Background
Methods
Study design
Data collection
Between-group analyses
Multivariable analysis adjustment: prediction of corticosteroid use and clinical outcomes
Propensity score adjustment
Baseline characteristicsa
| All patientsb (n = 607) | Not treated with corticosteroids (n = 327) | Treated with corticosteroids (n = 280) |
p Valuec
|
---|---|---|---|---|
Age, years | 47.4 (15.3) | 46.2 (15.2) | 48.8 (15.3) | 0.04 |
Female sex, n (%) | 315 (51.9) | 163 (49.9) | 152 (54.3) | 0.28 |
BMI, kg/m2
| 32.0 (10.4) | 31.8 (10.7) | 32.3 (10.1) | 0.62 |
APACHE II score | 20.6 (10.0) | 20.1 (9.7) | 21.2 (10.3) | 0.22 |
SOFA score, day 1 | 11.4 (3.7) | 11.3 (3.6) | 11.4 (3.8) | 0.70 |
SOFA score (cardiovascular), day 1 | 1.6 (1.5) | 1.6 (1.5) | 1.6 (1.5) | 0.91 |
PaO2/FiO2 ratio, day 1 | 155.1 (92.1) | 156.3 (91.8) | 153.5 (92.4) | 0.73 |
Tidal volume (ml), day 1 | 514.0 (131.9) | 515.8 (119.5) | 513.0 (144.2) | 0.84 |
Tidal volume per ideal body weight (ml/kg), day 1 | 6.2 (2.1) | 6.3 (2.1) | 6.1 (2.1) | 0.56 |
Positive end-expiratory pressure, (cmH2O), day 1 | 10.9 (4.6) | 10.7 (4.3) | 11.1 (4.9) | 0.39 |
Patients with any comorbidities (any) | 543 (89.5) | 280 (85.6) | 263 (93.9) | 0.001 |
Comorbidities per patient, n (%) | 3.2 (2.3) | 2.8 (2.2) | 3.7 (2.4) | <0.001 |
Pulmonary comorbidity, n (%) | 195 (32.1) | 67 (20.5) | 128 (45.7) | <0.001 |
Asthma, n (%) | 124 (20.4) | 42 (12.8) | 82 (29.3) | <0.001 |
COPD, n (%) | 100 (16.5) | 30 (9.2) | 70 (25.0) | <0.001 |
Cardiac disease, n (%) | 89 (14.7) | 40 (12.2) | 49 (17.5) | 0.07 |
Hypertension, n (%) | 195 (32.1) | 102 (31.2) | 93 (33.2) | 0.60 |
Obesityd, n (%) | 145 (23.9) | 74 (22.6) | 71 (25.4) | 0.43 |
Diabetes, n (%) | 155 (25.7) | 80 (24.5) | 75 (26.8) | 0.51 |
Immune suppressione, n (%) | 35 (5.8) | 10 (3.1) | 25 (8.9) | 0.002 |
Chronic renal insufficiencyf, n (%) | 53 (8.9) | 32 (9.8) | 21 (7.5) | 0.32 |
Dialysis dependence, n (%) | 14 (2.3) | 11 (3.4) | 3 (1.1) | 0.06 |
Autoimmune disease, n (%) | 10 (1.7) | 2 (0.6) | 8 (2.9) | 0.03 |
Cirrhosis, n (%) | 26 (4.3) | 14 (4.3) | 12 (4.3) | 0.99 |
Bacterial coinfection at admission, n (%) | 205 (33.9) | 118 (36.1) | 87 (31.1) | 0.19 |
Septic shock at admissiong, n (%) | 74 (12.3) | 33 (10.1) | 41 (14.6) | 0.09 |
Pregnant or postpartumh, n (%) | 30 (906) | 24 (14.7) | 6 (4.0) | 0.001 |
Cointervention | All patientsa (n = 607) | Not treated with corticosteroids (n = 327) | Treated with corticosteroids (n = 280) |
p Value |
---|---|---|---|---|
Mechanical ventilation, day 1 | 392 (66.2) | 212 (67.5) | 180 (64.8) | 0.48 |
Mechanical ventilation, any | 535 (89.2) | 275 (85.4) | 260 (93.5) | 0.001 |
Rescue oxygenation strategy, any | 98 (16.1) | 39 (11.9) | 59 (21.1) | 0.002 |
ECMO | 18 (3.0) | 10 (3.1) | 8 (2.9) | 0.88 |
HFOV | 61 (10.1) | 27 (8.3) | 34 (12.1) | 0.11 |
