Our study aimed to seek potential differences in the use of ambulatory physician services between people in different employment situations. The higher coverage of services among permanent full-time employees compared with other employees and the unemployed could not be attributed to needs arising from medical conditions or health impairments; on the contrary, the analyses showed a greater inequality between this group and the people occupying less stable labour market status when adjusted for clinical need. According to the analysis by service sector, permanent employees visited workplace physicians and private physicians more frequently, whereas fixed-term employees and the unemployed used public GP services. Corresponding differences were present but less pronounced in dental care. With ambulatory hospital visits no differences between employment statuses were found.
Reasons for inequality
Our results are in line with previous Finnish studies showing an increase in visits to physicians after re-employment  and a higher frequency of visits to public primary health care among the unemployed . Although previous studies (e.g. ) have reported socioeconomic inequality in Finnish hospital care, we found no evidence of this in ambulatory hospital visits. However, these studies concern quantity and quality of inpatient care, and our results, in fact, do not contradict their findings.
One explanation for our results is evidently the more comprehensive health service spectrum available to permanent full-time employees: in addition to GPs (in public health centres) they commonly have access to the occupational health care physician of the workplace, and they can also afford to visit various specialists (including dentists) in the private sector. From the perspective of the individual client the services certainly have additional value, but from the perspective of society the health care serving permanent employees may involve inappropriate and uneconomic overlapping.
The results concerning participants with a diagnosed disease also prompt the question of whether the coverage differences indicate marginalisation of the unemployed and fixed-term employees or medically unnecessary visits to physicians among permanent employees which may be related to employment. If we assume that absence of disease equals medical need of services the figures of Table 4 support the latter conclusion. If we assume that presence of disease equals the need, the former conclusion also gets support, in particular as regards fixed-term employees and the long-term unemployed. Thus, the answer would be 'both'. The reasoning is, however, complicated by the possibility of reverse causality, in other words, one cannot have a 'disease diagnosed by a physician' without visiting a physician, and the disease reported in the questionnaire may in fact be a consequence of the visit. Reverse causality might explain why adjustment for perceived health in the analyses of Table 3 lowers the odds ratios more than adjustment for disease.
The analysis of participants with cardiorespiratory and musculoskeletal problems serves to develop the discussion. According to national Current Care Guidelines, people with a chronic disease – in particular a cardiovascular or respiratory disease – should visit a physician at least once a year. No visit during twelve months therefore indicates obvious underuse of health services, or marginalisation from clinically relevant care. Table 6 illustrates that this is the case with fixed-term employees as well as with the unemployed. Moreover, the figures show that visits to health centres only partially compensate the inequality in use of primary health care created by occupational health and private services. The relatively high frequency of hospital outpatient visits made by these groups may also be interpreted as a compensatory action, but it may also indicate the need for more frequent specialist consultations because of a more severe condition. Finally, the conclusions remain uncertain since we do not know whether the visits had been made due to reported diseases or some other health problem.
In all, the study found considerable differences in the use of ambulatory physician services between people with different labour market statuses, and these differences grew when the mismatch between health status and use of services was taken into account. In particular, the health care of the long-term unemployed seems to be inadequate.
In the Finnish health service system, fixed-term employees and unemployed people seem to end up as clients of public health centres rather than any other sectors. The GPs in public health centres have received training to meet the needs of various client groups, but the question can still be raised as to whether they are aware of the specificity of their working-age clientele, whether they are competent, and whether postgraduate medical education should be available for handling the health problems of patients in more or less 'atypical' career situations.
The results may also reflect the importance of labour market status as a non-medical factor influencing health service need. It is evident that the care-seeking decisions of the unemployed are affected by less immediate needs to justify the sick role. In the case of a fixed-term or a part-time employee there may be reluctance to adopt the sick role, as it may risk the job contract or future career. The frequency and the nature of medical consultations may also differ due to differences in clinical needs such as symptom alleviation and rehabilitation. The need to optimise physical and mental fitness is probably more urgent in working life than in the everyday life of an unemployed individual, and feelings of hopelessness and indolence may affect the priority and perception of seeking health care. Moreover, an unemployed individual who feels ill may be reluctant to consult a physician for fear that the moral connotations of the employment situation become explicit, or for fear that a document of poor health may have an unfavourable influence on employment prospects. Correspondingly, non-permanent employees may perceive increased job insecurity to be a consequence of seeking health care.
The Finnish practice of sickness absence certification generates a need for 'clinically unnecessary' visits, in particular among permanent employees, who have more sickness absences than non-permanent ones . The unemployed are also expected to present a certificate for being 'absent from the dole' when ill, although they seldom do this and, then, only in cases of longstanding disability.
