跳到正文
原文
Nature 健康护理主题·· 2026-10-06AI 评分59

左位数字偏倚与眼科手术选择:800万患者队列中白内障手术率在年龄跨十年时上升

Cognitive heuristics in the operating room: the role of ‘left-digit bias’ in ophthalmic surgical selection

AI 导读

一项基于TriNetX、纳入8,156,608名40至85岁患者的回顾性队列研究显示,年龄跨入新十年时较年轻组的白内障超声乳化手术几率更低,48/49岁对50/51岁、58/59岁对60/61岁、68/69岁对70/71岁的OR分别为0.956、0.935和0.903,而十年中段的相邻年龄组比较均未达显著性阈值。

正文

Introduction

Cognitive heuristics are simplified mental ‘shortcuts’ that promote efficient decision-making in time-sensitive, complex situations [1]. However, they may also introduce systematic errors or unconscious biases [2]. Cognitive heuristics can be applied to clinical and surgical settings [3], where cognitive biases have been associated with adverse effects on surgical performance and patient outcomes [4].

Left-digit bias (LDB) is a cognitive heuristic in which decision-makers place disproportionate weight on the leftmost digit of a number, potentially leading to shifts in judgment at numerical thresholds. LDB has been well documented in psychology, marketing, and behavioural economics; for example, in ‘99-cent pricing’, consumers may perceive a $12.99 item as meaningfully cheaper than one priced at $13.00 [5].

Though clinicians frequently practice high-stakes decision-making, LDB has not been extensively investigated in many procedural subspecialties. In surgical settings, clinicians may utilise heuristic approaches to process complex information and guide time-sensitive management [6]. For procedural specialties, LDB may skew perception of patient age, introducing a cognitive bias that can impact clinical judgment and the likelihood of intervention. For instance, patients aged 59 vs. 60 years may be viewed differently and receive varying clinical recommendations, though the true difference in age is clinically insignificant.

Studies across several surgical fields suggest that age-related LDB may influence clinical decision-making. An oncologic study by Melucci et al. found that patients with potentially curable rectal cancer who were 60 or 80 years old were less likely to receive guideline-adherent chemotherapy than those aged 58/59 or 78/79, respectively [7] Similar age-boundary patterns have been reported in traumatic brain injury, with higher craniotomy rates among 79- than 80-year- olds [8] In localised prostate cancer, 69-year-olds were more likely to undergo radical prostatectomy than 70-year-olds [9] Furthermore, patients aged 78/79 undergoing lumbar spinal stenosis treatment were more likely than those 80/81 to undergo invasive decompression with instrumentation [10].

Findings suggestive of LDB in other surgical subspecialties support investigation into whether similar patterns occur in ophthalmic care. Although cognitive heuristics have been examined in ophthalmology as potential contributors to diagnostic bias [11], LDB has not been specifically investigated in ophthalmic surgical utilisation. Cataract surgery is particularly relevant because it is among the most common procedures in the United States, exceeding 3.5 million cases annually and projected to increase by more than 70% by 2036 due to an aging population and rising cataract prevalence [12, 13]. Cataract surgery is also elective, minimally invasive, and generally well-tolerated, allowing greater flexibility in surgical timing and potentially creating an opportunity for age-related heuristics to influence surgical decision-making.

There are no established guidelines recommending a specific age range for surgical intervention in adult ophthalmic care, allowing for individualised care based on clinical judgment and shared decision-making. The American Academy of Ophthalmology recommends cataract surgery when visual function no longer meets patient needs and surgery is expected to improve quality of life, regardless of chronological age [14]. Given that clinicians’ recommendations can lead to potentially life-altering interventions, understanding age-related patterns in ophthalmic surgical timing may help identify modifiable factors in clinical decision-making. This study aims to assess whether patterns consistent with LDB are present in the utilisation of common ophthalmic procedures, including phacoemulsification, corneal transplantation, and glaucoma procedures.

Methods

Data source

Data were obtained from the TriNetX U.S. Collaborative Network, a federated network aggregating de-identified electronic health records, including diagnoses, procedures, laboratory values, and healthcare utilisation data, from approximately 129 million patients across participating United States healthcare organisations. All analyses were performed within a single analytic session on July 28, 2026, to ensure internal consistency within the continuously updated platform.

