Risk factor profiles and visual outcomes of non-arteritic anterior ischemic optic neuropathy in younger Chinese patients

Abstract

Background

Non-arteritic anterior ischemic optic neuropathy (NAION) predominantly affects the elderly, although onset before age 50 is not uncommon. Systemic vascular diseases are the main risk factors in elderly patients, whereas ocular structural abnormalities may be associated with early-onset NAION. This study compared risk factor profiles and visual outcomes between young NAION patients and a selected contemporary comparison group of older patients consisting of consecutively enrolled cases within a predefined study period in China, to assess age-related differences.

Methods

In this single-center retrospective cohort study, patients diagnosed with NAION enrolled between May 2022 and February 2026 were stratified into a young group (aged < 50 years, n = 102) and a selected contemporary comparison group of older patients (aged ≥ 50 years, n = 111). Crucially, this older comparison group comprised consecutively enrolled cases within this predefined study period, rather than representing a random sample of all eligible older NAION patients. All patients underwent 6-month follow-up. Demographic, systemic, and ocular data, including best-corrected visual acuity (BCVA), visual field index (VFI), mean deviation (MD), and pattern standard deviation (PSD), were collected. Intergroup and intragroup comparisons were performed, and multivariable linear regression using a hybrid selection strategy (mandatory clinical covariates plus univariate screening) identified independent predictors of visual recovery.

Results

The young group included more males, had fewer bilateral cases, and exhibited higher smoking rates compared with this older comparison cohort (all P < 0.05). Conversely, this older comparison cohort had a higher prevalence of hypertension, cardiovascular disease, and peripapillary hyperreflective ovoid mass-like structure (PHOMS) (all P < 0.05). Inferior hemifield defects predominated in young patients, whereas diffuse defects were more common in this older comparison cohort. Significant within-group improvements in BCVA, VFI, MD and PSD were observed only in the young group (all P < 0.01), with between-group differences reaching significance only for MD. Multivariable analysis showed that younger age, shorter onset-to-presentation time, and better baseline BCVA were factors associated with greater BCVA recovery, though these associations should be interpreted cautiously given study limitations.

Conclusions

Younger and older NAION patients exhibit distinct risk profiles and visual field defect patterns. While younger patients demonstrated significant within-group visual improvement, only the change in mean deviation (MD) differed significantly between groups, necessitating cautious interpretation of age-related recovery differences.

Introduction

Non-arteritic anterior ischemic optic neuropathy (NAION) is a common acute optic nerve disorder in individuals over 50 years of age and represents a leading cause of irreversible vision loss in the middle-aged and elderly population [1]. Although NAION predominantly affects this demographic, cases in patients under 50 years of age do occur [2]. Currently, no proven effective treatment exists for NAION, and visual loss is typically permanent. Consequently, the cornerstone of existing management strategies involves identifying and controlling potential modifiable risk factors to mitigate the risk of fellow-eye involvement.

The exact pathogenesis of NAION remains elusive; however, it typically manifests in individuals over 50 years of age who present with a small optic cup, a crowded optic disc, and coexisting vascular risk factors [3]. Common risk factors include systemic vascular conditions, such as hypertension, diabetes, hyperlipidemia, obstructive sleep apnea (OSA), and smoking, as well as local anatomical anomalies like optic disc drusen (ODD) [4, 5]. In younger patients with NAION, systemic vascular risk factors are often less prominent, and the clinical presentation is frequently characterized by the presence of ODD-associated disc crowding. While peripapillary hyperreflective ovoid mass-like structure (PHOMS)—a sign of axoplasmic stasis—has been reported in association with younger cohorts [5], emerging evidence suggests that PHOMS are dynamic structures whose visibility is highly dependent on the disease stage, potentially shrinking or resolving as optic disc edema subsides. Consequently, observed variations in PHOMS prevalence across different age groups may reflect differences in symptom duration and disease phase rather than age itself serving as a direct determinant. Given that the differences in risk factor profiles and visual outcomes between younger and older NAION patients remain poorly defined, this study aims to characterize the distinct risk factor profiles and visual outcomes of young NAION patients in China, by comparing them with a selected contemporary comparison group of older patients enrolled consecutively within a recent defined time window, using a retrospective cohort analysis.

