A Latin American, Portuguese and Spanish consensus on a core communication curriculum for undergraduate medical education

Background To present learning outcomes in clinical communication for a Core Curriculum for medical undergraduate students in Latin America, Portugal and Spain (LAPS-CCC) and to establish an expert network to support a transnational implementation. Methods Through an iterative process, an international group of 15 experts developed an initial set of learning outcomes following a review and discussion of relevant international and local literature. A two-round Delphi survey involving 46 experts from 8 countries was performed. Quantative and qualitative analisis permited the definition of the final consensus. Results The initial proposal included 157 learning outcomes. The Delphi process generated 734 comments and involved the modification, deletion and addition of some outcomes. At the end of the process, a consensus was reached on 136 learning outcomes grouped under 6 competency domains with a high overall acceptance (95.1 %). Conclusions The learning outcomes of this proposal provide a guide to introduce, support and develop communication curriculae for undergraduate medical studies in the countries involved or in other Spanish- or Portuguese-speaking countries. Electronic supplementary material The online version of this article (doi:10.1186/s12909-016-0610-8) contains supplementary material, which is available to authorized users.


Background
In healthcare, communication between health providers and patients and their families has become a core competency for achieving quality objectives [1]. Research has shown positive associations not only between physicianpatient communication and important healthcare outcomes [2][3][4], but also between specific educational strategies and the acquisition and application of communication skills by students and health providers [5][6][7].
There are a number of documents defining good practice in physician-patient relationship and that can be useful for planning and developing training programmes and appropriate assessment strategies in this area [8][9][10][11]. In addition, several interesting and well-structured proposals have been published in Canada, UK, Germany and Europe [12][13][14][15]. These proposals have been reached by consensus and are intended to be reference-guides for the learning outcomes (LO) on communication that should be included in the

Methods
In brief, a Scientific Committee made up of 15 members which were established ad hoc, undertook a review of the literature on the subject. They exchanged relevant information on the matter and later implemented a variant of the Delphi method in order to reach consensus of an international panel of 51 experts on clinical comunication. A Steering Group (CGL, RRM and FCM) located at the Universidad Francisco de Vitoria (Madrid) took charge of the project organisation and coordination on behalf of the Scientific Committee. The surveys were carried out electronically through a specific website accessible to panelists.
Preliminary phase of the project: forming the scientific committee and preparing the survey contents SC members collected and shared any literature on conceptual frameworks, curriculum proposals, teaching guides, training programmes and institutional educational reports from their respective countries, relevant to physician-patient communication teaching and LO for medical studies. A significant number of documents were obtained not only from SC member countries, but also from other Latin American countries not represented in SC . A selection of the information was disseminated to all SC members together with a key set of international articles and consensus documents [8][9][10][11][12][13][14][15]37]. The SC members were asked to review and analyse this material and to make comments and their own proposal of communication LO. With all the information received, a first expanded list of learning outcomes LO was drawn up.
At this stage, the SC members also discussed a conceptual proposal on person-centred care that was realistic for clinical practice and applicable for teaching in our context [55]. This proposal identifies and highlights several dimensions or competency domains considered relevant in the context of healthcare relationships (Fig. 1). The final version obtained from this discussion was used for clustering and hierarchizing the LO subject to consensus [54]. The LO were phrased according to Bloom taxonomies [56] and defined as skills that are demonstrable, focusing on observable results [57]. SC members also received information on what was considered a LO, specifying how they should be worded.
Thus, a first draft of 157 LO was obtained and grouped under 6 general competency domains. SC members discussed this first version and proposed possible changes. The set of proposals for modification, deletion or addition of new LO or relocation of LO within the communication domains as well their supporting comments (227) were received by the Steering Group. A synthesis of these inputs was returned up to three times to the SC members, who discussed and approved a final version with 143 items, which was used to start the Delphi survey.

