This cross-sectional survey is the first to study the attitude and perceptions of insurance physicians towards evidence-based medicine and clinical practice guidelines. We performed the same study in a group of Belgian occupational health physicians  and will rely on the results of this parallel study and results from foreign studies in other medical fields for comparisons.
The response rate was average but acceptable, (48.7%) probably due to the fact that insurance physicians are not as familiar with evidence-based medicine. Further qualitative research would be a good complement to this study to determine the possible reasons for this low response rate and for validating the results. Despite the fact that we invited all physicians employed at public health insurance organizations or the NIDHI, it may be possible that the results are not representative for the whole population of insurance physicians. The survey was self-administered; physicians who are more positive or more familiar with EBM were probably more inclined to fill in the questionnaire. The category of insurance physicians privately employed were not part of our population. We are aware that this group could have had different opinions about EBM and clinical practice guidelines. The EBM knowledge questions did not represent actual knowledge but perceived knowledge which had its limitations and could be biased.
The reliability of the questionnaire was not quantified but logic checks were included. However, the logic checks gave no satisfactory results, probably because a relatively large share of physicians had no strong opinion about EBM and CPG's, supported by the relatively high percentage of neutral responses in the Likert scales of the attitude questions. Despite this, the majority of insurance physicians were still positive about EBM and CPG's, which echoes earlier studies [5, 9] and the results of our parallel study in the group of Belgian occupational health physicians . A positive correlation existed between the general attitude towards EBM and the general attitude towards clinical practice guidelines (rho = 0.396 - p = 0.05). No differences in attitude were observed between insurance physicians and medical inspectors of the NIDHI.
No relationship was observed between the general attitude towards EBM and general EBM knowledge. Better knowledge of EBM did not increase the awareness of its importance, supported by the systematic review of Coomarasamy et al . 18.1% became familiar with EBM during their basic training, all of them except one are older than 34 years while the EBM course was only recently integrated in the basic curriculum. These results could indicate that respondents may have had different interpretations of the meaning of an EBM course or the meaning of EBM itself.
45.7% of physicians perceived their searching skills as good to perfect while roughly the same percentage could not formulate a PICO question (51.4% had none to a little knowledge), were not familiar with MeSH terms (42.9%), and could not use methodological filters (46.7%), etc. It is not clear whether they considered their searching skills to be part of their EBM skills or not. The results could be explained by the conclusions of Sackett  who maintains that physicians think they are already practising EBM while in reality they are not.
Insurance physicians assess work incapacity and applications for health care reimbursement. For both activities it is important to correctly appreciate the diagnostic and therapeutic choices made by the treating physician. It therefore seems logical that the medical directors of social insurance institutions promote and enhance EBM skills among the social insurance physicians. Our results demonstrate that there is still a great need both for training in EBM and implementation of EBM in practice.
Personal characteristics of the respondents were not correlated to attitude and knowledge of EBM. The use of EBM is not related to age but rather personal conviction and practical possibilities.
Lack of time and lack of EBM were the most important barriers which echoes earlier studies [5, 11]. The reported barrier in terms of lack of evidence is not typical for all medical specialities but was also observed in our parallel study . Perceived lack of evidence was expected in this population given the difference in social insurance and workers' compensation legislation between countries. Although research is growing in insurance medicine, this perception could be partly solved by informing physicians of existing evidence in other medical fields which could be used to support their disability evaluations. It is remarkable that 55.2% of physicians indicated legal factors as a potential barrier. The tasks of the insurance physician are determined by Belgian legislation. The characteristics of the tasks of the insurance physician and the legal criteria leave little room for interpretation which is perceived as a potential barrier by many physicians. However, this is only partly true, as an insurance physician has to be able to correctly assess each medical condition based on good evidence.
The reported high impact of the conclusions of the literature on daily work (electronic as well as on paper) is somewhat in contrast to the low frequency of use of electronic databases. Roughly 10% admitted using electronic biomedical databases several times a week while 50% of the respondents reported that they had searched the literature to solve a specific problem for the last time last week. The same discrepancy in results was found in our parallel study in the group of occupational health physicians . The results here led us to presume that sources of literature other than electronic bibliographic databases were used to solve specific medical problems. This is to be expected considering the age of the physicians and the fact that journals were reported as the most important information source.
The majority of physicians gather their information here and there, and the access to high-quality information sources was low. It is possible we may have over- or underestimated the access to information sources because respondents may have interpreted 'access to' as 'awareness of', e.g. no-one reported having access to Bandolier while the information on the Bandolier website is free.
Efforts should be focused on improving personal access to electronic databases and the internet at the location where medical advice is provided; physicians with access should be encouraged to regularly search and use the literature and corresponding electronic databases. Clinical practice guidelines could be a way of making the evidence directly useful for insurance physicians, considering the correlation between the attitude towards CPG's and EBM plus the positive attitude towards CPG's.