What left-digit bias means for a patient's age
Left-digit bias is a cognitive shortcut in which the leftmost digit of a number carries disproportionate weight in judgment. The paper's own illustration comes from pricing: a $12.99 item can feel meaningfully cheaper than one priced at $13.00, even though the gap is one cent. Applied to age, a patient of 59 and a patient of 60 differ by one year, but the first digit changes from 5 to 6.
The study asks whether that digit change shows up in ophthalmic surgical selection. If clinicians anchor on the decade rather than the exact year, procedure rates should shift when patients cross 50, 60, or 70, and should not shift at comparable boundaries inside a decade. That contrast between decade boundaries and mid-decade comparisons is the paper's central test.
How the cohort and matching were built
The authors used the TriNetX U.S. Collaborative Network, a federated electronic health record network aggregating de-identified records from approximately 129 million patients across participating US healthcare organisations. All analyses ran in a single session on July 28, 2026.
Eligible patients were 40 to 85 years old, had a documented routine general health visit as the index event, at least one ophthalmic examination, a diagnosis code specific to the surgical outcome being studied, and at least one year of follow-up. Patients with a prior history of the same procedure before the index date were excluded.
Patients were grouped into consecutive 2-year cohorts from 40/41 through 84/85, giving 23 cohorts. Each adjacent pair was matched 1:1 on demographic characteristics, healthcare utilisation, systemic comorbidities, and ocular diagnoses and procedures. The authors report that all absolute standardised mean differences after matching were below 0.10.
Three decade-boundary comparisons were prespecified: 48/49 versus 50/51, 58/59 versus 60/61, and 68/69 versus 70/71. All other adjacent 2-year comparisons served as mid-decade controls. This design matters because a genuine age trend should appear in both sets; a left-digit pattern should appear mainly at the decade transitions.
Cataract surgery odds rise at each decade line
Of 8,156,608 eligible patients, 218,816 (2.68%) underwent at least one included ophthalmic procedure within one year of the index event. Phacoemulsification accounted for 201,591 of these patients, or 92.1% of those who had any included procedure.
At every prespecified decade boundary, the younger cohort had lower odds of phacoemulsification. The matched rates were 3.64% at ages 48/49 versus 3.80% at 50/51 (OR 0.956; 95% CI 0.926–0.986), 3.83% at 58/59 versus 4.09% at 60/61 (OR 0.935; 95% CI 0.909–0.961), and 4.54% at 68/69 versus 5.00% at 70/71 (OR 0.903; 95% CI 0.879–0.927). All three were significant at P < 0.001.
The direction is consistent, and the odds ratio moves further from 1.0 at each successive boundary, from 0.956 to 0.935 to 0.903. In absolute terms, however, the gaps are small: roughly 0.16 to 0.46 percentage points in matched procedure rates. The paper itself describes the observed effect sizes as modest.
Mid-decade comparisons show no such step
The mid-decade comparisons provide the counterweight. Across adjacent 2-year cohorts from 44/45 through 74/75, phacoemulsification odds ratios were near 1.0, and none met the prespecified significance threshold of P < 0.001.
This is the result that separates a left-digit interpretation from a simple age gradient. Phacoemulsification rates generally rose with age across the full range, from 3.54% at ages 40/41 to a peak of 5.26% at 74/75 before declining in older cohorts. A smooth age trend would predict steps at mid-decade boundaries too. The paper reports steps only where the leading digit changes.
The authors also examined patients younger than 60 who underwent phacoemulsification and found higher rates of type 1 and type 2 diabetes and consistently higher mean haemoglobin A1c than among those who did not have surgery, with all P < 0.001. That comparison describes the clinical profile of operated patients; it is not a test of the decade-boundary hypothesis.
Glaucoma and corneal procedures diverge from cataract surgery
The decade-boundary pattern was not general across ophthalmic surgery. For minimally invasive glaucoma surgery, patients aged 48/49 had 12.4% lower odds than those aged 50/51 (OR 0.876; 95% CI 0.814–0.940), but no significant differences appeared at 58/59 versus 60/61 or 68/69 versus 70/71.
Glaucoma drainage implants showed no significant differences at any decade boundary. Neither penetrating keratoplasty nor endothelial keratoplasty (DSAEK/DMEK) showed a significant difference at any decade boundary.
The mid-decade results for these procedures were mixed rather than clean. For MIGS, younger cohorts had lower odds at 54/55 versus 56/57 (OR 0.937) and at 56/57 versus 58/59 (OR 0.924), but higher odds at 60/61 versus 62/63 (OR 1.041). For glaucoma drainage implants, the only significant mid-decade comparison was at 60/61 versus 62/63 (OR 0.927). For DSAEK/DMEK, younger cohorts had higher odds at three mid-decade comparisons.
The authors interpret this divergence as reflecting differences in surgical indication. Cataract surgery is elective, depends on patient-reported symptoms, and involves shared decision-making about timing, whereas glaucoma and corneal procedures may be guided more closely by objective clinical measures such as urgency and extent of disease. That explanation is the authors' interpretation, not a measured mechanism in the data.
What the design can and cannot attribute
The study is a retrospective cohort analysis of de-identified records. It can show that procedure frequency changes at decade boundaries after matching on measured covariates; it cannot show that clinicians consciously weighed the leading digit of age. The authors state that the underlying mechanisms cannot be determined.
Several alternative explanations remain compatible with the data. Patients just past a decade may differ in unmeasured ways, including symptom severity, visual acuity, cataract density, patient preference, insurance details, or referral patterns. The matching covered demographic characteristics, healthcare utilisation, systemic comorbidities, and ocular diagnoses and procedures, but not granular clinical measurements.
The authors call for more granular clinical data, qualitative surveys of ophthalmic surgeons and patients, and validation in other large databases and diverse healthcare settings to distinguish left-digit bias from patient, clinician, or disease-related factors. Individual-level data are not publicly available; aggregate data may be accessed by qualified researchers through the network under a license agreement.
Where the decade-boundary signal matters most
The paper's most decision-relevant comparison is internal: phacoemulsification shows a consistent step at all three decade boundaries with no matching mid-decade steps, while glaucoma drainage implants and both corneal transplant procedures show no decade-boundary differences at all. MIGS sits in between, with a single significant boundary at 48/49 versus 50/51.
That split points to a specific priority rather than a general warning about bias. The signal concentrates in the highest-volume elective procedure in the dataset, where 201,591 of 218,816 operated patients had phacoemulsification, and where surgical timing is negotiated rather than dictated by an urgent clinical finding.
The practical threshold the paper supports is awareness at the decade transition itself. A clinician evaluating a 49-, 59-, or 69-year-old for cataract surgery has no clinically meaningful age difference from a patient one or two years older, yet the data show lower odds of surgery on the younger side of each line. The authors note that as the population ages and demand for cataract extraction increases, even small differences in surgical timing could accumulate at the population level. Whether that accumulation occurs is a question this dataset raises, not one it answers.
Sources used in this answer
Cognitive heuristics in the operating room: the role of ‘left-digit bias’ in ophthalmic surgical selection
Younger patients had lower odds of phacoemulsification at all three prespecified decade boundaries, with no significant mid-decade comparisons, while glaucoma drainage implants and corneal transplants showed no decade-boundary differences.