Mapping De-Implementation Strategies To Identied Determinants of Low-Value Statin Prescription for Primary CVD Prevention in Primary Care

Despite clear recommendations supporting healthy lifestyle promotion interventions for the primary prevention of CVD in low-risk patients, a considerable number of these people continue to receive inappropriate statin prescriptions. The present study reports on the structured process based on theory and evidence carried out for the design of de-implementation strategies to reduce the inappropriate prescription of statins and to increase the promotion of healthy lifestyles, in CVD prevention practice of primary care professionals for patients with low cardiovascular risk. A phase I formative study following a structured theory-informed process combining the Theoretical Domains Framework (TDF) and the Behavior Change Wheel (BCW) was conducted, comprising: semi-structured interviews (n=5) with primary care professionals to delimitate and dene the problem in behavioral terms; focus groups (4 groups with 21 physicians; 1 group with 6 patients) to identify the determinants of potentially inappropriate prescribing [PIP] of statins and healthy lifestyle promotion actions; mapping of behavioral change interventions operationalized as de-implementation strategies for addressing identied determinants; and consensus techniques for the prioritization of strategies based on perceived effectiveness, feasibility and acceptability. corporate accompanied for an audit/feedback system regarding CVD and helping and decision support tool incorporated into the CVD calculator in electronic clinical

the PRECEDE-PROCEED model [12], Implementation Mapping [13], or the Behavior Change Wheel (BCW) [14]. Though there is a wealth of recent literature on the development of intervention or implementation strategies to facilitate the uptake of innovative or evidence-based practices, the application of behavioral science theory for the development of de-implementation interventions is scarce [15][16][17].
The DE-imFAR (from the Spanish for DE-implementation of low-value pharmacological prescribing) study aims to carry out a structured, evidence based and theory-informed process involving the main stakeholders (health managers, professionals, patients, and researchers) for the design, deployment and evaluation of targeted de-implementation strategies for reducing low-value pharmacological prescribing [18]. Speci cally, the low-value practice in question is the pharmacological prescription of statins in primary prevention of cardiovascular disease (CVD) in low-risk patients (<10% CVR according to the REGICOR equation). Based on the evidence and the practice recommendations established by this evidence, it is recommended not to start treatment with statins in this population; further, in primary prevention of CVD, the promotion of healthy lifestyles through diet, physical activity and cessation of smoking should remain the preferred activities [19,20]. Despite these clear recommendations, the consumption of statins in the Basque Country has grown in recent years caused, in part, by an approach to cardiovascular risk (CVR) focused on the control of lipid levels and the use of medications. Thus, according to the descriptive observational study carried out by our group as part of the DE-imFAR project based on data in electronic health records (EHR) on the inappropriate prescription rate of statins in patients aged 40 to 75 years, without CVD, with high cholesterol but a CVR <5% (REGICOR), a prevalence of potentially inappropriate prescribing [PIP] for statins of around 16% and an incidence of new inappropriate prescriptions of 12.6 per 100,000 people/year has been estimated.
Furthermore, 60% or more of the EHRs of these people with PIP do not have a record of being given advice on physical activity or a healthy diet. Likewise, 49% did not receive preventive advice on giving up smoking [data not published].
In this paper we will report on phase I of the De-IMFAR study, which aims to design and develop de-implementation strategies to favor the reduction and/or abandonment of low-value prescribing of statins in primary prevention of CVD, addressing the main determinants (barriers and facilitators) of this clinical practice in the speci c context of the Basque Health care system (Osakidetza). To achieve this objective, De-imFAR applies a systematic, comprehensive and evidence-based framework for the design of de-implementation strategies, the Theoretical Domains Framework (TDF) [21,22] and the BCW [14]. Speci cally, this rst phase is based on conducting a formative research process to: i) understand the problem of statin PIP in primary prevention of CVD and de ne it in behavioral terms; ii) identify the main determinants of this clinical practice (e.g. at personal, inter-personal, organizational, social level) that must be addressed to change this behavior, and iii) map potential de-implementation strategies. In a second phase of the study, the feasibility and effectiveness of one or more de-implementation strategies will be evaluated to reduce the PIP of statins and strengthen the recommended clinical practice, of promoting healthy habits, in primary prevention practice on CVD in low-risk patients by health professionals and Primary Care teams. A working group composed of experts in the design of implementation strategies, methodologists, pharmacists, qualitative researchers, clinicians and health service managers carried out a structured and theory-informed process -speci cally, the BCW [14,23] -to map implementation strategies seeking to address the key determinants of low-value pharmacological prescribing in CVD primary prevention. In short, the BCW was used to identify, select, adapt and de ne possible behavioral change interventions operationalized as de-implementation strategies to address the prioritized determinants of PIP of statins in CVD primary prevention. This process involved eight steps grouped into the three following stages:

