Information giving is an important element in the quality of care: patients need information in order to understand their condition[38
]; to acquire a feeling of control, necessary for successful self-management[38
]; and to participate in medical decision-making [39
]. In 2002 the GPs showed a greater amount of information giving Nevertheless, our data also show a shadow-side in physician's changed behaviour. The 2002 physicians were less engaged in partnership building, for instance by asking for patients' opinions, asking for clarification of patients' words, or giving explicit structure to the consultation. They also expressed less often their concern for the patients' medical condition. All in all, the general practitioners from our 2002-sample seem to be more task-oriented than the GPs from the 1986-wave who asked more questions and sought more interaction with their patients.
These findings are in line with the results on patients' side of the communication. All in all, the hypertension patients from the 2002-wave made a substantially smaller contribution to the consultation than their 1986-counterparts, mainly in the process-oriented domain They asked fewer biomedical questions and engaged less in partnership building with the general practitioner, for instance by asking for clarification. Moreover they talked substantially less about what was bothering them. These findings are contrary to expectations, based on the theoretical literature about patient autonomy. In recent publications, it has been argued that patient participation is important for hypertension patients, because, once on medication, the hypertension patient will largely have to manage him/herself [39
]. Following the literature on patient autonomy, a rise in patient question-asking and process-oriented interventions would have been expected, as modern patients are said to want more information and to be more actively engaged in the medical consultation. For this controversy several explanations are possible. In the first place, it can be argued that patients who visit their general practitioner with hypertension are usually older and therefore don't fit into the model of the modern autonomous patient. Older people are indeed known to have less preference for shared decision making than younger patients[10
]. This could be either an age-effect or a cohort-effect[10
]. However, since no age differences existed between the two samples and a cohort effect would indeed have led to higher patient participation in the 2002 group, age cannot be an explanation for diminished patient participation in our recent sample. Theoretically, it is possible that general practitioners in the 2002 sample were so comprehensive in their information giving that patients had no remaining questions. However, this is quite unlikely. It is at odds with the evidence that doctors consistently underestimate patient's desire for information and that they are not good at eliciting patient preferences[10
]. In a recent Dutch study, one third of all general practice patients left the consultation room with unasked questions, 30 % blaming lack of time, 19 % unclear information from the general practitioner and 21 % reported to be too stressed to ask all questions [42
]. In a recent qualitative study, only four out of 35 patients said all they wanted to say during the consultation[43
]. Maybe today, hypertension patients are monitored more extensively than sixteen years ago and are called in to visit the general practitioner more often for a bodily check-up including blood pressure taking. It is, however, questionable if an increase in the number of visits actually decreases the number of questions asked by patients per visit. Perhaps the most plausible explanation for the low patient participation found in our study is that hypertension patients simply have not turned into active, independent and emancipated consumers, as we are led to believe they would; at least, not during medical consultations. This does not mean that these patients are not autonomous people outside the consultation room. Indeed outside the consulting room patients seem full of ideas on their medical condition and opinions of medical treatment [44
While this could explain why patient contribution in the medical consultation is low, it does not
explain why patients' level of activity is now lower than sixteen years ago. Maybe shifts in physician's behaviour could provide an explanation. Most authors agree that physicians set the agenda for the consultation, and patients follow [7
]. Doctors need to create opportunities for patients to feel comfortable in expressing their real worries[45
]. 'Being able to talk' has been found to be the most important element in a consultation [10
], and it is up to doctors to encourage patients to reveal more of what's on their minds[45
]. It has been demonstrated that general practitioners' style of listening to the patient will influence what the patient says[46
]. Now, the main shift in doctor communication behaviour that we found in this study is a shift from process-oriented towards task-oriented communication – mainly biomedical information giving. Information-giving fits within the first need of what George Engel[38
] has called: 'patients double need', i.e.: 'the need to know and understand', a cognitive need related to patients' own task-oriented coping efforts, such as involvement in medical decision making. Information giving clearly does not help to fulfil patients' second need: 'the need to be recognized and understood', an affective need related to patients' emotional coping efforts, such as revealing worries and concerns [45
]. Both elements, i.e. supporting patients in shared-decision making for instance by providing as much information as patients need versus facilitating patients to reveal complex agendas, are central in the concept of patient-centeredness [22
]. Our study shows that Dutch GPs have increased their information giving, but might have lost some of their former capacity to let patients talk along the way. A last possible explanation has to do with the entrance of the computer into the consultation room. None of the GPs from the first cohort were using a computer, while all of the GPs our more recent sample did. And as we have seen, they spent a considerable of amount on computerized record-keeping. Whereas patients used to continue talking when the physician wrote down his findings on paper in the first wave, they tended to remain silent when the physician was using his computer in the second wave. Moreover, it could well be that the computer further contributed to the already more businesslike atmosphere of the recent consultations. Margalit el al[47
] found that physician's gazing at the monitor was inversely related to physician engagement in psychosocial questioning and emotional responsiveness as well as to patient socioemotional and psychosocial exchange during the visit and that it diminished the dialogue between physician and patient. It could be that GPs are not yet completely used to computerized record-keeping, and need some time to adjust to this new 'third party' in the consultation room. However, it seems that both research and education should pay more attention to the influence of the computer on the course of medical consultations in order to minimize the disruptive and maximize the beneficial effects of this new companion of doctor and patient.
Strengths and weaknesses of the study
A strong point of this study is that communication patterns of both physicians and patients could be compared over a fifteen year period, using exactly the same methodology: observation of videotaped, real-life General Practice consultations with hypertension patients. The patient characteristics of both samples were comparable, thus adding to the plausibility that the differences in results between the two time periods are indeed a reflection of shifts in communication. Another strong point is that all statistical analyses have been guided by clear hypotheses, which were articulated before this study started.
The study also has some weaknesses. In the first place, different coders coded the two samples. However, all coders had been extensively trained according to the same training protocol, and were supervised by the first author. The 1987-manual was used for coding the first sample, the 1993-update for coding the second sample. The definitions of the codes in the 1993 update were not different in content from those in the 1987 manual, only more explanation and examples had been added to the 1993 manual. The complete text of both manuals can be requested from the authors. The high reliability scores of RIAS in both samples, similarly as found in earlier international RIAS studies, give us additional confidence that it is not very likely that the differences in results should be ascribed to the different coders.
Another potential weakness of the study is the fact that the two samples differed in the number of patients per GP. In the first study, there were several hypertension patients per GP, while in the second study only one hypertension patient per GP was included. As a consequence, consultations in the first sample were clustered around doctors, which asked for a multilevel approach, while in the second cohort no multilevel approach was needed. After consultation of a statistical expert, we decided to do a multilevel analysis, that controlled for the clustering in only in the first cohort. It is important to note that the statistical approach that we used yielded almost identical results than the conventional single-level approach. Seen the data structure and the distribution of de dependent variables, we think that the multi-level Poisson model provides the most accurate estimates of the differences in communication in both periods.