Statistical analyses
Outcomes were described for the phase preceding the COVID-19 pandemic
(01/01/20-12/03/20) and several COVID-19 phases based on preventative
measures in the Netherlands (see Fig. 1). These were: the first lockdown
(13/03/20-31/05/20), release of most restrictions (01/06/20-13/10/20),
the second ‘partial’ lockdown (14/10/20-14/12/20), the second ‘strict’
lockdown (15/12/20-19/01/21), and the lockdown with a curfew
(20/01/21-12/03/21). Psychosocial outcomes during the COVID-19 phases
were compared to pre-COVID-19 using linear mixed effects regression
analyses, to account for repeated measurements, with domains of HRQoL,
fatigue and caregiver distress as dependent variables and the COVID-19
phases as independent dummy variables. Clinical distress during the
COVID-19 phases was compared to pre-COVID-19 using logistic generalized
estimating equation modelling with an exchangeable covariance structure.
Models were corrected for demographic and medical variables (age, sex,
time since diagnosis, treatment status [on or off active treatment]
and CNS tumor diagnosis). A p-value of 0.05 divided by the number of
analyses conducted on (sub)scales of a questionnaire (e.g. for PedsQL
generic: 0.05/6=0.008) was considered statistically significant.