Methods
1st stage -Understand the behavior step 1) de ne the problem in behavioral terms; step 2) select the target behaviors; step 3) specify the target behaviors; and step 4) identify what needs to change.
2nd stage -Identify intervention options step 5) select intervention functions; step 6) select the speci c behavior change techniques. 3rd stage -Identify implementation procedures step 7) select strategies and speci c intervention techniques; step 8) select the execution mode of the intervention.
1st stage -Understand the behavior: step 1) de ne the problem in behavioral terms; step 2) select the target behaviors; step 3) specify the target behaviors; step 4) Identify what needs to change With the goal of de ning the problem in behavioral terms (step 1), we conducted a semi-structured interview with a sample of Family Physicians (FPs) (n=3) and Practice Nurses (n=2) in order to identify the overall behavioral scenario and break down the chain of behaviors and concomitant non-behavioral (e.g., contextual) elements. The interview script was centered on determining how physicians address and manage the clinical encounters related to CVD prevention, and what the main steps taken are. Three members of the working group independently reviewed the recordings of the interviews, and identi ed and proposed a set of possible target behaviors. Subsequently, based on the information compiled from the interviews and using matrices and exercises proposed by the BCW [23], the working group proceeded to vote and discuss until agreement in order to select (step 2) and to specify (step 3) the nal target behaviors most likely to lead to the desired behavior change, to be the focus of the next steps.
In order to explore the practice determinants of statin PIP in primary prevention of CVD and of the previously identi ed and selected target behaviors in the context of Osakidetza, a qualitative study comprising two sub-studies was designed and carried out: focus groups with FPs and one focus group with patients. The main goal of this qualitative inquiry process was to identify the main facilitators of the selected target behaviors related to the low-value pharmacological prescribing practice and the barriers to the provision of recommended clinical practice for primary prevention of CVD, i.e. the systematic and active promotion of healthy lifestyles.
For the recruitment of the healthcare professionals, emails were sent to all the Health Centers in the Ezkerraldea-Enkarterri-Cruces (n = 83) and Barakaldo-Sestao (n = 123) areas with a brief explanation of the project and the invitation to participate in it. Of the total number contacted, it was possible to recruit 21 FPs. Four focus groups were developed, two for each HD, with between 4 to 7 attendees in each group. Moreover, patients with statin PIP were identi ed in the patient lists of the participating professionals (n = 11), and consent for contacting them was requested from the doctor they were registered with. Finally, a discussion group made up of 6 patients who agreed to participate was held.
The groups were led by two researchers with experience in qualitative research methods, as well as knowledge of the clinical eld and the study objectives. The focus groups were audio recorded and transcribed verbatim. Informed consent for all participants was obtained prior to any research procedure. The script of the focus groups explored in-depth potential determinants with questions formulated to cover each of the TDF dimensions [22], and it was developed by researchers with experience in behavioral change and implementation research, together with clinicians with expertise in the primary prevention of CVD in primary care settings. Further, in order to facilitate the analysis, a coding scheme regarding the TDF dimensions and their relative constructs was also developed. Two researchers independently reviewed and coded the transcripts and iteratively discussed possible discrepancies until reaching a consensus. As a result, facilitators and barriers for PIP and the provision of advice in promoting healthy habits were identi ed in the group of professionals. In the case of patients, the following aspects were explored: how the pharmacological treatment was started; if it was a decision made in conjunction with the doctor; how they were informed; what factors could determine this action (preference or health problem, and at patient, professional, health center level), patient comfort with treatment, and so on.
2nd stage -Identify intervention options: 5) select intervention functions; 6) select the speci c behavior change techniques.
The goal at this stage was to identify the behavior change techniques for each of the agreed determinants of selected target behaviors in our context. Two researchers from the working group proceeded to map each of the identi ed barriers and facilitators grouped in the TDF domains with intervention functions, policy categories, and behavior change strategies most likely to produce a change, using the process established by the BCW [23].
3rd stage -Identify implementation procedures: 7) select strategies and intervention techniques; 8) select the mode of execution of the intervention.
Final de nition, packaging and selection of previously identi ed de-implementation strategies were carried out through a participatory consensus process, involving the working group as representatives of the main stakeholders. Further, a clear layout of the techniques to be applied (i.e., the actual content of the interventions, their possible formats and modes of execution) was speci ed for each of the possible interventions identi ed through this structured mapping process. In addition, and with the aim of helping to determine the format and mode of delivery of identi ed de-implementation strategies, examples of interventions and strategies identi ed in the literature were studied.
Finally, in order to prioritize the de-implementation strategies derived from the mapping conducted, a poll process using the LimeSurvey platform involving FPs who collaborated in the Focal Groups was carried out. Speci cally, they assessed the potential effectiveness, acceptability and feasibility of each identi ed strategy. Those considered potentially effective while highly acceptable and feasible for enacting behavior change were prioritized as the nal set of speci c strategies, to be contained in at least one broad de-implementation strategy seeking to reduce low-value pharmacological prescribing in the primary prevention of CVD.

