Table 2 Development and psychometric properties of the 11 included instruments

Physician Satisfaction Scale (Shore, 1986) [54, 76, 83]

Delphi method with family physicians to develop first 43-item version on 4 sub-scales. Tested on 49 physicians. The scale was then reduced to 16 items on two sub-scales and tested back on 131 physicians from Family Medicine, General Internal Medicine and Paediatric programs.

Content validity:

-efforts were put in the development phase of the instrument to ensure validity of the items (consultation with Delphi method).

Construct validity:

-factor analysis confirms two factors (average loading for patient-related: 0.71 and average loading for contextual: 0.58)

-the instrument did not discriminate between different residency programs, geographical location or years of training.

Internal consistency:

-Cronbach alpha for global scale: 0.85 (patient-related subscale: 0.89 and contextual subscale: 0.63)

Physician Mental Workload (Bertram, 1992) [55, 56, 59, 74]

A previous version of the instrument was constructed through discussion with physicians and from a preliminary literature search. It was tested in two different hospital settings and revisions led to a 10-item version also presented on a visual analogue scale. The present instrument is a 6-item adaptation of this previous one. It was tested on 22 residents, who in all saw a total of 92 patients during an afternoon clinic session. It was tested with residents and physicians in practice, internal medicine and very few in paediatric residency

Broad domain of human performance research and measurement approaches employed in the field of human factors research. It encompasses motivational, social, attitudinal, and organizational factors as well as human capability assessment, information processing and decision making and stress effects on performance.

Content validity:

-efforts have been made in the development of the first version of the instrument to ensure face validity by consulting physicians, and formal content validity by literature review. The process of selection of the items included in the present version is not described.

Construct validity:

-correlates with: fatigue: r = 0.42, mean experience, r = -0.65, resident self-rated quality: r = -0.67, third observer's overall quality rating, r = -0.18, personal interaction factor score: r = -.04, technical performance factor score: r = -0.38

-does not correlate with: total number of patients seen, proportion of new patients, patient complexity, personal interaction performance, overall ratings by faculty members and age of the residents.

-does not discriminate between female and male residents nor among postgraduate years.

Note: In order to not violate the assumption of independence between observations, the unit of analysis chosen was the resident, and an associated average score per resident was used with patient-specific measures.

Internal consistency :

-Cronbach alpha: 0.80 (unadjusted for non-independence of observation)

-Inter-items correlation: mean: 0.45 (SD.: 0.19)

Physician Frustration in Communicating with patients (Levinson, 1993) [72]

A group of experts developed an initial set of 32 items corresponding to common problems encountered by physicians in their encounters with patients. This was pilot-tested on 107 physicians of diverse trainings. A second version of 42 items on 8 sub-scales was distributed to 931 physicians, and was reduced to 39 items, and this version was completed by 1076 physicians. Final version consists of 25 items on 7 sub-scales.

Broad domain pertaining to the quality of the communication and the relationship between patients and their physicians as important pathways to both the medical outcome and satisfaction of both parties.

Content validity:

-efforts were put into the development phase of the instrument to ensure validity of the items (x consultation of experts) Construct validity:

-factor analysis confirms 7 factors. Mean respective factor loading for all items is 0.68 (SD = 0.10)

-the instrument discriminated between younger and older physicians (i.e. younger physicians had higher scores on all subscales meaning they felt more frustrated than the older physicians) and between primary care physicians and specialists on two subscales: too many problems and feeling distressed (i.e. primary care physicians had higher score than specialists). Greater time spent in primary care was associated with higher scores on several subscales. Convergent validity was shown with physicians' general level of satisfaction and the percentage of visits they reported as being frustrating correlating with higher scores on most subscales.

