Factors influencing implementation of a patient decision aid in a developing country: an exploratory study

Background Most studies on barriers and facilitators to implementation of patient decision aids (PDAs) are conducted in the west; hence, the findings may not be transferable to developing countries. This study aims to use a locally developed insulin PDA as an exemplar to explore the barriers and facilitators to implementing PDAs in Malaysia, an upper middle-income country in Asia. Methods Qualitative methodology was adopted. Nine in-depth interviews (IDIs) and three focus group discussions (FGDs) were conducted with policymakers (n = 6), medical officers (n = 13), diabetes educators (n = 5) and a nurse, who were involved in insulin initiation management at an academic primary care clinic. The interviews were conducted with the aid of a semi-structured interview guide based on the Theoretical Domains Framework. The interviews were audio-recorded, transcribed verbatim and analyzed using a thematic approach. Results Five themes emerged, and they were lack of shared decision-making (SDM) culture, role boundary, lack of continuity of care, impact on consultation time and reminder network. Healthcare providers’ (HCPs) paternalistic attitude, patients’ passivity and patient trust in physicians rendered SDM challenging which affected the implementation of the PDA. Clear role boundaries between the doctors and nurses made collaborative implementation of the PDA challenging, as nurses may not view the use of insulin PDA to be part of their job scope. The lack of continuity of care might cause difficulties for doctors to follow up on insulin PDA use with their patient. While time was the most commonly cited barrier for PDA implementation, use of the PDA might reduce consultation time. A reminder network was suggested to address the issue of forgetfulness as well as to trigger interest in using the PDA. The suggested reminders were peer reminders (i.e. HCPs reminding one another to use the PDA) and system reminders (e.g. incorporating electronic medical record prompts, displaying posters/notices, making the insulin PDA available and visible in the consultation rooms). Conclusions When implementing PDAs, it is crucial to consider the healthcare culture and system, particularly in developing countries such as Malaysia where concepts of SDM and PDAs are still novel.


Background
Shared decision-making (SDM) is part of patient-centred care whereby patients and clinicians decide on a treatment together. This is particularly relevant when the decision is preference-sensitive [1]. One way to promote SDM is to use a patient decision aid (PDA), which provides information about the decision, available treatment options, benefits and risks of each option and ways to clarify patient values [2]. In a Cochrane Review of 115 randomized controlled trials, PDAs have been proven to be effective in involving patients in SDM and improving their decision quality [2]. However, the adoption and implementation of PDAs in clinical practice remains poor [3][4][5]. One study has found that only about 10% of eligible primary care patients actually received PDAs despite clinic-wide PDA adoption [4].
There are many factors that influence the implementation of SDM and PDAs. Studies have highlighted barriers such as time constraints, healthcare professionals' attitude, perceived legitimacy of the PDA, lack of applicability due to patient characteristics, clinic capacity, processes of care and the healthcare environment [6][7][8]. Among the facilitators were provider's motivation, provision of training and skills development for providers, identification of a clinical champion, introduction of a system to identify eligible patients to use PDAs ahead of clinical consultations, positive impact on the clinical process and patient outcomes [7,8]. However, the relative influence of these factors varies across different countries.
As most of the studies on barriers and facilitators to implementation of PDAs are conducted in the west [6][7][8], the findings may not be transferable to Asian countries, most of which are developing countries, and issues such as socio-cultural barriers (language barrier and physician paternalism) and lack of resources (infrastructure or technology development) may be more significant compared to developed countries. Nevertheless, studies have shown that patients in Asia want to be involved in SDM [9,10] and there are an increasing number of PDAs being developed in the region [11,12]. However, no Asian studies have reported the implementation of these PDAs.
Malaysia is a developing upper middle-income country in Southeast Asia [13]. It has a population of 28.3 million, comprising three main ethnic groups namely Malay (67.4%), Chinese (24.6%) and Indian (7.3%) [14]. The implementation of SDM in Malaysia is hampered by several factors including the multi-cultural and language diversity of the population and the lack of patient involvement in healthcare [15]. In Malaysia, although the national language is the Malay language, not all Chinese or Indian patients are fluent in this language. Furthermore, each ethnic group has their own cultural norms and beliefs when it comes to health. Thus, doctors and patients may be mismatched linguistically as well as culturally. As an attempt to advance SDM, several Malaysian PDAs have been developed for patients who are making a decision on insulin use [12], early breast cancer [16] and early prostate cancer treatment [17]. These PDAs are available in the Malay, Chinese, Tamil and English language to facilitate decision-making as they will be able to read and understand the PDA in their preferred language. Furthermore, these PDAs explored cultural beliefs of the major ethnicities in Malaysia. Nevertheless, formal implementation of these PDAs has not been conducted. Therefore, this study aims to use a locally developed insulin PDA (http://dmit.um.edu.my/?modul=DMIT_PDA) [12,18] as an exemplar to explore the barriers and facilitators to implementing PDAs in a developing country.