Certification of sickness absence is a concrete example of health care functioning as an institution of social control . As a part of society, health care, and in particular occupational health care, is bound to serve society's dominant ideas and values. In Western capitalist societies the comprehensive health services of the permanently employed core work force may be interpreted not only as benevolent promotion of their wellbeing, but also as exploitative maintenance of their production capacity. Correspondingly, the buffer work force and the unemployed could be seen as marginalised from these services because their contribution to the production – and consumption – of commodities is less important. On the other hand, in an employment society unemployment is perceived as deviance, but for an individual it also means 'freedom' from work. Indeed, the present findings are consistent with the possibility that unemployment may free an individual from unnecessary medicalisation and the domination associated with it, as well as iatrogenic health problems.
As well as viewing the relationship of medical encounter with labour market status in terms of conflict theory, it can be considered through a number of less macrosociological and less structural frameworks. The analogy of the sick role with the role of the unemployed is obvious. An unemployed citizen has also failed to comply with social expectations, is dysfunctional for the social system, and needs to be controlled and regulated. The role of unemployed legitimates withdrawal from a social obligation – in other words work – and the unemployed individual is exempted from responsibility, that is to say (s)he is not blamed for his/her inability to keep or get a job. However, these rights are granted only on condition that (s)he shows motivation and co-operates in getting re-employed. (S)he is required to utilise relevant employment services and enrol as a job seeker at the labour force bureau. It is evident that illness and unemployment, when occurring simultaneously, have a new significance for an individual, but we may also ask whether the roles are partly interchangeable, in particular in the case of chronic illness and/or unemployment. Thus, we may ask whether health services and employment policy services are separate or partly alternative social systems for adapting deviant individuals. Furthermore, 'double deviance' on the part of the client may impact on the practices of professionals providing health and employment services.
The above analogy is by no means specific to the Parsonian, structural-functionalist framework. Utilising the Foucauldian concepts  health as well as employment services are constituents of the 'panopticon', or the set of social institutions used for the surveillance of citizens and for the execution of expertise and professional power. Indeed, the relative marginalisation of the unemployed from the health services may partly indicate that they tend to avoid medicalisation of their problems, and that they also have more freedom with regard to the sick role and to the social control and surveillance carried out by the health care institutions.
Physician consultation is both quantitatively and qualitatively the most important contact between an ill citizen and the health care system. There are also contacts with nurses, physiotherapists, psychotherapists and other health professionals, as well as with unofficial service providers, but it is unlikely that differences in the use of these services compensated the differences observed in this study. Rather, nurse contacts are most common in occupational health care, and there are also a large number of physiotherapists.
The investigation regarding visits to physicians in specified organisations may omit some contacts such as those taking place in student health services and military health care, but the vast majority of these groups were excluded from the study. Thus, the rate of visits to physicians used in this study may be considered a valid measure for the use of and access to health care.
Studies show that respondents remember recent visits to physicians fairly well, and that there is no association between demographic or health variables and the tendency towards discrepancy between self reported and registered visit rates .
With a response rate of 80% it is reasonable to assume that the respondents represent the population recruited in the HeSSup study and the follow-up. The low participation at baseline is analysed in detail elsewhere : the unemployment rate of the participants differed significantly from the expected (8.6% vs. 9.3%, odds ratio 0.92, confidence interval 0.88–0.97), but in the light of absolute figures this, as well as the differences in sex, marital status, education and indicators of health, could be considered as acceptable. Moreover, the proportion of unemployed respondents (6.9%) corresponded fairly well to the falling national unemployment figures  at the time of the second phase survey. Thus, although we cannot be sure that the respondents were a representative sample of the Finnish population, we may reasonably conclude that the sample was not too biased from the viewpoint of our study questions.
Smaje and Le Grand  present an in-depth discussion of the factors affecting the utilisation of health services, in particular of medical need as a confounding factor in comparisons of utilisation among population groups. The multivariate analysis applied in this study is similar to Smaje and Le Grand's model with the exception that, instead of ethnicity, our grouping was based on labour market status. We controlled separately for perceived health and reported disease as factors affecting service need. The results lend support to earlier evidence that people may be more likely to report long-term illness if they have recently visited a physician . Therefore, adjustment merely for reported disease as a measure of medical need might bias the analysis. This is why perceived health may be considered as a better indicator of the clinical need for services. Utilisation of three variables – in addition to self-rated general health, breathlessness and depression – makes the adjustment more comprehensive than in previous studies, although the results were quite similar (odds ratios not shown) when these variables were introduced separately into the analyses.