Data collection, processing, deidentification, and transfer complied with applicable data protection regulations. TriNetX is ISO 27001:2013 certified and operates under a comprehensive information security framework to safeguard personal and health data. The study used only de-identified data, involved no direct patient contact, and accessed no identifiable private information. This study was approved by the Albert Einstein College of Medicine Institutional Review Board and conducted in accordance with the Declaration of Helsinki.

Study design and cohort selection

This retrospective cohort study evaluated whether ophthalmic surgical procedure rates differed between adjacent 2-year age cohorts at decade boundaries. Eligible individuals were aged 40 to 85 years and were grouped into consecutive 2-year cohorts (40/41 through 84/85 years).

Eligibility required documentation of a routine general health visit, which served as the index event, and at least 1 ophthalmic examination. Comprehensive eye examinations were identified using CPT and HCPCS codes for dilated ophthalmic evaluation (Table 1). The general health visit was selected as the index event to capture a broad population of adults with baseline healthcare engagement at a given age. Patients were required to have a diagnosis code specific to the surgical outcome under study prior to the corresponding procedure and have at least 1 year of follow-up after the index event. For each surgical outcome, patients with a history of the same procedure before the index date were excluded.

Table 1 CPT and HCPCS-coded ophthalmic examinations, systemic, and ocular covariates.

Full size table

The prespecified decade-boundary comparisons were ages 48/49 vs. 50/51, 58/59 vs. 60/61, and 68/69 vs. 70/71. These comparisons evaluated changes in procedure rates when the leftmost digit of age advanced to the next decade. The remaining comparisons between adjacent 2-year cohorts were evaluated as mid-decade comparisons to distinguish findings at decade boundaries from gradual age-related changes in surgical incidence. All diagnostic and procedural codes used for cohort construction and outcome classification are detailed in Table 2.

Table 2 Coded diagnostic descriptions corresponding to CPT-coded ophthalmic surgery types.

Full size table

Baseline characteristics and propensity score matching

Patients in each adjacent age-cohort comparison were matched in a 1:1 ratio using propensity scores. Baseline covariates included sex, race, ethnicity, healthcare access/utilisation measures, relevant systemic comorbidities and ocular diagnoses, and prior procedures during the 2 years preceding the index event. Systemic and ocular covariates are outlined in Table 1. Healthcare access and utilisation covariates included the number of outpatient visits and ophthalmology services and procedures (TriNetX curated code 1012793). These variables were selected to reduce differences in healthcare engagement, underlying health status, and clinical factors associated with the likelihood of undergoing ophthalmic surgery and were applied uniformly across all analyses.

Propensity score matching was performed using the research platform’s integrated logistic regression algorithm, applying greedy nearest-neighbour matching. Covariate balance was assessed using standardised mean differences, with an absolute standardised mean difference (SMD) of 0.10 or greater interpreted as meaningful imbalance.

Outcomes and follow-up

Ophthalmic surgical outcomes were identified using CPT codes listed in Table 2. The primary outcome was phacoemulsification. Additional outcomes included MIGS, GDI, trabeculectomy, iridotomy, trabeculoplasty, PK, and endothelial keratoplasty (DSAEK/DMEK). Procedures were assessed during the 1-year period after the index event. The overall number of patients undergoing any included ophthalmic procedure was also reported. Patients who underwent >1 included procedure were counted once.

Among patients younger than 60 years, rates of type 1 and type 2 diabetes and mean haemoglobin A1c levels were compared between patients who underwent phacoemulsification and those who did not.

Statistical analysis

Procedure counts and proportions were calculated for each 2-year age cohort. Younger and older adjacent cohorts were compared using chi-square tests, and odds ratios (ORs) with 95% confidence intervals (CIs) were calculated for each surgical outcome. An OR below 1 indicated lower odds of surgery in the younger cohort relative to the adjacent older cohort. The same analytic approach was applied to decade-boundary and mid-decade comparisons across all procedure types. Rates of type 1 and type 2 diabetes were compared using chi-square tests, and mean haemoglobin A1c levels were compared using independent-samples t-tests. A 2-sided P < 0.001 was considered statistically significant. Descriptive procedure rates were calculated using the full eligible cohorts, whereas ORs and comparison-specific procedure rates were calculated after 1:1 propensity score matching.

Power analysis

An a priori power analysis was performed using estimated 1-year phacoemulsification rates of 3.73% among adults younger than 60 years and 4.47% among adults aged 60 years or older. Based on a 2-sample comparison of proportions with a significance level of 0.001 and 80% power, 24,411 patients per cohort (48,822 patients overall) were required. The available sample exceeded this threshold by several orders of magnitude, confirming adequate power to detect age-related differences in phacoemulsification rates.