Study design
This study adopted a qualitative methodology to explore the factors influencing implementation of the insulin PDA. In-depth interviews (IDIs) and focus group discussions (FGDs) were conducted.

Study setting and participants
Data collection was carried out at an academic primary care clinic at the University of Malaya Medical Center, an urban government teaching hospital. In Malaysia, the majority of diabetes patients are managed in the public sector where healthcare services are subsidized by the government. The primary care clinic is an outpatient clinic which accepts patients from all over Malaysia, and therefore, it has a high patient load with patients from various backgrounds. On average, there are 25 doctors in the clinic, two diabetes educators (DE) and 26 registered nurses (RN). Generally, patients in the clinic do not get to decide which doctor to see; hence, they may not consult the same doctor during the follow up.
There are three groups of healthcare professionals involved in insulin initiation in the clinic: doctors, DEs and RNs. In the clinic, only doctors (i.e. family medicine specialists) and medical officers were allowed to prescribe insulin to patients. There is a diabetes education centre located in the clinic where three DEs, who are nurses, provide patient education on diabetes, blood glucose monitoring and insulin initiation and injection techniques after receiving referrals from doctors. To qualify as a DE, RNs need to complete an advanced diploma in diabetes education. In the course, the RNs are trained on diabetes education, treatment and nursing of diabetes patients (including on insulin initiation). However, training on SDM and PDA is not part of the curriculum as the insulin PDA has not been formally introduced or implemented in the healthcare system in Malaysia. In contrast to DEs, RNs' tasks are to assist doctors in the consultation rooms. Of 26 individuals who were approached to participate in the study, one staff nurse refused due to time constraints.
This study used purposive sampling to recruit participants who were healthcare policymakers (HPMs) and healthcare providers (HCPs) working in the medical centre [19]. The inclusion criteria for HPM were those who were responsible for or involved in making decisions on whether a particular health intervention should be implemented in the hospital, while for HCPs were those who were involved in advising patients about insulin initiation. The HPMs recruited included the directorlevel hospital manager, Endocrine and Primary Care Medicine policymakers and hospital matrons. They set standards of care and implement programmes to improve diabetes care in the hospital. While they perform managerial tasks, they are also practicing HCPs. HCPs included the doctors, DEs and RNs. In order to achieve maximum variation, we recruited participants from different socio-demographic backgrounds (ethnicity, gender) and included those with and without experience in using the PDA. The PDA was previously pilot-tested with a small group of HCPs in the clinic to test its acceptability, and a few of them had continued using it after the pilottesting phase had ended. HPMs were not involved in the pilot study.

