Organising medication discontinuation: a qualitative study exploring the views of general practitioners toward discontinuing statins

Discontinuing medication is an important medical practice for a variety of reasons. It is a tool in reducing harms from taking drugs (e.g. side effects or drug-drug interactions), polypharmacy, reducing health system expenditure and improving evidence based medicine by discontinuing treatments that no longer have any evidence or indication [1, 2]. These challenges are compounded by an increasing prevalence of patients with several chronic diseases and several risk factors in high-income countries. These patients can easily have 20 indicated or recommended drugs on a medication list, and the challenge of prioritization and discontinuing medication, when necessary, becomes an important one [3]. Patients themselves have indicated they are interested in having more of their medication discontinued [4, 5]. There has been a growing interest in reducing the harms from taking drugs that has mainly focused on overdiagnosis and ‘the art of not doing’ [6–8]. We suggest that this interest in reducing harms caused by drugs would benefit from extending its focus to discontinuing medication.

Despite there being a tremendous focus on how to prescribe medication, there has been very little focus on the process of discontinuing medication [9, 10]. Patients are still being prescribed statins during end of life or palliative care, despite there being no benefit and some harm [11, 12]. Quantitative research has shown that discontinuation can be safe, that discontinuing multiple drugs is feasible and that it benefits patients and healthcare systems [1, 13, 14]. A feasibility study among elderly patients found that 4–5 drugs per patient could be discontinued, that 88 % of patients reported a global improvement in health and that only 2 % of patients needed to be restarted [14]. In another trial of discontinuation reductions in mortality, referrals to acute care and drug costs were reduced [1]. As a result of this research, there have been calls for a more proactive approach to discontinuing medication [15, 16].

However, there has been very little qualitative research that examines the mechanisms of discontinuing medication and how it is actually done in practice. Most of the qualitative research has primarily focused on the patient [17, 18], despite the general practitioner (GP) being an important actor. GPs (also known as family physicians) are the biggest prescribers of medication and are often those that have, willingly or unwillingly, the overall responsibility of a patient’s medication [3]. Therefore the process of discontinuing medication as a medical practice in primary care needs to be better understood.

Clinical decisions about discontinuation are often made in situations where the relation between causes and effects are unclear for individual patients. Such situations can be described as ambiguous, or a problem of too many interpretations that leaves the GP feeling uncertain about how to proceed [19]. For example, it may be unclear if the current treatment has any therapeutic value [20] or it may be unclear how to apply the clinical guidelines to a specific patient [3]. Whilst characteristics of ambiguous discontinuing situations have been described, this study is interested in how GPs respond to such situations as little is currently known about it.

Throughout the study we use the term ‘discontinuing’ rather than ‘deprescribing’ because it is a richer analytical description of the process. For example, patients cannot deprescribe their medication, but they can discontinue their medication. So even though this study’s focus is the professional, it is still important to allow for a patient’s agency in the process.

Interviews were on average one hour, recorded and transcribed afterwards. A pilot study of five interviews was done to develop and refine the interview guide. The interviews began by asking the GPs for recent examples of discontinuing statins so that they could be as concrete as possible about their reflections and action. If they could not give any recent examples, the themes from the next part of the interview guide were used. These included how the GPs kept themselves up-to-date with knowledge of prescribing and discontinuing and how they organised discontinuation in their practice.

Participant observations were done in three general practices. Each practice was observed for one day, totalling approximately 30 consultations. The observation days were chosen beforehand together with the GP so that these days were more likely to contain consultations involving discussions about medication. The GPs in the observed practices were interviewed beforehand, and in between consultations the GPs often explained their thought processes and their considerations of the preceding consultation. Notes were taken during the observations, and after the observations supplementary notes were recorded together with reflections on the observations of the day [31]. Participant observations generated data on the process of discontinuation and how it unfolded over time, and thus it complemented the interview data on the GPs’ reasoning and attitudes on discontinuation.

The method of analysis used was the Gioia method [32]. Here the data was coded inductively and two levels of data were developed: first order concepts and second order themes. The first order concepts generated were the initial codes and categories that closely reflect the original informant vernacular. Second order themes were developed by introducing the theoretical concepts of cues and attention to the first order concepts, so that analytical similarities and differences were grouped together and the themes offered a more theoretical perspective on the data. The study was inspired by sensemaking theory, a process based perspective that focuses on how people create meanings that shape their actions (and vice versa) [19, 33]. So the focus was on individuals rather than institutions. During the process of data collection and analysis, the two concepts of cues and ambiguity emerged as the most relevant theoretical concepts of sensemaking, and so were introduced to the data to further refine the findings. In this sense, the theoretical concepts acted as a ‘sensitising instrument’ that increased our sensitivity to analytically relevant data that otherwise may have been missed had if strictly inductive approach was used [34].