Results

Procedure rates

Across 23 consecutive 2-year age cohorts, 8,156,608 patients met eligibility criteria, of whom 218,816 underwent at least 1 included ophthalmic procedure within 1 year of the index event. Phacoemulsification was performed in 201,591 patients, accounting for 92.1% of patients who underwent an included procedure. For each comparison between adjacent 2-year age cohorts, 1:1 propensity score matching achieved adequate balance across included baseline covariates, with all absolute SMDs <0.10. The proportion of patients undergoing any included surgery increased from 2.01% at ages 40/41 to a peak of 3.79% at ages 78/79, followed by a decline among the oldest cohorts. Phacoemulsification rates increased from 3.54% at ages 40/41 to 3.83% at ages 58/59 and reached a peak of 5.26% at ages 74/75 before declining at older ages. MIGS rates ranged from 0.24% to 1.41%, with the highest rate observed at ages 78/79, while GDI rates ranged from 0.13% to 0.43%. PK rates ranged from 5.36% to 16.20%, and DSAEK/DMEK rates ranged from 6.11% to 15.04% (Fig. 1).

Fig. 1: Baseline Selected Ophthalmic Procedure Rates Across 2-Year Age Groups.

Surgery rates for cataract surgery (A) and glaucoma and corneal procedures (B) were calculated in the full eligible cohorts across age groups from 40/41 through 84/85.

Phacoemulsification procedures

A) Expanded mid-decade analysis

Across expanded mid-decade age group comparisons, phacoemulsification rates generally increased with age (Fig. 1). Comparisons between adjacent mid-decade cohorts from ages 44/45 through 74/75 yielded ORs near 1.0, with none meeting the prespecified significance threshold of P < 0.001 (Fig. 2).

Fig. 2: Phacoemulsification Procedure Rates and Odds Ratios Across Adjacent 2-Year Age Cohorts.

Procedure rates and odds ratios for phacoemulsification were calculated in the propensity score-matched adjacent age cohorts: A 44 to 55 years old, B 54 to 65 years old, and C 64 to 75 years old, demonstrating a significant change only when the left-digit advances (boxed in red). D Surgeons were less likely to perform phacoemulsification on younger patients immediately before the left-digit increases numerically, as evidenced by a lower OR at each turn of decade.

B) Two-year age end-of-decade cohort comparisons

By contrast, at several age thresholds when the left-most digit of a patient’s age advanced numerically (48/49 vs. 50/51, 58/59 vs. 60/61, and 68/69 vs. 70/71), patients in the younger age cohort were consistently less likely to undergo phacoemulsification compared to their older counterparts. In the propensity score-matched cohorts, phacoemulsification occurred in 3.64% of patients aged 48/49 compared with 3.80% of those aged 50/51 (OR, 0.956; 95% CI, 0.926–0.986), 3.83% aged 58/59 compared with 4.09% aged 60/61 (OR, 0.935; 95% CI, 0.909–0.961), and 4.54% aged 68/69 compared with 5.00% aged 70/71 (OR, 0.903; 95% CI, 0.879–0.927; all P < 0.001) (Fig. 2).

C) Diabetes rates across younger patient cohorts undergoing phacoemulsification

Mean haemoglobin A1c levels and rates of type 1 and type 2 diabetes were compared between patients younger than 60 years who underwent phacoemulsification and those who did not. Across all age cohorts from 40/41 through 58/59 years, patients who underwent phacoemulsification had higher rates of diabetes, with 43–48% having type 2 diabetes and 11–24% having type 1 diabetes. Mean haemoglobin A1c levels were consistently higher among patients who underwent phacoemulsification (all P < 0.001).

Glaucoma procedures

A) Expanded mid-decade analysis

The mid-decade comparisons for MIGS and GDI procedures were variable and did not demonstrate a consistent overall trend. For MIGS, patients in the younger cohort had 6.3% lower odds of undergoing surgery at ages 54/55 vs. 56/57 (OR, 0.937; 95% CI, 0.906–0.971) and 7.6% lower odds at ages 56/57 vs. 58/59 (OR, 0.924; 95% CI, 0.882–0.969), but 4.1% higher odds at ages 60/61 vs. 62/63 (OR, 1.041; 95% CI, 1.018–1.061; all P < 0.001). For GDI, the only significant mid-decade comparison occurred at ages 60/61 vs. 62/63, with lower odds among patients in the younger cohort (OR, 0.927; 95% CI, 0.891–0.963; P < 0.001) (Fig. 3A, B).

