Analysis of Sickness Absence Using Poisson Regression Models

Only available on StudyMode
  • Download(s) : 133
  • Published : March 18, 2012
Open Document
Text Preview
ANALYSIS OF SICKNESS ABSENCE USING POISSON REGRESSION MODELS David A. Botwe, M.Sc. Biostatistics, Department of Medical Statistics, University of Ibadan Email:

ABSTRACT Background: There is the need to develop a statistical model to describe the pattern of sickness absenteeism and also to predict the trend over a period of time. Objective: To develop a statistical model that adequately describes the pattern of sickness absenteeism among workers. Setting: University College Hospital (UCH), Ibadan, Nigeria. Methodology: A retrospective study involving a review of sickness records of all workers in UCH between January and December 2003 was carried out. Data were extracted from the staff records of the Staff Medical Services Department. Independent samples t-tests and one-way analysis of variance tests were used to test for statistically significant differences in the mean number of spells and duration between various groups of workers. Poisson regression models were fitted to describe the pattern of the number of spells of sickness. Results: Out of 3309 workers, 240 had records of sickness absenteeism, giving a prevalence rate of 7.3%. The mean spells of sickness was 3 spells per absentee per year, while the mean duration of absence was 4 days per absentee per year. Females had a significantly higher number of spells than males (p = 0.009) and longer duration of absence than males (p = 0.015). No statistically significant differences were observed in the mean number of spells between junior staff and senior staff, although it was slightly higher in the former. The Poisson regression model showed that sex, staff category and occupation are predictors of the number of spells of sickness, while age and marital status are not. A linear relationship was observed between the duration of absence and the spells of sickness. Conclusions: The variation in the spells of sickness shows that type of work and sex differences have significant influences on sickness absenteeism. However, age and marital status are not contributing factors of sickness absenteeism, though there may be slight differences between the various groups. The number of spells of sickness follows the Poisson distribution, and the Poisson regression model is adequate to describe and to predict the pattern of sickness absenteeism.


INTRODUCTION Sickness absence can be defined as absence from work attributed to illness or incapacity. In spite of this there is a school of thought that sickness absence must never be considered as a reliable indicator of true morbidity. Some workers may absent themselves from work and collect sick certificates from doctor-friends just to cover up. Therefore, by way of monitoring the health of workers and filtering out malingerers, many organizations provide health services for their employees in their working premises (Taylor and Pocock, 1981). It is asserted that sickness absenteeism cannot be understood if it is viewed as a simple function of ill health or other individual factors, such as job dissatisfaction. Sickness absenteeism should rather be regarded as a coping behaviour that reflects an individual's perception of his/her health (or illness) and is a function of a number of factors at different levels, primarily the combination of job demands and coping possibilities at the job (Kristensen, 1991). However, the rate of sickness absenteeism is an important index for assessing the health status of workers in an establishment. A high rate of sickness absenteeism among workers is of great concern to management who may consider it an indication of possible occupational hazard or an expression of lack of job satisfaction by their staff (Bamgboye and Adeleye, 1992). Sickness is widely recognized as the most important cause of absenteeism, accounting for almost two thirds of all absences from work (Williams, 2002). A number of studies conducted on this issue have indicated that the causes of employee absences...
tracking img