Development of a Prediction Model for the Risk of Infection in Patients with Aplastic Anemia: Survival Analysis in Recurrent Events.

Abstract

Background: In patients with aplastic anemia (AA), infection-related complications are the leading cause of mortality. However, limited knowledge about the predictive factors for infection in these patients exists. Thus, this study aimed to evaluate risk factors for infection and develop a risk prediction model for the occurrence of infection in patients with AA.

Materials and methods: Between January 2004 and December 2020, 206 patients with AA ≥15 years of age were included in this study. Survival analysis using recurrent event methodologies was conducted to identify predictive factors associated with infection, including the Anderson and Gill model; Prentice, Williams, and Peterson (PWP) Total Time model; PWP Gap Time model; marginal model; and frailty models. The best model was determined using backward stepwise regression, and internal validation was performed using the Bootstrapping method with 500 re-samplings.

Results: With a median follow-up of 2.95 years, the incidence rate of infection among patients with AA was 32.8 events per 100 person-years. The PWP Total Time model revealed that cirrhosis comorbidity, lymphocytes ≥80%, and previous infection increased the risk of infection, while bone marrow cellularity ≥20% offered protection. The bone marrow cellularity, lymphocyte percentage, previous Infection, cirrhosis, and hematocrit (BLICH) model was generated to predict the risk of infection. The internal validation showed a good calibration of this model.

Conclusion: Cirrhosis, lymphocytes ≥80%, previous infection, and bone marrow cellularity <20% are risk factors for infection in patients with AA. The BLICH model can predict the risk of infection in these patients.