Skip Navigation


Occupational Medicine Advance Access originally published online on March 15, 2008
Occupational Medicine 2008 58(5):355-360; doi:10.1093/occmed/kqn029
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
58/5/355    most recent
kqn029v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Lucey, S. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lucey, S. P.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2008. Published by Oxford University Press on behalf of the Society of Occupational Medicine. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Can pre-placement health assessments predict subsequent sickness absence?

Suzanne P. Lucey

Whitefriars Occupational Health Department, United Bristol Healthcare Trust, Whitefriars, NHS, Lewins Mead, Bristol BS2 8BQ, UK

Correspondence to: Suzanne P. Lucey, Whitefriars Occupational Health Department, United Bristol Healthcare Trust, Whitefriars, NHS, Lewins Mead, Bristol BS1 2NT, UK. Tel: +44 117 9282223; fax: +44 117 9283840; e-mail: sluceymed{at}yahoo.co.uk


   Abstract
Top
 Abstract
Introduction
 Methods
 Results
 Discussion
 Conflicts of interest
 References
 
Background Sickness absence is a growing economic problem, due largely to the financial losses it incurs. The ability to identify employees likely to take greater than average sickness absence may provide managers with useful information at the pre-placement stage.

Aim To confirm whether specific risk factors identified at the pre-placement health assessment could predict subsequent sickness absence.

Methods A total of 400 National Health Service pre-placement health questionnaires were analysed to allocate employees to low-, medium- or high-risk categories for subsequent sickness absence, using the risk table developed by C. J. M. Poole (Can sickness absence be predicted at the pre-placement health assessment? Occup Med (Lond) 1999; 49:337–339) [1]. Subsequent sickness absence was analysed to assess if there was an association between the allocated category and sickness absence taken.

Results Mean sickness absence hours per 1000 h worked were 22.5 (95% CI 18.2–27.2) in the low-risk group, 33.6 (27.2–40.7) in the medium-risk group and 44.7 (25.1–69.9) in the high-risk group (analysis of variance, P ≤ 0.002), demonstrating a statistically significant difference in sickness absence taken in subsequent years.

Conclusions The results confirmed Poole's hypothesis that future sickness absence can be predicted at the pre-placement health assessment. Certain risk factors, namely female sex, age, smoking, history of at least two previous episodes of low-back pain and previous days sickness absence identified at pre-placement assessment, predict a greater than average subsequent sickness absence. However, the best model using identified risk factors only predicted 10–12% of the variation in sickness absence.

Keywords      Occupational health questionnaires; pre-placement assessment; sickness absence; risk factors


   Introduction
Top
 Abstract
 Introduction
Methods
 Results
 Discussion
 Conflicts of interest
 References
 
There has been a steady rise in sickness absence over the past few years and this has been a cause of concern for the government and related bodies, such as the Health & Safety Executive, the Confederation of British Industry, the Department of Work and the Department of Health [2]. The ability to identify employees likely to take higher than average sick leave at the pre-placement stage may provide managers with valuable information and this has been attempted by various authors in the past [1,3–6].

Several factors have been suggested by Poole as having an association with morbidity and hence sickness absence, and these are a history of smoking, a high body mass index (BMI), previous depression, previous back pain, current sickness absence, past sickness absence, ischaemic heart disease and anterior knee pain intervention [1]. He hypothesised that employees could be grouped into one of three risk categories for future sickness absence, based on the presence or absence of these risk factors with Group 1 being low risk, Group 2 medium risk and Group 3 high risk.

The aim of this study, which included a literature search for each of the above risk factors and sickness absence prediction [6–28], was to test this hypothesis by identifying individual risk factors that could predict a greater than average level of sickness absence and determining whether specific risk factors collectively could predict a greater than average future sickness absence level.


   Methods
Top
 Abstract
 Introduction
 Methods
Results
 Discussion
 Conflicts of interest
 References
 
This was a retrospective cohort study of 400 consecutive employees commencing employment at a National Health Service (NHS) Trust from January 1998 to June 2000 with a 3- to 6-year follow-up period. Leavers prior to December 2003 were excluded. The personnel department provided information on all subjects eligible for inclusion. The local NHS Ethics Committee gave ethical approval for the study.

Of the first 400 employees, I excluded 24 as either there was no record of a pre-placement assessment or their case notes could not be accessed. The characteristics of this excluded group were not further assessed but there was no reason to believe they differed greatly from those of subjects included in the study. In order to collect data on 400 employees in total, I included a further 24 subsequent recruits.