Fig. 3: Odds Ratios for Glaucoma and Corneal Procedures Across Adjacent 2-Year Age Cohorts.

A Patients aged 48/49 years had lower odds of undergoing MIGS procedures than patients aged 50/51 years. Significant differences were also observed for 54/55 vs. 56/57 and 56/57 vs. 58/59 age-group comparisons, suggesting that these findings were not exclusively observed at decade-boundary transitions when the leftmost digit advanced. B–D For GDI and keratoplasty procedures, differences in the odds of undergoing surgery were not consistently associated with advancement of the leftmost age digit.

B) Two-year age end-of-decade cohort comparisons

Glaucoma surgery rates were generally comparable between younger and adjacent older cohorts at decade boundaries. Patients aged 48/49 had 12.4% lower odds of undergoing MIGS procedures than those aged 50/51 (OR, 0.876; 95% CI, 0.814–0.940; P < 0.001), while no significant differences were observed at 58/59 vs. 60/61 or 68/69 vs. 70/71 (Fig. 3A). GDI procedures demonstrated no significant differences at decade boundaries (Fig. 3B). Among the additional glaucoma procedures, including trabeculectomy, iridotomy, and trabeculoplasty, no clear end-of-decade pattern was demonstrated, although isolated significant differences were observed. Patients aged 48/49 had 15.4% lower odds of undergoing trabeculectomy than those aged 50/51 (OR, 0.846; P < 0.001). Patients aged 58/59 also had lower odds than those aged 60/61 of undergoing iridotomy and trabeculoplasty (OR, 0.822 and 0.873, respectively; both P < 0.001).

Corneal transplantation procedures

A) Expanded mid-decade analysis

PK demonstrated no significant differences across the mid-decade comparisons. For DSAEK/DMEK, patients in the younger cohort had higher odds of undergoing surgery at ages 60/61 vs. 62/63 (OR, 1.043; 95% CI, 1.022–1.071), 64/65 vs. 66/67 (OR, 1.048; 95% CI, 1.024–1.073), and 66/67 vs. 68/69 (OR, 1.036; 95% CI, 1.013–1.051; all P < 0.001). No consistent pattern was observed across the remaining mid-decade comparisons (Fig. 3C, D).

B) Two-year age end-of-decade cohort comparisons

Neither PK nor DSAEK/DMEK demonstrated a significant difference at any decade boundaries (Fig. 3C, D).

Discussion

Our results suggest that phacoemulsification may exhibit age-related patterns in surgical utilisation, consistent with LDB. Specifically, patients in younger age cohorts had lower odds of undergoing phacoemulsification at several decade boundaries (48/49 vs. 50/51, 58/59 vs. 60/61, and 68/69 vs. 70/71), whereas no consistent end-of-decade pattern was observed for glaucoma or corneal procedures. The consistent end-of-decade pattern observed for phacoemulsification, together with the absence of significant mid-decade differences, is consistent with LDB, although age-related healthcare utilisation, referral patterns, and clinical characteristics may also contribute.

Although cognitive heuristics and biases have been recognised in ophthalmic decision making, LDB has not been specifically investigated in ophthalmic surgical utilisation to our knowledge [11]. Our study provides an initial hypothesis-generating exploration of whether age-related patterns consistent with LDB may occur in ophthalmic surgery. While these findings do not establish LDB as the underlying mechanism, they identify a potential area for further investigation into how cognitive heuristics may interact with clinical and patient factors in determining surgical timing.

Phacoemulsification procedures

Phacoemulsification rates increased steadily with patient age, consistent with the age-related prevalence of cataracts. However, across several age cohorts, when the leftmost digit of a patient’s age advanced into the next decade (i.e. 48/49 to 50/51), patients at the beginning of a new decade were significantly more likely to undergo surgery than those at the end of the previous decade. These discontinuities were observed at three specific decade boundaries rather than uniformly across the age distribution, supporting a pattern consistent with LDB.

