Age-related left-digit bias in selection of patients referred for autologous and allogeneic hematopoietic transplantation: An analysis of 877 patients referred to a single center

Abstract Left-Digit Bias (LDB) is a well described cognitive phenomenon that results in greater emphasis of the left-most digit of a continuous variable when making decisions (Sokolova et al, J Marketing Research 2020 v 57: p771). Age-related left-digit bias (ARLDB) can result in preferential treatment of patients just below versus just above a change of left digit for age (e.g. age 69 vs age 70) despite such patients being biologically similar. ARLDB has been shown to significantly influence the decisions made by physicians when allocating medical care e.g. for rectal cancer surgery (Melucci et al Surgery, v 172: p 851) coronary grafting (Olenski et al NEJM 2020, v382: p778) prostate cancer surgery (Brant et al 2022, J Urology, v208: p997) and selection of organs from cadaveric donors for transplantation (Ali Husain et al Clin Transplant 2022, v35, p e14284). ARLDB has not been previously studied in hematopoietic cell transplantation (HCT). To assess if ARLDB was influential in the selection of patients who undergo HCT, we analyzed all consecutive patients referred to our center for consideration of autologous transplantation (autoHCT) for myeloma (MM), non-Hodgkin lymphoma (NHL) or Hodgkin lymphoma (HL), or allogeneic transplantation (alloHCT) for acute myeloid leukemia or acute lymphoid leukemia (AML/ALL) between 1998 and 2024. Patients included were aged between 67 and 72 at the time of their initial consultation. We compared the rate of transplantation in patients aged 67-69 (group 1) vs patients aged 70-72(group 2) at the time of their consultation and compared outcomes of patients transplanted in the two age groups. Data were extracted from our institutional database where they had been prospectively entered. The study population consisted of 877 patients, including 573 referred for autoHCT and 304 patients referred for alloHCT. Age groups 1 and 2 comprised 473 (54%) and 407 (46%) patients respectively. Patient characteristics were not significantly different between groups 1 and 2: male sex - 56% vs 61%; race- non-Hispanic white 66% vs 71%, black 25% vs 21%; diagnosis – MM/NHL/HL 68% vs 62%, AML/ALL 32% vs 38%; distance from transplant center – median 25 vs 26 miles; household income by zipcode – median $79.6K vs $79.8K; year of consultation 1998-2014 ( 37% vs 36%), 2015-2019 (31% vs 34%), 2020-2024 (32% vs 30%), median time from referral to consultation 19 vs 17 days, respectively. The Cochran–Mantel–Haenszel (CMH) test was used to associate proportion of referred patients proceeding to transplant with age group while accounting for clustering by transplant physician. 244 of 473 patients (51.6%) in group 1 vs 162 of 404 patients (40.4%) in group 2 proceeded to HCT following consultation (p<0.001). When analyzed separately by diagnosis and type of HCT, 198 of 321 (61.7%) of patients with MM/NHL/HL referred for autoHCT in group 1 proceeded to HCT vs 138 of 252 (54.8%) patients in group 2 (p=0.06). For patients with AML/ALL referred for alloHCT, 46 of 152 patients in group1 (30.3%) vs 24 of 152 (15.8%) patients in group 2 proceeded to HCT (p=0.003). For multivariable analysis, the generalized linear mixed model (GLMM) including transplant physician as the random effect factor was used to associate factors with proceeding to transplant. Group 1 patients were significantly more likely to proceed to HCT than group 2 patients (OR 1.62, p=0.001). Other significant factors associated with likelihood of proceeding to HCT were referral for autoHCT vs referral for alloHCT (OR 5.40, p<0.001), race – black vs white (OR 0.71, p=0.06), household income > $90K vs <65K (OR 1.52, p=0.024). For patients who underwent HCT from groups 1 and 2, three-year estimates of outcomes for autoHCT: overall survival (OS) 77% vs 83%, disease-free survival (DFS) 52% vs 55% and alloHCT: OS 57% vs 65% and DFS 59% vs 56% were not significantly different. These data show that when assessing the likelihood of proceeding to HCT, left-digit bias was demonstrable for patients aged 67-69 vs 70-72 referred to our center. The effect was more prominent in patients referred for alloHCT than autoHCT. Patients transplanted in the two age groups did not show significant differences in post-transplant outcomes. Awareness of this phenomenon may help prevent disparate access to HCT based upon biologically insignificant differences in age