Results
1st stage -Understand the behavior.
Steps 1, 2 and 3. De ne the problem in behavioral terms, select the target behaviors and specify the target behaviors.
A set of 5 semi-structured interviews with FPs (n=3) and practice nurses (n=2) with recognized expertise and experience in CVD prevention was carried out with the purpose of de ning the problem in behavioral terms and generating a list of candidate target behaviors. In short, physicians were asked to describe in detail the steps taken when addressing CVD primary prevention clinical encounters. After reviewing the recordings and performing the exercise established by the BCW, the working group rst de ned the target behavior as (Table 1): Patients with low CVD risk (<10) aged ≥40 years old in men and ≥45 years old in women Target behavior Reduce the prescription of statins in the context of the primary prevention of CVD in low-risk patients (REGICOR <5%) and favor the adoption and implementation of the recommended intervention, the promotion of healthy habits (regular physical activity, healthy diet and cessation of smoking) at any opportunistic or programmed screening or addressing of CVD risk factors in health center visits.
Reduce the prescription of statins in the context of primary prevention of CVD in low-risk patients (REGICOR <5%) and favor the adoption and implementation of the recommended intervention, the promotion of healthy habits (regular physical activity, healthy diet and giving up smoking) at any opportunistic or programmed health center visit for screening or addressing CVD risk factors.
Following this, an effort was made to breakdown the target behavior into the chain of behaviors involved and the concomitant precipitating factors (Appendix Table 1). Three precipitating factors for the practice of primary prevention of CVD were identi ed: i) alarm systems integrated within the EHR prompting the ful llment of the PAPPs: Preventive Activities Program; ii) the presence of an elevated cholesterol level within an analytical test; or, iii) the presence of a prescription initiated or suggested by another professional (specialized or private). Regarding the preventive action behaviors by physicians and nurses, 7 main steps were identi ed, ranging from the initial general approach for CVD primary prevention focused on CVD risk and the cholesterol level to the enactment of the decided treatment or intervention, the options being the prescription of a statin, the delivery of a healthy lifestyle promotion intervention, or both. Then, the working group, using the tools proposed by the BCW within this step, prioritized the following speci c behavior, described according to who needs to do what, when, where, how often, how and with whom, as that most likely to bring about change: The FP considers options and makes the clinical decision on intervention/treatment to be provided, based on the result of the CVD risk estimation, on knowledge and heuristics in relation to the recommended practice, their attitudes, expectations and abilities, and other contextual factors (time, work overload, organizational norms, decisional fatigue, etc.).
Step 4. Identify what needs to change.
Numerous determinants, facilitators of the inappropriate statin prescription and barriers towards healthy lifestyle promotion emerged from the focus groups with healthcare professionals and patients. Determinants were identi ed from the quotes extracted from the focus group guided by a pre-speci ed coding. Table 2 presents some examples of quotes classi ed by the domains of the TDF. Apart from one TDF dimension, Optimism, all the rest of the dimensions were covered in the FPs' discourses (See Table 2 for extracted quotes): "Then also, the issue of the reliability of the guidelines is an issue... the sensitivity and speci city you have when making a decision... the issue of cholesterol is quite controversial." (K_Q2) "Cholesterol levels have been very variable, and we didn't know if it was necessary to treat this in primary or secondary prevention, but then it became clear that it was in secondary, not in primary, that diabetics are in secondary, and if they're not... there we've also had a bit of trouble and so that could also be the cause of this prescription" (K_Q3) "I think that we have to be clear about that at least, that there's no evidence for giving statins, unless there's a family history, yes." (K_Q4) "We are seeing that there are other added risk factors, there are diseases that we are seeing that have a greater risk of having that disease, rheumatism for example, but some other things aren't. In the analysis that you have made of Osakidetza, this might be there or not, but you probably haven't been able to see if they have a family history of sudden death, you cannot see if in addition to this they have other diseases that have to do with greater risk, which are being seen today. We don't see many of these."(K_Q5) Skills "..For us it is also easier to prescribe a pill… It's simple, I ask you to take a test in two months, and ask 'Is everything okay? Does anything hurt? See you next year' and, that's it, it was a test and two appointments." (Sk_Q1) "We have a training de cit in terms of the prescription of physical exercise and the prescription of nutrition in general and if you have some training it is because you have asked for it, because you have read about it, because you have shown interest. I believe that the way we are working, it is very complicated in the appointment with the patient, with the time we have and all the things we have to do…" Beliefs about capabilities "It is much harder to change the habits of someone who comes to have their cholesterol tested if they are about 40 or 45, with settled habits that are di cult to change… that's harder than, 'Give me a pill and I am going to do it quickly', and I have peace of mind." (Cap_Q1) "Walking progressively without getting tired, that works for everyone. I am not ready to prescribe physical activity. I think we can, but it is not effective." (Cap_Q2) "This age group is people who are working and do not come to consult you except when they are sick for some reason, so they often pass under the radar. You ask them for a test, and their cholesterol is skyrocketing, but you don't get them to come to a consultation to see where they are failing, to be able to treat changes in habits... it is di cult to make them come to the health center, and it is also di cult to get them to make the changes... I think that there is a lot we don't see." (Cap_Q3) Beliefs about consequences "And the decision is always going to be, just in case, I'm going to give it to them. And then you also defend yourself just in case. " "Also, in the real world, statins are a spectacular, very effective drug. I have 270 cholesterol, I go on a diet or exercise and I get down to 240 and that's that. However, if I take the pill, after 3 months I am at 200 "(Con_Q2) "On the one hand we have the problem on both sides, we who nd it more work and have a reward in the medium to long term in terms of results, and on the other hand what the user wants is immediacy now. They've come to ask us to solve it now." (Con_Q3) "Patients also hear that statins are bad, that they can cause diabetes and brain hemorrhages... some stop taking them because they have heard that it can cause some problems, or there have been people who for muscular reasons have had to stop taking them and take others... there was one statin that came out and they had to withdraw it from the market... all of these are little things... but, well…" (Con_Q4) Motivation, goals, intent "My experience is that maybe you have been saying to the patient for 2 or 3 years, 'You have to take exercise, go for a walk'… and they always look for an excuse, 'I can't because of my work…', so in the end you say, 'Well, leave it then' and you give up." (M_Q1) "In the end it depends on the conviction that you have, if you are more convinced, you will dedicate more time. Personal conviction and what you want."(M_Q2) Memory, attention, decision making "We doctors are inert by de nition. Clinical and therapeutic inertia is part of our makeup. We are very inert, whether to prescribe or to stop prescribing."(MAD_Q1) "Often, when you are not sure, the most normal thing that we doctors learn is to see something and prescribe, as that it is the fastest thing we have…. so we don't have to explain… it's easier to give medicine than to explain." (MAD_Q2) "You are seeing patient 141, you are already tired, and someone has made an appointment for you to give them statins, they tell you that if something happens to them you will be responsible... And on top of it all, at that time of day you have low blood sugar... I ask you how you would manage that situation." (MAD_Q3) "...the matter of the asterisk, and what happens when we see one... just today someone came to me with cardiovascular risk of 3 or 4, and had an LDL that was almost 190. This was a young woman of 40, with low cardiovascular risk, and she asked me if she had to take something for it." (MAD_Q4) "And one thing, they should take away the asterisks, as we spend a lot of time explaining asterisks when we shouldn't have to." (MAD_Q5)

