This study included 4019 physicians who regularly had patient consultations involving sickness certification, and the majority of these professionals indicated that they needed to develop their knowledge and skills in relation to the process of certifying sick leave. However, there was substantial variation between different clinical settings, even after adjustment for demographic and work-related factors. A larger percentage of physicians in primary care and psychiatry reported a need for more knowledge and skills compared to other categories of physicians. Also, specialists expressed less need than non-specialists with respect to learning more about sickness certification. Nevertheless, the majority of all physicians reported that they did need CPD, especially regarding the roles of different stakeholders in sickness certification and how to assess patients' work capacity and optimal length and degree of sickness absence.
Strengths of this study include the large study population, the high response rate, and the many detailed questionnaire items about the aspects of interest. Another strength is that physicians in many different clinical settings were included - not only GPs, psychiatrists and orthopaedics, as has been the case in the majority of previous studies of physicians' sickness certification [1, 6, 12, 15, 16]. To what extent the results can be generalised to the whole of Sweden is uncertain. However, the two counties that were included comprise both large rural areas and very populous urban areas (with 2.3 million of the 9.1 million inhabitants in the entire country). Also, the present study population represented 24% of all physicians in Sweden, and, inasmuch as the response rate was high, the results can probably be generalised to at least the southern and central parts of the country, where 85% of the population lives. The variations between different types of clinics/practices are probably larger than the disparities between different parts of the country. The variation in sick-leave rates between different regions in Sweden does not necessarily indicate that physicians differ regarding their need for CPD. Generalisation to physicians in other countries cannot be done without considering the disparities in social insurance systems and in medical education. Despite that, studies conducted in other nations have also indicated that physicians feel they need more knowledge in insurance medicine [1, 6, 13]. The physicians' sickness certification tasks are basically the same in different welfare nations, as well as over the last decades. Moreover, development of the questionnaire employed in the current investigation involved a large number of clinicians and researchers focused on the discipline of interest, which, along with the fact that we obtained responses from a considerable number of participants, has provided an extensive base for claiming face validity.
As in all questionnaire studies, the data we acquired merely reflect the respondents' own assessments (here regarding their need for competence in sickness certification), which means that we do not have information on their actual proficiency in the areas of interest. The physicians' self-assessments in this context covered two aspects: (1) the competence they believed to be required for the tasks and (2) their ratings of their own competence in relation to the presumed requirements. It should be mentioned that some studies have indicated that the relationship between self-assessed and observed measures is weak [17, 18]. However, we found that assessments of the knowledge gap varied substantially between the different groups of physicians included in our investigation.
Our results show that physicians need training in sickness insurance in a broad sense (Table 2). More than 50% indicated having a large or fairly large need for CPD about the roles of the various stakeholders, including themselves, and different types of compensation other than sick pay, as well as sickness insurance rules. This has not been shown in earlier studies, which have focused on more specific issues. Furthermore, four out of ten physicians reported the need for CPD regarding the following: how to assess patients' work capacity and work demands, as well as optimal length and degree of sickness absence; how to make plans of action for the sick-leave period, and how to handle conflicts with patients about sick leave. These observations support similar results found in earlier qualitative studies [10, 13].
Previous research on physicians' knowledge and skills in insurance medicine has mainly considered the quality of the sick notes issued, not the physicians' own assessments of their knowledge regarding such tasks . Clearly, investigations that consider physicians' own assessments of whether they want to learn more, and, if so, what they actually want to learn, would provide information that can be used to enhance future medical education and CPD.
It is noteworthy that most of the physicians in the present investigation reported that they had little need to improve their knowledge and skills in issuing sickness certificates. A systematic literature review as well as other studies [1, 5] have established evidence that the quality of sick notes is often low from the perspective of the stakeholders who use such documents as a basis for determining whether patients are entitled to sickness benefits, that is, the patients' employers and the SIO. In one of those investigations , only 27% of the certificates provided clear assessments of both the medical disorder and the functional capacity of the patient, and 7% of the specialists' certificates were illegible. However, filling out sick notes was not considered to be very problematic by a majority of the physicians in our previous study , which suggests that the inferior quality of certificates is not a consequence of a lack of knowledge, but instead has other causes, such as low demands for quality from the SIOs . On the other hand, a qualitative study performed in Norway  revealed difficulties associated with communication of assessments to the SIO, as indicated by this quotation from one of the interviewed physicians: "It's difficult to present it on a dotted line."
