Understanding inequalities in mental health by family structure during COVID-19 lockdowns: evidence from the UK Household Longitudinal Study

Table 1 When measures were taken

Measures^a	April 2020 analysis N = 10,516	Jan 2021 analysis N = 6,893
Pre-exposure confounders
Sex	2018/2019^b	2018/2019^b
Age (in Feb 2020)	2018/2019^b	2018/2019^b
Ethnicity	2018/2019^b	2018/2019^b
UK country	2018/2019^b	2018/2019^b
Education	2018/2019^b	2018/2019^b
Exposure
Pre-lockdown family structure	2018/2019	Sep 2020
Post-exposure (pre-lockdown) confounders
Longstanding illness	2018/2019^b	2018/2019^b
Occupational class (NS-SEC)	2018/2019^b	2018/2019^b
Poverty	2018/2019^b	2018/2019^b
Pre-lockdown Loneliness	2018/2019	Sep 2020
Smoking	2018/2019	Sep 2020
Moderate–high risk drinking (AUDIT-C)	2017/2018	Sep 2020
Pre-lockdown mental health (GHQ-12)	2018/2019	Sep 2020
Lockdown confounders
Change from pre-lockdown family structure	April 2020 & 2018/2019	Sep 2020 & Jan 2021
Keyworker status	April 2020	Jan 2021
Shielding status	April 2020	April 2020–Jan 2021^c
Mediators
Active employment	April 2020	Nov 2020 & Jan 2021^d
Financial strain	April 2020	Nov 2020^e
Childcare/home-schooling	April 2020	Jan 2021
Caring	April 2020 & 2018/2019^f	Nov 2020 & Jan 2021^g
Loneliness	April 2020	Jan 2021
Outcome
Mental health (GHQ-12)	April 2020	Jan 2021

^aAssumed causal ordering runs from top to bottom. Although the assumed causal ordering does not exactly match the timing of measurement, especially for the January 2021 analysis, to maintain comparability between the two analyses, we assume that the measures taken are adequate proxies for characteristics that can affect each other in the assumed causal direction, regardless of when measured
^bIf information was missing, the most recent data from previous annual surveys were used for these characteristics
^cShielding status was based on ever having been advised to shield since the start of the pandemic
^dEmployment status from November was used first, with status from January used where information from November was missing. There was no question on furlough in November, so furlough information was taken from January
^eMeasure not available in January 2021
^fInformation on caring for others outside the household was taken in April 2020 and combined with information on caring for others within the household from the 2018/2019 survey
^gInformation on caring for others outside the household was taken from the November 2020 survey and combined with information on caring for others within the household from the January 2021 survey

ISSN: 1744-859X