TDF Dimension
Extracted quotes Environmental context, Resources, constraints "Sometimes, most of the time, we don't have enough time, and the time factor is important for everyone I think, to tell them, to try to convince them." (E_Q1) "...I think that the pressure of attending patients may have too much in uence on the matter of prescription." (E_Q2) "The Regicor does not mean you stop being a doctor, you have to continue being a doctor, just like we use the stethoscope as a tool. And the problem of the risk scale is good for the population, it is very good for population risks, but not for individuals, they weren't designed for that." (E_Q3) "Well, that allows me to put if the patient is in primary or secondary prevention, if they have anxiety or not, are stressed or not... that allows me to modulate those risk modi ers, and gives me peace of mind in both senses. This patient doesn't need statins, I'm sure, and that one does need statins, certainly." (E_Q4) "My nurse does it very well. I am very lucky, she is a highly trained woman who does it very well. So I delegate some things to her. But unfortunately, nowadays she is not always there, and not all nurses are trained... " (E_Q5) "… But it has to be at another level, multidisciplinary, health policies, health policies, lifestyle, which do not necessarily have to be based at the health center. It should also be involved but should not be the greatest weight and we should invest more in health policies especially in these types of people, the population base with least risk but who in the end are the ones that we can really prevent getting ill." (E_Q6) "This age group includes people who are working and do not come to consult you except when they are sick for some reason, so they often pass under the radar. You ask them for a test, and their cholesterol is skyrocketing, but you don't attract them to a consultation to see where they are failing, to be able to treat changes in habits. That is the problem that I think we have in this age group. With older people who come to the health center more often, it's much easier. But with people who are at work... it is di cult to make them come to the health center, and it is also di cult to get them to make the changes... I think that there is a lot we don't see." (E_Q7) "It is very di cult to get hold of them and to continue to call them in to make them get tests, like cholesterol, as they don't think much about prevention, because nothing hurts, and on top of that you restrict them a little, and in their life it is di cult for them to make those changes of habits so they don't come." (E_Q8) Social and Professional Role and Identity "I've had the experience of stopping a patient's statins, and the endocrinologist asked them what the family practitioner thought they were doing, taking them off statins... and then in the end the endocrinologist or the cardiologist put them back on them." (Rol_Q1) "You see that a patient who has been to the ... endocrinologist or ... a patient who is seen in oncology, then comes to us in a state because they tell them that the doctor has to lower their cholesterol. These colleagues have a completely different view from ours, that this is a disease, and it can be important, except for very high numbers, which is a separate issue. The cardiologist who sees patients every day with heart attacks and things like that is much more likely to prescribe statins than we are, who see that much less." (Rol_Q2) "This work is a bit beyond our usual work, but it should be a bit, it should direct us to giving a good prescription for physical education, where we can do this, or where there can be a good health provider who works in this way." (Role_Q3)