Study instrument
The interviews were conducted with the aid of a semistructured interview guide (Table 1), which was developed based on the Theoretical Domains Framework (TDF) [20] and a literature review on barriers and facilitators to implementing PDAs [7]. The TDF is an overarching framework of 14 theoretical domains synthesized from behaviour change constructs found in 33 behaviour change theories ( Table 2). The TDF was selected as the theoretical basis for this study because implementation success is largely dependent on the behaviour change of the users involved on whether to adopt or reject the implementation innovation. The TDF is comprehensive as it not only looks into the impact of rational and cognitive process of an individual but also looks into emotional factors [21] as well as external, organizational factors such as working environment and resources that may influence them on whether or not to adopt a new behaviour or innovation such as the practice of SDM and use of PDA. Furthermore, the TDF has been used in many clinical behaviour change implementation research projects [22], which is why it was felt to be appropriate for this study to understand the factors influencing implementation of insulin PDA.

Data collection process
HPMs and HCPs who fulfilled the study inclusion criteria were invited to participate in the study. Prior to the interviews, eligible participants who consented to participate in the study were given information about the insulin PDA, its objective and content, the concept of SDM and the various modalities (booklet, tablet, website) available. A video, which demonstrates how the insulin PDA can be used during a consultation, was shown to the participants to give them an idea on how the insulin PDA can be used in the clinic. The participants were also encouraged to think about other ways of implementing the insulin PDA in their practice.
The interviews were conducted between December 2015 and March 2016 by the researchers (WTT, YKL, CJN) who asked the participants 'open' and 'probing' questions to explore the factors influencing implementation of the insulin PDA in their practice. All the interviews Outer contextthe healthcare system • Are you confident/positive/optimistic that the current healthcare system is capable to successfully implement the insulin PDA despite any difficulty? Yes/No, why? (beliefs about capabilities; optimism) • What can be done by the higher-level authority to ensure the success of the implementation of the PDA? (reinforcements) • In your opinion, how can we make the implementation of the insulin PDA sustainable?
were face-to-face and were conducted in the clinic consultation rooms during rest times in between clinical consultations, or at the office of the participants. The interviews lasted from 50 to 90 min. This study was conducted in the university hospital setting, and three of the researchers (WTT, YKL, CJN) were known to the participants as they are based at this university. WTT is a PhD student, YKL is a lecturer, while both CJN and PYL are clinical lecturers who specialize in family medicine. All researchers are experienced in conducting qualitative research. As WTT did not know any of the participants prior to the interviews, WTT conducted most of the interviews (8/12) as YKL and CJN refrained from interviewing participants whom they knew. PYL did not conduct any interview. The interviews were audio-recorded, and field notes were written to document the content of and reflections on the interviews. The data collection ceased when data saturation was achieved; that is when the barriers or facilitators that emerged from the data became repetitive and there were no more new findings.

Data analysis
All the interviews were transcribed verbatim and checked for accuracy before being imported into Nvivo software for data analysis. The researchers (WTT, YKL, CJN, PYL) familiarized themselves with the first three transcripts and coded the transcripts independently using a thematic approach. The transcripts were read and coded line by line. A code is a short text which represents the meaning of the text segment. Codes that have similar meaning were grouped together to form a category, and later the categories were compared and merged into bigger themes [23]. Researchers met to discuss the categories and themes which emerged from their individual analysis. Any discrepancies in the categories and themes were resolved through consensus before finalizing the coding framework, which was used to analyze the remaining transcripts by one of the researcher (WTT). Any new codes and categories that emerged were added to the list of themes and categories upon consultation with the research team. The written field notes and interview reflections were triangulated with the results to ensure that the findings were correctly interpreted and no information were missed out and to supplement the interview findings. By comparing data from multiple sources, we hope to enhance the credibility of the findings [24].

Ethics approval
This study received ethics approval from the University of Malaya Medical Centre Medical Ethics Committee (reference: MECID.NO: 20158-1600).

Socio-demographic and practice profile of participants
Nine IDIs and three FGDs were conducted with 25 individuals who participated in the study: the hospital policymakers (n = 6), medical officers (n = 13), DEs (n = 5) and a RN. Two of the FGDs were conduced with medical officers (n = 7; n = 6). Another FGD with the nursing policymakers was conducted with smaller number of participants (n = 3) as they were the only three senior nursing staff at management level who were felt to be able to provide feedback on the implementation of the insulin PDA. There were six participants who have had experiences in using the insulin PDA (Table 3).

