A wealth of opportunities: an ethnographic study on learning to deliver high-value, cost-conscious care.

Objective: Since physicians behaviour determines up to 80% of total healthcare expenditures, training residents to deliver high-value, cost-conscious care is essential. Residents acknowledge the importance of high-value, cost-conscious care-delivery, yet perceive training to be insufficient. We designed an observational study to gain insight into how the workplace setting relates to residents high-value, cost-conscious care-delivery. Design: This ethnographic study builds on 175 hours of non-participant observations including informal interviews, 9 semi-structured interviews and document analysis. Setting: Department of obstetrics and gynaecology in an academic hospital in the Netherlands. Population or sample: 21 gynaecology residents. Methods: Iterative analysis process of fieldnotes, interview transcripts and documents, including open-coding, thematic analysis and axial analysis by a multidisciplinary research team. Results: Residents rarely consider health care costs, and knowledge regarding costs is often absent. Senior consultants guide residents while balancing benefits, risks and costs, with or without explicating their decision-making process. Identified learning opportunities are elaboration on questions raised concerning high-value, cost-conscious care, checking information about costs that are used in discussions about high-value, cost-conscious care, and having a more open and explicit discussion about high-value, cost-conscious care. Conclusion: Our study emphasizes that the opportunities and potential to train residents to deliver high-value, cost-conscious care in the workplace are present. The challenge resides in capitalizing on these opportunities. We suggest departments to consult external experts to facilitate discussions regarding high-value, cost-conscious care to contribute to informal learning and to create a workplace setting in which high-value, cost-conscious care-delivery is prioritized. Funding: none Keywords: medical education, high-value, cost-conscious care, residency training, ethnography.

Results: Residents rarely consider health care costs, and knowledge regarding costs is often 17 absent. Senior consultants guide residents while balancing benefits, risks and costs, with or 18 without explicating their decision-making process. Identified learning opportunities are 19 elaboration on questions raised concerning high-value, cost-conscious care, checking 20 information about costs that are used in discussions about high-value, cost-conscious care, and 21 having a more open and explicit discussion about high-value, cost-conscious care. 22 Conclusion: Our study emphasizes that the opportunities and potential to train residents to 23 deliver high-value, cost-conscious care in the workplace are present. The challenge resides in 24 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https: //doi.org/10.1101/19011916 doi: medRxiv preprint Introduction 1 Rising healthcare expenditure significantly impact the sustainability of healthcare systems. 1-3 2 Limiting costs whilst ensuring the quality of care is therefore high on the political and medical 3 agenda. 4 Since physicians' behaviour determines up to an estimated 80% of total healthcare 4 expenditures, shaping the behaviour of future physicians is a potentially valuable intervention 5 for ensuring delivery of high-value, cost-conscious care (HV3C). [5][6][7][8][9] HV3C is defined as 'care 6 that aims to assess the benefits, harms, and costs of interventions' that leads to 'care that adds 7 value' for the individual patient and the population in general. 10 HV3C is characterized by 8 three key principles: 1) it avoids wasteful care that has no added value for the patient, 9 therewith reducing unnecessary costs; 2) it aligns with the personal preferences and needs of 10 the patient; and 3) it is based on the latest evidence. Workplace based learning in the area of HV3C is increasingly investigated 9,20 and results 24 indicate that residents perceive HV3C training to be insufficient. 9 These findings are worrying, 25 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19011916 doi: medRxiv preprint since workplace based learning covers the majority of residency training 21 and lessons from 1 graduate medical education persist long into practice. [22][23][24] We aimed to build on and expand this 2 research by obtaining a real-life perspective on the workplace setting. Therefore, we designed 3 an observational study to gain insight into how the workplace setting is related to residents' 4 HV3-delivery. 5 6 Methods 7

