Methods for analysing individual changes in sick-leave diagnoses over time

摘要: Several methodological challenges arise when attempting to analyse individual data on changes of sick-leave diagnoses over several years. Sick-leave spells for a person can recur, have different diagnoses, and both these aspects are dependent previous episodes, the numbers repeated periods vary across subjects, standard statistical methods not valid variables nominal scales, e. g. diagnoses. Objective: Our aim was ascertain whether number pattern in associated with future disability pension (DP) test analysis measurements data. Participants: Data from 12-year prospective cohort study 8000 213 persons aged 25-34 who, 1985, had new spell andgt;= 28 days back were used. Methods: We used entropies, uncertainty coefficients adjusted measurements, transition matrices examine that occurred during follow up. Results: In 12 years 22% granted DP they changed diagnosis less frequently more often musculoskeletal than others. The variation degree dependence between consecutive DP. Conclusions: Many tools statistics based linear require numerical variables, but such discrete as diagnosis. cases, it be beneficial use applied information theory.