Screening process
We screened 2,187 hospitalised patients who tested positive for influenza A RNA. Overall, 693 immunocompetent adult and adolescent patients hospitalised with FluA-p were included in the final analysis (Fig. 1). Among these patients, 38.1% (264/693) were infected with A(H1N1)pdm09 and 11.0% (76/693) were infected with A(H3N2). In addition, 50.9% (353/693) of patients were infected with an unclassified influenza subtype.
Overview Of FluA-p Patients
Overall, 39.2% (272/693) of patients were above 65 years old and 66.5% (461/693) of patients were male. 35.1% (243/693) of patients had a history of smoking. Cardiovascular disease (19.6%), diabetes mellitus (13.3%) and cerebrovascular disease (10.4%) were the most common chronic medical conditions. Respiratory rates ≥ 30 beats/min and altered mental status could be seen in 17.5% (121/693) and 4.6% (32/693) of patients, respectively. Only 1.2% (8/693) of patients had SBP < 90 mmHg. 26.9% (172/639) of patients had pO2/FiO2 ≤ 250 mmHg (Table 1).
Table 1
Comparison of clinical features between deceased and survival patients
Variable | Total (n = 693) | Deceased group (n = 136) | Survival group (n = 557) | p-value |
Male (n, %) | 461 (66.5) | 92 (67.6) | 369 (66.2) | 0.757 |
Age ≥ 65 years (n, %) | 272 (39.2) | 60 (44.1) | 212 (38.1) | 0.195 |
Obesity (n, %) # | 48 (6.9) | 0 (0.0) | 48 (8.6) | < 0.001 |
Pregnancy (n, %) | 8 (1.2) | 0 (0.0) | 8 (1.4) | 0.338 |
Smoking history (n, %) # | 243 (35.1) | 68 (50.0) | 175 (31.4) | < 0.001 |
Comorbidities (n, %) | | | | |
Cardiovascular disease# | 136 (19.6) | 48 (35.3) | 88 (15.8) | < 0.001 |
Diabetes mellitus | 92 (13.3) | 14 (10.3) | 78 (14.0) | 0.253 |
Cerebrovascular disease | 72 (10.4) | 10 (7.4) | 62 (11.1) | 0.195 |
COPD # | 40 (5.8) | 3 (2.2) | 37 (6.6) | 0.047 |
Asthma | 19 (2.7) | 2 (1.5) | 17 (3.1) | 0.222 |
Chronic kidney disease | 16 (2.3) | 6 (4.4) | 10 (1.8) | 0.139 |
Malignant solid tumor | 16 (2.3) | 0 (0.0) | 16 (2.9) | 0.193 |
Clinical and radiologic characteristics (n, %) | | | | |
Respiratory rates ≥ 30 times/min | 121 (17.5) | 25 (18.4) | 96 (17.2) | 0.752 |
Confusion# | 32 (4.6) | 32 (23.5) | 0 (0.0) | < 0.001 |
SBP < 90 mmHg | 8 (1.2) | 0 (0.0) | 8 (1.4) | 0.338 |
Leukocytes > 10 × 109/L # | 118 (17.0) | 42 (30.9) | 76 (13.6) | < 0.001 |
Lymphocytes < 0.8 × 109/L # | 299/677 (44.2) | 120 (88.2) | 179/541 (33.1) | < 0.001 |
Hb < 100 g/L# | 69 (10.0) | 34 (25.0) | 35 (6.3) | < 0.001 |
ALB < 35 g/L# | 58/639 (9.1) | 12/131 (9.2) | 46/508 (9.1) | 0.970 |
BUN > 7 mmol/L# | 183/685 (26.7) | 97 (71.3) | 86/549 (15.7) | < 0.001 |
BG > 14 mmol/L | 8 (1.2) | 0 (0.0) | 8 (1.4) | 0.288 |
Arterial PH < 7.35 # | 120/639 (18.8) | 60 (44.1) | 60/503 (11.9) | < 0.001 |
pO2/FiO2 ≤ 250 mmHg # | 172/639 (26.9) | 28 (20.6) | 144/503 (28.6) | 0.061 |
Multilobar infiltrates# | 546 (78.8) | 120 (88.2) | 426 (76.5) | 0.003 |
Pleural effusion # | 120 (17.3) | 36 (26.5) | 84 (15.1) | < 0.001 |
Coinfections (n, %) # | 265 (38.2) | 84 (61.8) | 181 (32.5) | < 0.001 |
Early NAI use (n, %) # | 232 (33.5) | 60 (43.4) | 172 (30.9) | 0.003 |
Systemic corticosteroid use (n, %) # | 132 (19.0) | 60 (44.1) | 72 (12.9) | < 0.001 |
