The data file “FD-trygd” links Statistics Norway’s demographic and education data with files from NAV about disability pensions for the period 1 January 1992–31 December 2003 for all persons living in Norway [8, 9].
We chose a sample aged 30–55 years in 1992 because they were eligible for the disability pension over the whole period, with the oldest persons reaching retirement age after the study period and the youngest being old enough to get disability pensions for reasons other than inborn conditions. Those receiving the disability pension at 1 January 1992 were excluded, as were those not living in Norway on 31 December 2003.
The dependent variable was cumulative incidence of disability pensions from 1 January 1992 to 31 December 2003, taken from the FD-trygd file without missing cases. In Norway, all disability pensions are assessed, wholly or partly, by the national social security system, NAV.
Place of Origin
We used the official definition of immigrants based on the country where the parents were born, namely in Norway, other Western countries, East European countries or developing countries . Persons with one Norwegian parent were defined as ethnic Norwegians; otherwise, the more affluent of the parents’ respective countries of birth was used. We dichotomized the population into Westerners and non-Westerners, the first group being ethnic Norwegians and immigrants from Western Europe, North America, Japan, Australia and New Zealand. In our population, 92% of non-Western immigrants were born in their country of origin, i.e. they were first generation immigrants.
We used education as a proxy for occupational level to assess risk of receiving a disability pension. Statistics Norway records all Norwegian education outcomes, while immigrants have their self-reported education registered by immigration authorities. This is not always done, and in our study the education levels of 7.1% of non-Western males, 7.3% of non-Western females, 2.7% of Western immigrants and 0.3% of ethnic Norwegians were not recorded. Because a large proportion of non-Western immigrants are in unskilled jobs , we imputed basic education, which corresponds with unskilled work, to the missing cases.
Bivariate associations were tested by chi-square statistics and t tests (Tables 1, 2, 4). Poisson regressions were used for computing relative risks of receiving a disability pension adjusted for age, education and age at pension receipt (Table 3).