["Index for rehabilitation eligibility" for screening social insurance workers].

The principle "priority of rehabilitation over early retirement" might be realised by a screening by which employees in need of rehabilitation are detected in time and rehabilitation measures are purposively started. With the "Index of Rehabilitation Need" we continued our efforts to develop an applicable screening tool on an epidemiological basis. To this end (1) longitudinal data were established by repeating an epidemiological survey of a population sample in the Nordenham/Brake region (T0 = 1975/76, T1 = 1992/93); (2) T0-variables were identified which correlated significantly with the events of early retirement and/or rehabilitation in the period of T0 to T1 (98 cases of early retirement/357 controls; 127 cases of rehabilitation/200 controls; 185 cases of early retirement of rehabilitation/270 controls) using bivariate and multivariate regression analysis; (3) significant T0-variables were used to construct a questionnaire index (based on self assessment of symptoms/complaints, consumption of medicaments, smoking, and work load--16 items), a medical examination index (based on clinical/laboratory findings and medical diagnoses--10 items), and an overall index (sum of both indices--26 items); (4) the index values were calculated for cases of early retirement of rehabilitation and controls of the cohort (185/270), for each index significant differences between cases and controls tested, and the screening characteristics of the overall index analysed; (5) possible reasons for incorrect classifications were examined using a subsample of cases and controls (n = 96/78), for which additional data on medical and work history, stressful life events, and attitudes towards rehabilitation had been collected. All indices showed significant differences between cases of early retirement or rehabilitation and controls. These differences proved to be stronger with the questionnaire and overall indices (p < .0000 each) than with the medical examination index (p < .0006). The overall index did not detect 18% of the cases in need of rehabilitation (false negatives). The proportion of the false positives was 14%; sensitivity and specificity amounted to 57% and 76%. The analysis of the subsample revealed only two possible and plausible reasons for incorrect classifications: the time span between the first survey and the year of early retirement as well as injuries. The index detected cases of early retirement or rehabilitation more easily where the time span between T0 and the year of early retirement was shorter. The index cannot detect cases of early retirement and rehabilitation caused by injuries between T0 and T1, since it is based on chronic disorders and stresses to be the reason for both events. With respect to the sensitivity and specificity of the index the relatively long prediction period needs to be taken into consideration--between T0 and the time of the events there could have been a period of up to 17 years. However, the objective of a screening is not to predict the long-term outcome but to preselect persons who are likely to need rehabilitation and should be invited to a socio-medical examination in order to clarify their rehabilitation need and to start appropriate rehabilitation measures. The chance to detect true positive candidates and to exclude false negative candidates is essentially higher when the measurement of the predictors and the examination are carried out at the same time as has been shown in a former study. With regard to further proceedings we suggest to apply the index in a screening and to investigate the cost effectiveness and other aspects of the screening in a demonstration project.