The corresponding ORs (0.956, 0.935, and 0.903) represent relatively small individual patient-level effects, corresponding to approximately 4.4%, 6.5%, and 9.7% lower odds of undergoing phacoemulsification among patients in the younger age cohorts, respectively. Given the exploratory nature of this study, these findings should be interpreted cautiously as hypothesis-generating rather than definitive evidence of LDB. However, although these absolute differences were modest, the high volume of phacoemulsification performed in the U.S. could mean that small differences in surgical likelihood may have measurable population-level effects.

Several characteristics of cataract surgery may create circumstances in which age-related cognitive heuristics could influence surgical timing. Unlike many other ophthalmic pathologies, cataracts rarely require urgent intervention, and cataract extraction is typically elective, with surgical indications primarily driven by patient-reported symptoms and perceived impact of visual impairment on daily life [15]. Cataracts also tend to progress gradually, allowing flexibility in surgical timing through shared decision-making between patients and clinicians. Cataract surgery is generally a well-tolerated ambulatory procedure, often under local anaesthesia with mild sedation, and has an excellent safety profile in even advanced age groups, with low complication rates [16] and rapid visual recovery often within the first postoperative week [17].

These favourable characteristics, combined with the age-related nature of cataract development, may create circumstances in which elective cataract surgery is perceived as more appropriate at older ages. Transitions into a new decade (i.e., 58/59 to 60/61 years) may coincide with subtle shifts in how clinicians and patients conceptualise age, despite minimal clinical differences between adjacent cohorts.

The observed LDB-consistent association between decade age boundaries and phacoemulsification rates may have important clinical implications. As cataract extraction has been associated with improved quality of life and reduced risk of injury from falls and motor vehicle accidents [15], patients should ideally be offered the opportunity to undergo surgical intervention regardless of age if clinically appropriate. Recognising potential age-related patterns in surgical utilisation may help clinicians prioritise shared decision-making and promote recommendations for cataract extraction that are guided by objective clinical data and patient-centred outcomes rather than age-related expectations alone.

Glaucoma procedures

In contrast to phacoemulsification, glaucoma procedures did not demonstrate a consistent decade-boundary pattern. This may reflect differences in the clinical indications and time-sensitive nature of glaucoma surgery. Glaucoma treatment decisions are typically informed by objective clinical markers such as intraocular pressure (IOP) and visual field changes. Because these factors may vary across glaucoma procedures, we considered MIGS and GDI separately.

Glaucoma drainage implants (GDI)

GDIs are indicated in advanced or rapidly progressive glaucoma requiring substantial IOP reduction to prevent further optic nerve damage. The decision to perform GDI procedures is largely guided by disease severity, progression, and target IOP, which may limit the extent to which age-related patterns influence surgical timing. Consistent with this, we observed no significant decade-boundary differences in the likelihood of undergoing GDI procedures.

Minimally invasive glaucoma surgery (MIGS)

MIGS is generally used earlier in the glaucoma disease course to improve IOP control and is often performed concurrently with cataract surgery. Although MIGS provides less IOP reduction than other glaucoma procedures, its favourable safety profile often makes it an ideal option for patients with milder disease [18, 19]. MIGS did not demonstrate a consistent decade-boundary pattern, although significant differences were observed in several mid-decade comparisons. This suggests that age-related differences in MIGS utilisation may be variable rather than concentrated at decade boundaries. Because MIGS is typically performed concurrently with cataract surgery [20], its timing may reflect cataract-related indications in addition to glaucoma-specific factors, complicating interpretation. Given the absence of a consistent decade-boundary pattern, glaucoma severity may play a greater role in MIGS timing than age-based thresholds. Future studies distinguishing combined from standalone MIGS procedures could further clarify these observed patterns.

Corneal transplantation procedures

Similar to glaucoma procedures, corneal transplantation did not demonstrate a consistent decade-boundary pattern, although significant mid-decade differences were observed for DSAEK/DMEK. Higher overall surgical rates for PK and DSAEK/DMEK relative to phacoemulsification, MIGS, and GDI likely reflect their smaller, diagnosis-restricted eligible denominator rather than greater surgical volume [21,22,23]. These findings suggest that corneal transplantation timing may be more closely associated with the extent of underlying corneal pathology [24] than with perceived patient age.

Although age generally does not preclude corneal transplantation, age-related differences may influence postoperative outcomes. Among patients older than 65, DSAEK and PK show comparable graft survival and rejection rates across age cohorts [25]. However, older DMEK recipients may have a higher prevalence of comorbidities such as pseudophakic bullous keratopathy that increase risk of graft failure [26]. Thus, while age may influence outcomes and perioperative considerations, the absence of a consistent decade-boundary pattern suggests that age thresholds alone are unlikely to drive corneal transplantation timing.

