Table 1.
The demographic and clinical characteristics of patients with SFTS with and without HLH.
Fig 1.
Comparison of the discriminatory performance of individual early clinical and laboratory variables for HLH complicating SFTS.
Comparison of the areas under the receiver operating characteristic curves (AUCs) for early clinical and laboratory variables measured within 3 days after virologic confirmation. Each bar represents the AUC with its 95% confidence interval for an individual predictor distinguishing SFTS patients who developed HLH from those who did not. Variables are ordered from highest to lowest AUC, with markers at the top indicating the strongest discriminatory ability.
Fig 2.
Nomogram for bedside early risk stratification of secondary hemophagocytic lymphohistiocytosis in patients with severe fever with thrombocytopenia syndrome.
This nomogram is intended for use during the early hospital evaluation of patients with laboratory-confirmed SFTS. After locating the patient’s value for each predictor on the corresponding axis, draw a vertical line upward to the Points axis to assign points, sum the points across all predictors, and project the total vertically to the Predicted Probability of HLH scale to estimate individual risk. Higher total points indicate a higher probability of HLH and may justify closer monitoring, repeated HLH assessment, and earlier specialist consultation. Predictor coding: For Tmax, categories 1–3 correspond to ≤38.0 °C, 38.1–39.0 °C, and ≥39.1 °C, respectively. For ferritin, categories 1–3 correspond to <2000 ng/mL, 2000–6000 ng/mL, and >6000 ng/mL, respectively. For splenomegaly, 0 indicates absence and 1 indicates presence.
Fig 3.
Discrimination, calibration, and clinical utility of the SFTS–HLH prediction model in the derivation and internal validation cohorts.
Panel (A) shows the receiver operating characteristic (ROC) curve of the multivariable model in the training cohort, with the area under the ROC curve (AUC) quantifying discrimination; panel (B) displays the corresponding calibration plot comparing predicted and observed risks of secondary HLH in the training cohort; panel (C) shows the ROC curve of the model in the internal validation cohort; and panel (D) shows the calibration plot in the internal validation cohort after intercept-and-slope recalibration. Panels (E) and (F) display decision-curve analyses in the training and internal validation cohorts, illustrating the net clinical benefit of using the prediction model across a range of threshold probabilities compared with “treat-all” and “treat-none” strategies.
Fig 4.
External validation of the SFTS–HLH prediction model.
(A) Receiver operating characteristic (ROC) curve of the prediction model in the external validation cohort, with the area under the ROC curve (AUC) quantifying discrimination. (B) Calibration plot comparing predicted and observed probabilities of secondary HLH in the external cohort. (C) Decision-curve analysis in the external validation cohort, showing the net benefit of applying the model across a range of decision thresholds relative to “treat-all” and “treat-none” strategies.