Materials and methods

Ethical approval

This retrospective study was approved by the Ethics Committee of Jinan Mingshui Eye Hospital (Approval No. 2025-005-01) and adhered to the tenets of the Declaration of Helsinki. As this was a retrospective study utilizing de-identified patient data, the requirement for informed consent was waived by the ethics committee.

Study population and sampling strategy

The study period spanned from May 2022 to February 2026, during which a total of 550 patients with NAION meeting the inclusion criteria were identified from the electronic medical record system of Jinan Mingshui Eye Hospital. The cohort was stratified into young (< 50 years) and older (≥ 50 years) subgroups based on age at symptom onset. While the young group included all eligible patients identified throughout the entire study period, the older subgroup was defined as a selected contemporary comparison group (n = 111), comprising patients consecutively enrolled within a specific recent timeframe (October 2024 to February 2026) rather than a random sample from all eligible older patients across the full study duration. This time-stratified approach was adopted due to the substantial volume of records and the labor-intensive nature of detailed chart review, which rendered a comprehensive abstraction of all eligible older patients across the entire 4-year period unfeasible given resource constraints. The final analytical sample included 102 young patients and 111 older patients, with both groups completing a standardized follow-up with a primary endpoint defined as 6 months after symptom onset. Consequently, the findings regarding the older cohort should be interpreted as applying specifically to this selected contemporary comparison group.

Diagnostic criteria and exclusion of mimicking conditions

The diagnosis of NAION was established based on typical clinical features, including sudden, painless vision loss, altitudinal visual field defects, and optic disc edema [6]. Crucially, arteritic anterior ischemic optic neuropathy (AAION) was systematically excluded in all patients through a standardized protocol comprising: (1) assessment of systemic symptoms suggestive of giant cell arteritis (GCA) (e.g., scalp tenderness, jaw claudication, fever, and weight loss); (2) laboratory investigations including erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and platelet count (PLT); and (3) rheumatological evaluation when clinically indicated. For patients with persistent clinical suspicion despite normal inflammatory markers, or when further evaluation was deemed necessary, temporal artery ultrasound and/or biopsy were performed as appropriate. Patients presenting with elevated inflammatory markers or any suspicion of GCA were strictly excluded from the study.

Specific protocols for the young cohort

For patients aged < 50 years, a standardized diagnostic protocol was implemented to systematically exclude optic neuritis and other demyelinating or inflammatory etiologies. This protocol included: (1) magnetic resonance imaging (MRI) of the brain and orbits with gadolinium enhancement to assess for demyelinating lesions ; (2) evaluation of ocular pain, particularly pain on eye movement ; (3) assessment of color vision deficits, interpreted as supportive rather than exclusionary evidence given that color vision loss may also occur in NAION ; (4) visual evoked potential (VEP) testing to detect subclinical conduction delays, utilized as an adjunctive tool rather than a definitive exclusionary measure due to the frequent presence of VEP abnormalities in NAION; and (5) serological screening for autoimmune and inflammatory markers, including aquaporin-4 antibody (AQP4-Ab) and myelin oligodendrocyte glycoprotein antibody (MOG-Ab) where clinically indicated. Inclusion in the young cohort was determined by a comprehensive synthesis of clinical features—specifically the absence of ocular pain, the presence of characteristic altitudinal visual field defects, and optic disc hemorrhage—alongside negative MRI findings for demyelination. Inclusion did not require negative results across all functional tests. Only patients meeting these refined criteria were included to ensure diagnostic specificity.