Future directions and limitations

Further research is needed to clarify the potential role of LDB in ophthalmic surgery, particularly for common elective procedures where surgical timing may be influenced by several patient and clinical factors. Studies using more granular clinical data could help determine whether these patterns reflect age-related cognitive heuristics, variation in clinical decision-making, patient preferences, or other factors influencing surgical timing. For instance, qualitative surveys of ophthalmic surgeons and patients could evaluate awareness of LDB and explore factors influencing the timing of cataract surgery. Validation studies using other large databases and diverse healthcare settings could further assess the generalisability of these findings.

Our study has several important limitations. The TriNetX database allowed for cohort matching by sex, race, ethnicity, healthcare access and utilisation, relevant systemic comorbidities, and ocular diagnoses and prior procedures. Nevertheless, residual confounding from unmeasured clinical and healthcare-related factors remains possible. Because TriNetX does not capture clinician recommendations, patient preferences, disease severity, or reasons for performing or deferring a procedure, the underlying mechanisms cannot be established. The observed pattern should therefore be interpreted as consistent with, rather than proof of, LDB. Moreover, TriNetX does not capture patient-level insurance or payer type, precluding direct verification or matching of Medicare enrolment status. Although the study was adequately powered, procedure-specific sample sizes varied between cohorts and needed to be accounted for in our analysis. Lastly, the retrospective design is susceptible to misclassification errors from inaccuracies in ICD diagnostic and CPT procedural coding.

Conclusions

Our findings suggest that phacoemulsification may exhibit greater susceptibility to age-related patterns in surgical utilisation, consistent with LDB, compared with other ophthalmic procedures. This difference may reflect the elective nature of cataract surgery, reliance on patient-reported symptoms, and the role of shared decision-making in determining surgical timing. In contrast, the timing of corneal and glaucoma surgeries may be more strongly guided by objective clinical measures.

The observed pattern of phacoemulsification utilisation raises the possibility that age-related cognitive heuristics may contribute to variation in surgical timing, despite minimal clinical differences between adjacent age cohorts. Although the underlying mechanisms cannot be determined, potential variation in surgical timing warrants consideration. Greater awareness of potential age-related patterns consistent with LDB in ophthalmic surgery may help physicians recognise factors that could influence treatment timing and support patient-centred care. As our population ages and demand for cataract extraction increases, even small differences in surgical timing could have meaningful cumulative effects on visual outcomes and patient satisfaction at the population level.

What is known about this topic

  • Left-digit bias is a cognitive heuristic in which the leftmost digit of a number disproportionally influences judgment. In medical decision-making, this may affect perceptions of patient age despite clinically negligible differences between adjacent ages.

  • Left-digit bias has been associated with treatment decisions in other medical and surgical contexts, including orthopaedic surgery and oncology.

  • Whether left-digit bias exists in ophthalmic surgical decision-making has not been previously studied.

What this study adds

  • In a cohort of 8,156,608 eligible patients, including 201,591 who underwent phacoemulsification, younger patients had lower odds of undergoing phacoemulsification at all three prespecified decade boundaries, whereas no corresponding mid-decade comparisons were significant. These findings are consistent with a potential left-digit bias effect, although the observed effect sizes were modest.

  • Left-digit bias was generally not observed for common glaucoma and corneal transplantation procedures. This may reflect differences in surgical indications and timing, where objective measures such as clinical urgency and extent of disease guide treatment more closely.

  • These findings identify a potential age-related pattern in ophthalmic surgical utilisation, particularly in high-volume elective procedures like phacoemulsification, and provide a basis for further investigation using more granular clinical data to distinguish left-digit bias from patient, clinician, or disease-related factors.

Data availability

Individual-level data are not publicly available because of patient privacy protections and data use agreements. De-identified, population-level aggregate data supporting the findings of this study may be accessed through the research network by qualified researchers who establish an appropriate license agreement.

References

  1. Marewski JN, Gigerenzer G. Heuristic decision making in medicine. Dialogues Clin Neurosci. 2012;14:77–89. https://doi.org/10.31887/DCNS.2012.14.1/jmarewski.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Gigerenzer G, Gaissmaier W. Heuristic decision making. Annu Rev Psychol. 2011;62:451482 https://doi.org/10.1146/annurev-psych-120709-145346.