Selection of the international expert panel (October 2013)
Eligible experts, i.e. individuals who are widely recognised by peers for their extensive knowledge, ("epistemic" expertise), and/or performance ("performative" expertise) [58] were sought. For the selection of experts, a snowball sample technique was used [59] starting from the personal contacts of the SC members. The new "eligible experts" identified were also asked to nominate additional candidates. Thus, a list of 104 "eligible experts" were drawn up. In order to reduce the final size of the panel and select the most suitable candidates, each "eligible expert" was asked to choose from this list, those they considered as most renowned in this field (in relation to specific aspects such as academic leadership, impact publications, research experience, etc.) and to classify each "eligible expert" into one of the following three categories: unquestionable expert, possible expert and without opinion on the candidate's expertise. Focusing especially on those that received classification as unquestionable or possible expert from several different institutions/countries, the final panel was constituted, including 51 experts from Argentina, Brazil, Chile, Colombia, Spain, Mexico, Peru and Portugal: 37 physicians (4 planners and/or healthcare managers in public and private medical institutions and 1 representative of an organisation for patients' defence), 5 psychologists and 2 pharmacists, 1 philosopher and 1 PhD in Social Sciences.

Delphi survey phase
The "modified" Delphi technique [60] avoids the iteration of an excessive (and undetermined) number of survey rounds to achieve stability in groupal responses, allowing a free exchange of views in a faceto-face meeting celebrated between two sole confidential surveys. Although this method is more feasible than the original Delphi method [61], the local meeting defeats one of its cornerstones (anonymity of participants to prevent bias). To avoid this, we used a widely spread and useful variant of the modified Delphi technique with the following characteristics: limited number of rounds (two), no face-to-face encounter, participation of the panel members facilitated by providing detailed feedback between rounds and anonymous survey scores and written views of each participant [62][63][64]. Between the two rounds, panelists were informed about the group's responses to the first survey (statistical and graphical description) and received anonymous comments and clarifications that each participant had made. After reviewing this information, they participated in the second round in which they made further assessment on those LO that did not reach consensus in the first round. The electronic circulation of surveys accelerated the process and facilitated monitoring of those who did not reply on schedule. Close follow-up by the technical office was paramount in the process. This process took 12 weeks (October to December 2013). (Additional file 1 with COREQ checklist, providing more methodological details).

Analysis of the consensus degree Quantitative analysis
The panelists assessed each LO using an ordinal 9 level Likert scale, according to the format developed at UCLA-Rand Corporation [65]. The response categories on this scale are described by semantic descriptors in three regions (1-3 = "disagree"; 4-6 = "neither agree nor disagree"; 7-9 = "agree"). For each LO, the respondent could define their individual opinion in detail, by choosing one of the nine available points. To analyse the group opinion and type of consensus reached, we used the position of the median scores (in which region) and the level of agreement reached by the respondents. Thus, the median score determined the group opinion as "disagreement" on the LO if the median ≤ 3, "agreement" if the median ≥ 7. If the median score was in the 4-6 region, the group opinion was considered "doubtful". Consensus about one LO was defined when there was panel concordance: that is, when the number of experts scoring outside the region containing the median (the outliers), was less than a third of the respondents. Panel discordance was defined when the scores of a third or more of the panelists were in the region [1][2][3], and another third or more in the region [7][8][9]. In those LO where there was neither panel concordance nor discordance, the group opinion was considered as "undetermined". Any LO which did not obtain either a favourable or unfavourable consensus (ie all LO with doubtful, discordant and undertermined opinions), was included in the second Delphi round for re-assessment by the panel. Any LO that no obtained consensus, or obtained a favourable consensus but more than 25 % respondents' scores outside the region including median was defined as inappropriate.

Qualitative analysis
All comments made by panelists during the Delphi survey were analysed and a list of codes (or categories) was derived inductively from the comments. Three reviewers (RRM, CGL and FCM) independently analysed the comments in more detail and encoding all comments, grouping them in these initial categories or creating new ones. Discrepancies arising from this coding were resolved by verbal agreement between reviewers.

Results
46 (90 %) experts concluded the two rounds of the Delphi survey. During this process, the panelists made 734 comments (695 for the first round and 39 for the second). At the end of the process, consensus was obtained for 136 (95.1 %) LO. Table 1 (original Spanish and Portuguese version in Additional files 2 and 3) displays these LO, detailing for each one, the mean score, the median, the number of outliers (expressed as a percentage of all respondents) and the interquartile range distribution of scores. Table 2 describes the 34 LO with a mean score greater than 8.34 (the highest 25 %). Most of these items had a maximum of 5 % of experts who disagreed; items 11, 103, 104, 105 and 128 had almost 7 % who disagreed, and in items 12, 30, 43, 62, 63, 74 and 87, 100 % of panelists agreed. The LO which were defined as inappropriate (no obtained consensus, or obtained a favourable consensus but more than 25 % respondents' scores outside the region including median) are shown in Tables 3 and 4 shows the evolution of objectives from the initial proposal during the Delphi process. Except for category E (Communication through different channels) where a higher percentage of outcomes (57 %) had a level of discordance of between 10 and 20 %, all domains of competence got a high concordance among experts.