During the analysis we also looked for things that surprised or puzzled us in the data. For example, when GPs were asked about what prompted them to consider discontinuation, they often quickly and easily replied because of side effects. We could have stopped there and labelled ‘side-effects’ as cues, but we sensed that the ease with which they said it hid a lot of the work that had to go into generating this stable and medically accepted label ‘side effects’. Hence we wanted to find out more about how these ‘side effects’ arose as an issue in the first place and this eventually led to one of the study’s key themes (of patient and record cues). During the analytical process, the techniques of negative cases, constant comparisons and peer debriefing were used to improve the quality of the analysis [32, 35].

This description of discontinuing medication as something often talked about, but rarely done, was frequent among GPs.

And the GP sometimes forgot to discuss the possibility of discontinuation with the patients, especially if the patient did not mention it:

Sometimes you continue without thinking about it. The patient doesn’t come in and ask [for discontinuation]. They ask for the same medication. And you think they benefit from it. [GP6; p.4]

Thus, a mixture of a strong culture of prescribing and the difficulty of identifying relevant patients for discontinuing medication led to a sense of discontinuing medication being challenging to do.

This study contributes to the literature on discontinuing medication by exploring the perspectives of GPs. Generally, discontinuing institutionally recommended drugs, in this case statins, was found to be rare, but there were some GPs who discontinued systematically, despite the ambiguity surrounding the outcome. Nearly all instances of reported discontinuing had elements of dissonance. Those GPs that discontinued systematically created situations of dissonance deliberately. In doing so, they created a deeper culture of reflecting on the medication their patients take.

Creating this culture through situations of dissonance was done in at least two ways. Firstly, by systematically arranging a variety of different check-up consultations such as visits after hospital discharges or medication reviews of elderly home patients. Secondly by actively eliciting the patient’s experience of taking drugs, including the effects of individual drugs, the overall drug burden and patient’s concerns about taking specific drugs. The consequence of using both strategies was to produce more discontinuation cues, which in turn increased the likelihood of enacting discontinuation. This study complements previous research by showing the importance of using patient’s experiences in decisions to discontinue [20]. The study also shows that discontinuing medication is more likely to occur when organised proactively, rather than reactively, e.g. after a patient suffers from some kind of harm.

The reasons the GPs gave for discontinuing medication echo the reasons physicians give for abstaining from guideline recommended drugs [27]. For example the GPs sometimes choose to discontinue/abstain from treatment because of concerns about the harms of the drug, or because of problems with a high burden of treatment [27, 36]. We suggest that to frame the choice of GPs to discontinue guideline recommended drugs as an inability to follow guidelines with negative normative consequences or as ‘inertia’ is an unproductive one. Instead, a more fruitful approach could be to understand discontinuing medication as clinically relevant care and to acknowledge and support it as such. As one study on clinical inertia concluded, ‘much inaction may be clinically appropriate care’ [37]. Instead of castigating GPs for a lack of action, more effort should be directed towards understanding why GPs choose not prescribe a drug or to discontinue a drug.

Another interesting point from this study is how some GPs saw ambiguous situations as a reason to act and trial discontinuing with the patient over a period of time, rather than a reason not to act. This strategy of trialling to address the ambiguity of discontinuation has the effect of producing more cues for the GPs to use in their decision-making. This supports Weick’s [19] claim that reducing ambiguity requires access to more varied cues and the ability to argue for one interpretation over competing alternatives. In the case of discontinuing, GPs can use the information produced by the trial, e.g. if the harms disappear or not, to decide together with the patient if they should confirm or abort the ‘half-decision’ to discontinue.

Our findings suggest that the discontinuation of medication is a continually evaluative process (e.g. trialling, restarting at a later date) that depends on the patient’s situation. This approach is in line with the literature around situated judgment [9], that decision making draws on different kinds of justifications and is not solely determined by the amount of information available [38]. What these results show is how important it is that GPs feel more certain about the choice of discontinuing and that they can argue for that choice. So GPs can never act with complete certainty and discontinuing will remain an iterative process requiring a variety of strategies to deal with ambiguity.

The process of discontinuing medication was found to be a complex and ambiguous process for GPs in general practice. Discontinuation cues created situations of dissonance between the option of discontinuing or continuing the patient’s medication. If the high level of ambiguity meant the GP was in doubt about whether to discontinue or not, the strategy of trialling discontinuation for a period of time to generate more information before committing to the decision with the patient could be used. Suggestions for the future policy are centred on sustained reflection around discontinuing medication, by systematically flagging it up through a culture of systematically scheduled check-ups and generating information by explicitly and regularly eliciting the patient’s experience of taking drugs. The usefulness of these options needs to be explored more thoroughly if discontinuing medication is to reduce avoidable drug-related harms to patients and improve their quality of life.

CVD, cardiovascular disease; GP, general practitioner.