Social
In uences "Cholesterol doesn't hurt, but it is so well-known that people are terribly afraid of it. On the other hand, they are not afraid of weighing 100 kilos, or smoking, or not exercising, but cholesterol is something objective… "(SI_Q1) "Maybe the message of the media has a lot of in uence, maybe we should try to change it, so that people become more aware of what cardiovascular risk means, as they're not aware. I think that's where we spend most time, explaining it to them. " (SI_Q2) "I believe that, on this issue, unlike other health issues, people come with a very preconceived idea, because there is pressure. In fact, when people do some tests, the rst thing they ask you when they come for the results, is how high their cholesterol is." (SI_Q3) "But I am referring to the advertising in which exercise, healthy food is being promoted more… that is what needs to be promoted. In the past, people didn't know much about exercise, but now they are a little more aware. Another thing is to get them to do it on a regular basis. That is what is di cult for the patients." (SI_Q4) "For the patient, when you explain these dietary hygiene measures, it's like you aren't telling them anything... 'What did the doctor tell you? Nothing, the usual...' So it has little weight and little value for them, it's like not telling them anything. However, if you give them a pill and send them to have tests, that's different." (SI_Q5) "Sorry, I have to go now. I signed up for a congress to prescribe exercise, and they didn't accept me. I was amazed. The reply from the person in the department where I applied was: "That is not a primary medicine matter." I was amazed. To cap it all, I was the rst at that time." (SI_Q6) "We travel thanks to the pharmaceutical companies and we go to congresses thanks to the pharmaceutical companies and inadvertently there is always some contact in some way because they have given us training, which our company didn't do..." (SI_Q7) "I suppose these are the questions that (patients) often keep asking themselves, due to ignorance of the professionals, due to pressure from pharmaceutical companies, the media ... and they think that if you don't take it you will have a heart attack, sure." (SI_Q8)

Skills
Differential required skills of alternative behaviors, statin prescription versus healthy lifestyles promotion, due to their perceived or experienced ease/di culty seem to be, on the one hand, a facilitator of an inappropriate prescription and on the other, a barrier to the recommended practice to be provided, especially regarding the prescription of physical activity (Sk_Q1,Q2).

Beliefs about capabilities
The main determinant related to capabilities is the low perceived con dence in prescribing healthy lifestyles, a clinical practice considered di cult in itself as compared to prescribing a statin (Cap_Q1, Q2). This problem is augmented by the di culties faced by professionals to tackle healthy lifestyle promotion actions as a means of preventing CVD in low-risk patients, who are not usually frequent attenders (Cap_Q3).

Beliefs about consequences
The fear of negative consequences of not treating seemed to be a powerful driver of inappropriate prescribing (Con_Q1). This "defensive medicine" was also enhanced by the perceived effectiveness of statins in decreasing cholesterol levels (Con_Q2). Obtaining such a positive clinical result in the short term contrasted with the long term (and somewhat unperceived) bene ts of healthy lifestyle promotion actions (Con_Q3). The adverse effects associated with statins seemed to be a potential barrier to statin prescription (Con_Q4).

Motivation, goals, intent
The mentioned scarcity of positive expected results from healthy lifestyle promotion actions has derived in the low motivation of professionals (M_Q1). Actual intention in the form of action plans or goals, both for not prescribing statins and also for providing healthy lifestyle promotion interventions, is seen as a necessary condition for endorsing guideline-concordant CVD primary prevention efforts (M_Q2).

