Authors’ response to: Relationship between leukopenia and mortality among patients with haematological malignancies

To the editor,

We thank Caibao et al. [1] for taking the time to comment on our study “Sepsis mortality among patients with haematological malignancy admitted to Intensive Care 2000–2022: a binational cohort study" [2], recently published in Critical Care.

We agree that the relationship between leukopenia (a surrogate for neutropenia) and sepsis mortality in patients with haematological malignancies is complex. In our retrospective cohort study of patients with sepsis and haematological malignancies, we fitted a single mixed effects multivariable logistic regression model to identify risk factors for mortality. In this model we included an interaction term between leukopenia and haematological malignancy (leukopenia x HM status), each coded as a binary variable, with leukopenia defined as total white cell count < 1.0 × 10⁹ cells/L. In our model output we report the marginal effect of neutropenia in the presence or absence of haematological malignancy. This choice was made for interpretability and clinical relevance. For clinicians, the impact of neutropenia on mortality risk for septic patients with haematological malignancy is important to guide prognostication, escalation of treatment, and recognition of non-neutropenic patients in local guidelines for management of sepsis.

As noted by Caibao and colleagues, we further observed that leukopenia was associated with mortality in a univariable model, both in the presence and absence of haematological malignancy, but not in a multivariable model. Formal assessment of collinearity was performed using variance inflation factor (VIF) and all included variables had VIF < 10. This indicates that while crude mortality was higher in the neutropenic group, after adjustment for confounders including age, illness severity and year of admission, this was non-significant. Our findings are in keeping with previous studies reporting that neutropenia alone is not necessarily predictive of mortality after confounders including illness severity are accounted for [3, 4].

The complex relationship between leukopenia and mortality in septic patients with haematological malignancies warrants further study. Based on existing data, neutropenic patients should not be assumed to have a poorer prognosis, and sepsis in non-neutropenic patients should not be under-estimated.