1. Health measures in prediction models for high sickness absence: single-item self-rated health versus multi-item SF-12.
- Author
-
Roelen, Corne A.M., Heymans, Martijn W., Twisk, Jos W.R., Laaksonen, Mikko, Pallesen, Ståle, Magerøy, Nils, Moen, Bente E., and Bjorvatn, Bjørn
- Subjects
CHI-squared test ,CONFIDENCE intervals ,GOODNESS-of-fit tests ,LONGITUDINAL method ,QUESTIONNAIRES ,RESEARCH funding ,STATISTICAL sampling ,SICK leave ,T-test (Statistics) ,RECEIVER operating characteristic curves ,DATA analysis software ,DESCRIPTIVE statistics ,ODDS ratio - Abstract
Background: Self-rated health (SRH) has been found to predict sickness absence (SA). The present study investigated the effect of replacing single-item SRH by a multi-item health measure on SA predictions. Methods: Longitudinal study of 2059 Norwegian nurses with assessments in three waves each separated by 1 year. Health was measured by single-item SRH and multi-item SF-12 in waves 1 and 2. SA was self-reported in all three waves and high SA was defined as more than or equal to 31 SA days within the last 12 months. Predictions of high SA by a model including age, prior SA and single-item SRH were compared with predictions by a model including age, prior SA and multi-item SF-12. Both models were bootstrapped to correct for over-optimism and prospectively validated for their predictions in a new time frame. Results: 1253 nurses (61%) had complete data for analysis. The SF-12 model predicted the risk of high SA more accurately (X
2 = 4.294; df = 8) and was more stable over time than the SRH model (model X2 = 14.495; df = 8). Both prediction models correctly discriminated between high-risk and low-risk individuals in 73% of the cases at wave 2 and in 71% of the cases at wave 3. Conclusions: The accuracy of predictions increased when single-item SRH was replaced by multi-item SF-12, but the discriminative ability did not improve. Single-item SRH suffices to identify employees at increased risk of high SA. [ABSTRACT FROM AUTHOR]- Published
- 2015
- Full Text
- View/download PDF