I obtained the pre-employment forms from occupational health records and analysed them to place employees into a low-, medium- or high-risk category for sickness absence, using the risk category table (Table 1). The risk factors used to define each risk category were as follows: current sickness absence at pre-placement assessment, previous sickness absence spells and duration, history of back pain, history of depression, history of ischaemic heart disease, smoking status, BMI and anterior knee pain interventions. If an employee had more than one risk factor, he/she was placed in the higher category. I also recorded the gender and age of potential employees to determine if either was a risk factor for future sickness absence.


View this table:
[in this window]
[in a new window]

  		Table 1. Risk factors for high-, medium- and low-risk categories

 
The information was obtained from the pre-placement questionnaire completed by all prospective employees, which requested details of any medical conditions, including the dates of symptoms, the number of episodes and treatment to date. A nurse interviewed all prospective employees with patient contact and any employee who declared a medical condition or who provided incomplete information on the pre-placement form.

I undertook a pilot study of 20 questionnaires and obtained sufficient information to answer all questions except for previous sickness absence in the past year. Our pre-placement form requested a self-declaration of sickness absence over the last 2 years and this study therefore assessed prior sickness absence over a 2-year period rather than sickness absence per year for the last 2 years as described by Poole. Information about current life crises was not assessed at pre-placement and this factor has therefore been excluded in the study. Prospective employees with poorly controlled chronic illness or terminal illness were also not used in the study because the low numbers involved were unlikely to generate significant results. I calculated BMI for each person using information given in the pre-placement health questionnaire.

The maximum employment period studied was 6 years and the minimum period studied was 3 years and 6 months.

I analysed each employee's subsequent sickness absence over the defined period to determine any association between the risk category and sickness absence. Managers recorded details of the sickness absence of each employee from the date of commencement until the end of December 2003 and this was held electronically by the trust's workforce analyst. Sickness absence data was recorded as sickness absence episodes and sickness absence hours (corresponding to the actual contracted hours of work lost to sickness). The study therefore concentrated on two measures of sickness absence, sickness absence episodes and sickness absence hours.

For statistical analysis, the values were standardised so that differences in length of employment and part-time status could be taken into account. The standardised outcomes, sickness absence episodes per year and sickness absence hours lost per 1000 h worked were transformed by a square root transformation to give them an approximately normal distribution. One-way analysis of variance (ANOVA) was used to compare the sickness absence outcomes between the risk categories. Other variables were tested for their association with sickness absence episodes or sickness absences hours using t-tests, ANOVA or correlation coefficients as appropriate. Multiple regression models were constructed using the variables found to be significant in the univariate analysis to identify significant independent prognostic factors. The power of the study to detect differences between the risk categories was calculated, based on the category sizes and variability observed in the study.


   Results
Top
 Abstract
 Introduction
 Methods
 Results
Discussion
 Conflicts of interest
 References
 
The mean number of sickness absence episodes per year was 1.63 (range 0–9.6, 95% CI 1.49–1.77). The mean number of sickness absence hours was 28.3 h per 1000 h worked (range 0–436, 95% CI 24.5–32.3).

Table 2 shows the mean, minimum and maximum sickness absence episodes and hours for each risk category. The mean number of sickness absence episodes per year increased with risk category. The results for the ANOVA test (P < 0.001) indicate that the mean values in each risk category are significantly different from each other. Post hoc comparisons between pairs of risk categories show that means of only the low–medium pair are significantly different from each other (P < 0.01). The most common risk factors in the high-risk group were a previous history of more than three spells of sickness absence per year (20 cases), a BMI >35 kg/m2, three or more episodes of previous low-back pain, symptomatic ischaemic heart disease and three or more episodes of depression.


View this table:
[in this window]
[in a new window]

  		Table 2. Mean, minimum and maximum sickness absence episodes and hours for each risk category group

 
The mean value for the sickness absence hours per 1000 h worked increases with risk category. The results for the ANOVA test (P < 0.01) indicate that the means in the risk group are significantly different from each other. Post hoc comparisons between pairs of risk categories shows that means of the low–medium and low–high pairs are significantly different from each other (P < 0.05). The power of the ANOVA calculation for risk category was calculated based on the category sizes, means and standard deviations of the values found in the study. The power of the calculation for sickness absence episodes per year was 92% and of that for sickness absence hours per 1000 h worked was 89%.

