Introduction

Severe fever with thrombocytopenia syndrome (SFTS) is an emerging zoonosis caused by a novel bunyavirus. It was initially reported in the rural areas of China in 20091 and has also emerged in Japan2, Korea3, Vietnam4, and Thailand5. Patients with SFTS usually present with fever and thrombocytopenia, accompanied by some nonspecific symptoms such as dizziness, headache, nausea, vomiting, abdominal pain, and myalgia. And some critically ill patients could develop multiple organ dysfunction syndrome, such as encephalitis, heart failure, respiratory failure, severe pancreatitis, acute kidney injury (AKI), and disseminated intravascular coagulation (DIC), with a case fatality rate ranging from 2.5 to 30% in various epidemic regions6,7,8.

Recent studies have reported the high prevalence of co-infections in SFTS and pneumonia is a predominant co-infection, which should be of adequate concern9,10. As is universally accepted, respiratory failure is one of the most severe complications of pneumonia, which is generally associated with adverse outcomes in some infectious diseases11,12,13. Early diagnosis and timely control of respiratory failure would significantly reduce the mortality of SFTS patients with pneumonia. However, currently, there is no clinical predictive model to identify the population that has a higher risk of developing respiratory failure, and there is even no research to explore the risk factors for respiratory failure development in SFTS patients with pneumonia.

In this present study, we aimed to illuminate the prevalence, clinical characteristics, and outcomes of respiratory failure in SFTS patients with pneumonia. Then we identified the independent predictors of respiratory failure development, which were adopted as parameters of a nomogram to predict the respiratory failure development in SFTS patients with pneumonia.

Patients and methods

Patients

A total of 167 consecutive SFTS patients with pneumonia admitted to the Department of Infectious Disease, Zhongnan Hospital of Wuhan University between August 2016 and August 2024 were retrospectively included in our cohort (Supplementary Fig. 1). Patients were assigned into the no respiratory failure and respiratory failure groups according to respiratory failure development. This study was approved by the Ethics Committee of Zhongnan Hospital of Wuhan University (2024178 K).

Diagnostic criteria

The criteria for diagnosing SFTS were as follows: febrile patients (temperatures of 37.3 °C or higher for over 24 h) and decreased platelet count, laboratory-confirmed SFTSV infection by detection of viral RNA in serum via reverse transcriptase polymerase chain reaction. Respiratory failure was defined according to the Berlin Definition of Acute Respiratory Distress Syndrome (ARDS) and supplemented by the need for invasive mechanical ventilation (IMV) or non-invasive ventilation (NIV) due to hypoxemia unresponsive to supplemental oxygen therapy. Specifically, patients met at least one of the following criteria: (1) Hypoxemia: PaO2/FiO2 ratio ≤ 300 mmHg (or SpO2/FiO2 ≤ 315 for non-arterial blood gas cases) despite oxygen delivery ≥ 10 L/min via face mask or high-flow nasal cannula (HFNC ≥ 30 L/min). (2) Ventilatory support: Initiation of IMV or NIV (e.g., continuous positive airway pressure [CPAP] ≥ 5 cmH2O or bilevel positive airway pressure [BiPAP]) for respiratory distress, hypercapnia (PaCO2 > 50 mmHg with pH < 7.35), or clinical deterioration14.

Patients were excluded if they met one or more of the following reasons: (1) patients without pneumonia during hospitalization, (2) elective intubation or presence of respiratory failure on admission, (3) the presence of preterminal comorbidities (heart disease New York Heart Association III–IV, severe chronic obstructive pulmonary disease, chronic renal failure), (4) loss to follow-up, (5) patients with incomplete data.

Data collection

The medical records of patients with SFTS were reviewed, data of demographics, comorbidities, clinical manifestations and laboratory tests results including leukocyte count and percentage, neutrophils count and percentage, lymphocyte count and percentage, platelet count, hemoglobin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), albumin, alkaline phosphatase (ALP), gamma glutamyl transpeptidase (GGT), lactate dehydrogenase (LDH), amylase, lipase, total cholesterol (TC), triglyceride (TG), blood urea nitrogen (BUN), creatinine, cystatin-C, sodium, potassium, myoglobin, creatinine kinase (CK), creatinine kinase myocardial b fraction (CK-MB), troponin I, brain natriuretic peptide (BNP), prothrombin time (PT), international normalized ratio (INR), prothrombin activity (PTA), activated partial thromboplastin time (APTT), thrombin time (TT), fibrinogen, D-dimer, C-reactive protein (CRP), procalcitonin, interleukin-6 (IL-6), erythrocyte sedimentation rate (ESR), serum amyloid protein A (SAA), ferritin, SFTSV viral load, occult blood test (OBT), complications including invasive pulmonary aspergillosis (IFA), nosocomial infections, myocarditis, AKI, acute pancreatitis, rhabdomyolysis, DIC, haemophagocytic lymphohistiocytosis (HLH), shock, systemic inflammatory response syndrome (SIRS), and survival time were collected. The normal range of the above-mentioned laboratory parameters was shown in Supplementary Table 1.

