Transfer in ICU of febrile neutropenic patients: prospective validation of a prognostic score identifying 'high-risk' patients

Objective

Infectious complications in neutropenic febrile patients are associated with a high mortality rate. A previous case–control study in postchemotherapy neutropenic patients identified three early risk factors for ICU transfer (abnormal chest X-ray, positive blood cultures, and C-reactive protein ≥ 120 mg/l), and a prognostic score based for ICU transfer on the presence or not of at least one of these variables was then performed. Our objective was to validate accuracy, adequacy and reliability of this score with another cohort of febrile neutropenic patients.

Design

A prospective consecutive cohort study over 18 months in a tertiary care hospital.

Results and measurements

Ninety patients have been included; 57.7% were men, and mean age was 50 ± 19 years. The majority of them were hospitalized for acute myeloblastic leukaemia (AML) (78 patients), the others for acute lymphoblastic leukaemia (ALL) (12 patients). Twenty-four (26.7%) patients were referred in the ICU and nine (37.5%) of them died in the ICU. The following data were associated with an increased risk of transfer in the ICU during their stay in the hematology ward: ALL, more than three previous episodes of febrile neutropenia, microbiological documentation, clinical pulmonary location, and abnormal chest X-ray. When applied to this prospective cohort, sensitivity, specificity, negative and positive predictive values of our prognostic score were, respectively, 100%, 56.1%, 100% and 45.3%.

Conclusion

No prognostic score focused on identification of 'high-risk' neutropenic patients has yet been validated. Moreover, different severity-of-illness scores used in the ICU and recently tested in hematology wards have failed to predict accurately patients at 'high risk'. These results confirmed the validity of this score for eliminating patients without risk of transfer in the ICU, but also for identifying a 'sensitive' population at higher risk of complication and transfer in the ICU. The use in haematology wards of such a prognostic score, associated with an optimization strategy, should be prospectively tested on febrile neutropenic patients in a randomized controlled trial, with outcome as the primary objective.