Qualitative analysis
All panelists made comments during the Delphi process and by the end of the survey, 734 comments had been received (695 in the 1st round and 39 in the second) and were classified into 14 categories. Most of the comments highlighted the relevance and importance of the LO to be included in undergraduate medical curricula. Other comments were related to: (i) Practical implementation of LO in undergraduate teaching, (ii) Clarification of the LO meaning, avoiding cultural misunderstanding, (iii) Non-exclusive communication nature of LO (iv) LO related to the clinical experience and personal maturity of 2. Explains the professional and patient relationship models (focused on the professional, on the patient, on tasks, on the process, mixed…).  The student… 37. Critically assess the scientific findings on transferring information to patients and the implications in clinical practice.   54. Explores the patient's needs, resources (information, autonomy, trust, responsibility, psychological features…) and willingness to get them involved in the decision making.

Discussion
The LAPS-CCC consensus reached in this study has 136 LO grouped into 6 domains considered important for clinical communication. This consensus was created by 46 experts from 8 Spanish and Portuguese-speaking countries and refers to "nuclear communication learning outcomes for medical students". This consensus is intended as a proposal for the medical schools in LAPS cultural context to be used to facilitate the designing and development of their own clinical communication curricula. The review and discussion of literature, especially that provided by the SC members relative to their countries, confirmed the importance of creating a LAPS consensus: there is a lack of guidelines for communication teaching in LAPS, many universities do not an explicit curriculum for communication or are just developing communication skills teaching and there is great diversity in how communication is being taught. As in other consensus proposals that inspired this work [12][13][14][15], this consensus proposal considers that effective clinical communication is determined by those communicational skills related directly with the physicianpatient interaction, as well as skills to communicate with the family, teamwork, and for reflective practice. Communicating through various communication channels and in specific contexts also requires specific skills [13,54]. The grouping of the LO into these domains reveals some of the particular features of this proposal. The higher number of LO [65] included in the general "patient communication" competency domain points to the importance of patient-centred care: many of these LO are related to communication tasks that have been shown to be important in helping students understand the purpose of the medical consultation and in achieving consultation goals in clinical practice [9][10][11][12][13]. Another distinguishing aspect of our consensus is the numer of LO [14] related to "communication with the family". Although there are significant differences in the role of the family in the different LAPS countries depending on the degrees of "Europeanization" or the type of community (indigenous, rural, urban) the values of loyalty, respect, solidarity and reciprocity among family members are deeply rooted in the LAPS cultures and have been contrasted with North European and North American 129. Approaches some of them in a sensitive and constructive manner by applying and adapting the core communication skills and using the specific strategies and skills that each may require.   individualism. This implies the need to emphasize specific aspects of physician-patient/family relationship that this consensus reflects to a greater extent than others [66,67].
Most LO in the proposal are more attitudinal and behavioral than cognitive and to incorporate these LO into the medical curriculum will entail a number of practical challenges for our institutions. Experiential teaching methods will need to be prioritized [6,68] and wellstructured feedback using reliable and validated tools will be required [69][70][71]. For students to achieve these LO, they will need repeated exposure to a variety of clinical situations in which they can be observed and have time for learner's self-reflection. Teaching will need to include not only cognitive-behavioral strategies, but also reflective methods (i.e., reflective small groups, or directed reflection) [72], not as an isolated course, but throughout the curriculum, and with teachers who have been adequately trained [73,74]. This proposal of LO should be able to help many schools to design and modify their programmes in a more efficient way [27]. However, it is important to note, as many panelists did in their comments, that, despite the importance of introducing these LO, it may not be an easy task for young students, some of whom have little practical experience during the whole of their undergraduate programme. Many of these objectives require the student to grasp subtle aspects of the hidden curriculum and develop intuitive thinking [75], which may only be achieved by a slow incubation period into the milieu and frequent practice [76,77].
