Strengths and limitations
The questionnaire (Supplementary file
1) opens for the possibility of biased self-reporting, leading participants to give exaggerated accounts of socially desired behavior [
31]. This may also be a strength as there is no reason to suspect that the participants would report knowledge and skills they do not possess. The total anonymity of the survey could mitigate this bias, by allowing the respondents to express themselves more freely. Due to the importance of knowing distance from and affiliation to local hospital, information on age and gender of participants was not included in the survey, as these data could lead to identification of certain RGPs. The material gives no information of how these factors influences the answering, and challenges external validity. A response rate at 57% must be regarded as a strength as all RPGs in the county were invited, and GPs are known to typically have low response rates [
32]. The non-responders (43%) may, however, present problems of participation bias, with the risk of failing to capture the full range of views. One could suspect non-responders possibly to find the topic of palliative care less relevant, and to be less active and interested in the subject of study than those who did answer the survey, thus causing over-estimation of experience and knowledge among the RGPs [
33]. It has also been shown that GPs are less likely to respond to a survey the more time has passed since qualifying as doctors [
33]. This may have caused more experienced RGPs not to answer, thus causing an under-estimation of the amount of experience and skills of the RGPs in our material. A previous study has shown such a positive relationship between age of the GP and both confidence about being a key worker, and likelihood of providing end of life care [
34]. The total anonymity of respondents may have led to inappropriate mailing to RGPs that had already answered, causing some to answer the questionnaire twice. However, it is unlikely that many have taken the time to do this, especially as the reminder was sent shortly after the original deadline. Written comments were optional in the questionnaire and only a few respondents used this opportunity (Supplementary file 1). This may have resulted in only respondents with strong opinions commenting, and thus the results may not be representative for the total group. We still chose to include some comments in our results, thinking they convey attitudes and thoughts, apt to help in the interpretation of our findings. Although some missing data, this only caused minor alterations to our frequencies percentages results, and these are therefore given as valid percentages of those who answered. Due to sampling being restricted to one county, caution must be taken when generalizing from our findings. However, we do believe that the geographic spread of participants within the county is indicative of its representability. The area contains four hospitals of differing size, and is typical for many Norwegian counties, although lacking a larger university hospital. The data was collected in 2014. As there has not been structural changes to the palliative care services, competence requirements or general practice in Norway [
1,
10,
35], we have no reason to think that our main findings are no longer valid.
Findings in the light of current knowledge
Achieving death at home for those who wish is in many respects an ideal in palliative care [
36] and GP participation in the trajectory is one of many factors identified as facilitators for achievement [
15,
16]. The competence requirements and role assigned to the RGPs in the guideline is a means to govern RGP participation in the wanted direction [
30,
35]. We found that one third of the RGPs had not participated in terminal care at home the last three years, and that most who had participated had only experienced a few cases. The proportion of patients dying at home yearly in Norway is about 15% or less [
37], and a recent study showed that the potentially planned home deaths in Norway were 6.3% of all deaths [
38]. This means that an individual RGP will potentially experience a home death in their population, on average, about every two to three years, perhaps even less, as we found that near 40% of RGPs perceive that these patients are mainly handled by hospital specialists. Furthermore, we found that seeing few patients with palliative care needs, challenges the RGPs ability to maintain their competence in the field at the level required by the guideline. The finding is consistent with previous studies of procedural skills practice and training [
39‐
41] and in agreement with a previous Norwegian study by Austad et al. [
29], who found that GPs find it difficult to keep updated on guidelines for specific diseases that they do not see regularly.
The low guideline adherence among RGPs is also in agreement with previous studies of guidelines. There is a debate as to whether this may be due to lack of willingness of the GP [
27], or to the guideline content [
25,
28]. Comprehensive guidelines, also makes it difficult for the GPs to adhere [
29], and the GPs' situation of having multiple guidelines to follow simultaneously has been identified as one factor that may impede guideline adherence [
29]. The guideline for palliative care, however, differs from previously studied guidelines as it can be viewed as not diagnose-specific, thus representing a common pathway for multiple diseases at the end of life [
1]. Hence, there should be no mismatch between guideline and patients’ needs due to multitude of guidelines for single diseases, as previously described [
29]. Still, the RGPs seem unfamiliar with the contents of this guideline. Paradoxically, although recognizing the utility of forms like the ESAS-r for symptom assessment, the RGPs seemed to be reluctant to use them. They also seemed to recognize the utility of ACPs and available, updated medical information, but did not use them either. These paradoxes seem parallel to previous findings [
27]; the GPs report to acknowledge the value of guidelines, yet seeming unable to use them, and the relevance of guideline content itself may be questioned [
27]. Our data implies that the RGPs are not able to meet the competence requirement and maintain the skills they are expected to in the guideline, and it needs to be established how this affects the cooperation and division of labor within the health care services.