Emerging themes
Five themes emerged from the interviews, and they were lack of SDM culture, role boundary, lack of continuity of care (COC), impact on consultation time and reminder network.

Discussion
This study provides insights into factors influencing implementation of PDAs from a developing country, a setting which is scarcely reported. The factors revealed in our study are not too different from those commonly reported in the western countries such as lack of SDM culture [25][26][27], time constraints [6,25] and reminders. Less commonly reported factors which emerged from our study were role boundary and COC. This study highlighted a few important domains from the TDF that could influence SDM implementation in developing countries. For example, under the 'environmental context and resources domain' , we found that there was low awareness and receptivity to SDM among the HCPs, which then hindered the implementation of PDA in the Malaysian primary care setting. The participants of this study were doubtful if the concept of SDM is culturally acceptable in Malaysia due to HCP paternalism and patient's submissiveness towards doctors. This resistance to SDM would hinder the adoption of SDM and subsequently lead to non-use of the insulin PDA. There is a distinct contrast in SDM implementation between developed countries (such as the USA, UK, Germany, Netherlands, Australia, Italy and Spain) where SDM and the use of PDAs have already been implemented at policy and legislation levels [28][29][30][31] and developing countries in Asia see slower progress in adopting SDM and PDAs [15,32]. In Switzerland, however, hierarchical structures and asymmetric physician-patient relationship were reported to be barriers for SDM implementation despite doctors recognizing its importance [31]. This shows similarity to the hierarchical social pattern in Asian culture whereby doctors are placed at a higher societal stratum [33], and patients may therefore tend to consign their healthcare decisions to HCPs. Furthermore, doctors may not wish to offer a choice to patients as it is considered a good practice to initiate insulin in patients with T2DM who are unable to achieve glycaemic control despite taking maximal oral medications. Patients in the public health facilities may feel that they may not have control over their health decision as they only pay a nominal fee (i.e. 'I get what I pay for'). In addition, they do not have much freedom to choose the HCP they want to see. However, Asian studies have shown that many patients preferred an autonomous (active and shared) role in decision-making [9,10]. Thus, to facilitate the use of PDAs, the concept of SDM needs to be promoted among HCPs and patients in developing countries. Zhang et al. [34] highlighted that one way to increase patient involvement in making treatment decisions is to increase healthcare professionals' knowledge about this concept. HCPs also need to be trained on respecting patients' autonomy and how to engage patients in making decisions about their health care. The Patient Decision Aids Research Group has created continuous education such as online tutorials, interactive workshops, performance feedback and structured protocols in providing decision support [18]. Efforts are also needed to empower patient in being more involved in their healthcare, and one way to do that is to conduct public health campaigns [15]. In developed countries such as Australia and the UK, the ' AskShareKnow' [35,36] and the ' Ask three questions' [37,38] campaigns are conducted to encourage patients to ask their HCPs the three questions ('What are my options?' , 'What are the benefits and harms?' and 'How likely are these going to happen?') to increase patient involvement in healthcare. While it has been shown to improve patient-doctor communication in SDM in Australia [35], it is unsure if such campaign would be effective on countries with strong hierarchical social pattern like in developing countries such as Malaysia. Among other strategies that have been proposed to promote SDM in clinical practice, however, at the macro level in Malaysia are the following: (1) incorporation of SDM in clinical practice guidelines, (2) advocating the use of patient decision aids or other decision support tools in patient care, (3) inclusion of patient involvement in decision-making as a quality indicator and (4) payment/ reimbursement for practices which implement SDM or use decision aids [15].
Another factor influencing implementation that was raised was role boundary of HCPs in the clinic setting. Role boundary may act both as a facilitator or barrier for the implementation of the insulin PDA. Being clear of one's job responsibilities helped participants decide if they were the right person to use the insulin PDA. For example, all participants felt that doctors should introduce and use the insulin PDA, given their authority to prescribe insulin and familiarity with the patient's health profile. Relying solely on doctors would hinder an interprofessional team approach to using the PDA. Involvement of other HCPs besides physicians may help to disperse the work needed in providing decision support. For example, compared to primary care doctors who has to see high number of patients with various conditions, DEs would have more time to provide patient counselling as they only see diabetes patients. Many western studies have also shown HCPs other than doctors, such as nurses, social workers, psychologists and allied health professionals, who play a significant role in ensuring PDAs are implemented successfully by identifying eligible patients, contacting patients about the PDA and providing decision coaching [39][40][41]. Participants of this current study also suggested that nurses could remind doctors to use the PDA. To increase DEs' and nurses' involvement in insulin PDA implementation, they can be trained in decision coaching skills, which have been found to be effective in guiding patients to make an informed decision [42]. SDM and decision coaching training can be included in the advanced diploma in diabetes education and even basic nursing programmes. In addition, doctors also need to be trained to work with the nurses as a team to implement PDA in the clinic. A recent study showed that very few SDM training programmes adopt an interprofessional approach [43], which could help to utilize the strengths of each team members to create a more coordinated and efficient PDA implementation process.
