Analysis of influencing factors and predictive model construction for platelet transfusion efficacy in hematological patients

Background This study aimed to systematically analyze the independent risk factors for platelet transfusion refractoriness (PTR) in hematological patients, and to develop and validate a nomogram prediction model, thereby providing scientific evidence for personalized platelet transfusion strategies in clinical practice. Methods A retrospective cohort study was conducted involving 363 platelet transfusion episodes in hematological patients who received platelet transfusions at Deyang People’s Hospital between January 2023 and August 2023. Comprehensive clinical data and laboratory parameters were collected. Potential PTR-related factors were initially identified through univariate analysis, followed by multivariate logistic regression to determine independent risk factors. Using Rstudio software, a nomogram prediction model was constructed based on the identified factors. The model’s performance was rigorously evaluated through receiver operating characteristic (ROC) curve analysis, calibration curves, and internal validation using bootstrap resampling (1,000 repetitions) to assess discrimination, calibration, and clinical applicability. Results This study retrospectively analyzed 363 platelet transfusion episodes involving 131 hematological patients, the incidence of PTR was 30.85% (112/363). Multivariate logistic regression analysis revealed four independent risk factors for PTR: female gender (OR = 1.876, 95% CI: 1.147–3.067), transfusion frequency ≥ 10 times (OR = 2.552, 95% CI: 1.089–5.981), splenomegaly (OR = 3.170, 95% CI: 1.334–7.534), and antibiotic usage (OR = 2.177, 95% CI: 1.078–4.396) (all p 0.05). The predictive model demonstrated an area under the ROC curve of 0.673 (95% CI: 0.611–0.735), with specificity of 78.1%, sensitivity of 55.4%, Youden’s index of 0.335, and an optimal cutoff value of 0.320. Internal validation confirmed good consistency between predicted probabilities and actual observations. Conclusion We successfully developed and validated a PTR prediction model incorporating gender, transfusion frequency, splenomegaly, and antibiotic usage as key risk factors. This model exhibits promising clinical utility and can serve as an objective tool for optimizing individualized platelet transfusion protocols in hematological patients.