Physicians have an essential role in the sickness absence process; however, there is very little scientific knowledge about this. In the last years, some more studies have been published but the number still is scarce [1–6]. Possible negative consequences of being sickness absent have recently been highlighted [7, 8] as well as the importance of sickness certificates being issued with the same caution as other recommendations made by physicians to their patients, in order to avoid negative consequences [1, 9, 10].

In most countries there are two requisites for being entitled to sickness benefits; one must have a disease or an injury and this disease or injury must have affected one's work capacity [11]. In Sweden, as in most countries, all physicians may issue sickness certificates. However, writing the certificate is only one of the tasks that may be involved in a consultation where sickness-certification is considered. Those different tasks can be summarised as [1]:

In a systematic literature review of studies of sickness-certification practice, scientific evidence could be established for only two issues; that the sickness certificates had low quality and that the physicians found sickness-certification problematic [1, 11]. The Swedish Council on Technology Assessment in Health Care (SBU) concluded that the number of studies was low and that sample sizes were small and often very biased. Moreover, SBU stated that larger studies were warranted to be able to generalize results and to get more detailed information, as basis for interventions.

In this research area it is especially crucial to also use the physicians' perspective on the sickness-certification task and related problems, in order to be able to tailor interventions along the line of their experiences and wishes. Moreover, most studies in this area have included only general practitioners (GPs) [2, 3, 11–27]. Some studies have compared the frequency and the severity of problems among GPs with that of other physicians [1, 5, 6, 28–30]. Nevertheless, the knowledge about sickness-certification practice among other physicians than GPs is still scarce.

Therefore, we in a previous questionnaire study included physicians also from other clinical settings [6]. One finding was that consultations involving sickness-certification were equally, or even more, frequent among other specialists and clinics than in primary health care (PHC)/among GPs [6, 29]. This was unexpected and we have, consequently, here pursued our research about this in a larger study, including all the physicians in Sweden.

The aim was to gain detailed knowledge about physicians' work with sickness-certification regarding frequency of different types of situations and severity of related problems, in general as well as in specific clinical settings.

A cross-sectional questionnaire study was conducted. The questionnaire was administered to all physicians in Sweden, a country with 9.3 million inhabitants. The study population was the 36,898 physicians of all ages who lived and mainly worked in Sweden in October 2008. They were identified using a register of all physicians, held by the company that in Sweden has the most complete such register; Cegedim AB.

A comprehensive questionnaire of 163 questions about various aspects of sickness-certification practice and related work issues was developed, based on a previous questionnaire, research, and contacts with many different clinicians and researchers in the area [29, 31]. The questionnaire was tested in a pilot study with a random sample of 100 physicians (62.0% response rate).

In this study we included the 14,210 participants who were below the age of 65 and had consultations concerning sickness-certification at least some times in a year.

Of all the responding physicians below the age of 65, 67.4% had consultations considering sickness-certification at least a few times a year. That rate was higher for non-specialists than for specialists; 73.6% and 65.1%, respectively (Table 1). In a number of clinical settings, nearly all physicians had such consultations at least a few times a year; PHC (97.4%), infectious diseases (97.0%), occupational health service (96.4%), oncology (96.0%), and rheumatology (99.0%).

From here on, only those 14,210 physicians <65 years of age who at least sometimes per year had sickness-certification consultations were included in the analyses. More than one third of the physicians in twelve types of clinics had such consultations more than five times a week (Figure 1). The highest proportion of physicians who had such consultations more than five times a week were physicians in orthopaedic clinics and occupational health service. Notably is that also pain management, oncology, psychiatry, rehabilitation, rheumatology, and neurology, together eight different clinics, had much higher rates of sickness-certification consultations than those in PHC.

The association between the proportion of physicians in different clinics having sickness-certification consultations more than five times a week and the proportion having problems with these work tasks at least once a week is illustrated in Figure 2.

In most clinics, the proportion of physicians having many such consultations was positively associated to the proportion finding them problematic. However, the physicians in oncology clinics and in PHC diverged. A high rate of the physicians in oncology clinics had sickness-certification consultations but a relatively low rate found them problematic, while the contrary was found for PHC.

This is, so far, the internationally largest questionnaire study of physicians' sickness-certification tasks and problems, regarding type of situations they encountered and problems they experienced, including all physicians working in Sweden. Sickness-certification consultations were far more frequent than anticipated - more than two thirds of the physicians had such consultations, and of those over 80% had that at least once a week. Moreover, the rate of having such consultations was much higher in many other clinical settings than in PHC, e.g. in orthopaedics and in oncology. This has not been studied in other countries. Obviously, the previous focus on GPs regarding these tasks needs to be reconsidered. Nevertheless, a much higher rate of the physicians in PHC experienced these tasks as problematic. The majority of all physicians rated the task to assess magnitude of work incapacity and the prognosis of such incapacity as problematic. More detailed knowledge, regarding the broad tasks involved in sickness-certification consultations, has been called for [32, 33]; some of that is provided here.

