Table 3.

Long COVID symptoms at month 6 by vaccination status: observed and model-based estimates

	Descriptive statistics	Observed		Logistic regression
		Odds ratio (95% CI)	P value	Odds ratio (95% CI)	P value
Base Case: ≥ 3 symptoms
Boosted vs Unvaccinated	10 (14.9%) versus 54 (44.6%)	0.22 (0.10, 0.47)	< 0.001	0.36 (0.15, 0.87)	0.019
Primed vs. Unvaccinated	27 (37.5%) versus 54 (44.6%)	0.74 (0.41, 1.35)	0.332	0.60 (0.27, 1.34)	0.296
Boosted vs Primed	10 (14.9%) versus 27 (37.5%)	0.29 (0.13, 0.67)	0.003	0.59 (0.21, 1.65)	0.459
Sensitivity: ≥ 2 symptoms
Boosted vs Unvaccinated	14 (20.9%) versus 66 (54.5%)	0.22 (0.11, 0.44)	< 0.001	0.30 (0.13, 0.70)	0.003
Primed vs. Unvaccinated	35 (48.6%) versus 66 (54.5%)	0.79 (0.44, 1.41)	0.425	0.64 (0.30, 1.39)	0.370
Boosted vs Primed	14 (20.9%) versus 35 (48.6%)	0.28 (0.13, 0.59)	0.001	0.46 (0.18, 1.20)	0.140

Note for logistic regression model: the logistic regression model for number of post-COVID ≥ 3 versus < 3 used GEE estimation with unstructured correlation matrix. Covariates were variables for time, vaccination status and interaction of time by vaccination status, as well as covariates of participant pre-COVID-19 symptom onset score, sociodemographic characteristics (age, sex, regions, social vulnerability, race/ethnicity, high risk occupations), previously tested positive for COVID-19, severity of acute illness (number of symptoms reported on index date), and immunocompromised status