TABLE 4.

Sensitivity analyses, relationship of cannabis use status to SARS‐CoV‐2 infection

Cannabis use status	Sensitivity analysis 1: using initial cannabis use status rather than time‐varying status		Sensitivity analysis 2: excluding time when people were current or former smokers		Sensitivity analysis 3: case‐positive, control‐test‐negative approach a
	Adjusted hazard ratio (95% confidence intervals)				Adjusted odds ratio (95% confidence intervals)
	Model 1 b	Model 2 c	Model 1 b	Model 2 c	Model 1 b	Model 2 c
Current	0.71 (0.60, 0.83)*	0.71 (0.60, 0.83)*	0.59 (0.46, 0.76)*	0.59 (0.46, 0.77)*	0.73 (0.60, 0.90)*	0.76 (0.61, 0.93)*
Recently quit	0.92 (0.81, 1.04)	0.97 (0.85, 1.10)	0.92 (0.81, 1.05)	0.96 (0.84, 1.09)	0.77 (0.68, 0.87)*	0.83 (0.73, 0.94)*
Non‐user	1.00 (reference)	1.00 (reference)	1.00 (reference)	1.00 (reference)	1.00 (reference)	1.00 (reference)

^a Logistic regression model that includes all COVID tests during follow‐up, stratified by (conditioned on) calendar date of the test. This analysis only includes people who received at least one test. The cannabis status of people who tested positive on a given day is compared to the cannabis status of people who tested negative on the same day.

^b Model 1 is adjusted for age (21 categories: < 19, one for each year of age up to 39, 40+), race/ethnicity, neighborhood deprivation index, insurance payor and primary Kaiser Permanente Northern California (KPNC) facility.

^c Model 2 is adjusted for all covariables in model 1 plus body mass index, pre‐existing diabetes and hypertension, parity and tobacco smoking status.

^*Significant at P ≤ 0.05.