Employees who work 55 or more hours per week had a higher risk of stroke compared with those who work standard hours, according to a cumulative random-effects meta-analysis of published and unpublished data.
Lead researcher Mika Kivimäki, PhD, of University College London, and colleagues noted that the increased risk of stroke did not vary between men and women or by geographical region. They added that long working hours were associated with incident coronary heart disease, but the association was weaker than the one found with stroke.
The researchers, who published their results online in The Lancet on Aug. 20, searched the Embase and PubMed databases and identified prospective cohort studies that assessed long working hours and cardiovascular disease. They also examined unpublished individual-participant data from European prospective cohort studies.
“Combining estimates from published studies and unpublished data allowed us to examine the status of long working hours as a risk factor for coronary heart disease and stroke with greater precision and a more comprehensive evidence base than has previously been possible,” the researchers wrote.
In all, the meta-analysis included 25 studies from the U.S., Australia, Finland, Denmark, Sweden, the Netherlands, Belgium, Germany, the U.K., Northern Ireland and Israel.
The researchers had data from 603,838 men and women for the coronary heart disease analysis and 528,908 men and women for the stroke analysis. All of the participants were free from coronary heart disease and stroke at baseline.
They followed the coronary heart disease participants for a mean period of 8.5 years and the stroke participants for a mean period of 7.2 years.
After adjusting for age, sex and socioeconomic status, people who worked at least 55 hours per week had an increase in coronary heart disease and incident stroke. However, the researchers noted the evidence for coronary heart disease was not as robust as that for stroke.
They also mentioned a few study limitations, including that a large proportion of the unpublished data was from a data source that relied on a convenience sample that may contribute to availability bias. In addition, participants self-reported their work hours once, so their hours may have changed over time.