. 2016 Sep 17;6:91. doi: 10.1186/s13613-016-0193-x

Table 3.

Logistic regression models in the total ICU population excluding patients admitted after elective cardiac surgery

	Outcome: 30-day mortality^a (n = 6480)				Outcome: hospital mortality^b (n = 5184)
	n	% outcome (%)	Odds ratio (95 % CI) versus normochloremia	p value	n	% outcome (%)	Odds ratio (95 % CI) versus normochloremia	p value
Chloride category
Normochloremia	3382	10.2			2707	24.3
Hypochloremia	434	16.4	1.02 (0.71–1.48)	0.91	329	43.8	1.55 (1.13–2.13)	0.007
Moderate hyperchloremia	1456	8.5	1.16 (0.88–1.51)	0.30	1188	18.4	0.96 (0.77–1.20)	0.72
Severe hyperchloremia	1208	14.6	1.81 (1.32–2.50)	<0.001	960	29.0	1.49 (1.14–1.96)	0.003

	n	% outcome (%)	Odds ratio (95 % CI) versus intermediate SIDa	p value	n	% outcome (%)	Odds ratio (95 % CI) versus intermediate SIDa	p value
SIDa category
Intermediate SIDa	2982	9.6			2382	20.5
Low SIDa	1598	9.4	0.89 (0.68–1.16)	0.40	1260	21.5	0.94 (0.70–1.15)	0.60
High SIDa	1900	14.6	1.05 (0.83–1.32)	0.71	1542	34.9	1.35 (1.11–1.65)	0.002
			Area under ROC 86.7 %				Area under ROC 83.5 %
			Maximal VIF 2.26				Maximal VIF 2.39
			Tolerance 0.68				Tolerance 0.68

SIDa apparent strong ion difference (excl. lactate), ROC receiver under operating characteristics curve, VIF variance inflation factor

^aConfounders adjusted for in the model: sodium (high vs. normal*, low vs. normal*), SAPS-3, admission reason, RIFLEcrea*, RIFLEurine, lactate, heart failure*, COPD*, pCO2* (low, normal, high), albumin, SIG, potentiometry type* (*p > 0.05)

^bConfounders adjusted for in the model: sodium (high vs. normal*, low vs. normal*), SAPS-3, admission reason, RIFLEcrea, RIFLEurine, lactate, heart failure, COPD, pCO2* (low, normal, high), albumin, SIG*, potentiometry type (*p > 0.05)