Statistical analysis

Categorical variables were shown as numbers (percentages) and were compared by the Chi-square test or Fisher’s exact test. Continuous variables were shown as the means ± standard deviations for data with a normal distribution or medians with interquartile ranges (P25-P75) for data with a non-normal distribution, which were compared by the Student’s t-test or Mann–Whitney U test, respectively. To identify independent predictors of respiratory failure development while adjusting for confounding effects of other variables, variables that were significant in the univariate analysis (P < 0.05) were included in the initial multivariate model. Before fitting the multivariate model, we also calculated the variance inflation factor (VIF) for each variable in the multivariate model. A VIF value greater than 5 was used as a cut-off to identify variables that were highly collinear with other variables in the model. When multicollinearity was detected, we carefully evaluated the variables involved. In some cases, we removed one of the highly correlated variables based on clinical relevance. A stepwise backward elimination approach was applied. The variable with the non-significant P value (P > 0.05) was removed from the model. The model was refitted, and the process repeated until all remaining variables were statistically significant (P < 0.05). The area under the receiver operating characteristic curve (AUC) was evaluated for the discrimination of the nomogram. The goodness of fit of the nomogram was assessed by using the Hosmer–Lemeshow test combined with the calibration curve. Decision curve analysis (DCA) was used to evaluate the clinical utility of the nomogram. All data were analyzed with SPSS statistical analysis software version 26.0 (IBM), R statistical analysis software (version 4.4.1, Vienna, Austria), and P < 0.05 (two-sided) was considered statistically significant.

Results

Demographics, comorbidities, and clinical manifestations of SFTS patients with pneumonia according to respiratory failure development

A total of 167 consecutive SFTS patients with pneumonia were reviewed, including 122 patients in the no respiratory failure group and 45 patients in the respiratory failure group. Compared with patients in the no respiratory failure group, patients in the respiratory failure group had more presence of neurological manifestations including coma, lethargy, and confusion. The incidence of respiratory symptoms such as cough, sputum, dyspnea, and anorexia in the respiratory failure group was significantly higher than in the no respiratory failure group. Additionally, chills, abdominal pain, and diarrhea were significantly more common in the respiratory failure group compared with those in the no respiratory failure group. However, there were no significant differences in age, the proportions of male patients, and frequencies of comorbidities including diabetes mellitus, cardiovascular diseases, chronic pulmonary diseases, and chronic liver diseases between the two groups. Moreover, no significant differences in the frequencies of dysphoria, convulsion, chest distress, nausea, vomiting, diarrhea, haemorrhagic events, fever > 38 °C, headache, dizziness, and myalgia were observed between the two groups (Table 1).

Table 1 Comparison of demographics, comorbidities, and clinical manifestations of SFTS patients with pneumonia according to respiratory failure development.
Full size table

Laboratory test results, complications, and death of SFTS patients with pneumonia according to respiratory failure development

Among these laboratory parameters, serum levels of albumin, TC, and fibrinogen in the respiratory failure group were significantly lower than in the no respiratory failure group. In contrast, ALT, AST, ALP, GGT, LDH, BUN, creatinine, cystatin C, potassium, amylase, lipase, CK, CKMB, myoglobin, troponin I, APTT, TT, D-dimer, procalcitonin, IL-6, SAA, and ferritin levels were significantly higher in the respiratory failure group. Higher SFTSV viral load was also detected in the respiratory failure group than in the no respiratory failure group. No significant differences were observed between the two groups for the remaining parameters (Table 2).

Table 2 Comparison of laboratory test results of SFTS patients with pneumonia according to respiratory failure development.
Full size table

In terms of complications, patients in the respiratory failure group showed more presence of IFA, nosocomial infections, myocarditis, AKI stage 2 or 3, rhabdomyolysis, shock, and SIRS than in the no respiratory failure group. However, there were no statistical differences between the two groups in the incidence of acute pancreatitis, DIC, and HLH. Most importantly, the mortality rate of patients in the respiratory failure group was significantly higher than that of patients in the no respiratory failure group (Table 3).