Time constraints are a universal barrier in implementing SDM as well as PDAs [6,44]. Two studies reported longer consultation time with PDA use compared to usual care [45,46]. However, some studies indicate otherwise and one strategy that can be used to address the time issue is to highlight to HCPs that the use of PDAs does not always require more time compared to usual care [2,47,48] and that informed decision may be achieved more quickly over time [49]. While the first consultation using the PDA may take a longer time, insulin decision-making over subsequent consultations may be shorter, and this reduces delay in decisionmaking. The participants of this study also felt that previsit usage of the insulin PDA might help to save time. Brackett et al. [50] have tested a pre-visit model and found that in addition to time saving, physicians also reported that discussions during a consultation were more focused and helped to address patient values and preferences, which may otherwise be overlooked. It also facilitated decision-making as patients arrived at the clinic appointment more prepared to discuss and make a decision [50]. One strategy used to allay fears over increased time was by asking a mentor or peer expert to demonstrate how the insulin PDA can be incorporated into a standard clinical consultation [51,52] without taking more time and to emphasize that the delivery of PDA only becomes easier and possibly faster with training and practice.
Another important aspect of the implementation of the insulin PDA is the need to follow up patients (ideally by the same HCP) to ensure that there is continuity in the delivery of the insulin PDA. The use of PDAs is not a one-off event as SDM is a continuous interactive process between HCPs and patients; hence, COC is crucial in ensuring effective delivery of the insulin PDA. The lack of COC in the use of PDAs may cause difficulties for patients to raise or discuss issues pertaining to the PDAs that were brought up in the previous consultation [27]; it also prevents rapport-building between HCPs and patients that could facilitate informed decision-making. In Asia, COC is a challenge in healthcare delivery due to high patient load, lack of manpower, time constraints [53,54], lack of family physicians and uncoordinated referral mechanisms [55]. COC is also difficult particularly in an academic healthcare setting, where there is a high turnover of staff after the medical trainees have completed their training [53,56,57]. To address this barrier, systematic documentation of the use of PDA and its discussions with patients can help to facilitate communication between different doctors who follow up the same patients. Furthermore, new HCPs need to be educated on decision support and PDA use. Such training can be introduced during the orientation programme for new HCPs, having a resource person that provide more information about the PDA in the implementation setting as well as having on-going training and support to familiarize HCPs with the concept of SDM and the use of PDAs.
'Forgetting to use' has been reported as a barrier in the implementation of PDAs [58,59] even after provision of training and continuing medical education credits to encourage use among HCPs [58]. Participants of our study highlighted a reminder network to address the issue of forgetfulness as well as to trigger interest in using the insulin PDA. The use of EMR and IT system has been raised as an effective reminder platform in the implementation of SDM and PDA; for example, they can be used to identify eligible patients for PDAs [44] before their clinic visit, to prescribe PDAs as well as to cue for PDA use [60]. While HCPs have reported they would be more likely to use the PDAs if they were reminded through EMR [4], some felt that an electronic, interactive decision aid linked to a computerized reminder system may not necessarily be better compared to traditional paper resources because of technical issues. The integration of PDA reminders into the EMR will require technology support, which can be a significant barrier in developing countries that are still using paper-based systems [61]. If technology cannot be adopted to facilitate implementation, peer support would be another useful form of reminder, which was also raised by participants in our study. For peer support to be effective, the working culture needs to be one that holds the belief that the use of PDA is the preferred practice style. Efforts are needed to create awareness on the benefits of PDAs so that HCPs will be willing to use them. Physician champion plays an important role in creating awareness about the availability and the importance of PDAs, and encourage the staff to use them; this has been reported as one of the key factors in the successful implementation of PDAs [62]. Another reminder that was raised was having posters and notices to promote the insulin PDA. Promotional brochures placed at exam rooms have been implemented to increase patients' interest in decision aids [4,26]. However, while it prompted discussions about PDAs, it did not significantly lead to physicians prescribing PDA to patients [26]. More studies are needed to look at the effectiveness of having promotional materials such as posters or notices on HCPs in remembering to use as well as on patients to initiate discussions with HCPs on PDAs. Ultimately, HCPs and patients need to be made aware that long-term patient engagement in self-management for chronic conditions like diabetes is crucial in improving diabetes control and outcome. This would then obviate the need for reminders. The use of PDA should be integrated as part of the training on longitudinality of care whereby HCPs and patients should be made aware that insulin is an option as diabetes progresses and that PDAs are available to help them make this decision.

