Table 3:

Cox model estimates for outcomes of antiviral therapy initiation, HHV-6 testing and HHV-6 detection incorporating propensity score weighting for baseline variables

	Adjusted hazard ratio (95% confidence intervals)
Antiviral initiation	therapy
Cohort
Pre-LTV cohort	Ref
Post-LTV cohorta
< 4 weeks post-HCT	0.57 (0.38–0.87)
≥ 4 weeks post-HCT	0.14 (0.10–0.20)
Acute GVHD grade
0–1	Ref
≥ 2	2.17 (1.71–2.75)
Tested for HHV-6
Cohort
Pre-LTV cohort	Ref
Post-LTV cohorta
< 6 weeks post-HCT	1.1 (0.68–1.78)
≥ 6 weeks post-HCT	0.36 (0.18–0.72)
Acute GVHD grade
0–1	Ref
≥ 2	1.63 (1.03–2.58)
HHV-6 detected
Cohort
Pre-LTV cohort	Ref
Post-LTV cohort	1.08 (0.44–2.62)
Acute GVHD grade
0–1	Ref
≥ 2	3.53 (1.31–9.49)

GVHD, graft-versus-host disease; HHV-6: human herpesvirus 6; LTV: letermovir

Adjusted Cox models used stabilized inversed probability of treatment weights (SIPTW) to balance baseline variables associated with the pre-LTV and post-LTV cohorts. Variables that were incorporated into IPTW weighting are summarized in Table S4. Acute GVHD was separately included in the model as a time-dependent variable.

^aDue to violations of the proportional hazards assumption for cohort, we added a time-interval-cohort interaction term to these models to compute separate cohort adjusted hazard ratios for time intervals in which the proportional hazards assumption seemed plausible; these time intervals were <4 and ≥4 weeks for antiviral therapy and <6 vs ≥6 weeks for HHV-6 testing.