The history of registered sickness absence predicts future sickness absence

Occup Med (Lond). 2011 Mar;61(2):96-101. doi: 10.1093/occmed/kqq181. Epub 2010 Dec 20.


Background: The history of sickness absence has been found to predict future sickness absence.

Aims: To establish the review period of historical sickness absence data that is needed to predict future sickness absence.

Methods: The individual number of days and episodes of sickness absence were ascertained for 762 hospital employees from 2004 to 2008 inclusive. Past sickness absence was included stepwise in ordinal regression models. The explained variance of the ordinal regression models reflected the extent to which future sickness absence could be predicted and was expressed in percentages calculated as Nagelkerke's pseudo R(2) × 100%.

Results: A total of 551 employees (72%) had complete data and were eligible for regression analysis. Days of sickness absence in the past year predicted up to 15% of future days of sickness absence. Adding the sickness absence data of the past 2 or 3 years did not further increase the predictability of days of sickness absence. Episodes of sickness absence in the past year predicted up to 25% of future episodes of sickness absence. The predictability of episodes of sickness absence increased to 30% when the past 2 years of sickness absence were included in the regression model, but did not further increase when sickness absence of the past 3 years was included.

Conclusions: Employees who are more likely to have an above average sickness absence can be identified from their history of sickness absence in the past 2 years.

MeSH terms

  • Absenteeism*
  • Adult
  • Forecasting
  • Humans
  • Middle Aged
  • Netherlands / epidemiology
  • Personnel, Hospital / statistics & numerical data*
  • Regression Analysis
  • Sick Leave / trends*
  • Time Factors