Statistical tests were carried out on the other variables in the database to see how closely they were related to the sickness absence outcomes. Age was the only continuous variable. There were four categories with two variables, namely smoking, gender, ischaemic heart disease and anterior knee pain. There were six categorical variables with more than two categories, namely BMI, low-back pain, depression, employment status at pre-employment, past sickness absence spells and past sickness absence duration.

The mean age in the study group was 42.8 years. Pearson correlation coefficients were calculated to measure the linear association between age and the sickness absence outcomes. In both cases age has a negative correlation coefficient, indicating that sickness absence decreases with age. The correlation between age and sickness absence hours is not significantly different from zero, but for age and the number of sickness absence episodes, there is a significant negative correlation (r = –0.16, P < 0.01).

Table 3 shows those variables for which there were significant differences in either sickness absence episodes or hours. Smokers had significantly more sickness absence than non-smokers and females had significantly more sickness absence than males. Employees who had taken >10 days sick leave in the previous year also had significantly higher sickness absence. Episodes of low-back pain predicted sickness absence hours only.


View this table:
[in this window]
[in a new window]

  		Table 3. Frequency and duration of sickness absence

 
BMI, depression, employment status and number of spells of previous sickness absence did not predict sickness absence episodes or sickness absence hours.

There was only one case with positive ischaemic heart symptoms and/or positive stress test identified and only two cases of anterior knee pain, giving insufficient numbers in these two categories for further statistical analysis.

There were six categorical variables with more than two categories and these were tested with ANOVA to determine if there was a significant difference between the categories. In some cases, the numbers in different categories were small and the categories were combined to give just two category variables.

Linear regression was used to assess how well risk category predicted sickness absence. This was done separately for both sickness absence episodes per year and hours per 1000 h worked.

If the risk category (i.e. classifying employees into low-, medium- or high-risk groups based on risk factors described) was used as the only predictive factor, only 3.3% of the variation in the sickness absence episodes was predicted. As this value was low, a multivariate regression model which included smoking status, previous days sickness absence, gender and age as well as risk category was also investigated (as these factors were found to be significant in the univariate analysis with sickness absence episodes per year). A stepwise procedure for this linear regression excluded factors that were found not to be significant. If the risk category factor was excluded, then using the model below, a prediction of 10.6% of the variation in sickness absence episodes could be made.

Formula

If the risk category was used as the only predictive factor, only 3% of the variation in the sickness absence hours was predicted. A multivariate regression model which included smoking status, previous days sickness absence, gender and low-back pain episodes as well as risk category was also investigated (as these factors were found to be significant in the univariate analysis with sickness absence hours). A stepwise procedure for this linear regression excluded factors that are found not to be significant. Risk category was excluded and the final model below predicted 11.9% of the variation in sickness absence hours.

Formula


   Discussion
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
Conflicts of interest
 References
 
This study demonstrated a statistically significant difference for subsequent sickness absence rates between three risk category groups determined by the presence of one or more risk factors identified at pre-placement health assessments, confirming Poole's hypothesis that it is possible to predict future sickness absence at a pre-placement assessment in this way.

Analysis of individual risk factors demonstrated a statistically significant decrease in sickness absence episodes with increasing age, as previously demonstrated by Roelen et al. [4]. This may be due to the small numbers of older workers in this study as the subjects were in their initial years of employment and hence a younger workforce. The study demonstrated that smokers have significantly more sickness absence (episodes and hours) compared with non-smokers, consistent with the findings of other studies [5,9–11]. Females had significantly higher rates of sickness absence (episodes and hours) compared with males. This is again consistent with data from other studies [5,29,30].

Differences in sickness absence rates linked to BMI were not statistically significant. This may be because absence due to obesity may not occur within the short follow-up period of 3–6 years in this study. Longer follow-up might be needed to confirm the findings of other authors [5,12,13]. A history of low-back pain symptoms was linked to a statistically significant increase in sickness absence hours, consistent with results of other studies [19–23]. A number of studies have found that a history of depression is a predictor of future sickness absence [3,6,14,16]. Despite it being a common illness, only five employees in the study admitted having more than two previous episodes. The small number may have accounted for the lack of statistical evidence to support an association between previous depression and subsequent sickness absence. Results were not statistically significant for episodes or hours for current sickness absence at pre-employment, possibly due to the small numbers of employees in the ‘unemployed’ group. The study supports the assertion that past sickness absence duration predicts future sickness absence as also demonstrated by Duijts et al. [3] and Nieuwenhuijsen et al. [6]. The low numbers of ischaemic heart disease cases may be due to the low percentage of males in the study (15%) and also the age of the study subjects. The low numbers of knee pain cases may also reflect the relatively low age of the study subjects.