Nitric oxide | 52 (8.6) | 21 (6.4) | 31 (11.1) | 0.04 |
Prone ventilation | 11 (1.8) | 3 (0.9) | 8 (2.9) | 0.07 |
Antibiotic treatment, day 1 | 382 (63.9) | 199 (60.9) | 183 (65.4) | 0.25 |
Antibiotic treatment, any | 588 (96.9) | 317 (96.9) | 271 (96.8) | 0.91 |
Neuraminidase inhibitor treatment, day 1 | 316 (52.1) | 166 (50.8) | 150 (53.6) | 0.49 |
Neuraminidase inhibitor treatment, any | 564 (92.9) | 294 (89.9) | 270 (96.4) | 0.002 |
Vasopressor treatment, day 1 | 251 (41.4) | 131 (40.1) | 120 (42.9) | 0.49 |
Vasopressor treatment, any | 334 (55.0) | 171 (52.3) | 163 (58.2) | 0.14 |
Time-dependent variable adjustment using marginal structural models
Results
Characteristics of patients receiving corticosteroids or not
Medication variable | Median (q1, q3) or n (%)a
|
---|---|
Corticosteroid treatment, n (%) | 280 (46.1) |
Corticosteroid prescribed, n (%) | |
Prednisone | 189 (34.4) |
Methylprednisolone | 177 (32.2) |
Hydrocortisone | 161 (29.3) |
Dexamethasone | 22 (4.0) |
Cortisone | 1 (0.2) |
Duration of corticosteroids, days | 7.0 (4.0, 13.0) |
Dose, hydrocortisone equivalents per dayb (mg) | 227 (154, 443) |
Dose, hydrocortisone equivalents per day (mg/kg) | 3.1 (1.7, 5.8) |
Duration between onset of critical illness and corticosteroid initiation, days | 0.0 (0.0, 3.0) |
Duration between hospital admission and corticosteroid initiation, days | 2.0 (1.0, 8.0) |
Duration between onset of ventilation and corticosteroid initiation, days | 1.0 (0.0, 3.0) |
PaO2/FiO2 before corticosteroid administrationc (cmH2O) | 140 (91, 220) |
Positive End-Expiratory Pressure before corticosteroid administrationc (cmH2O) | 10.0 (8.0, 14.0) |
SOFA cardiovascular score before corticosteroid administration | 2.0 (0.0, 3.0) |
Corticosteroid use
Corticosteroid treatment | ||||
---|---|---|---|---|
Characteristics and cointerventions | Univariable analysis, OR (95 % CI) |
p Value | Multivariable analysis, OR (95 % CI) |
p Value |
Age (per 1-year increase) | 1.01 (1.00, 1.02) | 0.04 | 1.06 (0.99, 1.02) | 0.41 |
Female sex | 1.18 (0.86, 1.63) | 0.31 | 1.33 (0.92, 1.92) | 0.13 |
APACHE II score (per 1-point increase) | 1.01 (0.99, 1.03) | 0.21 | 1.01 (0.99, 1.03) | 0.37 |
Number of comorbidities per patient (per 1 increase) | 1.17 (1.09, 1.26) | <0.001 | 1.04 (0.93, 1.15) | 0.51 |
Pulmonary comorbidity | 3.24 (2.27, 4.63) | <0.001 | 3.82 (2.48, 5.87) | <0.001 |
Cardiac disease | 1.51 (0.96, 2.38) | 0.07 | 1.00 (0.55, 1.81) | 0.99 |
Immune suppressiona
| 3.09 (1.46, 6.54) | 0.003 | 3.58 (1.59, 8.06) | 0.002 |
Septic shock at admissionb
| 1.52 (0.93, 2.48) | 0.09 | 2.20 (1.25, 3.88) | 0.006 |
Bacterial coinfection at admission | 0.80 (0.57, 1.13) | 0.19 | 0.71 (0.48, 1.05) | 0.09 |
Outcomes
In-hospital mortality | ||||
---|---|---|---|---|
