Exposure Therapy

The primary endpoint was evaluated between each risankizumab dose and placebo using the Cochran–Mantel–Haenszel test, adjusting for stratification factors (prior exposure to anti-TNF therapies and baseline worst Hurley stage). Missing efficacy values during the double-blind period were handled using nonresponder imputation incorporating multiple imputation to handle missing data due to COVID-19 as the primary approach for categorical endpoints and mixed-effect model repeat measures for continuous endpoints. All safety analyses were performed on the safety populations, defined as all patients who received at least one dose of study drug in the respective study period. The number and percentage of patients experiencing TEAEs were tabulated using the Medical Dictionary For Regulatory Activities (version 24.0) system organ class and preferred terms; severity and relationship to the study drug were assessed by the investigator.

The patient population for this study included adults (aged ≥ 18 years) with moderate-to-severe HS. Eligible patients were required to be diagnosed at least 1 year before the baseline visit and have a total abscess and inflammatory nodule (AN) count ≥ 5 at baseline, HS lesions in ≥ 2 distinct anatomical locations, a draining fistula count ≤ 20 at baseline, and inadequate response to oral antibiotics for the treatment of HS. Patients were ineligible to participate if they had exposure to biologic agents blocking IL-12/23, IL-23, or IL-17 within the past 6 months; prior exposure to anti-TNF therapies (except those for the treatment of HS that demonstrated inadequate response); relevant medical conditions (such as hepatitis B, hepatitis C, HIV, or tuberculosis); or if they were pregnant or breastfeeding.

We characterized the study population using median (interquartile range (IQR))
and frequency (percent) as appropriate. We compared the MS versus monoADS groups
using χ², Wilcoxon, or Kruskal–Wallis tests. We performed bivariate
analyses with respect to the HRQoL of the parent, accounting for clustering at
the individual level using random effects specifications (Appendix e1).^{19 (link)}We constructed three multivariable regression models using random effects to
evaluate whether familial health conditions or SEP modified the relationship
between the child’s MS diagnosis (vs. monoADS) and parental HRQoL. Using random
effects specifications, we calculated the variance in parental HRQoL that is due
to within-participant variation (time-variant factors that fluctuate between
HRQoL assessments) separate from the variance in parental HRQoL due to
between-participant variation (time-invariant factors that do not change within
participants). For each multivariable model, the outcome of interest was
parental HRQoL. Model 1 assessed the effect of an MS diagnosis and ⩾1 family
health condition(s). Model 2 assessed the effect of an MS diagnosis and ⩾1
comorbidity among the child with MS or monoADS. Model 3 assessed the effect of
an MS diagnosis and low SEP. For each model, we created a categorical variable
with four levels combining the two dichotomous predictors of interest to
ascertain the joint effect separate from the individual effects. For example,
the categories for Model 1 were: MS and a family health condition, MS without a
family health condition, monoADS with a family health condition, and monoADS
without a family health condition (reference group).^{20 (link),21 (link)} We adjusted for the
following time-invariant covariates: number of children in the family (count),
sex of the child (male as reference group, binary), age of the child at symptom
onset (years, continuous), length of stay in hospital for first attack (days,
continuous), child’s birth country (born outside Canada as reference group,
binary). Number of children in the family was modeled as time-invariant because
few changes were reported. We also adjusted for time-variant covariates: the
child’s age at the time of HRQoL assessment (years; continuous), the child’s
HRQoL (continuous), and the presence of functional neurological impairments
(normal or mild impairments as reference group; binary). We adjusted for country
of birth to account for the large proportion of internationally educated
immigrants in Canada who report working in occupations for which they are
over-qualified.^{22 (link)} We also adjusted for low SEP (high SEP as the reference
group, binary; Models 1 and 2), health conditions among the parents or siblings
(no health conditions as the reference group, binary; Model 2), and family
health conditions (no health conditions as the reference group, binary; Model 3)
as time-invariant covariates. Tests for reverse causality between the parent and
child’s HRQoL were negative.²³ The joint effect of the
predictors was categorized as synergistic when it was greater than the sum of
the individual effects and statistically significant. We report R2 values to
assess the proportion of variability in parental HRQoL explained by the
independent variables.
Since our multivariable modeling methods excluded participants for whom
covariates are missing or not applicable, we did not include variables unique to
children with MS (e.g. relapses and exposure to disease-modifying therapies
(DMTs)); these variables are reported to support generalizability of our
findings.