If risk category alone was used to predict sickness absence, it only predicted 3.3% of the variation in the sickness absence episodes. The regression model which best predicted sickness absence episodes included smoking status, gender, age and previous days sickness absence. The predictive factor increased from 3.3 to 10.6%.

If risk category alone was used as the only predictive factor, only 3.0% of the variation in the sickness absence hours is predicted. The linear regression model which best predicted sickness absence hours combined smoking status, previous days sickness absence, gender and low-back pain. This model predicted 11.9% of the variation in sickness absence.

The study had a number of limitations that need to be considered.

Employees who left the trust during the study period were excluded. The study group was therefore a survivor population and the ‘leaver population’ may have included a percentage of employees who left the trust because of sickness or had more risk factors for sickness absence than the survivor group.

The pre-placement questionnaires were self-completed and, therefore, may have been inaccurate as subjects may have under-reported previous illnesses either intentionally or because of recall bias. Intentional misrepresentation may be more relevant for certain factors such as depression, which is a common condition and associated both with sickness absence and a perceived stigma.

Some of the factors used to categorise employees were easily defined, e.g. smoking status, while others involved a subjective assessment of the number of previous episodes of a condition. Height and weight was self-declared and results may be biased to decreasing BMI, particularly in women, as there may be a tendency to overestimate height and underestimate weight.

The socio-economic class of employees was not determined in this study and this could potentially have biased results.

The study confirms Poole's hypothesis that it is possible to predict future sickness absence at the pre-placement health assessment, demonstrating a statistically significant difference between three risk category groups in relation to sickness absence taken in subsequent years. Certain risk factors, namely age, female sex, previous days sickness absence, smoking history and a history of at least two previous episodes of low-back pain identified at pre-placement assessment, predict a greater than average subsequent sickness absence; however, the best model identified using these risk factors predicted only 10–12% of the variation in sickness absence. The study therefore also highlights the current limitations of performing pre-placement health assessments for the purpose of identifying employees who may take greater than average subsequent sickness absence.


Key points
  • Results confirmed Poole's hypothesis that future sickness absence can be predicted at the pre-placement health assessment.
  • Female sex, age, smoking, history of at least two previous episodes of low-back pain and previous days sickness absence, identified at pre-placement assessment, predict a greater than average subsequent sickness absence.
  • The best model using identified risk factors only predicted 10–12% of the variation in sickness absence.

 


   Conflicts of interest
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Conflicts of interest
References
 
None declared.


   References
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Conflicts of interest
 References
 

  1. Poole CJM. Can sickness absence be predicted at the pre-placement health assessment? Occup Med (Lond) (1999) 49:337–339.[CrossRef][Medline]

  2. Barham C, Begum N. Sickness absence from work in the UK. Labour Market Trends. April 2005.

  3. Duijts SFA, Kant IJ, Landeweerd JA, Swaen GMH. Prediction of sickness absence: development of a screening instrument. Occup Environ Med (2006) 63:564–569.[Abstract/Free Full Text]

  4. Roelen CAM, van der Pol TR, Koopmans PC, Groothoff-Johan WAU. Identifying workers at risk of sickness absence by questionnaire. Occup Med (Lond) (2006) 56:442–446.[CrossRef][Medline]

  5. Labriola M, Lund T, Burr H. Prospective study of physical and psychosocial risk factors for sickness absence. Occup Med (Lond) (2006) 56:469–474.[CrossRef][Medline]

  6. Nieuwenhuijsen K, Verbeek JHAM, de Boer AGEM, Blonk-Roland WB, van Dijk FJH. Predicting the duration of sickness absence for patients with common mental disorders in occupational health care. Scand J Work Environ Health (2006) 32:67–74.[Web of Science][Medline]

  7. Harrison DA, Martocchio JJ. Time for absenteeism: a 20 year review of origins, offshoots and outcomes. J Manage (1998) 24:305–350.[CrossRef]

  8. Sydsjo A, Alexanderson K, Dastserri MA, Sydsjo G. Gender differences in sick leave related to back pain diagnoses. Spine (2003) 28:385–389.[CrossRef][Web of Science][Medline]

  9. Bush R, Wooden M. Smoking and absence from work: Australian evidence. Soc Sci Med (1995) 41:437–446.[CrossRef][Web of Science][Medline]