Research design 8
We conducted an ethnographic study to gain insight into how the workplace setting is related 9 to residents' HV3C-delivery. Ethnography is a qualitative research methodology, commonly 10 used to study cultures, by which data is collected through fieldwork. 25,26 In this case fieldwork 11 consisted of non-participatory observations, informal interviews, semi-structured interviews 12 and document analysis to generate insight into the sample population's views, motives and 13 actions in the real-life setting. 25,27 For a more elaborate description of ethnography as a 14 methodology and insight in the iterative process of data-collection and data-analysis, we 15 would like to refer to Appendix A. 16 17 Setting 18 The fieldwork was conducted in an academic hospital in the Netherlands at a Department of 19 Obstetrics and Gynaecology. At the time of the study, a national government-subsidized 20 project was running aimed at increasing awareness of HV3C during residency training 28 . This 21 project provided educational material and e-learning opportunities to stimulate residents to 22 voluntarily set up projects to improve HV3C delivery in their department. During our data-23 collection no formal education was present to teach residents HV3C-delivery. 24 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19011916 doi: medRxiv preprint 7 1 Subjects 2 All 21 resident physicians working in the gynaecology department at the time of our study 3 participated. Prior to the study they were informed about the research by letter and attended a 4 short presentation during the morning handover, in which we explained the primary research 5 goal as 'we want to gain insight into how the workplace setting is related to residents' HV3C-6 delivery'. Written informed consent was obtained from all participating residents. The Ethical 7 Review Board of the Netherlands Association for Medical Education approved this study on 8 August 24, 2017, under file number 881. 9 10 Data collection 11 The second author (LMJ) shadowed gynaecology residents in daily practice for 175 hours 12 during the period of three months. LMJ accompanied the residents during patient consultations, 13 surgical procedures, deliveries and medical rounds, and attended meetings, e.g. morning reports, 14 educational activities and (multidisciplinary) seminars, as well as lunch and coffee breaks. 15 These non-participant observations (i.e. the researcher was not involved in care delivery) 16 focused on residents' behaviour, conversations and situations regarding care delivery. 17 Observations were followed by unstructured and ad hoc informal interviews with the aim of 18 gaining thorough comprehension of the context and rationale of the observed moments. 29,30 19 Additionally, semi-structured interviews were held (by LMJ and LAS) with nine purposively 20 selected participants to gain additional information about specific observed situations and the 21 department in general from representatives of the observed population. Participants were 22 sampled based on the level of involvement in discussions and conversations regarding HV3C, 23 both residents and staff-members with high and low levels of involvement were selected. All 24 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19011916 doi: medRxiv preprint 8 interviews were audio-recorded and transcribed verbatim. Document analysis of hospital and 1 national protocols and patient charts was executed to see, when residents or supervisors referred 2 to protocols, whether their claims were supported as well as whether the decision-making 3 process was retraceable. 4 5 Observed behaviour and conversations were thickly described, written up in detailed, context-6 sensitive and locally informed fieldnotes about the observed events 25,31 . Data collection 7 involved iterative cycles of data collection and analysis, combining observations, interviews 8 and analysis together with the research team (see Appendix B). 25,30 The iterative process of 9 data collection discontinued when data-saturation was reached, meaning data-analysis did not 10 result in new concepts and an adequate understanding of key concepts was gained. 26 Data 11 saturation was reached after approximately 175 hours of observation and 9 semi-structured 12 interviews. 13 14 Data-analysis 15 As is common in ethnography, the data were analysed iteratively (see Appendix B) starting 16 with line-by-line open coding by LMJ, LAS and GB. 25,32 The next steps were thematic and 17 axial analyses, in which we identified central themes and relationships between the various 18 themes. As a result of team discussions, we gradually adapted the focus of the observational 19 themes and conducted semi-structured interviews in order to deeper our understanding of the 20 data. Additionally, FS checked if the observations represented national care delivery and not a 21 specific regional setting. We are aware of the influence of the researchers' background on the 22 process of data-collection and data-analysis, to increase transparency we included a 23 reflexivity-paragraph in Appendix A. 24 25 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19011916 doi: medRxiv preprint

Results 1
The consideration of benefits, risks and health care costs in daily practice 2 When protocols, guidelines or regional agreements failed to provide clear direction towards 3 care delivery, or when new scientific information became accessible, health care professionals 4 needed to decide which care delivery was considered 'appropriate'. Health care professionals 5 then considered pros and cons, including concerns for patients, consequences of a missed 6 diagnosis, respect for autonomy of patients, burden of testing or treatment, influence of 7 quality of life and long-term health benefits. We rarely observed attention for health care 8 costs, or conversations regarding who would eventually pay for those costs. In cases in which 9 costs were included in the process of balancing pros and cons of health care services, the 10 absence of knowledge regarding costs became apparent. Questions regarding financial costs 11 arose when residents and senior consultants discussed a scientific article, reviewed the 12 protocol, or discussed the necessity of structural testing. Observations demonstrated that 13 financial reasoning did not influence the process of decision making. We did not observe 14 occasions in which health care costs were checked as a result of such discussions. 15