Handling of bilateral cases

To ensure the statistical independence of observations, only one eye per patient was included in the primary analysis. For patients with bilateral involvement, the right eye was designated as the index eye; for those with unilateral disease, the affected eye was selected. Consequently, data from the fellow eye were excluded to prevent pseudoreplication and minimize inter-eye correlation. This selection strategy aligns with the independence assumption required for our statistical models.

Data collection and treatment protocol

Clinical data were retrospectively collected for both groups, including gender, age, smoking history, alcohol consumption, systemic comorbidities (e.g., hypertension, diabetes), and medication history. For patients with acute NAION presenting with poor baseline best-corrected visual acuity (BCVA) and no severe contraindications, systemic glucocorticoid therapy was initiated within 2 weeks of symptom onset. The clinical presentation commonly included a relative afferent pupillary defect (RAPD) and sectoral or diffuse optic disc edema. Treatment regimens comprised either oral prednisone (1 mg/kg/day for 14 days) or intravenous methylprednisolone (250 mg/day for 3 days) followed by oral prednisone (1 mg/kg/day for 11 days). Both protocols included gradual tapering based on clinical response and resolution of disc edema. Treatment allocation was non-randomized and guided by clinical judgment; consequently, any observed changes in visual outcomes are susceptible to confounding by indication and should be interpreted as associative rather than causal. For diabetic patients, blood glucose levels were closely monitored, and antidiabetic regimens were adjusted as needed to manage hyperglycemia.

Ophthalmic examinations and outcome measures

Comprehensive ophthalmic examinations were performed, including BCVA, non-contact tonometry, RAPD, and slit-lamp biomicroscopy to evaluate anterior segment status and media clarity. Lens opacity was graded as clear, mild, moderate, or severe according to standardized clinical criteria, and the presence of pseudophakia was documented. Fundus examination was conducted, followed by automated perimetry using the Humphrey Field Analyzer 3 (model 840; Carl Zeiss Meditec, Inc., Dublin, CA, USA) with the 30 − 2 test strategy and the SITA Fast algorithm. Reliability criteria were defined as fixation losses < 20%, false-positive errors < 15%, and false-negative errors < 15%. Optic disc morphology was assessed using optical coherence tomography (OCT) (VG200D Intalight, Dream 150 system), with specific attention paid to small optic cup dimensions, PHOMS, and ODD. At the final follow-up, outcome measures including BCVA, visual field index (VFI), mean deviation (MD), and pattern standard deviation (PSD) were recorded. BCVA was measured using a standard logarithmic visual acuity chart and converted to the logarithm of the minimum angle of resolution (LogMAR) for statistical analysis.

Statistical analysis

Statistical analyses were performed using SPSS version 27.0 (IBM Corp., Armonk, NY, USA). Data normality was assessed using the Kolmogorov–Smirnov test. Continuous variables with non-normal distributions were presented as median (interquartile range, IQR), and categorical variables as number (%). Group comparisons for baseline characteristics were conducted using the Mann–Whitney U test or Pearson’s chi-square/Fisher’s exact test, as appropriate. Paired non-normal data were analyzed using the Wilcoxon signed-rank test. To directly compare visual improvement between groups, change scores (Δ) were calculated for LogMAR BCVA, VFI, MD, and PSD (defined as final follow-up values minus baseline values). Between-group differences in these change scores were assessed using the Mann–Whitney U test.

Multivariable linear regression analysis was performed on the entire cohort to identify independent predictors of changes in visual function outcomes. An initial candidate pool of 21 baseline covariates was established, encompassing demographic factors, onset-to-presentation time, lens status, baseline visual function metrics, systemic comorbidities, and optic disc parameters.

We constructed multivariable linear regression models using the enter method (forced entry). Variable inclusion followed a pre-specified strategy combining clinical prior knowledge with data-driven screening: (1) covariates deemed clinically essential a priori—including age, sex, lens status, onset-to-presentation time, baseline visual acuity, baseline visual field status, steroid use, diabetes, hypertension, cardiovascular risk factors, and optic disc parameters—were included as mandatory covariates regardless of their univariate significance [7]; and (2) among the remaining candidates, variables demonstrating statistical significance in univariate analyses (P < 0.1) were identified as additional predictors. All variables meeting either criterion were simultaneously entered into the final models to ensure comprehensive adjustment for confounders without algorithmic exclusion.

