Table 2. . Logistic regression models for the probability of acute SARS-CoV-2 infection and death of inpatients with COVID-19.

Logistic regression model	Estimate	Standard error	Wald stat.	Lower CL 95.0%	Upper CL 95.0%	p-value
I. Univariate logistic regression model for acute COVID-19 as outcome
Intercept	5.067	0.773	43.00	3.55	6.58	<0.0001
sDPP4 activity (U/l)	-0.158	0.024	42.32	-0.21	-0.11	<0.0001
II. Univariate logistic regression model for death as outcome in COVID-19
Intercept	1.4819	0.837	3.134	-0.159	3.123	0.0767
sDPP4 activity (U/l)	-0.1160	0.0380	9.322	-0.191	-0.042	0.0023
III. Multivariate logistic regression model† for death as outcome in COVID-19
Intercept	-5.3327	2.2481	5.62681	-9.73897	-0.9265	0.0177
Age (years)	0.0920	0.0282	10.62594	0.03669	0.1473	0.0011
sDPP4 activity (U/l)	-0.0975	0.0394	6.12794	-0.17474	-0.0203	0.0133

I. Model outcome: acute SARS-CoV-2 infection *(WHO ordinal scale: 3–8) versus plasma donor recovered from COVID-19 and ‘before-COVID-19 pandemic’ control groups combined.

II. Model outcome: death versus survival in hospitalized acute COVID-19 patients.

The univariate logistic regression models for 19 other relevant predictor candidates are indicated in Supplementary Table 1.

III. Model outcome: death versus survival in hospitalized acute COVID-19 patients.

^†Additional adjustments (13 laboratory predictors and 5 established clinical risk factors) on the relationship between sDPP4 activity and mortality are reported in Supplementary Table 2.

CL: Confidence limit; sDPP4: Soluble DPP4.