Physician Satisfaction with Primary Care Office Visits (Suchman, 1993) [84, 85]

The development of this instrument was achieved within a larger initiative, "The Collaborative Study of Communication Dynamics". This initiative was organized by the Task Force on Doctor and Patient of the Society of General Internal Medicine that was conducted at 11 sites in North America. Members of this group included well-known experts in the field of patient-doctor interaction and communication. The instrument was tested with 124 physicians (35 residents, 60 general internists and 3 family physicians) who saw a total of 550 patients.

Content validity: face validity is considered in that the items of the scale share common ground with previously published measures

Construct validity:

-factor analysis reveals 4 distinct factors, but since these construct domains were not predicted at first in a theoretical framework, this analysis provides weaker support for the construct validity of the instrument

-a number of patient characteristics were significantly associated with the sub-scales. For example, emotional distress of patients was negatively correlated with all satisfaction dimensions except the time dimension. Satisfaction with the patient doctor relationship sub scale was the most important determinant of global satisfaction (R² = 39%) while the adequacy of data collection process was the second most important determinant (R² = 4%).

Note: Non-independence of observations was taken into account: a bootstrapping technique was used to create 10 replication samples of n = 124 and factor analysis was then performed 10 times.

Internal consistency

-Cronbach alpha for the 19 specific items (excluding the general satisfaction question): 0.82

-Cronbach alpha for all 20 items: 0.84

Collaboration and Satisfaction about Care Decisions (Baggs, 1994) [49–53, 75]

This instrument is based on a conceptual model for collaboration for conflict resolution. It was developed from an initial 2-item version, the Decision About Transfer, a literature review on the subject and opinion of experts in collaborative practice and of practising professionals in the field. It was pilot tested on a convenience sample of 32 nurses and 26 residents in an intensive care unit.

Thomas (1976) conceptual model of collaboration for conflict resolution and organisational theory by Thompson (1967).

Content validity:

-literature review on the subject and opinion of experts in collaborative practice and of practising professionals in the field.

Construct validity:

-factor analysis confirms a single factor (Eigen value of 4.5, no other higher than 1) that explains 75% of the 6 specific collaboration items variance. Mean factor loading for the six specific collaboration items was 0.87 (SD.: 0.04).

-convergence of a combined score of the six specific collaboration items with a combined score of the two satisfaction items: r = 0.66

Criterion validity:

-correlation of the six specific collaboration items with the global collaboration question: r = 0.87

Note: Non-independence of observations was taken into account: factor analysis was performed with a sample size of 56 (i.e. all independent data entry points) and confirmed one factor for collaboration

Internal consistency:

-Cronbach's alpha: 0.93

-Inter-item correlations: 0.52 – 0.83

Medical Communication Competence Scale (Cegala, 1998) [60–63]

Post-interview questionnaires in clinical setting as well as self and other evaluation of communication competence by 15 family practice residents inspired the development of a first version of 56 items. Six physicians scored each item for their importance to communication competence during a medical consultation. Best items constituted the 37 items final version. A corresponding patient instrument was also pilot-tested concomitantly. Hence, these two instruments were pilot-tested with 65 doctors and 52 patients who provided a total of 117 data entries.

Extensive theoretical review supports the development of the scale.

Content validity: efforts were put into the development phase of the instrument to ensure validity of the items (face validity by consultation of potential users)

Construct validity:

-Factor analysis supports construct validity, but complete loading data is missing.

-A cluster analysis using Euclidean distances among standardized item response was performed for each file. As hypothesized by the author, the results of both cluster analysis covered 4 clusters: information giving, information seeking, information verifying and socio emotional communication.

-a series of research questions in which within-sample comparisons (i.e. comparisons made within the physician data file and the patient data file) and between-sample comparisons (i.e. comparisons made between the physician and the patient data files) were shown to be consistent with the literature on doctor-patient communication. For example, doctors rated their socioemotional competence higher than their competence in information exchange than in any of the other information subscales paralleling poor competence in information exchange observed in previous researches.