The final composition of the multivariable models was as follows: For outcomes ΔLogMAR, ΔVFI, and ΔMD, the final models included age, sex, onset-to-presentation time, lens status, baseline BCVA ≥ 0.5 (standard logarithmic visual acuity chart), baseline LogMAR, baseline VFI, baseline MD, baseline PSD, systemic glucocorticoid use, diabetes mellitus, hypertension, cardiovascular disease, PHOMS in the affected eye, and ODD in the affected eye. For the outcome ΔPSD, the final model included all variables listed above, with the addition of cerebrovascular disease.

We employed multivariable linear regression models with visual function change scores (derived from baseline and follow-up measurements) as the dependent variables. To account for potential mathematical coupling and regression to the mean, baseline visual function metrics were included as covariates in these models. Given the exploratory nature of this study and sample size constraints, internal validation methods such as bootstrapping were not performed. In the results tables, only variables retained in the final multivariable models are presented; variables excluded during the selection process are omitted to prevent misinterpretation of missing data. Results are reported as unstandardized regression coefficients () with 95% confidence intervals (CIs) and two-tailed -values; statistical significance was set at .

Results

Baseline characteristics and risk factor profiles

The baseline characteristics of the study population are summarized in Table 1. Compared with the ≥ 50-year group, the < 50-year group had a higher proportion of male patients (62.75% vs. 45.05%, P = 0.010). Onset-to-presentation time (median 8.0 days in both groups, P = 0.702) and interval between bilateral onset (P = 0.392) were similar between the two groups. The < 50-year group showed lower rates of bilateral involvement (14.71% vs. 28.83%, P = 0.013), lower prevalence of RAPD positivity (45.10% vs. 59.46%, P = 0.036), and a different distribution of lens status, characterized by higher rates of clear lenses and lower rates of mild and moderate opacities (all P < 0.001), although no significant difference was found in severe opacity (P = 0.499). The younger group also had lower pseudophakia rates (1.96% vs. 9.91%, P = 0.015). Furthermore, this group demonstrated a higher proportion of eyes achieving BCVA ≥ 0.5 at baseline (59.80% vs. 40.54%, P = 0.005) and at final follow-up (65.69% vs. 40.54%, P < 0.001). Systemic glucocorticoid use was more common in the younger group but did not reach statistical significance (44.12% vs. 31.53%, P = 0.058).

Table 1.

Baseline demographics, clinical characteristics, and treatment of patients with NAION stratified by age group

Characteristic	< 50 years (n = 102)	≥ 50 years (n = 111)	Test statistic	P
Age (years), median (Q1, Q3)	45.00 (42.00, 48.00)	60.00 (55.00, 67.00)	U = 11322.0	< 0.001
Male, n (%)	64 (62.75)	50 (45.05)	χ²=6.694	0.010
Onset-to-presentation time (days), median (Q1, Q3)	8.00 (5.00, 15.00)	8.00 (5.00, 14.00)	U = 5489.5	0.702
Bilateral involvement, n (%)	15 (14.71)	32 (28.83)	χ²=6.165	0.013
Interval between bilateral onset (months)*, median (Q1, Q3)	4.00 (0.00, 18.00)	6.50 (0.00, 51.00)	U = 213.0	0.392
Right eye, n (%)	54 (52.94)	60 (54.05)	χ²=0.026	0.871
RAPD positive, n (%)	46 (45.10)	66 (59.46)	χ²=4.397	0.036
Lens status
Clear lens, n (%)	87 (85.29)	32 (28.83)	χ²=68.737	< 0.001
Mild opacity, n (%)	12 (11.76)	49 (44.14)	χ²=27.269	< 0.001
Moderate opacity, n (%)	1 (0.98)	17 (15.32)	χ²=14.118	< 0.001
Severe opacity, n (%)	0 (0)	2 (1.80)	Fisher’s exact test	0.499
Pseudophakia, n (%)	2 (1.96)	11 (9.91)	χ²=5.861	0.015
Baseline BCVA ≥ 0.5 (5 m SL), n (%)	61 (59.80)	45 (40.54)	χ²=7.890	0.005
Final BCVA ≥ 0.5 (5 m SL), n (%)	67 (65.69)	45 (40.54)	χ²=13.480	< 0.001
Systemic glucocorticoid use, n (%)	45 (44.12)	35 (31.53)	χ²=3.590	0.058

