies suggested that enthusiastic clinicians or ‘innovators’ achieved more with poor guidelines than what others achieved with better quality guidelines [24]. Also some dissemination strategies (e.g. educational outreach visits) work well in influencing prescribing in some settings [25] and not in other settings [26].

Many clinical guidelines fail to improve practice and a lot of inappropriate variation exists in cost and quality of care. Despite the availability of evidence, changes in prescribing are delayed and require planning and investments. For example, clear-cut messages for antibiotic prescribing are still to be implemented [27] and although there are numerous studies on how to improve prescribing, all reasons behind ‘nonscientific prescribing’ are not known [28].

Historically, little work has been done on how to influence general practitioners’ (GPs’) practice [29]. Most of behaviour change interventions are based on the naïve assumption that clinicians will change if they are given information (‘production-dissemination’ [30] or ‘information deficit’ [31], [32] models). According to these models, interventionists endeavour to improve quality of care by providing more information on safety, efficacy and cost-effectiveness of intended behaviours. The inevitable outcome of this approach has featured in abundance of information delivered to medical practitioners [33].

While there is evidence that neither demographic and social characteristics [34] nor organisational variables [35] predict physician prescribing, few studies have tested the theoretical models that acknowledge psychological processes which precede behaviour [32], [36]. Focus on GPs’ attitudes and personal beliefs may provide a better insight of prescribing. The importance of identifying underlying personal reasons for nonscientific prescribing is documented in the literature. A multi-centre trial studied physicians’ expressed reasons for nonscientific prescribing’ and concluded that ‘greater attention must be paid to physicians’ attitudes and motivations concerning suboptimal prescribing if programmes are to succeed in replacing these practices with more rational clinical decision making’ [31] (p. 577).

Underlying beliefs are better captured through qualitative studies [37]. Interviews could be particularly useful in identification of barriers to guideline implementation and in understanding how those barriers affect individual physicians [38]. A study found that GPs spoke freely of their prescribing and were ready to offer embarrassing observations during the interviews. They demonstrated that interviews were useful for studying prescribing in primary care. The study identified three models of change in prescribing: accumulation model, challenge model and continuity model [39]. Another British study of GPs’ and consultants’ reasons for change in clinical practice found that education was important in changing prescribing [40]. A qualitative study of 24 Scottish GPs assessed statins prescribing using semi-structured interviews. It concluded that GPs rarely critically appraised trial results, but evaluated the results in terms of their social and economic implications [41]. It suggested that compiled sources of evidence (e.g. evidence-based guidelines) could play a useful role. It implied that GPs were likely to rely on the trustworthiness of the source of the guidelines, and that GPs considered the implications of implementing clinical guidelines. Another qualitative study of prescribing, performed on 17 GPs, tried to identify variables that explained prescribing variance [42] and concluded that consultant's prescribing was among those variables. Salisbury and colleagues used qualitative and quantitative methods to audit three changes in prescribing (use of warfarin or aspirin for atrial fibrillation, angiotensin converting enzyme inhibitors for heart failure and substitution of trimethoprim for co-trimoxazole) and concluded there were no effects from clinical protocols [43]. They reported positive views of participants towards clinical protocols, but found that many did not use protocols [43].

There have been further qualitative studies to elicit broad issues such as ‘GPs views on use of guidelines’ or ‘their attitudes towards evidence-based medicine in general’ [44], [45]. These issues are too broad and the researchers would be more likely to get general statements in response. In contrast others studied specific diseases [41], [46], [47] and hence lacked the spectrum to compare different clinical conditions, guidelines and implementation strategies.

The qualitative study reported in here was conducted as part of a multi-part project that also included a systematic review of effective interventions to improve prescribing in primary care, and two national surveys of GPs’ prescribing intentions and beliefs and their actual prescribing [48], [49], [50]. For the qualitative study we focused on a small set of clinical conditions. This selective approach could provide the opportunity of comparing different clinical conditions, clinical guidelines and dissemination strategies [51]. We aimed to explore key themes for the implementation of prescribing recommendations, GP attitudes and beliefs about guidelines, and barriers to and facilitators of implementation.