The analysis of the interrelations between functional disabilities of claimants for a disability pension produced four main types of disabilities. On the basis of those four types of disabilities, we identified four subgroups of severely disabled claimants who were still able to work in normal jobs, and were relatively homogeneous in the main type of disablement. Within each of those subgroups, jobs that were regarded as accessible for them were classified into occupations. The occupations most frequently used for the claimants within each of the four subgroups were selected to become reference occupations. The reference occupations can be used to direct the rehabilitation efforts towards accessible work opportunities, and may be applicable in the assessment of the work capacity of claimants in regular jobs. In the search for a job for a claimant with a severe disability of one of the four types, and restricting the search to regular jobs without major adaptations, one should start with considering the reference occupations belonging to that type of disability. If the claimant is able to work in one of the reference occupations, one can subsequently look for further job opportunities, possibly with higher task demands, also depending of the characteristics of the client, e.g. education; work experience; motivation; and so on. On the other hand, if a claimant is not able to work in any of the reference occupations, it is unlikely that any other occupation is accessible for that claimant because of the severity of his/her disability and the higher task demands in those other regular occupations. Therefore, the reference occupations may be useful instruments to assess the possibilities of claimants to work in regular jobs at all, i.e. in jobs without major adjustments. The reference occupations were selected because their frequent use in disability assessments for one of the four subgroups, and therefore the corresponding work demands are very low. Especially for the severely disabled these occupations are the rare opportunities to work in regular jobs. Moreover, none of the work demands in these occupations reaches a level that forms a threshold for many claimants, so the general accessibility of these occupations is relatively high, also for people with other disabilities.
Four scale variables were computed to measure the four disability types. The Cronbach's Alpha of one of those scales was only 0.46, probably due to the low number of two variables on which the scale was based. For the application of the scale scores in the identification of subgroups of severely disabled claimants in this study, the low value of the Cronbach's Alpha of this scale was considered acceptable.
To identify severely disabled claimants with one of the four types of disabilities (but still assessed as being able to work), two threshold values were arbitrarily set at 50% and 66% probability of a full disability pension. Additional sensitivity analyses in the vicinity of the 66% threshold did not show a sudden change in accessible occupations (results not presented here), indicating that the choice of the threshold level did not dominate the selection of reference occupations.
The four subgroups of severely disabled claimants differed in some characteristics from each other and from the other claimants (Table ). The subgroups disabled in psychological abilities and autonomy were more represented in the two younger age categories, especially compared to the subgroup disabled in general physical abilities. In the two subgroups disabled in general physical abilities and in psychological abilities included a higher percentage of female claimants. The four disabled subgroups were better represented in two higher educational categories taken together than the group of all other claimants. However, the results for the four educational categories separately differed between the four subgroups severely disabled claimants. Although some of the differences between the groups in the three characteristics of Table were substantial, none of these were indications that the four subgroups of severely disabled claimants were an extremely deviating selection from the total group of claimants.
The second order factor analysis on 14 of the 15 first order factors resulted in four factors, which was less than the seven factors that were identified within the Norwegian NFAS [11
]. The Norwegian scale "holding/handling" showed a great resemblance to the scale "manual skill and grip strength". Within the NFAS, three other scales within the physical domain were distinguished, whereas in our second order factor analysis the remaining factors were combined to produce the scale "general physical abilities". Of the three factors in the mental domain of the NFAS, the scale "senses" resembled the scale "communication" of the first order factor analysis of the LFA, which we excluded from our analysis because of the lack of reliability. The remaining two scales in the mental domain of the NFAS differed from the two psycho-social scales of the LFA. This may be due to differences in the subject matter and the application between the two functional ability lists, although both lists were partly based on the ICF [14
]. The main differences were: an emphasis on everyday activities vs. work-related activities; self-reported disability vs. assessment by an insurance physician; and duration of the sickness spell preceding the disability assessment: six weeks vs. two years.
The similarities and differences between the factors found in our study and in the Norwegian study [11
] suggest limited possibilities to apply our results in other countries. It seems plausible that the relative prevalence of some types of disabilities in the work force will show similarities between countries, especially under similar conditions with respect to, for example, the level of prosperity; the nature of the jobs; the working population; the social security system; and so on. Rehabilitation professionals in other countries, consulted by disabled clients of such a similar type, may be inspired by our reference occupations in their advising about accessible occupations, although the task composition of specific jobs may differ between countries. In addition, the method for the selection of reference occupations may be applied in other countries, although the availability of a similar data set may be problem.