Table 3 Comparison of complications and death of SFTS patients with pneumonia according to respiratory failure development.
Full size table

Univariate and multivariate risk analysis for respiratory failure development

By univariate analysis, the presence of neurological manifestations, dyspnea, abdominal pain, diarrhea, myalgia, IFA, nosocomial infections, myocarditis, AKI stage 2 or 3, rhabdomyolysis, shock, and SIRS were identified as risk factors for respiratory failure development in patients with SFTS. Serum levels of ALT, AST, ALP, GGT, LDH, BUN, creatinine, cystatin-C, potassium, amylase, lipase, CK, CKMB, myoglobin, APTT, TT, CRP, IL-6, SAA, ferritin, viral load, and low serum levels of albumin, TC, and fibrinogen were also identified as risk factors. Among these parameters, neurological manifestations (aOR, 4.336; 95% CI 1.296–14.499; P = 0.017), nosocomial infections (aOR, 4.886; 95% CI 1.006–23.738; P = 0.049), AKI stage 2 or 3 (aOR, 11.304; 95% CI 2.988–42.765; P < 0.001), SIRS (aOR, 8.161; 95% CI 2.575–25.862; P < 0.001), serum levels of albumin (aOR, 0.852; 95% CI 0.740–0.981; P = 0.026), and CKMB (aOR, 1.009; 95% CI 1.001–1.017; P = 0.028) were proven to be independent predictors for the respiratory failure development on multivariate analysis (Table 4).

Table 4 Univariate and multivariate logistic regression analyses for respiratory failure development in SFTS patients with pneumonia.
Full size table

Predictive model for respiratory failure development in SFTS patients with pneumonia

Based on the univariate and multivariate logistic regression analysis, a nomogram incorporating six predictors was constructed to predict the probability of respiratory failure development in SFTS patients with pneumonia. In the nomogram, each predictor was given a corresponding score by projecting its status to the upper point scale (0–100) with a straight line. A total score was then calculated by summing up the whole scores of the six predictors. The predicted probability of respiratory failure development could be obtained by projecting the total score line straight to the probability scale line at the bottom (Fig. 1).

Fig. 1
Fig. 1
Full size image

Nomogram for predicting respiratory failure development in SFTS patients with pneumonia. To use, locate the patient’s neurological manifestations and draw a line straight upward to the “Points” axis to determine the score associated with neurological manifestations. Repeat the process for nosocomial infections, AKI stage 2 or 3, SIRS, albumin, and CKMB, and sum the scores together and locate this total score on the“Total Points” axis. Draw a line straight down to determine the probability of respiratory failure development.

The discrimination, calibration, and clinical utility of the nomogram

The AUC of the nomogram was 0.932 (95% CI 0.888–0.975), showing an elegant discriminative performance (Fig. 2A). The goodness-of-fit χ2 of the nomogram was 7.932 (P = 0.440), which indicated no evidence of poor fit. Moreover, the calibration plot of the nomogram also showed a good agreement between the predicted and actual probabilities (Fig. 2B). DCA is an appropriate method for assessing predictive models by the net benefit and the range of threshold probabilities. In this present study, the clinical utility of the nomogram was also assessed by DCA. The plots demonstrated that the nomogram showed a greater net benefit with a wider range of threshold probabilities for predicting respiratory failure development in SFTS patients with pneumonia (Fig. 2C).

Fig. 2
Fig. 2
Full size image

The discrimination, calibration, and clinical utility of the nomogram. (A) Area under the receiver operating characteristic curve (AUC) of the nomogram for predicting respiratory failure development in SFTS patients with pneumonia. (B) Calibration plot of the nomogram for predicting respiratory failure development in SFTS patients with pneumonia. (C) Decision curve of the nomogram for predicting respiratory failure development in SFTS patients with pneumonia.

Discussion

Accumulating evidence has shown that pneumonia is one of the most frequent co-infections and is associated with high mortality in patients with SFTS10,15. It is well known that respiratory failure is a serious complication of pneumonia, generally leading to death16. In this present study, we showed that the mortality rate of SFTS patients with pneumonia who developed respiratory failure was up to 77.8%, but all patients who did not develop respiratory failure survived. Early detection and immediate treatment of respiratory failure could be a potential strategy to improve the prognosis of these critically ill patients. Our study explored the clinical information of respiratory failure in SFTS patients with pneumonia and established a nomogram for predicting the risk of respiratory failure development. The DCA indicated that the nomogram presented a greater net benefit with a wider range of threshold probabilities (0.07–0.88) for predicting respiratory failure development. For example, if a SFTS patient with pneumonia has a predicted risk of respiratory failure development of 50% (within the 0.07–0.88 range), and there are inhaled medications available to reduce the risk of respiratory failure development, the clinician would discuss with the patient the potential benefits of starting the medication, such as fewer hospitalizations and improved quality of life, as well as the possible side effects. Based on this shared decision-making process, the clinician and the patient can jointly decide whether to initiate the treatment.

