Patients’ and clinicians’ expectations on integrative medicine Services for Diabetes: a focus group study

Background Difference of perspective between patients and physicians over integrative medicine (IM) research and service provision remains unclear despite significant use worldwide. We observed an exceptionally low utilisation of IM and potential underreporting in diabetes. We aimed to explore the barriers and recommendations regarding service delivery and research of IM service among diabetes patients and physicians. Methods A 10-group, 50-participant semi-structured focus group interview series was conducted. Twenty-one patients with diverse severity of disease, comorbidities and education levels; and 29 physicians (14 conventional medicine (ConM) and 15 Chinese medicine (CM)) with diverse clinical experience, academic background and affiliation were purposively sampled from private and public clinics. Their perspectives were qualitatively analysed by constant comparative method. Results Seven subthemes regarding barriers towards IM service were identified including finance, service access, advice from medical professionals, uncertainty of service quality, uncertainty of CM effect, difficulty in understanding CM epistemology and access to medical records. Patients underreported the use of CM due to the concern over neutrality of medical advice among physicians. Inconvenience of service access, frequent follow-up, use of decoction and long-term financial burden were identified as key obstacles among patients. Regarding research design, ConM physicians emphasised standardisation and reproducibility while CM physicians emphasised personalisation. Some CM-related outcome measurements were suggested as non-communicable. Both physicians acknowledged the discordance in epistemology should be addressed by pragmatic approach. Conclusion Key obstacles of CAM clinical utilisation are different between patients. Further assessment on IM should be pragmatic to balance between standardisation, reproducibility and real-world practice. Evidence-based IM programs and research should merge with existing infrastructure.


Background
Controversies on incorporating complementary and alternative medicine (CAM) into integrative medical (IM) care continue amid increasing utilisation and volume of evidence worldwide [1][2][3][4][5]. It was estimated that 25.9% of the general population in Europe in 2014 [6] and 35% of US adult population used CAM in 2015 [7], associated with higher education and income [6,7]. The main form of CAM used were body-mind therapies and herbal treatments, and Chinese medicine (CM) is one of the most utilised streams of CAM [6,7]. Patients with chronic 'incurable' conditions tend to perceive CAM as an efficacious alternative [8,9].
Hong Kong has a developed health system and the introduction of CM faces challenges similar to the western world [10,11]. While the lack of research-related capacity (mainly quality of evidence) and cultural believe has been repeatedly identified as key obstacles in existing literature [12][13][14], barriers and recommendations beyond these two domains requires further studies, especially from the implementation and organisational perspective [8,15]. Also, research on the comparison of the perspectives among patients and physicians, especially between physicians from CAM and conventional medicine (ConM) of different specialties is limited amid the global call on addressing misaligned expectations between stakeholders prior to the design of interventions [16][17][18][19].
Nevertheless, a service programme in Hong Kong revealed that only 20% of diabetes patients have ever considered CM as an option and only less than 2% have ever used CM as a treatment for diabetes or DKD which was way below the utilisation in other conditions (e.g. around 50% for cancer patients) [37]. Interestingly, some patients reported that ConM and CM physicians both discouraged IM care. The aristolochic acid-related herbal nephrotoxicity [38,39] and over-the-counter CM-associated increase all-cause mortality [40] may partly explain the reluctance of ConM -CM collaboration in DKD [12]. However, the exceptional underutilisation and potential underreporting of CAM use [41] in diabetes hinted at possible unidentified expectation mismatch between CM, ConM physicians and patients from existing literature that is specific to the condition and worthy of special attention.
We aimed to explore the barriers and recommendations regarding IM service delivery among patients and physicians of conventional medicine and CAM, and subsequently identify key areas for research, education and service provision.

Study design
We conducted a 10-group semi-structured focus group interview series on patients and physicians with constant comparative method. Participants were recruited from public and private clinics and teaching hospitals in Hong Kong.

