The Third Meeting of the Verona Network on Sequence Analysis

The fourth meeting of the Verona Network on Sequence Analysis

“Consensus finding on the appropriateness of provider responses to patient cues and concerns”

Verona, 3rd – 4th February 2006

Lidia Del Piccolo, Claudia GossSvein Bergvik

Department of Medicine and Public Health, Section of Psychiatry and Clinical Psychology, University of Verona.

On 3rd – 4th February 2006,twenty-three researchers from seven different countries (England, The Netherlands, Norway, Switzerland, Germany, Italy and USA) met in Verona to attend the fourth invitational workshop of the “Verona Network on Sequence Analysis”, hosted by the University of Verona, and organized by Christa Zimmermannand her colleagues.

The first part of the workshopwaschaired by Christa Zimmermann and Arnstein Finset. It was dedicated to the discussion of inter rater reliability in coding patient cues and concerns during patient-provider interactions, and establish a final consensus onthedefinitions of patient cues and concerns elaborated by the Network members in a collaborative process during 2005.The second part of the workshop, chaired by Lidia Del Piccolo and Cathy Heaven,focused on finding a common ground on what to consider an appropriate provider response to these patient expressions. The workshop was organised into five parallel sessions, each followed by a plenary session.

Discussion of inter rater reliability findings and implications for a final consensus on cue/concern definitions

Theinter rater reliability study of the consensus proposal was based on the work by the network in 2005 (Del Piccolo et al., 2005) and had involved fourteen participants, allcoding the same transcript of a cancer consultationprovided by Phyllis Butow.

In this proposal a cuehad been defined as “a verbal or non verbal hint which suggests an underlying unpleasant emotion and would need a clarification from the health provider”. It included five subcategories ofverbal cues: a. Words or phrases which suggest vague or undefined emotions; b. Verbal hints to hidden concerns; c. Words or phrases which describe physiological correlates (regarding sleep, appetite, physical energy, excitement or motor slowing down, sexual desire) of unpleasant emotional states; d. Unusual or affect loaded, emphasized or repeated mentions of issues of potential importance (life threatening therapy, stressful life events and social problems); e. Communication of life threatening diagnosis (e.g. “Doctor I have cancer”). Non verbal cues included: a. Clear expressions of negative or unpleasant emotions (crying) or b. Hints to hidden emotions (sighing, silence after provider question, frowning etc).

A concernhad been defined as a “clear and unambiguous expression of an unpleasant current or recent emotion where the emotion is explicitly verbalized, with a stated issue of importance for the patient (“I am so worried about my husband’s illness”; “Since the illness of my husband I feel very helpless”) or without (“I am so anxious”; “I am nervous”).”

Other three dimensions had been added to better qualify these definitions:

Repeated cues/concerns: a cue/concern is coded only once when repeated in a turn, whereas separate codes might occur in the same turn if the content is new. In subsequent turns, a repeated concern is always coded, whereas a repeated cue is coded only when the doctor had clearly ignored it by not addressing it or by missing the mark.
Cue/concern source: to identify if the cue/concerns is patient- or doctor-initiated. This was based on the assumption that the expression of cues and concerns facilitated by the health provider (doctor-initiated), is an indicator of the space given to patients to explicate their concerns, whereas the expression of patient-initiated cues and concerns is an indicator of how much patients have to take the initiative or to actively struggle to direct the health provider’s attention to specific worries.
Definition of current/recent importance. This because concerns of current or recent importance (related to illness or to other stressful issues) are known to be associated with emotional distress of clinical significance and are useful predictor variables. The inclusion of past concerns with uncertain current or recent importance invalidates this relationship.

