Background: Conflicting evidence underlies the controversial role of allogenic blood transfusion in recurrence of non–small cell lung cancer (NSCLC). Insufficient sample size and failure to measure effects of important confounders in previous studies contribute to the conflicting findings. To overcome these limitations, we applied robust statistics and weighted covariates in a large study cohort. Methods: Cox regression analyses were used to estimate the recurrence and survival in patients with NSCLC disease stages I through III who were transfused for a haemoglobin level less than 8.0 g/dL within seven days after surgical resection. Inverse probability of treatment weighting (IPTW) was used to balance covariates in the sequential cohort of patients receiving an incremental amount of blood. We applied restricted cubic spline functions to characterise dose-response effects of transfusion amount on recurrence and mortality. Results: A total of 209 (11.6%) of 1803 patients received transfusions. Over a median of 42 months after surgery (interquartile range 24.9–71.9), patients who received blood had a greater risk of early recurrence (IPTW-adjusted HR: 1.81, 95% CI: 1.59–2.06, P < 0.001) and all-cause mortality (IPTW-adjusted hazard ratio, HR: 2.38, 95% CI: 1.97–2.87, P < 0.001). A non-linear dose-response occurred between transfusion amount and recurrence or mortality. Conclusions: The greater risk of disease recurrence and early mortality after surgical resection in NSCLC patients who receive blood transfusion supports use of clinical strategies to reduce exposure. Further studies are needed to identify benchmarks to guide evidence-based practices.
ASJC Scopus subject areas