*: Only for patients with bilateral involvement (< 50 years: 1 unknown; ≥50 years: 6 unknown)

RAPD: relative afferent pupillary defect; BCVA: best-corrected visual acuity

The comparison of risk factor profiles between the two age groups is detailed in Table 2. No statistically significant differences were observed for the presence of ≥ 1 vascular risk factor, diabetes mellitus, hyperlipidemia, cerebrovascular disease, anemia, alcohol consumption, PHOMS in the fellow eye, or crowded optic disc in the fellow eye (all P > 0.05). ODD in the affected eye showed no significant difference between groups (P = 0.108). Differences in the distribution of certain risk factors were observed: the < 50-year group had a higher proportion of smoking history (OR = 2.619, 95% CI: 1.167 to 5.873, P = 0.017), while the ≥ 50-year group had higher proportions of hypertension (OR = 0.533, 95% CI: 0.303 to 0.936, P = 0.028), cardiovascular disease (OR = 0.308, 95% CI: 0.097 to 0.977, P = 0.036), and PHOMS in the affected eye (OR = 0.291, 95% CI: 0.163 to 0.520, P < 0.001).

Table 2.

Comparison of risk factor profiles between the two age groups

Risk factor	< 50 years (n = 102)	≥ 50 years (n = 111)	χ²	P	OR (95% CI)
≥ 1 vascular risk factor, n (%)	77 (75.49)	94 (84.68)	2.839	0.092	0.557 (0.281, 1.106)
Diabetes mellitus, n (%)	27 (26.47)	34 (30.63)	0.450	0.502	1.227 (0.675, 2.228)
Hypertension, n (%)	31 (30.39)	50 (45.05)	4.843	0.028	0.533 (0.303, 0.936)
Hyperlipidemia, n (%)	36 (35.29)	45 (40.54)	0.621	0.431	0.800 (0.459, 1.394)
Cardiovascular disease, n (%)	4 (3.92)	13 (11.71)	4.392	0.036	0.308 (0.097, 0.977)
Cerebrovascular disease, n (%)	3 (2.94)	7 (6.31)	1.345	0.246	0.450 (0.113, 1.790)
Anemia, n (%)	4 (3.92)	5 (4.50)	0.045	0.833	0.865 (0.226, 3.315)
Smoking, n (%)	21 (20.59)	10 (9.01)	5.731	0.017	2.619 (1.167, 5.873)
Alcohol consumption, n (%)	30 (29.41)	22 (19.82)	2.650	0.104	1.686 (0.896, 3.171)
PHOMS in affected eye, n (%)	26 (25.49)	60 (54.05)	18.015	< 0.001	0.291 (0.163, 0.520)
ODD in affected eye, n (%)	3 (2.94)	0 (0.00)	4.465	0.108	-
Crowded optic disc in fellow eye*, n (%)	77 (77.78)	62 (66.67)	2.962	0.085	1.750 (0.922, 3.321)
PHOMS in fellow eye, n (%)	8 (7.84)	5 (4.50)	1.034	0.309	1.804 (0.571, 5.706)

*: Denominator excludes eyes with optic atrophy or unknown status. <50 years: 2 eyes not measured due to optic atrophy, 1 eye unknown; ≥50 years: 17 eyes not measured due to optic atrophy, 1 eye unknown