Noninvasive ventilation (n,%) | 159 (22.9) | 71 (52.2) | 88 (15.8) | < 0.001 |
Invasive ventilation (n,%) | 158 (22.8) | 86 (63.2) | 72 (12.9) | < 0.001 |
Admittance to ICU (n,%) | 176 (25.4) | 92 (67.6) | 84 (15.1) | < 0.001 |
COPD: chronic obstructive pulmonary disease; SBP: systolic blood pressure; Hb: hemoglobin; ALB: albumin; BUN: blood urea nitrogen; BG: blood glucose; pO2/FiO2: arterial pressure of oxygen/fraction of inspiration oxygen; NAI: neuraminidase inhibitor. #: variables cited in the table above were the candidates which were entered into the multivariate logistic regression model. The bolded values are p-values < 0.05, which represented significant differences between survival group and deceased group. |
Table 2
AUC for mortality predictions in FluA-p patients
Variable | AUC | SE | 95% CI | Z statistic | p value |
FluA-p score | 0.908 | 0.016 | 0.881–0.931 | −− | Reference |
PSI risk class | 0.560 | 0.035 | 0.518–0.602 | 10.875 | < 0.001 |
CURB-65 score | 0.777 | 0.020 | 0.740–0.811 | 6.041 | < 0.001 |
AUC: area under the curve; SE: standard error; CI:confidence interval. |
Almost forty percent (38.2%, 265/693) of patients were coinfected with other community-acquired pathogens. Streptococcus pneumoniae (33.2%) was the most common coinfection, followed by Klebsiella pneumoniae (30.6%) and Staphylococcus aureus (20.4%) (Supplementary material 4).
All patients received antibiotic treatment within 48 hours after admission (Supplementary material 5), and NAI therapy during the course of the disease. Early NAI therapy and systemic corticosteroid use were administered in 33.5% (232/693) and 19.0% (132/693) of patients, respectively. 22.8% (158/693) of patients received invasive ventilation, 25.4% (176/693) of patients were admitted to the ICU, and the 30-day mortality rate was 19.6% (136/693) (Table 1).
There were no significant differences in the demographic characteristics, clinical features, approach to clinical management, and treatment outcomes between patients in the derivation and validation cohorts (Supplementary material 6).
Predicted and actual mortality in FluA-p patients stratified by CURB-65 score and PSI risk class
Supplemental material 7 shows the actual and predicted mortality rates stratified by PSI risk class and CURB-65 scores. For the 136 deceased patients, the proportions of patients with PSI risk I~Ⅴ were 38.2% (52/136), 8.8% (12/136), 5.9% (8/136), 47.1% (64/136) and 0% (0/136), respectively; the proportions of patients with CURB-65 scores 0–5 were 0% (0/136), 66.9% (91/136), 12.5% (17/136), 0% (0/136) and 0% (0/136), respectively.
Risk Factors For 30-day Mortality
Following the procedures described in the Statistical Analysis section, the following variables were entered into a backward stepwise logistic regression analysis: obesity, smoking history, cardiovascular disease, chronic pulmonary disease (COPD), confusion, leukocytes > 10 × 109/L, lymphocytes < 0.8 × 109/L, hemoglobin (Hb) < 100 g/L, albumin (ALB) < 35 g/L, blood urea nitrogen (BUN) > 7 mmol/L, arterial PH < 7.35, pO2/FiO2 ≤ 250 mmHg, multilobar infiltrates, pleural effusion, early NAI therapy, systemic corticosteroid use, and coinfections.