Memory, attention, decision-making
A repeated theme in physicians' discourse is the in uence of clinical inertia in decision-making favored by contextual factors such as lack of time and heavy workload (MAD_Q1,Q2). Pharmacological prescription is perceived to require less cognitive effort in a saturated clinical practice that leads to decisional fatigue. A defensive medicine mindset is always present when deciding upon treatments (MAD_Q3). Physicians also requested the removal of asterisks in patients' reports of analytics (i.e., an asterisk is placed alongside cholesterol measured level when value is equal or greater than 200 mg/dl) as this visual stimulus induces patients' concerns regarding cholesterol levels (MAD_Q4,Q5). Such markers incite cholesterol-control-focused clinical actions.
Environmental context, resources, constraints As previously commented, lack of time and the heavy workload experienced in Primary Care are the main obstacles for prevention efforts (E_Q1,Q2). Physicians also perceived that tools within the EHR are useful but limited for both estimating cardiovascular risk, for reminding and fomenting guidelineconcordant CVD primary prevention practice, and for restricting inappropriate statin prescribing (E_Q3, Q4). Teaming up with an involved Practice Nurse for sharing prevention efforts facilitated adequate healthy lifestyle promotion actions in primary prevention of CVD (E_Q5). Lack of external resources inside and outside the clinical setting (i.e., allied health professionals, community resources, etc.) limits the reach of prevention efforts, especially in low-risk young adults, as a non-frequent-user population (E_Q6-Q8).

Social and Professional Role and Identity
Lack of coherence in prescription criteria among the different healthcare professionals (i.e., cardiologists, neurologists and interns in addition to FPs) that attend the same patients dilutes responsible clinical practice (Rol_Q1, Q2). Uncertainties regarding limits in responsibility with respect to healthy lifestyle prescribing and fear of questioning each other's clinical decisions help to maintain inappropriate treatments (Rol_Q3).

Social In uences
Patients' lack of awareness together with a perception of low susceptibility and vulnerability regarding cardiovascular risk hamper physicians' primary prevention efforts (SI_Q1, Q2). In contrast, due to the importance given by the media and probably fueled by the pharmaceutical industry, cholesterol is "the bad guy" everybody is worried about and needs to be addressed (SI_Q3). Another ambivalence occurs with healthy lifestyles. On the one hand, the population seems to be more conscious about the overall bene ts of healthy behavior. But on the other hand, patients seem to have become so used to the message about the need to change to healthy habits that some prefer to take a "magic" drug in the belief that there is no need to change habits (SI_Q4, Q5). In fact, neither the internal context in the health system which does not prioritize healthy lifestyle promotion practice, nor the external context at societal level in uenced by media messages and the economic interests of the pharmaceutical industry targeting cholesterol reduction exclusively, are conducive to good CVD primary prevention practice (SI_Q6-Q8). Professionals also perceive that in certain sectors of the population, such as those with lower socio-economic status, the promotion of healthy habits, although being recommended practice, is very di cult to implement (SI_Q9, Q10).

Emotion
Mixed emotions are reported by physicians who mainly favor inappropriate prescribing. Professionals must make decisions in an emotional climate marked by uncertainty due to the variability of recommendations and limitations of the Clinical Practice Guidelines and fear of consequences of not treating (Em_Q1-Q3). The feeling of pleasing the patient coupled with peace of mind after prescribing statins and obtaining "positive" cholesterol results are factors that seem to weigh substantially on decision making (Em_Q4, Q5). In contrast, positive emotions associated with successful healthy lifestyle changes seen in patients are the only emotional asset that favors continuing the work of promoting healthy lifestyles (Em_Q6).

Behavioral Regulation
Professionals complain of a poor quality assessment culture in the healthcare system and of lack of standards and indicators established by the organization to anchor and guide clinical performance (BR_Q1). Data are needed to be able to re ect on performance and to be able to set goals, monitor progress and provide useful feedback, and the lack of access to such data prevents re ection and the establishment of objectives, both of which are seen as necessary to correct the problem of inadequacy in drug prescription (BR_Q2-Q4).

Reinforcement
In addition to the above-mentioned in relation to objectives and performance indicators, the results of the evaluations of indicators carried out by the organization do not translate into incentives/disincentives for professionals, which generates demotivation among those professionals willing to do things well (Re_Q1,Q2).
Moreover, we carried out one focus group with six patients. We must highlight that the majority of the participants indicated a lack of explanation in the treatment prescribed and their desire to be more involved in the treatment decision. Moreover, they believe that family history has a lot of weight in the decision and they are concerned about it. They report that only some professionals recommended healthy habits with or without prescription of statins. When we asked about their preferences for doing physical activity or taking a cholesterol drug in a context of low CVR, different positions arose: some prefer physical activity and others prefer to combine exercise and pharmacological treatment. Overall, they are satis ed with taking statins although they prefer not to think about the adverse effects.
2nd Stage. Identify Intervention Options Steps 5 and 6. Select intervention functions and speci c behavior change techniques The identi ed determinants in the form of "what needs to change" categorized in the COM-B and TDF dimensions were linked to the intervention functions guided by the BCW instructions and suggestions. Then, all potential policy categories were also identi ed. Lastly, potential BCTs were determined. Table 3 summarizes the mapping process conducted, linking practice determinants for PIP of statins (mainly facilitators) and for providing healthy lifestyle promotion interventions (mainly barriers), to intervention functions and policy categories, ending with potential BCTs for attaining the desired target behavior. For example, the lack of awareness among patients regarding the problem of inappropriate pharmacological prescription (Facilitator of the low-value practice) can be addressed through persuasion (Intervention function) and communication actions (Policy category) enacted by techniques focused on providing information about health consequences (BCT) of this low-value practice. Table 3. Mapping matrix of potential intervention functions, policy categories and behavior change interventions to identi ed determinants of inappropriate statin prescription and healthy lifestyle promotion categorized by COM-B/TDF dimensions identi ed from the qualitative study.