Clinical characteristics and cointerventions | Univariable analysis, OR (95 % CI) |
p Value | Multivariable analysis, OR (95 % CI) |
p Value |
Age (1-year increase) | 1.02 (1.01–1.04) | 0.002 | 1.02 (1.00–1.03) | 0.15 |
Female sex | 0.94 (0.61–1.38) | 0.77 | 0.92 (0.56–1.50) | 0.73 |
Corticosteroid | 1.82 (1.21–2.74) | 0.004 | 1.85 (1.12–3.04) | 0.02 |
APACHE II score (1-point increase) | 1.07 (1.05–1.11) | <0.001 | 1.07 (1.04–1.10) | <0.001 |
PaO2/FiO2 ratio (cmH2O), day 1 | 1.00 (0.99–1.00) | 0.004 | 0.99 (0.99–1.00) | 0.06 |
Asthma | 0.64 (0.37–1.11) | 0.12 | 0.64 (0.33–1.26) | 0.20 |
Autoimmune disease | 3.21 (0.85–12.13) | 0.09 | 3.02 (0.75–12.20) | 0.12 |
Bacterial coinfection at admission | 1.45 (0.96–2.20) | 0.08 | 1.20 (0.72–1.98) | 0.48 |
Analysis | Odds ratio (95 % confidence interval) |
p Value |
---|---|---|
Crude unadjusted analysis | 1.82 (1.21–2.74) | 0.004 |
Multivariate logistic regression analysis | 1.85 (1.12–3.04) | 0.02 |
Multivariate logistic regression analysis adjusted for propensity score to receive corticosteroids | 1.71 (1.05–2.78) | 0.03 |
Treatment groups matched on propensity to receive corticosteroids | 1.52 (0.90–2.58) | 0.12 |
Marginal structural model adjusting for baseline and time-dependent between-group differences over the first week of ICU admission until discharge or death | 0.96 (0.28–3.28) | 0.95 |
Variable | Rate ratio (95 % CI) |
p Value |
---|---|---|
Corticosteroid use | 0.96 (0.28–3.28) | 0.95 |
APACHE II score (1-point increase) | 1.07 (1.01–1.13) | 0.02 |
SOFA score, day 1 (1-point increase) | 0.97 (0.87–1.09) | 0.63 |
Age (1-year increase) | 1.00 (0.97–1.03) | 0.87 |
Female sex | 1.53 (0.65–3.83) | 0.32 |
Asthma (yes vs no) | 0.97 (0.27–3.52) | 0.96 |
Autoimmune disease (yes vs no) | 1.92 (0.54–6.81) | 0.31 |
Discussion
Conclusions
Key messages
-
While observational studies have reported significant corticosteroid-associated mortality, corticosteroids have remained a common treatment in influenza-related critical illness.
-
Among 607 patients with H1N1pdm09-related critical illness from 51 Canadian ICUs, corticosteroids were administered to 46.1 %.
-
Compared with patients who did not receive corticosteroids, patients who received corticosteroids had higher hospital crude mortality; however, the odds ratio association between corticosteroid use and hospital mortality decreased from 1.85 with multivariate logistic regression, to 1.71 after adjustment for propensity score to receive corticosteroids, to 1.52 after case matching on propensity score, and to 0.96 using marginal structural modeling to adjust for time-dependent between-group differences.
-
These findings highlight the challenges and importance in adjusting for baseline and time-dependent confounders when estimating clinical effects of treatments using observational studies.