Table 2.

The association between physical activity and work impairment due to absence and presenteeism (% of work time lost)

	(1)	(2)	(3)
MET-minutes per week
600–750	−0.015**	−0.012**	−0.008**
	(−0.011 to −0.019)	(−0.008 to −0.016)	(−0.004 to −0.012))
750–900	−0.022**	−0.017**	−0.011**
	(−0.016 to −0.028)	(− 0.013 to −0.021)	(−0.007 to −0.015)
900–1500	−0.024**	−0.018**	−0.011**
	(−0.02 to −0.028)	(−0.14 to −0.22)	(−0.007 to −0.15)
1500–2100	−0.031**	−0.024**	−0.016**
	(−0.025 to −0.037)	(−0.02 to −0.28)	(−0.012 to −0.02)
>2100	−0.033**	−0.027**	−0.018**
	(−0.029 to −0.032)	(−0.023 to −0.031)	(−0.014 to −0.022)
Controls
Sociodemographic	Yes	Yes	Yes
Work related	No	Yes	Yes
Health and lifestyle	No	No	Yes

**p<0.01, *p<0.05. n=117 240. 95% CIs in parentheses. Regression model in column 1 controls for age, gender, education, marital status, ethnicity, financial concerns, being a carer for a child or ill family member, engagement in voluntary or civic activities, and country-fixed and time-fixed effects; column 2 additionally controls for work-related factors, including working irregular hours, total hours of work per week, job position, income and stress at work; column 3 additionally controls for lifestyle, physical and mental health variables, including excessive alcohol consumption, smoking, psychological distress, chronic illnesses, sleep length and body mass index. All model specifications include organisational dummy variables, as well as dummy variables for the week, month and year of the given survey response. Estimates need to be multiplied by 100 to calculate the percentage point change in work impairment due to absenteeism and presenteeism. Results based on Table A.8 in Hafner et al.¹⁵

MET, metabolic equivalents of task.