Strengths of the study are the very large sample size, that all physicians (N = 36,898) working and living in Sweden were included, that all clinical settings were included, and the detailed questions about these tasks. The study group is large enough to admit sub-group analysis, e.g. regarding different types of clinics. Another strength, from an intervention perspective, is that these results were based on the physicians' own experiences of tasks and problems. They can, therefore, be of good use when e.g. targeting different types of competence development [13]. A limitation is the drop out of 39%. Nevertheless, the response rate can be considered high for this type of study and the study design admits analyses of bias in the dropout. Differences in dropout rates between physicians being board certified specialist and non-specialist i.e. not yet fully trained or registered specialist might have affected our results. Non-specialists were, as expected, younger. During training, they often change residence, also geographically, which might be one reason for the higher drop out. The non-specialists had a higher dropout rate and also reported more problems regarding sickness certification than the specialists, which might have lowered the crude OR for problems reported from clinics with higher proportion of non-specialists. That is, in some cases the ORs might be underestimated, even when adjusted for rate of registered specialists.

To sickness certify a patient is a common task in health care in Sweden as in many other countries. However, this recommendation can so far not be based on scientific evidence [1]. In a previous, smaller study a slight association was found between physicians having at least six sickness-certification consultations a week and rates of having problems with this at least once a week [6]. The results of the present, much larger study, goes in the same direction and we found similar associations in the majority of clinics, however, not for physicians in oncology and PHC. The proportion of physicians experiencing problems regarding sickness-certification in general, as for specific items, varied with type of clinic. The physicians in PHC had the highest ORs for experiencing problems, although they did not have the highest frequency of sickness-certification consultations.

Actually, also some physicians in geriatrics and child care had sickness-certification cases. This can be explained in at least three ways; one is that the participants were asked to indicate the type of clinic where they mainly worked, but some might also be clinically active in other types of clinics. The second is that some of their patients are adolescents, above the age of 16, and thus can be sickness absent and in geriatrics also some patients work in spite of old age. The third is that they might sickness certify parents of children or relatives of geriatric patients. So far, none have studied this.

We were surprised by the very high rates of consultations and also by the large variety in rates of physicians experiencing problems with these tasks. In Sweden, as in most welfare countries, specialists generally are to refer patients to PHC when treatment is finished or stabilised. It is an understanding that sickness-certification is to be monitored from PHC if a patient is referred to other clinics by the general practitioner in PHC. Nevertheless, very high rates of physicians in different clinics had such tasks very often. More studies are needed to verify these results also in other countries. However, other studies indicate that GPs generate about half of all sickness-certification, which is well in line with our results [1, 34]. An obvious issue here is whether the results can be generalized also to other countries. Sickness-certification practices have only been studied in very limited populations and mainly for GPs in other countries - thus, the situation might be similar there, however, that remains to be seen. Regarding other aspects of sickness-certification practice, results from different countries and from different time periods have been unexpectedly similar [11].

The most problematic part of sickness-certification seemed to be to assess the magnitude of the patient's work capacity. This result goes in line with some previous studies [19, 35, 36]. A variety of instruments for assessment of work capacity are used in different countries, however, scientific knowledge on the validity of them, their effects, and on possible implications for the work of physicians is warranted [37]. Overall, the highest OR for problems with work-capacity assessments was found among physicians in PHC and in psychiatry, rheumatology, and neurology which had a still higher OR for problems providing a long-term prognosis of work capacity. We have not found any other studies about this.

The majority of physicians had sickness-certification consultations every week and problems experienced regarding this varied substantially in frequency as well as severity between clinical settings. The physicians at a vast majority of the clinics regarded sickness-certification consultations as problematic and far more so among physicians in PHC in spite of that they did not have as many such consultations. So far, most interventions concerning sickness-certification have been targeted towards physicians in PHC/GPs. Other physician groups with high frequencies of consultations and/or problems were found in oncology, orthopaedic, psychiatric, pain management, and rheumatologic clinics. The results indicate the importance to take account of the variety of problems in physicians in different clinical settings experience, when planning interventions aimed at improving their work with sickness-certification of patients.

So far, most interventions regarding physicians' sickness-certification practices have been targeted towards PHC and GPs. Our results indicate that those physicians also to a great extent experience sickness-certification consultations as problematic. Nevertheless, also other physicians have high risks for experiencing them as problematic, e.g. in rheumatology, neurology, psychiatry, and orthopaedic clinics. Moreover, in several clinical settings the physicians had such consultations more often than the GPs. Thus, the results indicate that much can be gained through focusing on physicians also in other types of clinics when planning interventions to improve physicians' sickness-certification practice.