Beliefs about capabilities
Perceive that one is able and has the necessary skills to provide the healthy lifestyle promotion Perceive that statin prescribing is not such a simple (low skill) or safe practice Perceive that one is competent and con dent enough to carry out the CV risk screening process Perceive that one is competent and con dent enough to respond to the sporadic arrival of patients in the target population for CVD primary prevention (they come infrequently), through the promotion of good habits Perceive that statin treatment is not so easy for the patient (dosage) Have a sense of self-confidence in prescription of physical activity and other healthy habits The nal selection of previously identi ed de-implementation strategies was performed through a participatory consensus process involving the working group, all healthcare professionals involved in the discussion groups conducted and representatives at managerial level of the two HDs from the Basque Healthcare System.
Informed by previous studies in the eld and several team members' experience in the design of implementation strategies, the working group decided upon potential intervention formats and modes for their execution and drew up a list of 13 potential de-implementation strategies. Lastly, these potential strategies were sent back to all the health care professionals involved in the discussion groups and two health managers for their evaluation regarding three dimensions: acceptability, feasibility and potential effectiveness. Thirteen complete evaluations (13/23) were received that permitted the prioritization of the deimplementation strategies (See Table 4).

Discussion
This study aimed to report on the application of a systematic, comprehensive, theory-and evidence-informed framework to design potentially effective and feasible de-implementation strategies to favor the abandonment of low-value pharmacological prescribing in CVD primary prevention of persons with low CVD risk [18]. Speci cally, guided by the TDF and the BCW frameworks [14,[21][22][23], we have conducted a series of actions for identifying determinants of low-value practices and behavioral objectives as areas for improvement, which have helped us to design, operationalize and prioritize various de-implementation strategies. These actions or tasks were as follows: semi-structured interviews with a sample of Family Physicians in order to identify the overall behavioral scenario and break down the chain of behaviors and concomitant factors and to de ne the problem in behavioral terms; a qualitative study through discussion groups to identify the main determinants of the low-value pharmacological prescription in CVD primary prevention (e.g. personal, inter-personal, organizational, social level), and those relevant for promoting the implementation of the recommended practice (healthy lifestyle promotion interventions); mapping and operationalization of the de-implementation and/or implementation strategies; prioritization of strategies based on professionals' perceived acceptability, feasibility and potential effectiveness.
Avoiding or substituting proven potentially harmful, ineffective or ine cient medical practices is important for improving the quality of healthcare while ensuring sustainability of healthcare systems, which is why in recent years the interest in and the evidence base related to successful de-implementation strategies to favor the abandonment of low-value practices has grown quickly. Statins are among the most widely prescribed medications globally and are increasingly used to prevent cardiovascular disease (CVD) in people without CVD ('primary prevention'). However, statins have no or little value for the primary prevention of CVD in low-risk patients [19,20]. On the other hand, healthy lifestyle promotion interventions in clinical settings have been shown to be effective and are the preferred recommended practice, especially in low-risk patients.
From the growing scienti c evidence in implementation research it is known that factors determining the implementation of both evidence-based and inappropriate interventions in the clinical setting are multi-level, complex, and context speci c [2]. Consequently, the design of interventions should be carried out following a process of formal analysis of the objective behavior and its theoretically predicted mechanisms of action, all guided by models or theories that cover the entire range of possible in uences or determinants of the behavior in question [8][9][10][11]. Through the performed qualitative study with both main involved healthcare professionals (family physicians and nurses) and affected users (low-CVR-risk patients with PIP of statins), we have identi ed multilevel determinants of the target low-value practice within the context of two HD in our Basque public healthcare service. Almost all of the dimensions of the TDF have been called into play, as at least one practice determinant (barrier or facilitator) has been included in these dimensions. Some of the most consistently reported determinants among interviewed professional groups were the lack of time and external resources, preferences and characteristics of patients, limitation of available clinical tools and CPGs, social pressures, fears about negative consequences of not treating high cholesterol levels with drugs, and lack of skills and training of professionals in healthy habit promotion. Patients' main determinants were the lack of explanation of the situation during the medical appointment, the desire to be more involved in the treatment decision, belief and concern about family history in the decision, and the lack of healthcare professional's recommendation about healthy habits with or without prescription of statins.
The identi ed determinants are in line with other determinants identi ed or reported in previous studies regarding determinants of low-value practice and of low-value pharmacological prescription. Regarding determinants, uncertainty due to the variability and/or con ict of the guidelines with respect to the recommended practice, the pressures and demands on the part of the patients, the need for rapid and decisive action in response to the reasons for consultation and the desire to please the patients have been identi ed as interconnected motives that justify maintaining low-value practices in general [24].