Participants
Sixty-one invitations were sent out. One patient rejected due to family issues and another 2 rejected due to difficulties in allocating time for the study. Five CM physicians rejected due to work engagement and 3 did not respond to email invitations.
Fifty subjects were recruited (21 patients, 14 ConM physicians and 15 CM physicians) from October 2015 to June 2018. Patients aged 18 or above, diagnosed with diabetes / DKD and having follow-up at public outpatient clinics were included and sampled purposively from the consultations of general, renal and family medicine outpatient clinics of Queen Mary Hospital and Tang Shiu Kin Hospital to form 3 groups of 6-8 with diverse age groups, CKD stages, diabetes duration, comorbidities and education levels to enrich data coverage.
ConM physicians formed 3 groups of 3-6, and CM physicians formed 4 groups of 3-4. Physicians with experience in managing diabetes / DKD and were included and purposively sampled with diverse age groups, practicing duration, qualification and experience of ConM and CM, affiliating institutions, service sector (private / public) and involvement in research, administration and teaching.
Participants were recruited face-to-face, though telephone or by email. The purpose of interview was explained to the participants by the research coordinator (K.W.C.). The process of recruitment, interview and analysis were iterative until data saturation was observed.

Interview process and interview guide
The focus group interviews lasted 60-120 min (allowing at least 20 min time per participant per group for sufficient participation) and were conducted privately in meeting places near participants' workplace or residence. The interviews were facilitated by a moderator (P.W.L.) with 1) undergraduate public health training, 2) 2 years of in-depth interviews and focus group interviews moderation experience and 3) basic knowledge on CM, ConM and diabetes to ensure adequate skills and knowledge for conducting the interviews. The moderator was not involved in the conceptual design of the study, did not have prior contact with the study subjects, did not have conflict of interests with subjects and the research topic, and was informed about the group composition until 1 week before the interview. The identity of moderator was not disclosed to participants before interviews. He was working in public health administration sector but not involved in integrative medicine or diabetes issues in daily work during the period of study to avoid any potential conflict of interests. One to two clerical staffs were present at different interview sites for administrative support. The moderator recorded participants' demographics with simple questionnaire and mediated the interviews.
The discussion was built around consultation experience, concerns and expectations based on a semistructured interview guide (Supplementary File). The interview guide was piloted among other patients and physicians prior to the interviews. Minor ad-hoc adjustment was made to explore new themes. Patient group started prior to physician groups in each round of interview. In the first and early second round of interviews, the interviews focused on gathering detail experience in consultation encounters. As comments began to repeat in second round, emphases were put on contrasting previous groups' findings across subject groups (e.g. discussing findings of patient groups with physicians) and on seeking solutions. Data saturation was observed during the last round (patient and ConM: 3rd round, CM: 4th round) of interview in the respective groups and therefore sampling ceased [42,43].
Interviews were audiotaped and transcribed verbatim. Interviewees' identity was preserved by using individual codes. Brief notes on interview content and behaviour of interviewees were made real-time during interviews and subsequently combined with the transcripts. The transcripts were cross-checked between 2 independent researchers (K.W.C., P.W.L.).

Data collection and analysis
The transcripts were analysed using the constant comparative method [44] aided by simple software (Microsoft Word and Excel) for better access. The transcripts were first familiarised independently by 2 bilingual investigators (K.W.C., P.W.L.). Maximum codes on main themes and subthemes were generated independently to open up all fields during initial open coding. The transcripts were revisited to check for emerging ideas. The concepts and theories were refined, and the association of the coding was explored to form axial coding. A preliminary coding scheme was established with agreement by investigators. The preliminary findings were discussed with one representative from each group for validation and supplementation based on the interview content [45].
Core coding was confirmed for the dataset after observing no new themes at the last round of interview. The finalised coding scheme was applied to index and chart the whole dataset. Charted result was translated by a bilingual investigator (K.W.C.) when used as illustrative quotations to best preserve original meaning. Analysis was completed in October 2018. The analysis process is summarised in Fig. 1.

Demographics
Demographics of subjects are presented in Tables 1 and 2. Majority of patients had unsatisfactory glycemic control (71.4%), CKD stage 2-4 (95.2%) and albuminuria (90.5%). Eight (57.1%) ConM physicians were internists and 6 (42.9%) were family medicine specialists or general practitioners. 6 (42.9%) ConM physicians had informal CM education and all CM physicians (n = 15) had formal credit-bearing ConM education in anatomy, physiology, immunology, pathology and internal medicine as CM undergraduate programmes contained substantial portion of ConM elements in Hong Kong. Table 3 presented the CM and ConM education background of participants.

