I have often wondered whether long working hours are either necessary or efficient, and how far their purpose is something other than production, for example to signal to others how important or indispensable the person working them is. There is a certain pride involved: not holier, but more hard-working, than thou. Whenever I have worked long hours I have found it more difficult to think and concentrate at the end of the working day than at the beginning, but perhaps a sample of one is not strong evidence.
Long working hours have also been suspected of being bad for the health, especially for the heart. A recent paper in the Lancet examined the question of whether such working hours were statistically associated with heart attack or stroke. The authors (44 of them!) performed a meta-analysis – a summation of all previous research on the subject – and came to the conclusion that long working hours, defined as those more than 55 hours per week, increased the risk of stroke and, to a lesser extent, of heart attack.
Theirs was a labor of Hercules. First they had to trawl through an enormous number of papers in the medical literature to find data of sufficient quality for inclusion in the study. They also examined unpublished studies to exclude publication bias, which is the tendency to publish research with positive findings but not to publish that with negative findings. In all they examined 11,700 studies with regard to the possible relationship between stroke and hours worked, and 13,477 studies with regard to the possible relationship between heart attack and hours worked. They weeded out all but 25 studies as containing insufficient data for their purposes, which were to control for such obvious confounding factors as age, sex, smoking, blood pressure, diabetes, social class and so forth (people of higher social class tend to work longer hours, though length of hours worked is unlikely to explain their social ascendency). If there were a Nobel Prize for diligence, these authors – mostly Scandinavian – would certainly deserve to win it.
Over half a million people without evidence of cerebrovascular disease were followed up, on average, for just over seven years, and in that time 1,722 of them had a stroke. More than 600,000 people without evidence of ischaemic heart disease were followed up, on average, for eight and a half years, and in that time 4,768 of them had heart attacks.
After an enormous amount of statistical manipulation (by that word I do not mean to impugn the authors’ integrity), they discovered that the risk of having a stroke if people worked more than 55 hours was 1.3 times greater than if they worked “normal” hours (35 – 40 hours) and their risk of having a heart attack was 1.1 times higher. Furthermore they were able to show for strokes that there was a dose-response relationship between numbers of hours worked and risk of stroke (that is to say the risk of stroke increased in linear fashion with increasing numbers of hours worked), which is important because such a relationship is usually one, but only one, of the criteria of a causative relationship.
The authors admitted to certain limitations in their paper, for example that the assessment of the hours worked depended entirely on a single self-report. Strictly speaking, then, stroke was associated with a single self-report of working more than 55 hours a week. But the authors do not mention that a relative risk of 1.3 is itself not very great, and in fact means practically nothing. Even if the association were a causative one, which is by no means certain, the chances of a young man having a stroke are so small that a slight increase in the risk is of trifling importance.
Diligence in scientific research is necessary but not sufficient.