With regard to inappropriate prescription of drugs in general, a systematic review published by Anderson et al [25], in which the barriers and facilitators for inappropriate prescription are explored, highlights four aspects that facilitate or hinder professionals' decisions when faced with a possible pharmacological One peculiar aspect in this point is that, due to the clinical scenario addressed (the reduction of low-value prescribing of statins in CVD primary prevention where the promotion of healthy lifestyles is the alternative, recommended practice), this project has attempted to simultaneously identify determinants of both clinical practices. Although it may seem obvious, in such clinical scenarios, stress must be placed on identifying the factors that facilitate or maintain the lowvalue practice, and on the other hand, the barriers that impede the recommended practice [29].
With the main goal of designing and developing speci c strategies that address the speci c determinants of CVD prevention practice in Osakidetza, the main study action at the current stage has been to carry out a mapping process of simultaneous de-implementation and implementation strategies in order to reduce low-value practices (inappropriate statin prescribing) and promote the implementation of recommended practice (healthy lifestyle interventions), based on the determinants of routine practice reported by practitioners in the focus groups, following the procedure established by the BCW. The 13 deimplementation strategies that have emerged from these performed structured and theory-informed processes are all "old known" strategies and interventions. Speci cally, in the context of the reduction in the prescription of low-value statins, a certain effectiveness of dissemination strategies, informative web pages, and the implementation of electronic clinical practice guidelines has been observed, compared with routine practice [30][31][32][33][34]. However, dissemination strategies such as educational or training actions for professionals (webinars and workshops) were not effective by themselves, only when used as multicomponent strategies, combining workshops and informative web pages [30,31]. Finally, an intervention based on sending a scenario with a clinical case and audit/feedback to professionals [35] and techniques to aid decision-making, aided by communication of risk to the patient and audit/feedback, achieved good results in calculating cardiovascular risk and in adjusting the prescription [36-38].Though not innovative interventions or strategies, those identi ed are those that address the speci c determinants identi ed by the protagonists. Furthermore, the agents involved have prioritized the resulting potential deimplementation strategies after assessing their perceived acceptability, feasibility and potential effectiveness.
Finally, both actions, determinant identi cation and mapping of strategies, aim to target the speci c clinical behavior most likely to enable the desired change prioritized by the research group and professionals involved early in BCW steps 1 to 3: physicians' decision-making regarding the therapeutic option. Following a taxonomy of choice architecture techniques [39], all except two of the 13 identi ed strategies may be categorized as in uencing decision-making through three different modes: decision information, decision structure, and decision assistance. The other two consist in patient-mediated intervention and an interprofessional collaboration intervention in which FPs and their corresponding Practice Nurses agree, plan and organize how to collaboratively proceed in and provide CVD prevention intervention actions.

Conclusion
The study aims to contribute to the body of currently scarce literature available on practical de-implementation initiatives by providing detailed illustrations/explanations of our stepped, systematic approach to the design and development of targeted behavior change actions based on prominent available frameworks and theories, mostly from implementation science. Key research questions in implementation science also involve determining what implementation strategies should be provided, to whom, and when, to achieve optimal success in implementing evidence-based clinical practice. As the same paradigm must apply for de-implementation of low-value practices, we propose now to investigate the comparative effectiveness of some/different types or intensities of the prioritized strategies in phase II of the DE-imFAR project. The evaluative phase of our study will have the aim of increasing evidence on whether the speci c strategies that address determinants of recommended practice in CVD prevention, some similar to those evaluated in the few studies conducted to date, are also effective in our context. If the strategies explored are successful, health planners and managers will have the evidence needed to support the introduction of such structured strategies, informed by the application of methods and procedures of the emerging science of implementation and de-implementation.

Declarations
Ethics approval and consent to participate The research protocol has been approved by the Basque Country Clinical Research Ethics Committee (Ref: PI2019102, approved on 10/04/2019). The Primary Care Research Unit of Bizkaia is explicitly authorized by the Healthcare Management of the Basque Health System to extract and use data from its electronic health records for research purposes.

Consent to publication
Not applicable

Availability of data and materials
Since data supporting the present study will mostly concern routine data retrieved from the electronic health records of the Basque Health Service-Osakidetza, it will be only shared on justi ed request to the study guarantors.