The interview transcripton which to apply the consensus proposal was composed by a total of 218 units, 110 of the health provider and 108 of the patient, of these 33 were identified as patient cues or concerns, with a range of 6 to 16 classified units per rater. The discrepancy between the total number of units identified (33) and the number of units classified by each rater (6 to 16) evidenced the scarceoverlap among the raters’ coding. Among the five concerns identified, only one was unanimouslyclassified as such by all 14 raters (i.e. the sentence “And I’m unhappy with that”), and for nine of the 14 raters this was the only concern of the whole interview.The remaining units classified as concern had less than 50% of agreement. Overall, Cohen’s Kappa was 71,34.A total of 30 cues were identified (the range among raters was 4 to 15), but nonewere unanimously classified as a cue by all 14 raters. Cohen’s Kappa was fairly low(34,21).The reported data showed that concern codings were more reliable than cues codings, but that both concepts neededto be refined.

The task of the first parallel group session was to consider several critical issues that had to be resolved: the addition of some including/excluding criteria in the definition of cues/concerns, amore clear definition on the coding of repeated cues, and some pitfalls on the distinction between Doctor Initiated (DI) and Patient Initiated (PI) source.The parallell groups made suggestions on how to solve these issues, these were then discussed and integrated during the subsequent plenary session.

Regarding the including/excluding criteria in the definition of cues/concerns, three main issues were discussed. First, there was a debateabout the inclusion of “questions”containing cues or concerns. The decision was to code these questions byapplying the same rules of the consensus definitions, but also to identify them as questions to distinguish them from non question cues and concerns. Second, it was decided to include the affirmative responses to questions checking patients’ worries,among concerns (e.g. D: ”Are you worried about …?” P: ”yes”).Third, all groups agreed on the importance to consider the emotional level of cues. Anotherissue regarded the content of expressionswith an implicit shared understanding ofthe underlying emotion. All agreed that such expressions are difficult to code (e.g. only a mentioning of a word “cancer” can not be a defining criteria) without referring to their context. Criteria had therefore to be much more specific and partly dependent on the specific interests of research.

Regarding repeated cues, all agreed that repeated cues should be coded, to avoid the risk of losing important data, but that more specific criteria were needed in the definition of repeated cues and in the level of analysis (e.g. by turns or by utterances?)

Regarding the distinction between Doctor Initiated (DI) and Patient Initiated (PI) cue/concern source, it was suggested to rate PI whenever the patient in replyingto a health provider’s question goes beyond the topic about which he/she is invited to talk.

The plenary session concluded by delegating Lidia Del Piccolo, Arnstein Finset and Christa Zimmermannto prepare a final consensus document which would take into account the discussed observations and on which a more extensive reliability study would then be carried out.

The definition of appropriate provider responses to cues/concerns

During the 2005 workshop the discussion on appropriateness had raised more questions than answers (Del Piccolo et al., 2005). Working on transcripts of medical interviews, it was relatively easy to agree on whether the responses were appropriate or not, but more difficult to formuate defining criteria for appropriateness, because a variety of issues emerged. Indeed no conclusion was reached and the decision had to bepostponed to the 2006 meeting.

Among theunresolved issues were: Is a silence, an utterance such as "hmmm" or a facilitation after a cue/concern always appropriate? How long may be the delay of an appropriate response in terms of turns in order to be coded? Do we have to consider the effect of the provider response on the patient, in order to code it as appropriate? How to distinguish ‘dismissing’ from ‘understanding’, e.g. in the provider utterance “it isnormal to be anxious before a surgery”. In synthesis, it appeared that a definition of appropriateness based exclusively on categories of coding systems (such as open questioning, facilitation, empathic comments) was regarded by all as a too rigid approach and could not acknowledge the complexity of the problem. It was felt that we needed a specific multi axial classification system for appropriateness which would consider the match of patient/cues/concerns and provider response in terms of content, the aim pursued by the provider response (acknowledging, clarifying, exploring, understanding, etc, but always “hitting the mark”) and the skill level (technique).

Thefollowing dimensions were considered to be important for the definition of appropriateness or provider responses to cues/concerns:

-the conceptual framework or belief system of the health provider (biomedical, biopsychosocial, psychodynamic etc) which meant that one should share a common concept;

-the setting (emergency room, general practice);

-the different phases of the consultation (emotion handling by empathic responses might be inappropriate in a very early phase of the consultation, where reflecting comments (acknowledgements) might be more adequate.