Internal consistency for the doctor's scale (Cronbach alpha's)

- information giving: 0.86

- information seeking: 0.75

- information verifying: 0.78

- socio emotional communication: 0.90

Provider Decision Process Assessment Instrument (Dolan, 1999) [57, 64–67, 69–71]

Based on the construct of decisional conflict, this instrument is an adaptation of O'Connor's 16-item Patient Decisional Conflict Scale. Data were obtained on two sites from 14 residents, 7 physicians and one fellow in General Internal Medicine.

Ottawa Decision Support Framework.

Content validity: face validity assessed by asking participants for direct feedback.

Construct validity: moderately confirmed by negative correlation with two satisfaction items: satisfaction with the decision (Spearman's r = -0.58) and assessment of the quality of the decision (Spearman's r = -0.52).

Internal consistency:

-Cronbach alpha: 0.878

Note: In order to not violate the assumption of independence between observations, a bootstrapping approach was used. (i.e. 30 random samples consisting of one patient from each of seven physicians) Cronbach alpha was 0.90, 95%CI= 0.87 – 0.92.

Patient-Physician Discordance Scale (Sewitch, 2003) [77–82].

On the basis of a literature review, two domains were identified: patient's health status and the office visit. Two experts, a clinical psychologist and a gastroenterologist, were provided with a list of items recorded from the literature review and asked to select the top 10 items thought to be relevant to making treatment decision. A consensus was reached after a brief discussion.

Broad domain of patient-physician discordance.

Content validity: Based on the literature review and two experts.

For construct and criterion validity, data are provided only for the combination of the physician's and patient's questionnaires

Data are provided only for the combination of the physician's and patient's questionnaires.

Mutual Understanding Scale (Harmsen, 2005) [68]

This instrument was developed based on Kleinman's theory, a method of phasing or structuring of consultations by the physician (S.O.A.P. method) and a consensus method of decision-making called the Nominal Group Technique or expert-panel meeting

Kleinman's theory about the influence of culturally determined views on health beliefs and the necessity for physician and patient to demonstrate these views by exchanging explanatory models during the consultation.

Content validity: By using questions about different consultation aspects, known as GP standard of structuring the consultation, the complete consultation was covered.

For construct and criterion validity, data are provided only for the combination of the physician's and patient's questionnaires

Data are provided only for the combination of the physician's and patient's questionnaires.

Reasons for Treatment Selection Questionnaire (Linden, 2006) [73]

N/A

N/A

Questionnaire concerning the doctor-patient communication skills [58]

This pair of instruments was developed based on the Patient Centered Care method [98] and theories in the field of communication. Its authors drew on existing instruments and the communication skills expertise of 2 members of the steering group to create the pair of instruments. The initial instruments were administered to 4 specialists and 3 family doctors in Ontario, Canada, who, along with their patients, provided feedback. The final pair of instruments was tested with 16 family doctors and 22 specialists from 3 Canadian provinces. These doctors recruited a total of 1881 patients.

Patient Centered Care method [98] and theories in the field of communication.

Content validity: based on existing instruments and the communication skills expertise of 2 members of the steering group to create the pair of instruments.

Construct validity:

-Factor analysis was performed by using the whole set of 38 items (19 items in the doctor's questionnaire plus 19 items in the patient's questionnaire) to ascertain whether the patient and doctor items were 2 separate factors. Then by examining the data for patient and doctor separately, the authors ascertained if the process and content items accounted for separate factors.

The items on all 3 datasets (i.e. 19 items from the patient data alone, 19 items from the doctor data alone, and the combined dataset of 38 items) were separately intercorrelated using Pearson product) moment correlations.

Internal consistency:

-Cronbach alpha for the doctor and patient questionnaires: 0.70 and 0.69, respectively.

Number of patients per doctor required for a reliable assessment of the doctor's overall communication skills:

- The G analysis provided a G = 0.98 and 0.40 (standard errors of 0.003 and 0.02) for doctors and patients, respectively.