This study described the experiences of consultants, GPs and patients with referrals from primary care to medical specialist care. Overall, the majority of the three groups had positive experiences with the referrals. There was some variation regarding the consultants' experiences across their medical disciplines, with cardiologists having the least positive experiences. GPs (but not patients) had more positive experiences with the referral, if the referral had a diagnostic purpose. As assumed, GPs experiences were more positive if they initiated the referral. In contrast to our expectations, patients had more positive experiences if the referral had been initiated by the GP.
Previous studies revealed high and mostly unexplained variation across GPs regarding referrals [5–11]. But despite a very large variation, GPs referral rates do not correlate with the appropriateness of the referral . Consequently the referral rate represents an unsatisfactory indicator of quality . Our study focussed on the perceived value of the referral, including patient, GP and consultant views.
The evaluation by all involved goups, GPs, patients and specialists revealed quite good results. Some studies in other countries have shown less positive results regarding the value of referrals, but some of these studies tend to be small . Studies involving larger numbers of GPs revealed different estimations of appropriateness: with 90.4% of appropriate referrals Fertig et al. found a higher rate than in the present study .
It could be assumed, that GPs experiences were more positive with the referral, as our study revealed. Moreover, an interesting finding of our study, which was also contrary to our expectations, was not yet assesssed nor reflected in previous studies: Patient satisfaction is associated with the initiator of the referral. We can only speculate about factors leading to the unexpected results. We assume that patients expect optimal information about disease and treatment, but they respect GPs superiority in judging if a referral is necessary or not. Our findings contradict statements of critics of the gate-keeper role, saying that patient satisfaction increases if they initiate the referral. Previous findings indicating this, were shown in completely different settings, where the patients did not have a trustfull relationship to their GP . Regarding referrals, it seems that many patients may appreciate some guidance and co-ordination by the GP. GPs satisfaction was higher if the referral purpose was diagnostic than therapeutic. An explanation for this result could be that the aim of diagnostic referrals is to exclude serious diseases or to confirm suspected diseases. In most cases, the purpose is well defined and the results may help the GP to proceed in his course in case of confirmation as well as in case of exclusion. In contrast to referrals with a diagnostic aim, specialists treatments are nearly completely outside the GP's influence and may not always be perfomed according to his ideas.
The study has some limitations. Although the sample of referrals was reasonably large, it may suffer from selection bias as it was derived from a group of 25 GPs in a specific region. The study was based on self-report measures, which were not previously validated, but we believe that most questions were rather straightforward. In case of somatoform symptoms, the dissatisfaction of patients could be increased due to the fact that no organic cause has been found. It could be assumed that many of these referrals had been initiated by the patient. This could have been a source of bias, but we could only collect and link the data from consultants, GPs and patients regarding a certain referral process, but not with individual patient data. Additional information on diseases, comorbidities, social statuts, and former health care utilization would have increased the power of our data since it is known that these factors influence referrals . Due to the design of the study, we have no data about the outcome of consultations that were made directly by patients without visiting a GP first. Comparing these outcomes would increase the significance of our results but is associated with multiple problems of data collection.
This observational study suggests that referrals from primary to secondary care in Germany are reasonably appropriate and that satisfaction with referrals is high among GPs, consultants and patients. Nevertheless, GPs should try to meet the different consultants' needs for different information . For instance, detailed diagnostic and therapeutic information is essential for cardiologists but less important for orthopedic surgeons. Possible approaches for improvement could be frequent joint consultations or the possibility to discuss a referral with the consultant [15–18]. Sharing information on the patient, his history and medication is necessary to increase efficacy and to improve continuity of care . However, the fact that GPs as well as patients' satisfaction was higher if the referral was suggested or initiated by the GP supports the role of a qualified gate-keeper.