Skip to main content
. 2015 Apr 1;10(4):e0121691. doi: 10.1371/journal.pone.0121691

Table 4. Cox Proportional Hazards Models for Mortality Prediction of PQ Poisoning.

Univariate Cox Model Multivariate Cox Model
variable HR (95%CI) P value HR (95%CI) P value
Amount 6.91 (6.11–7.81) < 0.001 2.49 (2.15–2.88) < 0.001
Plasma 10.21 (9.17–11.49) < 0.001 4.52 (3.92–5.21) < 0.001
Urine 7.81 (6.86–8.89) < 0.001 2.04 (1.74–2.40) < 0.001
APACHEII 8.08 (6.99–9.33) < 0.001 3.80 (3.24–4.46) < 0.001
SOFA 5.23 (4.54–6.02) < 0.001 N/A 0.225
Ground glass opacity 14.34 (12.23–16.80) < 0.001 5.82 (4.77–7.09) < 0.001
Consolidation 2.74 (2.45–3.08) < 0.001 N/A 0.229
Pneumomediastinum * 6.09 (5.31–6.99) < 0.001 1.72 (1.49–1.98) < 0.001
“No obvious lesion” * 0.35 (0.30–0.42) < 0.001 N/A 0.240

Definition of abbreviations: PQ = paraquat; HR = harzard ratio; N/A = not applicable.

Note: * indicates variables were categorical using the presence of CT finding as a dichotomous variable; other variables were categorical according whether lesion volume ratio was above or below the optimal cutoff point.