Sick leave and death

“Sick leave 'link to early death'” is the headline on the BBC news website, suggesting that people who have “long spells of sick leave for psychiatric reasons” are twice as likely to die from cancer as healthier employees. A study in over 6,000 civil servants also found that “those who had taken a long period of sick leave had a 66% higher risk of early death”, the website adds.

There are some limitations to this analysis of data from a large study. Although it suggests that there was a 2.5-fold increase in likelihood of cancer death with absence for ‘psychiatric’ reasons, the actual number of people dying in this category was very small. Also, the definition of ‘psychiatric’ absences is not clear.

It is important for employers to keep accurate records of absences to be aware of their employees’ health and allow early identification of cases where extra support may be needed – either from healthcare professionals or in the work environment. Accurate records could also be an important source of data for further studies such as this.

Where did the story come from?

Dr Jenny Head and colleagues from University College London, Karolinska Institutet in Sweden, and the Finnish Institute of Occupational Health in Finland carried out this study. The research (the Whitehall II study) was funded by grants from the Medical Research Council, the British Heart Foundation, the Health and Safety Executive, the Department of Health, the National Institutes of Health, the Agency for Healthcare Policy Research and the John D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socioeconomic Status and Health. It was published in the peer-reviewed medical journal, the_ British Medical Journal._

What kind of scientific study was this?

This publication is based on an analysis of data from the Whitehall II study – a prospective cohort study of civil servants in the UK that started in 1985. The Whitehall study enrolled all London-based office staff aged 35 to 55 years from 20 civil service departments. Seventy-three per cent of those approached agreed to participate, leaving a final cohort of 10,308 (6,895 men and 3,413 women). Between 1985 and 1988, participants were screened for entry into the study, and computerised sickness absence records were reviewed for the participants since 1985. Absence records were available for 9,179 civil servants.

In this publication, the researchers were interested in whether medically certified sickness over a three-year period was linked to mortality, and whether the diagnosis behind people’s sickness absences affected this risk. Diagnostic codes for absence had been recorded by the civil service and the researchers in this study converted these codes into disease categories. These categories were based on the morbidity coding system of the Royal College of General Practitioners, but four extra categories were added (gastrointestinal, headache and migraine, neurosis and neurosis ill-defined). The three-year exposure period was defined as the three years after baseline screening (for those departments that had computer records of absence at the beginning of the study) and three years from January 1 1991 for departments who didn’t collect computerised records until 1991.

Data about deaths was available through the National Health Service Central Register mortality register. All-cause mortality and deaths from cardiovascular disease and cancer were recorded from the start of the participants’ three-year exposure period until September 30 2004. The researchers took into account other factors that may have an effect on mortality, including smoking, alcohol consumption, high blood pressure, self-rated health, presence of longstanding illness (diabetes, heart disease, respiratory illness, cancer etc), disability or infirmity.

The researchers used statistical methods to assess whether there was a link between the number of sickness absences in the three-year period and death from all or any cause. They also investigated whether the prediction of mortality was greater when they looked at the specific reason for work absence. Both of these analyses took into account (were adjusted for) other factors that may be linked with mortality, such as smoking, drinking and general health.

What were the results of the study?

Not all participants in the Whitehall II study were included in the analyses due to missing data or incomplete three-year exposure (i.e. the participant died or left the civil service). 3,830 participants were excluded and an investigation of their characteristics revealed that, as a group, those who were excluded had a lower mortality.

Of the 6,478 participants who were included in this study, 288 had died during follow-up. People with more than one medically certified absence (absence of greater than seven days duration) during the three-year exposure period were 1.7 times more likely to die than those with no such absences. They also found that the more absences a person had, the greater their risk of death.

When they looked at specific diagnoses, the highest risk for death was due to absences for ‘circulatory problems’ (4.7 times increased risk of death), followed by surgical operations (2.16 times increased risk), neurosis (ill-defined; 2.03 times increased risk), injury (1.66 times increased risk) and diseases of the respiratory system (1.63 times increased risk of death). There were a small number of people absent for ‘cancer’ and of those (10 people), their risk of death was 21.3 times more likely than those without absences, though the estimate was not very accurate given the small sample.

Researchers also looked at link between absence (overall and for particular reasons) and cause of death. They found that being absent overall was associated with cardiovascular and cancer deaths. Absence for infectious or parasitic disease, circulatory problems, respiratory illness and surgical operations were all significantly linked to cardiovascular death. When looking at reasons for absence and cancer-related death, only psychiatric absences and surgical operations were significantly linked. Further breakdown of ‘psychiatric’ absence into ‘neurosis’ and ‘neurosis ill-defined’ found that there was only a link with ‘neurosis ill-defined’.

What interpretations did the researchers draw from these results?

The researchers conclude that “knowing the diagnosis for medically certified sickness absence from work significantly improves the prediction of mortality”. They say that “unexpectedly, employees who had one or more absence for psychiatric reasons had a considerable 2.5-fold greater cancer mortality”.

What does the NHS Knowledge Service make of this study?

This study examines the link between absences from work (both generally and for specific reasons) and mortality overall and from cancer or cardiovascular causes. The finding that absence for psychiatric reasons is linked to an increased likelihood of death due to cancer is described as “unexpected”. The researchers did not explore why this might be the case.

There are several points to highlight in relation to this study, which should be kept in mind when interpreting the results:

• When broken down into the different categories, the absolute numbers of people who died were small. Only 12 people had psychiatric reasons for absence and died from cancer-related causes. This is a small number, and the results may have occurred by chance. The researchers acknowledge that sub-sample size is a problem, and that their results “need replication”. The small size means that the study was also underpowered to explore gender differences or differences across employment grade (a proxy indicator of socioeconomic status). 
• Also, when the researchers further explored ‘psychiatric’ reasons, i.e. breaking it down into its constituent ‘neurosis’ and ‘neurosis ill-defined’, they found that only ‘ill-defined’ neurosis was linked to cancer death. This definition included tiredness and stress, which may have nothing to do with mental illness (i.e. may be indicators of physical illness).
• The researchers also say that “the recorded diagnosis for a sickness absence may not cover all of the actual causes”. This is an important limitation, and is reflected by the finding that only 64% of recorded reasons for absence corresponded with GP diagnoses around that time. ‘Coexisting’ illness would not have been captured in the absentee records. 
• To determine whether the civil service codes for diagnosis were accurate, researchers obtained information from GPs for all absences greater than 21 days between 1985 and 1990, and assessed whether there was agreement between GP records and civil service codes. They found a 64% agreement. 
• In their analyses, the researchers took into account smoking and alcohol consumption, so they can be sure that the link between psychiatric absences and cancer mortality was not due to differences in drinking or smoking behaviour. However, there are other confounding factors that may be responsible for this, and these were not measured. The researchers suggest that one reason might be that depression interferes with help-seeking behaviour, delaying early detection and treatment of cancer. 
• Relationships between sickness and absence may be similar in other populations, but when generalising findings to other work groups it should be noted that these were all civil servants working in London. Typical personal and social lifestyle and working pressures may differ across occupations.

It is important for employers to keep accurate records of absences to be aware of their employees’ health and allow early identification of cases where extra support may be needed – either from healthcare professionals or in the work environment. Accurate records are also an important source of data for further such studies.