High-level themes
Seven high-level themes (barriers towards IM service, motivation to seek CM service, background knowledge on diabetes, experience of CM service, preferred model of integrative service delivery, evidence of IM and CM hospital) were identified with 25 subthemes. Data on barriers towards IM service and preferred model of integrative service delivery are summarised in Fig. 1. Additional quotes are summarised in Table 4.

Theme 1: barriers towards IM service
Seven subthemes were identified regarding barriers towards IM service, including finance, inconvenience of access, advice from medical professionals, uncertainty of service quality, uncertainty of CM effect, difficulty in understanding CM epistemology and access to medical records.
Finance Finance was identified as a major practical issue against the use of CM / IM by patients. CM service in Hong Kong is financed through insurance and out-ofpocket payment. Government subsidised over 90% for ConM service [46] and comparable CM service for diabetes / DKD operated by non-governmental organisations is around 100 times more expensive [47]. The chronic nature of diabetes / DKD intensified the impact of finance.
Inconvenience of access The inconvenience of access to CM service, including frequent consultation and the use of raw herbs was also frequently mentioned among patients, although majority of them (61.9%) were unemployed / retired. Patients had a longer follow-up period of 3-6 months for ConM services when compared to 1-2 weeks for CM. Although CM physicians anticipated that inconvenience would affect compliance, they believed it was inevitable due to the personalised nature of CM.
"CM is very troublesome. We need to treat the (herbal) medicine and attend consultation frequently. ConM doctors give you 3-month medication." (Patient 5) Advice from medical professionals While some patients were advised against CM by ConM physicians, they hesitated to comply as they believed ConM physicians could not decide the effect of CM. We sought to delineate the underlying mechanism of discouragement and the lack of understanding of CM effect (especially on the interactions of IM) emerged as the key reason and is further dissected. Uncertainty of service quality Uncertainty of service quality was a common concern among patients, ConM and CM physicians. Patients doubted CM physicians' academic qualification and questioned the experiencebased consultation of some CM physicians.
"They (traditional CM physician) prescribe medicine based on experience. I do not know the effect after taking their medication." (Patient 18) ConM physicians were concerned about the standardisation of CM, interaction between CM and ConM and the background nephrology knowledge among CM physicians.
"If they (CM physicians) can prove CM does not adversely interact with ConM and they are registered, I do not disagree." (ConM 12) "I advised them (patients) not to approach CM physicians who are not affiliated with hospitals or NGOs as I am not sure about their understanding on kidney disease or replacement therapy." (ConM 14) Uncertainty of CM effect All parties were uncertain about the efficacy of CM. A patient who experienced suspected CM nephrotoxicity suggested that the response to CM was personalised, as he experienced toxicity despite others' positive feedback. CM physicians also acknowledged that there may not be adequate evidence to inform specific management. However, there were ambivalent opinions on the study design for further research, especially on clinical trials.
"For CM, there is limited information on who is more responsive to treatment. You said you (another patient participant) got better after taking CM but I got worse. ConM is different." (Patient 18) "As CM takes into account many factors including geographical and climatic effect … CM may not be   However, ConM physicians suggested that it is hard to understand and translate CM-related clinical outcomes to their practice.
"(CM theory) is difficult to me. Even if they (CM physicians) wrote consultation notes, I could not understand … I do not understand how they monitor disease progress." (ConM 1) Access to medical records ConM physicians raised concern that the CM medical record, especially handwritten prescription, was often unreadable. This imposed a risk of mishandling of patients under CM service. Also, ConM physicians were worried that the restricted access of investigations among CM physicians may affect clinical management. However, CM physicians who advocated traditional practice were less relied on investigation while other CM physicians suggested that investigations would help alert them critical conditions that require prompt management. Consultation mode Majority of participants suggested that there was a need of physically combined service, preferably at the same institution. This would improve the confidence of patients and communication between physicians, leading to more efficient patient management. Patient also expected physicians to offer compromised IM treatment protocol as clinical solutions. Both ConM and CM physicians acknowledged that the combined treatment should be formulated according to the appropriate theory basis, i.e., CM treatment directed by CM theory and vice versa.
"If you reverse the role, you ask all ConM physicians to feel the pulse before prescribing ConM medicine, they will also feel uneasy. … ... We all do not want to follow a protocol that does not match our practice." (CM 9) "I think collaborative service has a great potential but theory integration almost looks impossible. Even if we have a pharmacy offering both ConM and CM, it does not mean that CM physicians would accept us (ConM physicians) to prescribe CM based on ConM theory. Vice versa, it would be hard for us to accept CM physicians prescribing mycophenolic acid because the patient is having 'qi deficiency'." (ConM 7)

