Do our talks with patients meet their expectations?

Factor analysis of both the pre- and post-visit lists of questions on preference and performance revealed 2 relevant subscales: an affect-oriented scale of 7 communication aspects and a task-oriented scale of 6 communication aspects (Cronbach’s alpha between 0.74 and 0.89).

We also used communication aspects from the original 4-point scale to present 4 new categories that compared and contrasted preferences and relevance. These categories included: important and performed; important and not performed; not important and performed; not important and not performed. In the multilevel analysis, we included the 2 subscales using the original 4-point scale.

Socio-demographic and practice variables were derived from the GP questionnaires in the Second Dutch National Survey of General Practice (2001).

Video observations

Nine observers measured verbal behavior during the videotaped visits using the Roter Method of Interaction Process Analysis (RIAS²⁶), a well-documented, widely used system in the US and Netherlands. This observation system distinguishes both affect-oriented (socio-emotional) and task-oriented (instrumental) verbal behavior of doctors and patients, reflecting the care and cure dimensions, respectively. The RIAS categories are mutually exclusive and exhaustive.

Affect-oriented communication consists of personal remarks, agreements, concerns, reassurances, paraphrases, and disagreements.

Task-oriented talk includes asking questions, giving information, and (only GPs) counseling about medical/therapeutic and psychosocial, social context and lifestyle issues, and process-oriented talk (instructions, asking for understanding).

After finishing the RIAS-coding, we calculated the total numbers of affect-oriented and task-oriented verbal behaviors separately for GPs and patients.

The relevance and performance items and the RIAS-categories all measured the affect-oriented and task-oriented aspects.

We used the Noldus Observer-Video-Pro computer program for the observation study,²⁷ including measurement of consultation length. The interobserver reliabilities were good to excellent: between r=0.80 to r=0.95 per category, except for personal remarks (0.72).

Patient-centeredness measured in 3 ways

The observers, using a 5-point scale, also rated the extent to which GPs communicated in a patient-centered way in 3 areas: patient’s involvement in the problem-defining process; patient’s involvement in the decision-making process; and doctor’s overall responsiveness to the patient.

Based on ratings in these 3 areas, we determined an overall magnitude of patient-centeredness (Cronbach’s alpha=0.75). Observers and the responsible researcher met weekly to validate the quality of rating. The same was done for the RIAS coding.

Controlling variables

For GPs, controlling variables were gender, age, and number of full-time equivalents (FTEs) working. For patients, GP and patient gender were included in the variable “gender-dyad”—male GPs/male patients, male GPs/female patients, female GPs/male patients, female GPs/female patients. Other patient variables were age; education (low=no/primary school, middle=secondary school, high=higher vocational training/university); health problems: somatic or psychosocial (ICPC chapters); overall physical health and mental health during the past 2 weeks; and consultation length.

Data analysis

We used descriptive and multilevel analyses. The intra-class correlations of the affect-oriented and task-oriented communication and patient-centeredness were significant (between .05 and 0.23), which made it clear that consultations of the same GP did indeed exhibit a greater degree of similarity than the consultations of different GPs. Therefore, multilevel analyses were necessary to account for the clustering of patients with the same GP.²⁸ We applied a significance level of ≤0.05 (2-sided).

Results

Response rate

The overall patient response rate was 88%. Analysis of non-responders’ gender, age, and type of insurance showed no bias resulting from patients’ refusal.

GP response rate was 72.8%. Respondents were representative of all Dutch GPs with respect to gender, age, working hours, practice experience (mean=15.6 years, SD=8.3, range=1–32), and location (58% in an urban area). More GPs worked in a partnership or group practice than in a solo practice. We analyzed the influence of the practice type on doctor-patient communication and deemed it insignificant.

Study population

GP and patient characteristics appear in TABLE 1. Among patients, 22% had little education, 62% had an average education, and 16% had higher education. Nearly 10% had a psychosocial problem. GP-patient gender dyads were as follows: 32.1% male GP/male patient; 45.3% male GP/female patient; 6.9% female GP/male patient; 15.8% female GP/female patient.

TABLE 1
General practitioner and patient characteristics (N GPs=142, N patients=1787)

Preference and performance of communication aspects

GPs good with affect-oriented communication aspects. Patients considered 6 of 7 affect-oriented communication aspects as very important (87%–96%, TABLE 2). The item “Doctor was empathic to me” was less important (61%) than items like “Doctor listened well to me” (96%) and “Doctor took enough time for me” (93%). We noted only a few discrepancies between preference and performance of the GPs’ affect-oriented behavior. If patients said beforehand that a communication aspect was important, the doctors nearly always performed that aspect. For instance, 87% wanted enough attention from the doctor and received it, while 99% of all patients received GP’s attention, whether it was important to them or not.