Table 3

Meta-analysis for the associations between being a key worker and outcomes

Outcome	n	Study heterogeneity			OR (95% CI)	P value	Egger’s test p value
Outcome	n	I²	Q	P value	OR (95% CI)	P value	Egger’s test p value
COVID-19 infection*	11 076	54.72%	6.63	0.085	1.43 (1.22 to 1.68)	<0.0001	0.801
Change in financial situation†	10 649	0.00%	1.73	0.631	0.32 (0.24 to 0.65)	<0.0001	0.534
Change in trust in government‡	10 236	29.57%	4.26	0.235	1.00 (0.89 to 1.11)	0.932	0.126
Change in conflict§	10 234	29.86%	4.28	0.233	1.19 (1.03 to 1.37)	0.016	0.137
Change in household composition¶	10 736	44.96%	5.45	0.142	0.98 (0.85 to 1.12)	0.743	0.073
Change in psychological distress**	9877	34.71%	4.60	0.204	0.95 (0.85 to 1.05)	0.320	0.293
Change in alcohol consumption††	10 269	62.46%	7.99	0.046	0.88 (0.79 to 0.98)	0.022	0.052
Change in smoking‡‡	11 076	24.09%	3.95	0.257	0.60 (0.44 to 0.80)	0.005	0.144
Change in sleep duration§§	10 371	61.75%	7.84	0.049	1.06 (0.94 to 1.19)	0.350	0.245

All analyses used a random-effects model meta-analysis. Significant p values which persisted at a false discovery rate of 0.05 are highlighted in bold.
*COVID-19 infection was coded as 0=no, 1=yes.
†Change in financial situation was coded as 0=same or better, 1=worse.
‡Change in trust in government was coded as 0=same or more, 1=less.
§Change in conflict during lockdown was coded as 0=less, 1=same or more.
¶Change in household composition during lockdown was coded as 0=no, 1=yes.
**Change in psychological distress during lockdown was coded as 1=an increase in the mental health score, 0=same or lower score.
††Change in alcohol consumption was coded as 1=more drinks per week during lockdown compared with pre-lockdown or 0=same or lower.
‡‡Change in smoking was coded as 0=same or less cigarettes smoked during lockdown compared with before lockdown, 1=more cigarettes.
§§Change in sleep duration was coded as 0=same or more hours slept during lockdown and 1=less hours slept during lockdown.