Discussion
This is the first focus group study comparing the expectation of patients and physicians regarding the use of IM service for diabetes, with the participation of family medicine, internal medicine and CAM physicians in addition to patients. We demonstrated that there was an g attended any Chinese medicine (CM) theory-related sessions including Chinese herbal medicine, acupuncture, moxibustion, orthopedics and health management h attended any biomedical or epidemiological sessions involving clinical medicine i defined as credit bearing sessions j defined as non-credit bearing sessions including health talks, seminars and other forms of information sessions. All CM physicians were exposed to formal conventional medicine (ConM) education included but not limited to anatomy, physiology, microbiology, immunology, pathology, diagnosis, pharmacology and internal medicine. Postgraduate doctoral clinical training was presented as a comparable qualification of specialist training of CM physicians as there is no CM specialist registration system in Hong Kong. Patients with informal CM and ConM educations were purposively sampled for second and third round. ConM physician with formal CM education were purposively sampled for second round expectation mismatch between patients (prefer physicians to design integrative treatment protocol and prefer biomarkers for disease monitoring), CAM physicians (believed diabetes patients focused on quality of life) and physicians of conventional medicine (prefer patients to make the choice of treatment).

Justification of study design
We investigated with a qualitative approach to explore unknown mechanism in the underutilisation and underreporting of CAM among diabetes patients and physicians and conducted focus group interviews to enhance interactions between participants with different experience. The interview guide was designed to explore more practical concerns and actual consultation experience from implementation perspective as existing literature is limited in this aspect. The guide was slightly tilted towards exploring how to integrate CM service into existing ConM system as majority of diabetes patients were receiving ConM as background treatment locally and internationally. The bias of interviewing skills and contents was minimised with an independent experienced moderator. Patients and physicians with different background hypothesised to offer divergent perspectives (e.g. complications for patients, training and work nature for physicians) were purposively sampled to enrich the content and coverage. As possible underreporting of IM utilisation was previously reported [41], we separated patients and physicians into different groups to encourage expression of comments and experience, and to minimise dominant opinion. Data saturation, defined by having no emerged new key themes, was reached in the third round of interview among patients and ConM physicians, and the fourth round of CM physician interview. This reconciled with previous review and empirical studies that 80% of themes could be captured in 2 to 3 rounds of interview and data saturation usually occurred in the 3 to 4 rounds [42,43]. Only two high level themes were presented as this was a large-scale focus group study and we are presenting the most relevant themes and subthemes to maintain focus in discussion.

The unaddressed and less documented preference and practical concerns
Existing literature regarding the obstacles of IM emphasised cultural believe and research-related capacity. Our study echoed as uncertainty of service quality and effect of CM were identified as shared concerns among all stakeholders as summarised in Fig. 1.
Our findings further highlighted the importance of mutual understanding between physicians of different streams as any advices regarded unreasonable and unexplainable by patients would result in non-compliance and underreporting. In addition to the existing understanding, practical issues including inconvenience of access, frequent follow-up and financial burden were stressed among patients in this series, which were less documented and is likely related to the chronic nature of diabetes. Besides, we found that ConM with exposure to CM knowledge preferred more pragmatic and clinical evaluation of IM service as they recognise there is a fundamental difference in epistemology between ConM and CM. We also documented more practical concerns from the medical community for the consideration of further IM study designs.