16
It is the end of the morning round and medical students, supervised by the resident-in-training 17 are critically appraising a procedure. The resident noticed differences between the value and 18 use of repetitive PCR-ratio (protein-to-creatine)  The influence of the senior consultant on the process of balancing benefits, risks and costs. 5 HV3C-delivery by residents was guided by senior consultants' preference. In some cases, the 6 senior consultant extensively 'walked residents through' the process of balancing all relevant 7 stakes that led to their proposed diagnostic or treatment plan. In other situations, the senior 8 consultant stated their opinion of what they considered to be appropriate care without further 9 explanation. Although residents not always understood nor agreed with the senior consultant, in 10 general they did not request insight in the senior consultant' considerations regarding what was 11 deemed appropriate. Residents tended to execute the advice of the senior consultant with little 12 countering of the consultants' judgment. 13 14 Variation between senior consultants was prominent and recognizable for residents and was 15 sometimes considered confusing especially in the eyes of junior residents. Senior residents used 16 these differences to get agreement on their own proposed plan by being deliberate in whose 17 senior consultant's advice they sought in certain cases, as mentioned by resident R17 in an 18 informal interview. Based on the analysis of our observations, informal and formal interviews, it became apparent 5 that there were three main areas for learning which could be improved. First, although residents 6 and staff-members raised questions regarding what constitutes high-value, cost-conscious care, 7 we observed that questions regularly remained unanswered or undeliberated. In these situations, 8 conversations were discontinued (for example by raising a different question) or paused ('we 9 should check that'), without revisiting raised questions or checking/changing the course of 10 action. Our observations also demonstrated questions followed by (collective) discussions, 11 explicating the process of balancing benefits, risks and costs of care delivery. We asked both 12 residents and supervisors how questions related to care delivery could be enhanced to facilitate 13 residents' learning. Residents shared that they were in need of a more structural review of 14 patient cases, followed by an organized overview of benefits, risks and costs of the case at hand. 15 This could be done for example during the morning handover when discussing a patient case. 16 Secondly, decision making often was based on assumptions about costs without efforts to check 17 them. This was observed for example when costs were considered 'high' yet actual amounts 18 were not known or the meaning of commonly used terminology (for example what do we mean 19 by 'bedrest') was not understood or defined. Thirdly, face-to-face conversations about care 20 delivery were absent in instances where the line of reasoning was not clear for either staff or 21 residents. Although senior consultants were sometimes openly critical to residents about how 22 colleagues balanced benefits, risks and costs of care delivery, this did not result in a face-to-face 23 conversation about the delivered care and consequently did not result in a take-home message 24 for residents. For example, critiques often ended with comments such as; "they probably had a 25 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19011916 doi: medRxiv preprint good think about it" or "let's continue with the next case". Supervisors confirmed that 1 especially formal meetings created interesting teaching opportunities but were hesitant to 2 deepen the discussion in those meetings. Supervisors motivated this by describing how they 3 aimed to maintain a safe learning environment (do not give the resident the impression their 4 decision was 'false') while also exposing residents to the uncertainty that is present in care 5 delivery (modelling that there is more than one 'correct' treatment decision). 6 7 8 Discussion 9 Summarising 10 Training residents to provide HV3C is essential for the sustainability of healthcare 11 systems. 2,5,9,10  The importance of knowledge-transmission for the training of high-value, cost-conscious care 21 delivery has been emphasized in previous research. 7,9 We observed that physicians aim to 22 make health care costs part of the discussion, yet lack information to do so highlighting the 23 importance of knowledge of health care costs in particular. Besides the lack of formal 24 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https: //doi.org/10.1101/19011916 doi: medRxiv preprint education, another known hurdle for knowledge transmission is the complexity of health care 1 economics and the lack of transparency of costs in health care systems. 33 2 3 The consultant as leading role model 4 The leading role of the senior consultant in health care delivery in general, 9,19,20 and HV3C 5 specifically 9,33 has been previously identified. Through observing role models' care delivery 6 residents learn professional behaviours, particularly from those senior consultants that match 7 their personal views. 34 Data-analysis demonstrates that senior residents intentionally seek 8 advice from senior consultants who are likely to share their views on HV3C-delivery. 9 Training opportunities lie in pursuing a workplace setting where senior consultants 10 deliberately assist residents in explicating their reasoning regarding HV3C, with or without 11 sharing the same view. This can be achieved by inviting residents to discuss, assess and 12 reflect on future and past care delivery. Practice variations between consultants and 13 departments could function as a starting point for these discussions and reflections, 35 challenging in a workplace setting that is considered to be hierarchical. 38 It is not surprising 23 that the current workplace setting is not familiar with discussing care delivery from a HV3C 24 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19011916 doi: medRxiv preprint perspective since issues of accessibility and affordably of health care are recent. Nevertheless, 1 current budgetary crises in most health care systems urge for brisk culture change. To initiate 2 a culture change, we suggest focussing interventions on senior consultants and other staff 3 members. Changing the overarching culture and stimulating residents to actively engage in 4 HV3C is desirable. Therefore, future research should investigate how departments can be 5 supported in developing a critical attitude towards care delivery in light of HV3C. Action 6 research 39 can be a suitable strategy to challenge conventional thinking about HV3C in the 7 clinical setting. In such design, integrating an independent external expert to regularly 8 question care delivery, and guide group discussions in an open and tolerant manner, may 9 speed up consciousness of value for patients, costs and the urge for bottom up system change. 10 This 'devils' advocate' could help to fill in the opportunities detected in this study; 11 elaboration on questions raised concerning HV3C, checking information about costs that is 12 used in discussions about HV3C, and having a more open and explicit discussion about 13 HV3C. The ultimate achievement would be that staff members and residents would take over 14 this role and challenge each other to underpin their decision-making process in HV3C or 15 structurally reflect on HV3C-delivery via group-discussions. 16 17

Strengths & limitations 18
The main strength of this study is the use of ethnography to gain insight in how the workplace 19 setting is related to residents' HV3C-delivery. A major strength of ethnography is the ability to 20 observe things that those who are routinely involved may not see as well as collecting data that 21 participant might not be willing to share with researchers. 26 Another strength is the rigorous 22 performance of our data collection and analysis, in compliance with quality criteria of 23 qualitative research. Our study has several limitations. Participating residents might have been 24 more favourable towards HV3C-delivery, as they were aware of the researchers' presence and 25 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi. org/10.1101/19011916 doi: medRxiv preprint