Dear Editor,

Humans rely on heuristics to simplify decision-making [1] and often use chronological age as a key factor. However, chronological age may not accurately reflect an individual’s physiological age and functional performance [2]. A notable heuristic in decision-making is left-digit bias, where humans tend to categorize decisions based on the left-most digit of a continuous variable (e.g., age, blood pressure) [3]. For example, items are often priced at $9.99, because the left digit bias makes $9.99 seem cheaper than $10. A literature search reveals that this phenomenon has not previously been studied in multiple myeloma (MM). We evaluated the presence of a left digit bias and the impact of age on decision-making in an administrative database in Ontario, Canada, by assessing differences in outcomes and utilization of therapies, specifically autologous stem cell transplant (ASCT), among patients with MM near the age of 70.

We conducted a retrospective, population-based observational study using administrative healthcare data from the province of Ontario. Since Ontario has a universal, publicly funded health system, access to hospital, physician, and cancer care expenditures, including chemotherapy agents and ASCT, is covered. Administrative healthcare datasets were linked using unique encoded identifiers and analyzed at ICES (formerly known as the Institute for Clinical Evaluative Sciences). ICES is an independent, non-profit research institute that uses deidentified anonymous healthcare and demographic data for health system research [4].

All adults (age ≥ 18) with a new diagnosis of MM (International Classification of Diseases for Oncology, 3rd Edition, histology code 9732) who received treatment within one year of diagnosis between January 2007 and December 2022 were identified. The Ontario Cancer Registry was used to identify patients with MM and the initial cohort creation.

Patient demographics were extracted, including age, sex, distance to nearest cancer center, comorbidities, previous cancer history, presence of end-organ damage (CRAB features) at diagnosis, and functional status at the time of diagnosis. The main outcome of interest was the use of an upfront ASCT within 12 months of MM diagnosis. In Ontario, treatment decisions are made at the time of diagnosis, as there are different funded pathways based on transplant eligibility, making the age of diagnosis relevant for decision-making [5]. Furthermore, there are no mandated age cutoffs to determine transplant eligibility; this decision is left to the discretion of the treating physician [5, 6].

We compared the baseline characteristics and transplant utilization for patients aged 65–75 at the time of diagnosis. A multivariate logistic regression analysis was performed to evaluate the effect of age, adjusting for other demographic characteristics.

Table S1 lists characteristics of the included patients, stratified by age of diagnosis ranging from 65 to 75. Patients were balanced across various demographic features. Due to the study period and the reimbursement policy in Canada at that time, the use of daratumumab was uniformly low in induction (less than 7% for all age groups), while the use of bortezomib was high (greater than 74% in all age groups). Characteristics did not differ substantially across the different age groups, as highlighted in Table S1.

Although there was not a decrease in ASCT utilization from 65 to 66, the use of ASCT within the first year of diagnosis decreased age 67 and older: 59% for those aged 65, 62% for those aged 66, 57% for those aged 67, 49% for those aged 68, 44% for age 69, 29% for age 70, 16% for age 71, 12% for age 72, 7% for age 73, and 2% for age 74 and 75. (Fig. 1). Even after adjusting for all available other covariates in a multivariate model (Table S2), there were reduced odds of receiving ASCT with yearly increment in age 67 and beyond, (from 65 to 66: OR = 1.05, 95% CI 0.75–1.48, 66 to 67: OR = 0.81, 95% CI 0.58–1.30, 67 to 68: OR = 0.72, 95% CI 0.53–0.99; 68–69: OR = 0.77, 95% CI 0.57–1.05; 69 to 70: OR = 0.52, 95% CI 0.37–0.72; 70–71 OR = 0.45, 95% CI 0.31–0.65; 71 to 72 OR = 0.75, 95% CI 0.49–1.15; 72 to 73: OR = 0.54, 95% CI 0.32–0.90, 73 to 74: OR 0.30, 95% CI 0.13–0.67, 74 to 75: OR = 0.91, 95% CI 0.32, 2.54). The odds of receiving ASCT did not drop dramatically from 69 to 70 (OR = 0.52) compared to 70 to 71 (OR = 0.45), indicating an absence of left-digit bias, although there was a more pronounced drop between 69 and 70 (OR = 0.52) compared to 68–69 (OR = 0.77).

Fig. 1
figure 1

Utilization of autologous stem cell transplantation within 1 year of diagnosis stratified by age at diagnosis.

Full size image

This study explored the impact of age and left-digit bias in decision-making and outcomes for patients with MM. Although we did not find clear evidence of left-digit bias in decision-making for MM, we observed a significant decline in the use of ASCT between the ages of 65 and 75. This decline persisted even after adjusting for comorbidities and performance status, indicating that chronological age plays a substantial role in decision-making for MM.

We use ASCT as an example to highlight differences in clinical decision-making. Our dataset indicates that chronological age can significantly influence treatment decisions, sometimes even at an institutional level [6, 7]. The decision to proceed with ASCT is highly nuanced, as the procedure incurs substantial short-term toxicity and an increased risk of secondary cancers, but also increases the duration and depth of remission [8]. Such decisions should be based on patient preferences, clinical factors, and individual patient characteristics, rather than arbitrary age cut-offs. Future work should explore whether such differences also lead to the use of triplets versus quadruplets, chimeric antigen receptor therapy, and bispecific antibodies.

Our study was unable to evaluate long-term differences in progression-free survival that could be influenced by variations in treatment decisions made at the time of diagnosis. The dataset used does not provide detailed information on aspects such as cytogenetic risk status, which is important information when addressing whether to pursue ASCT for patients. Exploring these phenomena would require datasets capable of addressing these specific questions. Additionally, we could not determine the reasons why individual patients did or did not undergo ASCT, and unmeasured variables may have influenced these decisions. Future qualitative studies can explore whether the decision to not pursue ASCT in such situations is led by the physician or by the patient.

In summary, our study did not demonstrate a left digit bias, but it revealed a decline in the use of ASCT as patients aged between 65 and 75 years, with a most pronounced decline between ages 69–71. This trend persisted even after accounting for comorbidities and performance status, highlighting the prominent role of chronological age in treatment decisions for MM. Considering that chronological age often inadequately reflects physiological health [2], as evidenced elsewhere, we suggest making decisions based on alternative measures of frailty and fitness rather than solely relying on chronological age.