Implications for practice, organisation and policy
Convenience of access was frequently quoted by patients as a key barrier while it was not concerned by CM physicians. IM would lead to increased number of clinical visits. While frequent follow-up may provide patients with the best-tailored management, the required effort would compromise patients' quality of life and increase the economic loss. Combined IM consultation under the same organisational structure was generally preferred by our subjects (both patient and physicians) as it enhances quality control, reduces time cost for consultation and facilitates sharing of medical records among physicians. Integrated clinics are increasingly common globally and provide a platform for physicians from different streams to design and evaluate IM protocols [48]. High quality decoctingfree herbal granules provide patients more standardised and user-friendly administration for long-term use [49].
In contrast to many existing IM models globally that the decision and liability of IM use was shifted to the patients, requiring patients to actively search and critique the evidence themselves, our data suggested that patients expected physicians of different streams to optimise the IM protocol and offer integrated evidence-based recommendations for their informed decision making. This will require centralised and concerted effort in evaluating IM efficacy with parallel mechanism of existing statutory bodies, for instance NICE (UK), FDA (US) and NMPA (China), if not integrating IM procedures into these existing infrastructures.
Interestingly, CM physicians from private sector believed the access cost of CM has driven patients to demand extra efficacy and quantifiable evidence when evaluating CM, which amplified the effect of lack of evidence, a well-known barrier. Patients reconciled that cost was one of the key barriers of using IM service. In Hong Kong, CM service is mainly financed through out-of-pocket payment and private insurance whereas ConM service is over 90% publicly subsidised. Comparable CM service of diabetes operated by NGOs is around 100 times more expensive in long run. Government policy on CAM service financing emerged as a decisive factor of implementation.

Implications for education and research
Physicians from both sides called for more basic IM knowledge from their counterparts for better communication. In China, most physicians are double qualified and practise ConM and CM simultaneously. While dual training may provide a comprehensive foundation for IM practice, there are criticisms on the ungrounded integration between ConM and CM on physician level as suggested by the interviewees. As many ConM and CM interventions have been shown to have interactions, IM service requires specifically designed IM-oriented research and evidence instead of simple integration of disaggregated evidence from either side. Also, dual training is difficult to propagate to health systems of other countries and requires substantial resources. The effectiveness and optimal ratio of one physiciandual medical practice model and one administrative systemdual streams of physician model in health systems also require further implementation studies.
Regarding research, both ConM and CM physicians acknowledged that research design should formulate according to the corresponding school of theory of ConM and CM to reflect actual practice. Nevertheless, ConM physicians emphasised more on standardisation and reproducibility while CM physicians focused more on personalisation on study design which highlighted the different epistemology between CM and ConM. Both CM and ConM physicians of different specialties proposed pragmatic trial as an agreeable solution which aligned with the trend of establishing real-world comparisons lately [19,43,50]. Patient selection, intervention, outcome measurement and analysis of the IM studies should take into consideration the realworld practice of different streams of medicine and patient preference.

Limitations
This study has several limitations. Since the aim of this series was to identify detailed barriers, concerns and expectations on IM diabetes management, the findings are context specific, similar to other qualitative studies. However, we believe the major findings could be carefully generalised to the IM management of chronic conditions in places where CAM is not part of the public health system, including most of the developed countries. Also, this series only identified possible mechanisms of social behaviour and further quantitative studies including survey are needed to better determine the magnitude of and prioritisation of the concerns. Lastly, the study period was long due to the availability of funding and difficulties in organising focus groups, especially for physicians, which is a commonly known challenge.

Conclusion
Inadequate explanation of CAM effect to patient may lead to underreporting. Practical issues on the access and finance of medical service emerged as major barriers to the delivery of IM service from the patients' perspective. Inadequacy in evidence of clinical effectiveness and the difficulties in interpreting CM theory remained as the major obstacles in the consideration of CM among ConM physicians. Pragmatic clinical evaluation is mutually accepted by both ConM and CM physicians. Research design should focus on the add-on effect of CAM and consider standardisation, reproducibility and the real-world practice. To enhance the utilisation, IM programmes that are evidence-based should merge with existing infrastructure.