Strengths and limitations of study
This study highlights implementation issues that are pertinent to developing countries where literature on barriers and facilitators to implementation of PDA is scanty. This study explored views from a diverse group of key stakeholders from policymaker, physician, nurse and DE, and this helps to elicit factors influencing implementation from various professional levels.
There are a few limitations in this study. First, there were participants who had not been exposed to or had no experience in using the insulin PDA; they may not be able to grasp the concept of the insulin PDA, which involves the process of shared decision-making between HCP and patients. Instead, they may have thought that the insulin PDA is just another educational material. There is a possibility that some issues pertaining to the implementation processes for PDA may have been omitted. The researchers tried to reduce this limitation by explaining to the participants the purpose of the insulin PDA, concept of SDM, its content, as well as showing a video of a mock consultation using the PDA before conducting the interviews. Second, the data collection was only conducted in an academic primary care clinic and did not include other healthcare settings such as public community health clinics or private practices. Hence, the findings of this study may not be transferable to other settings. Third, it is noteworthy that there was a preponderance of female participants in this study, as more female than male candidates are interested to pursue family medicine as a specialty in Malaysia. However, the researchers felt that the participants' sex may not have significant impact on the study findings given that the topic of this study is one that is not gender-sensitive. Fourth, only HPMs' and HCPs' views were gathered in this study. Patient's perspectives should also be taken into account, which are currently being explored in the next phase of this study. Lastly, there is a possibility that participants may have provided positive responses or socially desirable answers, as they did not want to criticize their own practice; researchers tried to offset this bias by reassuring the participants that confidentiality would be kept and that their participation in this study would not affect their work or career.

Conclusions
When implementing PDAs, it is crucial to consider the healthcare culture and system, particularly in developing countries such as Malaysia. This study highlights that to facilitate the implementation and effective use of PDA, there is a need to change physician's paternalistic attitude as well as to promote patient empowerment so that the concept of SDM can be embraced. The use of an interprofessional approach and reminders should also be considered when designing strategies for PDAs implementation.