Article Text
PDF
Statistics from Altmetric.com
Kivimaki et al1 undertook research to identify some determinants of sickness absence in Finnish hospital physicians between 1997 and 1998. This was a questionnaire survey sent to 816 physicians and a control group of 542 senior nurses employed in one of 11 hospitals in Finland. Social circumstances, work characteristics, and various measures of health were assessed by questionnaire and employers' registers were used to assess recorded sickness absence.
There are some limitations in the study design and subsequent conclusions that are not acknowledged in the text.
The response rate from the physicians was a disappointing 55%. Nearly half of the physicians approached did not participate in the study. This could bias the results considerably. The authors state that the response rate obtained in this study corresponds to that obtained in previous research. Seven references are cited as evidence. The response rates of doctors from six of the seven quoted papers are higher than 55% and are as follows: 82%2; 79% and 76%3; 63%4; 78%5; 58%6; and 87%–94%.7 The authors could have cited a further reference8 that was used later in the paper, but chose not to (response rate 80%).
The authors state that the recording of attendance is reliable in the Finnish public sector. This would be a very important factor in the context of this study of sickness absence in hospital physicians. The authors cite two references9,10 to back up their statement. These two studies both concern local government workers in Raisio, Finland between 1990 and 1995 with the main outcome measure being recorded medically certificated absence held on computer by an occupational health unit. Both of these papers simply state that all sick certificates must be forwarded for recording, but provide no other evidence that the recording of sickness absence is reliable. The possibility that doctors might underreport sickness absence, leading to incomplete recording, is not considered. This could add further bias to the study, thereby affecting the results and conclusions.
References
Author's reply
In his electronic letter, Murphy wrote that the 55% response rate from the doctors could have considerably biased the results reported in a study of sickness absence in Finnish hospital physicians.1 According to the results, absence rate for doctors is low. The study also suggests that poor teamwork and traditional psychosocial risk factors—such as work overload and low job control, contribute to long term sick leave among doctors.
As already reported in the article (362), hospital registers show that low response rate did not affect the findings on absence rates. In those who responded to the survey, the rate of short term (1–3 days) sickness absences was 38.0 spells per 100 person-years. The corresponding rate for the eligible population was 37.9 spells per 100 person-years. Regarding the rate of long term (<3 days) sickness absences, the rate was exactly the same for the respondents and the eligible population—20.2 long sickness absence spells per 100 person-years.
Bias due to low response rate seems also unlikely relative to psychosocial factors. In work units with a response rate lower than 55%, the mean scores of teamwork and job control were 3.5 and 4.0, respectively. In work units with a response rate of 55% or higher, the corresponding mean scores were also 3.5 and 4.0, respectively. In the overload scale, the mean score of 3.8 in the units with a low response rate did not differ from the mean score of 3.6 in the units with a high response rate (p=0.130).
Murphy acknowledges that incomplete recording of sickness absences could also affect the results and conclusions. However, at least the following points suggest that the recording of long term sickness absences of doctors was reliable.
Firstly, Finnish hospitals receive compensation for loss of salary due to sick leaves longer than 10 days from the Social Insurance Institution, a body subordinate to Parliament. To receive all the compensation to which the hospitals are entitled, they are motivated to keep strict records of sick leaves. This was especially true during the study period. Hospitals had a shortage of financial resources compared with the amount of services they were expected to provide.
Secondly, indicators of health problems—such as poor self rated health, presence of diagnosed chronic disease, and psychiatric morbidity (as indicated by caseness in the general health questionnaire)—were associated with long spells of sickness absence (see table 3 in the article).1 For doctors, these associations were as strong as or stronger than those found for the controls. Only the opposite case—if the associations had been weaker for doctors—could have been an indication of unreliable measurement of sickness absence of doctors.
Thirdly, the low recorded rate of absence for hospital physicians is well in line with other research. Results on self reported absences in a random sample of Finnish physicians suggest that physicians are on sick leave much less often than other employees.2 According to a British study, over 80% of physicians had worked even when they felt too unwell to carry out their duties to the best of their ability.3 A Norwegian study reports that 80% of the physicians had worked during an illness for which they would have sick listed their patients.4
In conclusion, evidence and additional analyses provide no support for the possibility that the issues listed by Murphy would have added bias to the study on sickness absence in Finnish hospital physicians.