    Article  Google Scholar 

  3. Aylmore H, Agarwal S, Marcus HJ, Pandit AS. Cognitive biases and heuristics in surgical settings: a systematic review. Ann Surg. 2025;282:946–953. https://doi.org/10.1097/SLA.0000000000006736.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Armstrong BA, Dutescu IA, Tung A, Carter D, Trbovich P, Wong S, et al. Cognitive biases in surgery: systematic review. Br J Surg. 2023;110:645–54. https://doi.org/10.1093/bjs/znad004.

    Article  PubMed  Google Scholar 

  5. List JA, Muir I, Pope D, Sun G. Left-digit bias at Lyft. Rev Econ Stud. 2023;90:31863237 https://doi.org/10.1093/restud/rdad014.

    Article  Google Scholar 

  6. Mangus CW, Mahajan P. Decision making: healthy heuristics and betraying biases. Crit Care Clin. 2022;38:37–49. https://doi.org/10.1016/j.ccc.2021.07.002.

    Article  PubMed  Google Scholar 

  7. Melucci AD, Loria A, Ramsdale E, Temple LK, Fleming FJ, Aquina CT. An assessment of left-digit bias in the treatment of older patients with potentially curable rectal cancer. Surgery. 2022;172:851–8. https://doi.org/10.1016/j.surg.2022.04.038.

    Article  PubMed  Google Scholar 

  8. Blumenfeld A, Sang HI, Baird R, Brenden M, Bjordahl P. 79 is the new 70: left digit bias in craniotomy. Am J Surg. 2022;224:1442–4. https://doi.org/10.1016/j.amjsurg.2022.10.008.

    Article  PubMed  Google Scholar 

  9. Brant A, Lewicki P, Wu X, Sze C, Johnson J, Basourakos S, et al. Impact of left-digit age bias in the treatment of localized prostate cancer. J Urol. 2022;208:997–1006. https://doi.org/10.1097/JU.0000000000002857.

    Article  PubMed  Google Scholar 

  10. Akosman I, Shafi K, Subramanian T, Cunningham M, Kim HJ, Lovecchio F. Left-digit bias in surgical decision-making for lumbar spinal stenosis. Spine J. 2024;24:1388–95. https://doi.org/10.1016/j.spinee.2024.03.007.

    Article  PubMed  Google Scholar 

  11. Hussain A, Oestreicher J. Clinical decision-making: heuristics and cognitive biases for the ophthalmologist. Surv Ophthalmol. 2018;63:119–24. https://doi.org/10.1016/j.survophthal.2017.08.007.

    Article  PubMed  Google Scholar 

  12. Chen SP, Woreta F, Chang DF. Cataracts: a review. JAMA. 2025;333:2093–103. https://doi.org/10.1001/jama.2025.1597.

    Article  CAS  PubMed  Google Scholar 

  13. Lin L, Liang Y, Jiang G, Gan Q, Yang T, Liao P, et al. Global, regional, and national burden of cataract: a comprehensive analysis and projections from 1990 to 2021. PLoS ONE. 2025;20:e0326263 https://doi.org/10.1371/journal.pone.0326263.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. American Academy of Ophthalmology. Cataract/Anterior Segment Summary Benchmark - 2024. Accessed January 22, 2026. https://www.aao.org/education/summary-benchmark-detail/cataract-anterior-segmentsummary-benchmark-2020

  15. Miller KM, Oetting TA, Tweeten JP, Carter K, Lee B, Lin BS, et al. Cataract in the adult eye preferred practice pattern®. Ophthalmology. 2022;129:P1–P126. https://doi.org/10.1016/j.ophtha.2021.10.006.

    Article  PubMed  Google Scholar 

  16. Perumal D, Dudley RA, Gan S, Boscardin WJ, Gill A, Gelb A, et al. Anesthesia care for cataract surgery in Medicare beneficiaries. JAMA Intern Med. 2022;182:1171–80. https://doi.org/10.1001/jamainternmed.2022.4333.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Cicinelli MV, Buchan JC, Nicholson M, Varadaraj V, Khanna RC. Cataracts. Lancet. 2023;401:377–89. https://doi.org/10.1016/S0140-6736(22)01839-6.