Our findings also indicate that the RGPs to a certain degree actively choose a different approach for various reasons. They seem to perceive that they have too few cases for the symptom assessment forms to be useful to them. At the same time they confirm that they commonly approach their patients through conversation and make use of their previous knowledge of the individual patient, consistent with the widely used patient-centered clinical method of general practice described by Levenstein et al. [
42]. A Norwegian study has shown that RGPs, and especially experienced RGPs, also rely strongly on person-related knowledge about their patients and that too much standardization in patient care plans can hinder genuinely tailored, individual treatment [
43]. This may indicate that the working methods described in the guideline, based on the specialist health services' way of doing it, do not harmonize with the more flexible person-centered approach and working methods in general practice, demonstrated in a previous study [
44].
The GP as entry-point and coordinator of primary care is a trait shared by many European countries [
45] and palliative care is one of the core values of general practice according to the WHO [
9]. Our findings may therefore be relevant to European and other countries with a similar health care structure. Implementation of generated medical knowledge by means of clinical guidelines is a widespread strategy in the world today and understanding of the barriers for implementation is important [
30]. Our findings challenge guideline content with respect to the complexity of the knowledge the RGP is expected to maintain when patient encounters are infrequent, and whether the recommendations in the guideline fits the working methods of general practice. The guideline [
1] could represent a common pathway for several diagnoses, but as it describes the patient population as consisting mainly of cancer patients, together with its origin and formal organization within the cancer care program, it may not seem relevant for general practice [
1]. Whereas cancer seem to be the most frequent patient group from the point of view of the specialist in palliative care units, frailty, organ failure and dementia dominate causes of death in primary care, confronting the GP with a large variety of trajectories [
46], that perhaps are difficult to standardize, as pointed out in a recent editorial by Mitchell and Murray [
47]. This also raises the issue of guideline applicability as a barrier for adherence [
48].
We found that only one respondent reported to participate regularly in joint meetings with the hospital specialists upon discharge from hospital, and that nearly 40% of the RGPs perceived that hospital specialists mainly handled their palliative patients. Discharge planning is an important task for the hospital based palliative teams [
49]. A customized approach is expected to be beneficial and should incorporate a clear “care transition” [
50]. The finding may imply that the specialist level do not act according to the guideline recommendations either [
1]. This is in agreement with a previous implementation study, who demonstrated low guideline adherence among hospital specialists when they were supposed to hand over tasks to GPs [
51].
In 2017, an evaluation of the palliative care services in Norway was performed [
52], the report describing the RGP as “on the sideline” of palliative care. Our finding that only one fifth of the RGPs do not see themselves as central in this work, and that most seem confident in the provision of palliative care, contrasts somewhat with this report. The finding that the RGPs to a high degree make themselves available out of normal work hours also challenges this report. To our knowledge, no other Norwegian health worker has been shown to make themselves available, in their spare time, and to such an extent, and this comes in addition to having high reported work hours in the first place [
53]. These findings are consistent with previous findings of GPs’ commitment towards cancer patients, and providing palliative care [
44,
54].
The impact of distance from hospital on the RGPs perceived role in palliative care is an interesting finding. A previous Norwegian study found that that some rural and small-town GPs contributed considerably to cancer care in their patients’ local communities [
55]. In a recent Danish study, they found that rural GPs were more secure in the administration of subcutaneous medication than their urban colleagues [
34], and in a Dutch study, rurally based GPs were more confident in administering emergency care than urban or metropolitan GPs [
56]. In the latter study, this was perceived as linked to proximity to the hospital emergency services, leading to the urban GPs being surpassed. In the case of palliative care, the hospital based palliative teams in Norway are meant to be ambulatory, acting as consultants supporting primary care [
1]. The teams should ensure equality of services regardless of geography, which is a widely recognized principal in health care organization [
57]. Although cultural differences between rural and urban RGPs may contribute [
56], our finding may also represent a distance decay effect [
58], pointing towards the possibility of unwarranted variation in the specialist service provision.