It is feasible to engage field GP participation in hospital-led studies that involve their patients. Based on our interviews, when GP participation is not planned from the beginning of the study, the GPs felt excluded and disdained. Psychology of commitment motivational techniques developed by social psychologists may be of use in regaining their interest. Our survey-based approach used many of these techniques and led to response rates greater than 50%, even when a time-intensive task (i.e., filling in a CRF) is required.
As mentioned in the introduction, there have been few studies on this topic reported in the literature. Some studies have assessed the recruitment of GPs into research networks [9, 18, 19] or studied the attitudes of GPs towards research [3, 20–22]. None have explored the commitment of field GPs to a study initiated by hospital practitioners. Nevertheless, our study confirmed that there is still a lack of interest in research and research culture [3, 20, 22] among some French GPs. French GPs who conduct academic work are mainly involved in pedagogy and learning; contrary to the literature [17, 19], 2 GP instructors in our study did not show any greater interest in research than other GPs. We suspect that the “lack of time” stated by many GPs is actually a lack of motivation to engage in hospital-led research, as studies have shown there is no explicit link between clinical workload and trial involvement . The exclusive fee-for-service remuneration of French GPs appeared to be a major barrier for participation (as shown in an Australian study comparing participation rates under a fee-for-service or capitation payment model ), and the stipend from the Regional Union of Self-Employed Practitioners (URMEL) was a determining incentive in getting GPs to participate [3, 10]. The response rate was within the expected range , at the lower limit when a costly task was required and the higher one when it was not, showing that the consequences of the initial feeling of discredit were blurred by the implementation of psychosocial techniques.
We acknowledge several limitations to this study. In our qualitative exploratory study, data saturation was reached for the study population, but as the main study had initially started in the large urban district of Lille, all the interviewed GPs were working there, with loose associations between GPs and academic specialists from the 3 academic hospitals. Outcomes might have been different if interviews had involved GPs from smaller urban districts (e.g., Calais, Dunkirk, Valenciennes) where relationships between GPs and hospital specialists might be stronger. Purposive sampling might have been a more adequate approach. The fact that the interviewers and coders were GPs themselves might have presented a preconception bias, though we avoided giving them background information on the main Diagest 3 study and we ensured that the transcripts were not coded by the interviewer.
In our cross-sectional analysis, the sample size could not be computed on the main outcome (participation rate in GPs), since it depended on the main study sample size and many “regular doctors” appeared to not know the patient enrolled in the Diagest 3 study. The inclusion of GPs in our study was based on their identification by the study subjects as their “regular doctor.” This method of recruitment revealed a lack of reliability among GPs, as one-third of the GPs were not able to provide data about the patients who had identified them as their GP. As of 1 January 2006, French people must be registered on their GP’s patient list, although they may change their GP as often as they want, and some French women receive primary care from their gynaecologist rather than from their GP. Social Security is not authorized to disclose the name of the actual regular doctor of a patient (national data processing authority), and for this reason, there is still no other means by which we can track or verify a particular subject’s “regular doctor,” except to rely on identification by the subject.
The questionnaire and its analysis grid were based on the outcomes of the qualitative exploratory study. Yet, participants do not answer a questionnaire the same way they participate in a face-to-face interview: they appear to be more conformist with regard to their group attachments . The first question in the questionnaire was supposed to allocate GPs to one of the 4 profiles, and the allocation then validated by their answers to subsequent questions. However, the answers given by the GPs did not follow this hypothesis, and the initial 4 profiles had to be redesigned into 9, and then redistributed on a pragmatic basis into 3, with GPs in the new engaged profile being more likely to participate in a hospital-led study. This redesign of the GP profiles prevented us from associating CRF response rates to each profile as initially planned, though a trend towards higher participation rates among the engaged GPs was observed. Our sample size was too small to conduct a factor analysis, and we had already pooled the 90 questionnaires sent with an attached CRF and the 27 sent without a CRF for the profile analysis. Future studies with larger sample sizes will include a factor analysis to support the validity of our profile analysis results.
Another weakness of this study is that the CRF data collected were retrospective, whereas those collected in the main study were prospective. This led to a higher level of missing data in our study. Nevertheless, the average value of each item collected on the CRF and its confidence interval were consistent with the values collected in the main study, indicating that the data collected on the CRF in our motivation study were of good quality. For example, the quality of data for blood pressure collected on the CRFs was assessed against data from the ESCAPE study ; measurements fell within the same range as that of other blood pressure studies in general practice in France.
The findings of the Diagest 3-GP motivation study will be useful as we design future hospital-led studies in which engagement of field GPs is needed. Specifically, we propose that at the enrolment of each subject, GPs are included in follow-up, and receive special training and financial compensation. Depending on the GP’s level of engagement, we also propose that a questionnaire similar to the one used for this study be developed based on the integrative model of behavioural prediction developed by Martin Fishbein . The questionnaire would be submitted to each GP and the data used to validate our typology. If our findings are validated in future studies, then we would recommend that methods to engage field GPs be included in the protocols of studies where the field GP is needed to lower the subjects’ drop-off rate.