With such challenges and the high number of objectives, one could question whether this proposal really defines a "minimum standard" for LAPS medical schools. Perhaps one of the most interesting and useful contributions of this proposal is the set of 34 objectives with the highest degree of agreement that we have called the"core of the core". Most of these LO (24/34; 70.5 %) belong to the generic competency domain of "doctorpatient communication skills" and its content includes general communication objectives to consider in the process of doctor-patient relationship and skills for the development of an interview. This type of LO are also the most valued in other consensuses [12,14,15]: the UK proposal, for example, locates these skills in the third ring of the "Communication Curriculum Wheel" emphasizing them as "the backbone of effective clinical communication curriculum" [13].
The challenge of reaching a consensus on LO in a structured way, incorporating the best expert opinion, and avoiding biases is elusive. Previous similar studies have predominantly used variants of the original Delphi and Nominal Group technique [14,15,78]. In this study, we apply a variant of the modified Delphi method, which replaces its traditional face-to-face interaction of the experts by sending them a detailed result report of the first survey (including statistical results, a remembrance of personal position and comments from other panelists in relation to each LO [62][63][64]. The reliability and validity of the results are conditioned by aspects such as the selection and participation of the  experts involved (number of participants, professional profile, level of expertise and type of participation in the process), the type of questionnaire proposed, the definition of a sufficient level of agreement and subsequent applicability [79,80]. This consensus reflects the views of the 46 subjects, but other people might have reached a different consensus. The key issue here is whether the participants are really authentic subject matter experts and with influence in their respective scopes. The technique used in the study for the selection of experts (snowball sample), allows the identification of a potential expert in a more comprehensive way (through an active search through potential experts networks and a consensus involving the re-selection/confirmation of a subset of experts by the experts themselves) than when this is done by direct election of an expert [81]. Thus, a biased selection (by awareness and desirability of the initial members of the SC) should have been mitigated [82]. So, in our study of 104 potential experts located in 9 countries, we selected 51 from 8 countries, and who represent a wide variety of medical specialties and teaching responsibilities; other healthrelated professions were also included. Although a representative of patients' associations was included, the presence of a larger number of patients or the inclusion of students may have enriched the panel findings. There does not appear to be a optimal size for a Delphi panel [83]. Although forecast error decreases as the sample size increases, in general a panel of more than 30 members is not usually efficient [84]. The number of participants in our consensus is greater than in other previous professional consensuses on communication [8], but lower than in the two major existing consensuses [14,15]. However, our panelists' participation during the process was higher than in these consensuses and can be considered high (90.2 %; 46/51).
One of the main strengths of the Delphi method lies in its iterative nature, which encourages participants to revisit their opinions and reflect on them once they have known others'. We feel that a highlight of our process was detailed anonymous feedback between rounds.
The extensive review of the literature included an exhaustive review of the most relevant recommendations and proposals in LAPS. Preliminary discussion on these and on the communication domains to consider the structure created an important common ground for the preparation of a list of LO. The criteria used to mark the consensus has been statistical and, though accepted, is a discretionary criterion. It is hard to imagine a better way to standardize and ensure the process and its results due to the risks involved in unstructured discussions [85].
The final validity of this consensus will be determined by the number of institutions and programmes that use it for designing or adapting their own communication curricula. In this sense, the experts involved can also play an important role in disseminating the proposal within their own institutions and in their areas of influence.

Conclusions
This study presents a core clinical communication curriculum with 136 LO for undergraduate medical students proposed by 46 experts from 8 Spanish and Portuguese-speaking countries reached through a structured technical consensus (variation of modified Delphi). Thirty-four of these LO have a high degree of consensus and could be especially useful as a basic framework for the development of undergraduate medical educational programmes in clinical communication.

Practical implications
The LAPS-CCC is a proposal that can be useful to increase awareness and disseminate patient-physician communication educational programmes in medical schools in the LAPS countries or in other countries with similar characteristics. The LAPS-CCC can guide the design of communication programmes for medical students, especially regarding the choice of content (LO) that each institution may consider as more suitable for their graduates,