Physician agreement on the diagnosis of sepsis in the intensive care unit: estimation of concordance and analysis of underlying factors in a multicenter cohort

Table 4 Parameters that vary significantly between different infection site subgroups

Parameter	Mean ± SD (non-respiratory)	Mean ± SD (respiratory)	p value¹
Number in group	48	49
Overall agreement	0.85 ± 0.05	0.62 ± 0.08	6.0E-09
Free-marginal κ	0.77 ± 0.07	0.43 ± 0.12	6.8E-09
No. of indeterminates (identified by discharge consensus assessment)	6/48 = 12.5%	15/49 = 30.6%	0.030
Lowest MAP	53.9 ± 16.4	64.4 ± 17.1	0.003
Max temperature	38.3 ± 1.0	37.7 ± 0.8	0.002
Min temperature	36.0 ± 0.70	36.4 ± 0.7	0.03
Log₂ PCT	2.58 ± 3.10	− 0.46 ± 3.82	0.001

Patients with identified sites of infection (N = 97) were stratified into the following subgroups: non-respiratory (abdominal + urinary + other; N = 48) and respiratory (pneumonia + non-pneumonia; N = 49). Parameters that varied significantly (p < 0.05) between these two groups were identified by statistical testing (t test for continuous parameters; two-proportions z-test for categorical parameters)
¹Two-tailed t test for all parameters except for overall agreement and number of indeterminates, for which a two-proportion z-test was run instead (www.vassarstats.net)