Table 3.

Univariable and multivariable Cox regression analyses of chest X-ray variables and formation of the chest X-ray score

Outcome: all-cause mortality	Hazard ratio (95% CI)	Wald X2	P value	Hazard ratio (95% CI)	β-coefficient	Score
Alveolar oedema
Absent	Referent	Referent	Referent	Referent	Referent	0
Present	1.20 (0.98–1.47)	1.71	0.09	1.04 (0.84–1.31)	0.05	1
Severe	1.67 (1.22–2.28)	3.20	0.001	1.35 (0.97–1.88)	0.30	3
Kerley B lines
Absent	Referent	Referent	Referent	Referent	Referent	0
Present	1.28 (1.04–1.59)	2.30	0.02	1.21 (0.98–1.51)	0.19	2
Cardiothoracic ratio
≤ 0.55	Referent	Referent	Referent	Referent	Referent	0
0.55–0.70	1.21 (0.96–1.52)	1.62	0.10	1.12 (0.89–1.41)	0.12	1
> 0.70/unmeasurable	1.84 (1.35–2.50)	3.84	< 0.001	1.60 (1.16–2.19)	0.46	5
Chest X-ray projection
Posterior-anterior	Referent	Referent	Referent	Referent	Referent	0
Anterior–posterior	1.13 (1.07–1.20)	4.28	< 0.001	1.47 (1.17–1.86)	0.38	4
Pleural effusions
Absent	Referent	Referent	Referent	Referent	Referent	0
Present	1.29 (1.05–1.59)	2.47	0.01	1.12 (0.89–1.40)	0.11	1
Pulmonary venous congestion
Absent	Referent	Referent	Referent	N/A	N/A	N/A
Present	1.06 (0.85–1.33)	0.53	0.60	N/A	N/A	N/A
Chest X-ray score
Chest X-ray score	1.10 (1.07–1.13)	6.15	< 0.001	N/A	N/A	N/A

The score was constructed using the beta-coefficients (log hazard ratio) of a multivariable model, containing only chest X-ray variables that were significantly related to all-cause mortality on univariable analysis (p < 0.1). These variables were: alveolar oedema, Kerley B lines, cardiothoracic ratio, chest X-ray projection and pleural effusions. For example, the beta coefficient for Kerley B lines from the multivariate analysis is 0.19—which was rounded to 0.2 multiplied by 100 to give 2 points

CI confidence intervals, N/A not applicable