  10. Ryan J, Zwerling C, Jones M. Cigarette smoking at hire as a predictor of employment outcome. Occup Environ Med (1996) 38:928–933.[CrossRef]

  11. Torres Lana A, Cabrera de Leon A, Marco Garcia MT, Aguirre Jaime A. Smoking and sickness absence among public health workers. Public Health (2005) 119:144–149.[CrossRef][Medline]

  12. Levy E, Levy P, Le Pen C, Basdevant A. The economic cost of obesity: the French situation. Int J Obes (1995) 19:788–792.[Web of Science][Medline]

  13. Moreau M, Valente F, Mak R, et al. Obesity, body fat distribution and incidence of sick leave in the Belgian workforce: the Belstress study. Int J Obes (2004) 28:574–582.[CrossRef][Web of Science][Medline]

  14. Hensing G, Brage S, Nygard F, Sandanger I, Tellnes G. Sickness absence with psychiatric disorders—an increased risk for marginalisation among men? Soc Psychiatry Psychiatr Epidemiol (2000) 35:335–340.[CrossRef][Web of Science][Medline]

  15. Kessler RC, Frank RG. The impact of psychiatric disorders on work loss days. Psychol Med (1997) 27:861–873.[CrossRef][Web of Science][Medline]

  16. Laitinen-Krispijn S, Bijl RV. Mental disorders and employee sickness absence: the Nemesis study. Soc Psychiatry Psychiatr Epidemiol (2000) 35:71–77.[CrossRef][Web of Science][Medline]

  17. Stansfield S, Feeney A, Head J, et al. Sickness absence for psychiatric illness: the Whitehall study. Soc Sci Med (1995) 40:189–197.[CrossRef][Web of Science][Medline]

  18. Thase ME. Long term treatment of recurrent depressive disorders. J Clin Psychol (1992) 53:33–44.

  19. van den Heuvel SG, Ariens GAM, Boshuizen HC, Hoogendoorn WE, Bongers PM. Prognostic factors related to recurrent low back pain and sickness absence. Scand J Work Environ Health (2004) 30:459–467.[Web of Science][Medline]

  20. Van der Weide WE, Verbeek JH, Salle HJA, Van Dijk FJH. Prognostic factors for chronic disability from acute low back pain in occupational health care. Scand J Work Environ Health (1999) 25:50–56.[Web of Science][Medline]

  21. Leo A, Elders MD, Heinrich J, Burdorf A. Risk factors for sickness absence because of low back pain among scaffolders. (2003) 28:1340–1346. Spine.

  22. Infante-Rivard C, Lortie M. Relapse and short sickness absence for back pain in the six months after return to work. Occup Environ Med (1997) 54:328–334.[Abstract/Free Full Text]

  23. Bartys S, Burton K, Main C. A prospective study of psychosocial risk factors and absence due to musculoskeletal disorders-implications for occupational screening. Occup Med (Lond) (2005) 55:375–379.[CrossRef][Medline]

  24. Walter PJ. Return to work after coronary artery bypass surgery. Eur Heart J (1988) 9:58–66.[Abstract/Free Full Text]

  25. Danchin N. Work capacity after myocardial revascularization: factors related to work resumption. Eur Heart J (1988) 9:44–47.[Abstract/Free Full Text]

  26. Lacaille A, Hogg RS. The effect of arthritis on working life expectancy. J Rheumatol (2001) 28:2315–2319.[Abstract/Free Full Text]

  27. Cutbill JW, Ladly KO, Bray RC, Thorne P, Verhoef M. Anterior knee pain: a review. Clin J Sports Med (1997) 7:40–45.[Web of Science][Medline]

  28. Palmer KT, Milne P, Poole J, Cooper C, Coggon D. Employment characteristics and job loss in patients awaiting surgery on the hip or knee. Occup Environ Med (2005) 62:54–57.[Abstract/Free Full Text]

  29. Gjesdal S, Bratberg E. The role of gender in long term sickness absence and transition to permanent disability benefits. Eur J Public Health (2002) 12:180–186.[Abstract/Free Full Text]

  30. Sydsjo A, Alexanderson K, Dastserri MA, Sydsjo G. Gender differences in sick leave related to back pain diagnoses. Spine (2003) 28:385–389.[CrossRef][Web of Science][Medline]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
58/5/355    most recent
kqn029v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Lucey, S. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lucey, S. P.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?