To our knowledge, this is the first time to report the prevalence, clinical characteristics, and prognosis of respiratory failure in SFTS patients with pneumonia. We revealed that patients who developed respiratory failure had significantly more presence of neurological manifestations, as well as higher serum levels of laboratory parameters referring to liver, heart, pancreas, kidney, coagulation dysfunction, and viral load than patients who did not. The extremely abnormal levels of these laboratory variables generally suggest multiple organ failure, even high mortality17,18. It is well known that neurological manifestations and high viral load are the independent risk factors for adverse outcomes of patients with SFTS19,20,21.

In this present study, neurological manifestations, nosocomial infections, AKI stage 2 or 3, SIRS, serum levels of albumin, and CKMB were shown as independent predictors for respiratory failure development in SFTS patients with pneumonia.

As mentioned above, neurological manifestations have been proven to be life-threatening complications of patients with SFTS and are often related to fatal outcomes22. Some studies have confirmed that SFTS patients with neurological manifestations are susceptible to respiratory failure development23,24.

Reddy et al. demonstrated that inpatients with cirrhosis who developed nosocomial infections had a higher incidence of respiratory failure than those who did not25. Moreover, a prospective multicenter study reported that nosocomial infections could be used to predict respiratory failure development in patients with cirrhosis26. In accordance with these studies, we revealed that nosocomial infections could be independent risk factors for respiratory failure development in the SFTS cohort. We previously discovered that the cumulative survival rates of SFTS patients at AKI stage 2 or 3 were significantly lower than those of SFTS patients without AKI or at AKI stage 1, and stage AKI 2 or 3 was considered as an independent predictor for in-hospital mortality of SFTS patients27. Another retrospective study found that SFTS patients with stage AKI 2 or 3 had a higher proportion of receiving mechanical ventilation for respiratory failure28. Additionally, Herrlich et al. identified kidney-released circulating osteopontin as a novel mediator to trigger lung endothelial leakage, inflammation, and respiratory failure in an AKI mouse model29. Consistent with these studies, we proved that stage AKI 2 or 3 could predict respiratory failure development in SFTS patients.

SIRS is a complex pathophysiologic response to an insult such as infection, trauma, burns, or other injuries30. It was reported that SIRS was highly frequent and was associated with high mortality in patients with SFTS31. A prospective multicenter study indicated that SIRS was significantly associated with the risk of acute respiratory failure after out-of-hospital cardiac arrest32. In our study, we confirmed that SIRS was an independent risk factor for respiratory failure development in SFTS patients with pneumonia. Serum albumin is the most abundant protein in plasma and represents the main carrier of endogenous and exogenous compounds33. Some studies have shown that low serum albumin can be used to predict the mortality of SFTS patients34,35. Wong et al. reported that hospitalized patients with cirrhosis who developed respiratory failure had lower serum levels of albumin, and in-hospital albumin use was an independent risk factor for respiratory failure development26. As one of the specific biomarkers of myocardial injury, CKMB has become an important criterion for diagnosing myocardial necrosis36. Some studies have reported that elevated serum levels of CKMB are associated with poor prognosis of SFTS patients37,38. In our study, we found that CKMB had a stronger association with the respiratory failure development, which may be due to its ability to capture a broader range of pathophysiological changes related to cardiac stress. This early release of CKMB may serve as an early warning sign of impending cardiac dysfunction, which can then lead to respiratory failure.

The main limitation of this study was the lack of external validation of the nomogram, which restricted our ability to confidently extend the applicability of the nomogram to broader and more diverse populations. Different cohorts may exhibit variations in demographic characteristics, disease prevalence, genetic backgrounds, and environmental factors, all of which could influence the performance of the nomogram. By not having an external validation set, we are unable to assess whether our results hold true across these various contexts, thus limiting the scope of our conclusions. Therefore, we plan to conduct multi-center cohorts that involve diverse patient populations from different geographical regions. This would allow for the collection of external validation data, enabling a more comprehensive assessment of the predictive of the nomogram. We also suggest exploring the use of publicly available large-scale datasets for external validation purposes, which could provide a wider range of samples and enhance the credibility of our research.

Conclusion

In conclusion, respiratory failure is associated with adverse outcomes including severe complications and death in SFTS patients with pneumonia. Neurological manifestations, nosocomial infections, AKI stage 2 or 3, SIRS, serum levels of albumin, and CKMB are independent predictors of respiratory failure development. We established a practical clinical scoring system, through which the risk of respiratory failure development in SFTS patients with pneumonia can be estimated. We can detect high-risk pneumonic patients and provide intensive medical intervention for patients as soon as possible, potentially reducing respiratory failure rate in SFTS.