    Article  PubMed  Google Scholar 

  18. American Academy of Ophthalmology. Reporting clinical endpoints in studies of minimally invasive glaucoma surgery - 2025. Accessed January 22, 2026. https://www.aao.org/education/clinical-statement/reporting-clinicalendpoints-in-studies-of-minimal

  19. Jayaram H, Kolko M, Friedman DS, Gazzard G. Glaucoma: now and beyond. Lancet. 2023;402:1788–801. https://doi.org/10.1016/S0140-6736(23)01289-8.

    Article  CAS  PubMed  Google Scholar 

  20. Gedde SJ, Vinod K, Wright MM, Muir KW, Lind JT, Chen PP, et al. Primary open-angle glaucoma preferred practice pattern®. Ophthalmology. 2021;128:P71–P150. https://doi.org/10.1016/j.ophtha.2020.10.022.

    Article  PubMed  Google Scholar 

  21. Xiao G, Tsou BC, Soiberman US, Prescott CR, Srikumaran D, Woreta FA. Keratoplasty in the United States: trends and indications from 2015 to 2020. Cornea. 2023;42:1360–4. https://doi.org/10.1097/ICO.0000000000003180.

    Article  PubMed  Google Scholar 

  22. Tan DTH, Dart JKG, Holland EJ, Kinoshita S. Corneal transplantation. Lancet. 2012;379:1749–61. https://doi.org/10.1016/S0140-6736(12)60437-1.

    Article  PubMed  Google Scholar 

  23. Stuart AJ, Romano V, Virgili G, Shortt AJ. Descemet’s membrane endothelial keratoplasty (DMEK) versus Descemet’s stripping automated endothelial keratoplasty (DSAEK) for corneal endothelial failure. Cochrane Database Syst Rev. 2018;6:CD012097. https://doi.org/10.1002/14651858.CD012097.pub2.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Fuest M, Ang M, Htoon HM, Tan D, Mehta JS. Long-term visual outcomes comparing Descemet stripping automated endothelial keratoplasty and penetrating keratoplasty. Am J Ophthalmol. 2017;182:62–71. https://doi.org/10.1016/j.ajo.2017.07.014.

    Article  PubMed  Google Scholar 

  25. Duman F, Kosker M, Suri K, Reddy JC, Ma JF, Hammersmith KM, et al. Indications and outcomes of corneal transplantation in geriatric patients. Am J Ophthalmol. 2013;156:600–.e2. https://doi.org/10.1016/j.ajo.2013.04.034.

    Article  PubMed  Google Scholar 

  26. Achiron A, Nahum Y, Tiosano A, Mashor RS, Tuuminen R, Elbaz U, et al. Descemet membrane endothelial keratoplasty outcomes between young and old graft recipients. Curr Eye Res. 2021;46:1460–6. https://doi.org/10.1080/02713683.2021.1899245.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

The authors would like to thank Jeffrey S. Schultz for his assistance with review, editing, and manuscript preparation. We also acknowledge the support of the Department of Ophthalmology at Albert Einstein College of Medicine for providing the resources and clinical environment that made this work possible.

Author information

Authors and Affiliations

  1. Department of Ophthalmology, Albert Einstein College of Medicine, New York, USA

    Kara Sangiuolo, Rachel Zhang, Jainam Shah, Lily Young, Zara Mian, Jee-Young Moon, Jeffrey Schultz & Anurag Shrivastava

Authors

  1. Kara Sangiuolo
  2. Rachel Zhang
  3. Jainam Shah
  4. Lily Young
  5. Zara Mian
  6. Jee-Young Moon
  7. Jeffrey Schultz
  8. Anurag Shrivastava

Contributions

KS, RZ, ZM, and AS contributed to the conception and design of the study. JS and JYM contributed to data collection, data analysis, and interpretation. KS, RZ, JS, and AS drafted the manuscript and created figures and tables. LY and AS critically revised the manuscript for important intellectual content. All authors reviewed and approved the final manuscript and agree to be accountable for all aspects of the work.

Corresponding author

Correspondence to Anurag Shrivastava.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Sangiuolo, K., Zhang, R., Shah, J. et al. Cognitive heuristics in the operating room: the role of ‘left-digit bias’ in ophthalmic surgical selection. Eye (2026). https://doi.org/10.1038/s41433-026-04926-y

Download citation

  • Received: 22 April 2026

  • Revised: 22 August 2026

  • Accepted: 15 September 2026

  • Published: 06 October 2026

  • Version of record: 06 October 2026

  • DOI: https://doi.org/10.1038/s41433-026-04926-y

来源:Nature 健康护理主题 · nature.com