Our analyses of associations revealed two groups of correlated questionnaire items, and based on those, we created two indices that we designated the knowledge index and the skills index. Here, the word knowledge refers to what we call "know-what", or explicit knowledge, indicating information that can be verbalised and communicated to others. By comparison, a skill can be described as "know-how" or implicit knowledge . For example, in a qualitative study , it was found that physicians described assessment of functional ability as being "in the back of our minds," and the authors called this phenomenon "tacit assessment."
The results of the present study indicated that the need for knowledge and skills was lower among those with longer professional experience measured as age, years in practice, or level of training (Table 5). This observation supports the Dreyfus and Dreyfus model  of development from novice to expert in a specific area, in which the expert relies on "pattern recognition" based on previous experiences, rather than explicit knowledge. However, in a study of the quality of sickness certificates , it was found that such documents more often lacked essential information when issued by specialists than when provided by those who were not yet specialists. Another study revealed that a larger number of sick notes were issued by more experienced physicians than by those with less experience . We noted a small increase in the need for knowledge and skills in physicians who handled a large number of sickness certification cases and had a high frequency of problems in handling such documentation, as well as those who had regular contacts with the SIO (Table 5). Thus, more frequent handling of sickness certification does not imply a lower need for knowledge, as might be assumed.
Having a workplace policy on sickness certification was associated with less need for knowledge and skills. However, this must be interpreted with caution, since no conclusions regarding the direction of this association can be drawn from our results.
For both the knowledge and the skills index, the mean percentage of maximum was significantly higher for physicians working in primary care and psychiatry (Table 6), and was significantly lower for those working in surgery. To find plausible explanations for these inter-clinic disparities, we adjusted our results for a number of possible confounders, but the differences remained (Model 4, Table 6). These dissimilarities might be due to differences in the patient categories visiting the various types of clinics. Using that perspective, we can say that the sickness certification task differs depending on where a physician works, and thus the need for CPD ought to vary between physicians in different clinical settings and therefore might involve somewhat different aspects. Other interpretations of the results are also possible, for example, the idea of different sickness certification "cultures."
In a previous study of the same questionnaire data , we found that the level of problems experienced in sickness certification also varied substantially between clinical settings. In addition, reports of problems came from the same types of clinics/practices where many physicians indicated the need for further knowledge and skills in corresponding areas. For instance, a large number of physicians working in primary care had problems with this task, and they also expressed the need for more knowledge.
Physicians in primary care or occupational health services differed more from the reference group mean in the skills index than in the knowledge index (Model 4, Table 6). This could imply that they already had relatively high levels of knowledge, but they were confronted with more difficult cases and therefore needed relatively more skills than knowledge. Most of previous interventions (half or full day courses arranged by the SIO) have mainly been directed towards GPs.
For one of the items (E in Table 2), none of the correlations with the other items reached 0.500, probably due to few physicians wanting more knowledge or skills in issuing sick notes. Two other items (G and J) were only weakly associated with the other items, and hence they were not included in any of the indices. These two items were among those that the lowest percentage of physicians indicated they wanted to learn more about (Table 2).
Implications for further research
Further research should be performed to elucidate the complex relationship between physician competence/experience and performance in insurance medicine, for example considering the quality of sick notes. Furthermore, studies are needed to explore the prerequisites for high quality in the sickness certification performed in different clinical settings.
Implications for practice
It is essential that stakeholders in this area recognise that physicians feel they need not only knowledge, but also skills, in insurance medicine to deal with aspects such as assessments of work capacity and handling conflicts. Furthermore, the social insurance system lacks transparency in the sense that it does not allow physicians to fully understand their own responsibilities or those of other participants in the sickness certification process. Measures should be taken to improve physicians' competence in insurance medicine, and this should be done in pre- and postgraduate training, as well as in CPD.