-Patient needs

-The synergy between verbal (content) and non verbal expressions (tone of voice, eye contact, body posture)

-Provider- and patient-shared knowledge. Sometimes a provider refers back to something which has been discussed in a previous visit, which may make a superficially defined inappropriate utterance appropriate in the eyes of the patientbut not in that of the rater. Patient background information, whether this is known by the provider,also can determine whether something is appropriate or not.

Participants worked in parallel groupstrying to identify on interview transcripts appropriate responses first to concerns and then to cuesand to give a definition of appropriateness.The groups adopted different approaches: one group focused more on procedural and conceptual caveats, whereas the othertried to propose a coding system.

Conceptual observations regarded the question of adoptingalternative concepts or expressions rather than “appropriateness” (i.e. Missed opportunity vs. Taken opportunity; Inhibitory vs. Facilitative) in order to avoid value judgments inherent in the terms appropriate and inappropriate, and to focus instead on what seems to be the function of the utterance. It was also underlined that: 1. The researchers’perspectiveof the function of the consultation may influence the ratings 2. The type of coding (utterance-by utterance, turn of speech) will influence what is coded, and therefore should be considered.3. Appropriateness should not be specified a priori, but should be hypothesized and tested empirically in relation to outcome measures. 4. Appropriateness should always be considered in relation to outcome andcontext (family background, individual characteristics, medical condition, shared knowledge, previous visits etc.).

Regarding the classification proposal of responses to concerns, seven anchoring points were identified for both content and emotion: Ignoring, Moving away - devaluating, Facilitate, Take notice - acknowledge, Explore (take further), Supportive comment, Not applicable.

During the plenary session all agreed that inappropriateness is relatively context independent. It seemed therefore easier to find a common ground on inappropriate responses rather than on appropriate responses.Proposalsof inappropriate responses to concerns were: blocking; moving away or shifting topics; downplaying, minimizing, ignoring, jumping to a solution, inappropriate reassurance. Reassurance and “facilitation” stimulated some discussion. “Reassurance” may be considered an interviewing behavior that will change according to the phase of the interview – if the issue or the emotion has not been dealt with, reassurance is an inappropriate and premature behavior, when following a clear description of what the feeling of the patient is, it could be appropriate. Also “facilitation”, defined by some as a passive listeningbehaviour, was not always considered as an appropriate response.Some struggled with the contradiction of the label facilitation (if an intervention is facilitating how can it be judgedas inappropriate?).

Examples of appropriate responses to concerns that were considered: try to find common grounds (on both emotion and content); acknowledging (agree, reflection); giving room for the patient; exploring and empathic responses.Proposals for inappropriate response to cues were similar to those for concerns, with the addition of leading questions and responses to superficial cue aspects. Appropriate responses to cues that were considered: answer a question; verify the point, facilitate; reflect, pick up the cue; explore; educated guessing, and in some cases legitimize and reassure.

A small group of participants (Jesse Jansen and William Verheulwith the help of Arnstein Finset, Cathy Heaven andLidia Del Piccolo) was delegated to reconsider the conceptual framework and to prepare a draft of a core definition of “appropriateness” that could conciliate the main aspects that emerged from the workshop reflections. This document will be presented during the EACH 2006 Conference in Basel.

Closure of the meeting

Jozien Bensingclosed the meeting on Saturday afternoon. She appreciated the collaborative atmosphere of the meeting and encouraged future collaborations by suggesting, in continuity with this year’s workshop, some stimulating research initiatives.

In synthesis, the two days workshop has been intensive and productive. Afinal consensus statement on cue/concerns definition was definitely stated and a new reliability study has been planned. Afirst consensus elaboration of how to classify health providers’ responses to patient cues and concerns was launched. In view of the next meeting in 2007, both initiatives will keep the Network busy also during this year, exchanging discussions and reflections on the controversial issues and,hopefully,finding solutions.

Reference

Del Piccolo L., Goss C. & Zimmermann Ch. (2005) The third meeting of the Verona Network on Sequence Analysis. Finding common grounds in defining patient cues and concerns and the appropriateness of provider responses. Patient Education and Counseling, 57, 241-244.