Canadian Respiratory Journal

Research Article

Association of Upon-Diagnosis Blood Eosinophilic Count with Frequency and Severity of Annual Exacerbation in Chronic Obstructive Pulmonary Disease: A Prospective Longitudinal Analysis

Table 4

The Cox proportional hazard model for evaluating the risk of annual exacerbation adjusted for COPD medications (bronchodilators, inhaler corticosteroids, domiciliary oxygen, and rehabilitation).


Variables	Units/levels	Univariate		Multivariate
Variables		Univariate		Model 1		Model 2		Model 3
Events		Annual exacerbation (any)		>1 annual exacerbation		>2 annual exacerbations		>3 annual exacerbation
Events		HR (95% CI)	value	HR (95% CI)	values	HR (95% CI)	values	HR (95% CI)	values

Gender	Male vs. female	1.30 (1.26–3.24)	0.02	1.14 (1.00–1.44)	0.09	1.23 (0.92−0.41)	0.21	1.13 (0.81–1.85)	0.29
BMI	Each 1 kg/m² increase	1.12 (0.92–1.58)	0.22	—	—	—	—	—	—
Smoke	Presence vs. absence	1.45 (0.88–4.62)	0.19	—	—	—	—	—	—
Pack-year	Each 10-unit increase	1.12 (1.00–1.96)	0.07	—	—	—	—	—	—
FEV1/FVC	Each 10-unit decrease	1.42 (1.13–2.65)	0.00	1.52 (1.18–2.75)	0.00	1.61 (1.11–2.03)	0.00	1.33 (0.90–2.63)	0.21
CAT score	Each 5-unit increase	1.23 (0.92–1.65)	0.25	—	—	—	—	—	—
Gold score	≥3 (3, 4) vs. <3 (1, 2)	1.35 (1.12–2.10)	0.03	1.32 (1.01–2.79)	0.00	1.39 (1.09–2.40)	0.00	1.19 (1.30–1.52)	0.03
Cor pulmonal	Presence vs. absence	1.23 (0.90–2.63)	0.17	—	—	—	—	—	—
Influenza vaccination	Absence vs. presence	1.41 (1.06–2.35)	0.04	1.31 (0.91–2.41)	0.08	1.21 (0.84–2.12)	0.25	1.29 (0.79–1.82)	0.21
Pneumococcal vaccination	Absence vs. presence	1.44 (1.21–2.26)	0.01	1.35 (1.20–2.62)	0.09	1.25 (0.88–2.63)	0.30	1.31 (0.92–1.51)	0.21
Leukocytosis	Presence vs. absence	1.09 (1.00–1.13)	0.35	—	—	—	—	—	—
Eosinophils	Each 50 cell/microliter increase	1.92 (1.39–2.51)	0.00	1.92 (1.39–2.51)	0.03	1.65 (1.41–2.29)	0.04	1.44 (1.33–2.15)	0.03

value <0.1 are bold in univariate analysis column. value <0.05 are bold in multivariate analysis columns. Variables with value <0.1 were entered into the multivariate model. The role of each variable in different cutoff episodes of exacerbation has been evaluated in 3 multivariate Cox models.