In the work disability assessment procedure, a job file with a total set of about 7,000 jobs was used to select suitable jobs from. Although this number of jobs was substantial, these jobs constituted only a limited part of the total of millions of regular jobs in the Netherlands. The size of the job file of the CAMS was limited because of the costs of the observations on location of the task demands of jobs within companies and of the scheduled updates of those observations. The coverage of the total of 1,211 occupational titles of the SCO-'92 by the occupations in the CAMS was relatively high in the three lower level occupations (coverage 47%; 39%; and 23%, respectively), whereas the occupations on the high level (14%) and the academic level (3%) were relatively sparsely covered (the second row of Table ). In the development and maintenance of the CAMS job file, this unbalanced composition of the job file was preferred because the lower level jobs were accessible to almost all claimants as far as the required educational level is concerned, while higher level jobs were only accessible to the higher educated claimants with the right specialization. However, the selection of only elementary and low level occupations as reference occupations was not only caused by the composition of the CAMS job file. Although the education requirements of high and academic level occupations that were included in the job file were all of a more or less general nature to avoid additional barriers caused by the required specialization, those educational requirements would only be met by a small part of the work force (and of the claimants). Some task demands in a number of higher-level occupations may be low, especially the physical demands. However, the requirements of those occupations are often very specific, and the higher educated claimants are not able to return to their former higher level jobs because of health problems, and probably similar problems will occur if they try to work in other higher level jobs of a comparable nature. Moreover, they do not have the right specialisation to function in completely different occupations, and therefore they miss the opportunity to apply their specific work ability, as developed in their education and their former job [18
]. To exaggerate: the long-term sick-listed employee, formerly working in a high-level job, is apparently not able to function in that job, and probably neither in similar jobs for which he/she is qualified, and for other high-level jobs he/she lacks (some of) the specific qualifications needed in those jobs. Therefore, it is probable that the (lower level) reference occupations will be the only opportunities for the high-educated claimants to work in a regular occupation. Hence, it is improbable that including more higher-level occupations in the job file would have resulted in the selection of a higher-level reference occupation. It is plausible that the composition of the sets of reference occupations, with only elementary and low level occupations, was caused by the additional barriers in the higher-level occupation for many claimants, leading to a low a priori
maximum coverage. The reference occupations at the elementary and low level are useful to explore the general work ability [18
The SCO-'92 occupational classification comprises 1,211 distinct occupational titles. Each occupational title covers jobs from a variety of companies and institutions, and therefore with a certain variation in specific working situations. A limited selection of these jobs within each occupation was included in the job file of the CAMS. This subgroup of jobs may not be a representative sample of all the jobs within the occupation in the Dutch work force, for example relatively low in the level of demands or required skills and education. However, each occupational title covers jobs from a variety of companies and institutions, and therefore with a certain variation in specific working situations. Not all of the jobs within the job file had been used by the LE in disability assessments of moderate severely disabled claimants. In these disability assessments, relatively demanding jobs had a lower probability of being used. Therefore, the choice of a reference occupation will probably be based on the use of a relatively accessible subgroup of jobs within the occupation in the job file. The description of the reference occupations, including the task demands, to be used in the newly proposed disability assessment procedure should be based on this accessible selection.
Our proposal for the set of reference occupations is based on the outcomes of real assessments of work disability in the public work disability insurance system in the Netherlands. The functional abilities of the claimants were used to look for dimensions within those abilities. In addition, the data of work disability assessments of severely disabled claimants within each dimension were used to identify jobs low in task demands in the corresponding dimension. This way of identification ensured the relevance of the reference occupations for the disabled employees involved. The large number of assessments makes the results stable, i.e. the probability is high that the same procedure would result in (almost) the same set of reference occupations, if it would be repeated in another cohort of claimants. However, it remains unknown whether these results can be generalised to other countries and/or social security systems. Nevertheless, the development of reference occupations might be very relevant for other countries, given the size and urgency of the problems in the work participation of disabled employees[1
]. These reference occupations can be applied as an alternative method for the assessment of the work capacity of disabled employees in regular occupations as well as in rehabilitation efforts for those employees.