A multivariate logistic regression model indicated that the following variables were significantly associated with 30-day mortality (see Fig. 2): BUN > 7 mmol/L (OR 1.604, 95% CI 1.150–4.492, p = 0.040), pO2/FiO2 ≤ 250 mmHg (OR 2.649, 95% CI 1.103–5.142, p = 0.022), cardiovascular disease (OR 3.967, 95% CI 1.269–7.322, p < 0.001), arterial PH < 7.35 (OR 3.959, 95% CI 1.393–7.332, p < 0.001), smoking history (OR 5.176, 95% CI 2.604–11.838, p < 0.001), lymphocytes < 0.8 × 109/L (OR 8.391, 95% CI 3.271–16.212, p < 0.001) and early NAI therapy (OR 0.567, 95% CI 0.202–0.833, p = 0.001).
Comparison Of Severity Scores For Mortality Prediction
In order to develop a simple and useful clinical predicting tool, relative weights were assigned according to the regression coefficient (β) of each categorical variable. Supplementary material 8 shows that the AUROC of the derivation cohort was 0.934 (95% CI 0.906–0.957), which was higher than the CURB-65 score (AUC = 0.813, 95% CI 0.772–0.850, p < 0.001 ) and the PSI risk class (AUC = 0.577, 95% CI 0.527–0.625, p < 0.001) (Supplemental Fig. 1). Supplementary material 9 shows that the AUROC of the validation cohort was 0.846 (95% CI 0.781–0.897), which was higher than the CURB-65 score (AUC = 0.681, 95% CI 0.604–0.752, p < 0.001 ) and the PSI risk class (AUC = 0.525, 95% CI 0.445–0.604, p < 0.001) (Supplemental Fig. 2). For the full sample of 693 patients, the AUROC was 0.908 (95% CI 0.881–0.931), which was higher than the CURB-65 score (AUC = 0.777, 95% CI 0.740–0.811, p < 0.001 ) and the PSI risk class (AUC = 0.560, 95% CI 0.518–0.602, p < 0.001 ) (Table 3). Table 4 shows the sensitivity, specificity and actual mortality associated with the FluA-p score (in the full sample of 693 patients). In accordance with the cut-score approach described earlier, patients were divided into high-risk and low-risk groups based on a cut-off value of 7. The Kaplan-Meier survival curves showed that 30-day mortality was significantly higher in patients with high-risk than for patients at low-risk (52.9% vs 2.1%, log rank test, p < 0.001) (Fig. 3).
Table 3
FluA-p score and actual mortality
Score | Actual 30-day Mortality (%, n) | Sensitivity | 95% CI | Specificity | 95% CI | +LR | -LR |
-2ཞ1 | 0 (0/120) | 100.00 | 97.3–100.0 | 0.00 | 0.0–0.8 | 1.00 | |
2 | 50.0 (8/16) | 100.00 | 97.3–100.0 | 25.05 | 21.2–29.2 | 1.33 | 0.00 |
3ཞ6 | 0 (0/237) | 94.12 | 88.7–97.4 | 26.72 | 22.8–30.9 | 1.28 | 0.22 |
7 | 24.4 (20/82) | 94.12 | 88.7–97.4 | 76.20 | 72.1–79.9 | 3.95 | 0.077 |
8 | 25.0 (8/32) | 79.41 | 71.6–85.9 | 89.14 | 86.0–91.8 | 7.32 | 0.23 |
9 | 71.4 (40/56) | 73.53 | 65.3–80.7 | 94.15 | 91.7–96.1 | 12.58 | 0.28 |
10 | 100 (8/8) | 44.12 | 35.6–52.9 | 97.49 | 95.7–98.7 | 17.61 | 0.57 |
11 | 81.3 (52/64) | 38.24 | 30.0–47.0 | 97.49 | 95.7–98.7 | 15.26 | 0.63 |
12 | NA | 0.00 | 0.0–2.7 | 100.00 | 99.2–100.0 | | 1.00 |
+LR: positive likelihood ratio; -LR: negative likelihood ratio. |