PHOMS: peripapillary hyperreflective ovoid mass-like structure; ODD: optic disc drusen

The distribution of visual field defect types at baseline is presented in Table 3. At baseline, the distribution of visual field defect types differed significantly between the < 50 years and ≥ 50 years groups. The two most common defect patterns in both cohorts were inferior hemifield defects and diffuse defects. Specifically, inferior hemifield defects were the predominant type in the < 50 years group (50.98%), whereas diffuse defects were more prevalent in the ≥ 50 years group (40.54%). The proportion of inferior hemifield defects was significantly higher in the < 50 years group than in the ≥ 50 years group (50.98% vs. 36.04%, P = 0.028). Conversely, diffuse defects were significantly more common in the ≥ 50 years group than in the < 50 years group (40.54% vs. 22.55%, P = 0.005). No statistically significant differences were observed between the two age groups regarding superior defects (P = 1.000), enlarged blind spots (P = 0.657), peripheral defects (P = 0.891), or quadrantanopia (P = 0.285).

Table 3.

Types of visual field defects at baseline

Defect type	< 50 years (n = 102)	≥ 50 years (n = 111)	Test statistic	P
Inferior hemifield defect, n (%)	52 (50.98)	40 (36.04)	χ²=4.838	0.028
Diffuse defect, n (%)	23 (22.55)	45 (40.54)	χ²=7.917	0.005
Superior defect, n (%)	2 (1.96)	3 (2.70)	Fisher’s exact test	1.000
Enlarged blind spot, n (%)	5 (4.90)	7 (6.31)	χ²=0.197	0.657
Peripheral defect, n (%)	5 (4.90)	5 (4.50)	χ²=0.019	0.891
Quadrantanopia, n (%)	15 (14.71)	11 (9.91)	χ²=1.141	0.285

Changes in BCVA and visual field indices within age groups

BCVA and visual field indices stratified by age group are summarized in Table 4. In the < 50 years group (n = 102), statistically significant changes were observed from baseline to final follow-up across all visual field indices and BCVA. While the median BCVA numerically remained unchanged at 0.22 LogMAR, the overall distribution showed a statistically significant shift toward improvement (P = 0.002), with a median change of -0.09 LogMAR (IQR: -0.21 to 0.00). Notably, 25.49% of eyes achieved a clinically meaningful improvement (≥ 0.2 LogMAR) (Table 5). The VFI increased from 72.50% to 79.00% (P = 0.008), MD improved from − 11.98 dB to -10.79 dB (P = 0.004), and PSD decreased from 10.16 dB to 8.98 dB (P = 0.001). In contrast, in the ≥ 50 years group (n = 111), no statistically significant within-group changes were detected. BCVA remained stable (median 0.40 to 0.40, P = 0.073), and there were no significant differences in VFI (62.00% to 60.00%, P = 0.532), MD (-14.96 dB to -15.61 dB, P = 0.589), or PSD (8.28 dB to 8.82 dB, P = 0.410).

Table 4.

Comparison of baseline and final follow-up BCVA and visual field indices within each age group

Group	Baseline	Final	T	P
< 50 years (n = 102)
BCVA (LogMAR), median (Q1, Q3 )	0.22 (0.10, 0.52)	0.22 (0.00, 0.52)	884.5	0.002
VFI (%), median (Q1, Q3)	72.50 (52.00, 85.00)	79.00 (51.75, 90.25)	1605.0	0.008
MD (dB), median (Q1, Q3)	-11.98 (-18.21, -8.43)	-10.79 (-19.18, -6.16)	1716.0	0.004
PSD (dB), median (Q1, Q3)	10.16 (6.93, 12.40)	8.98 (5.48, 11.44)	1627.0	0.001
≥ 50 years (n = 111)
BCVA (LogMAR), median (Q1, Q3 )	0.40 (0.22, 1.00)	0.40 (0.22, 1.00)	1280.0	0.073
VFI (%), median (Q1, Q3)	62.00 (28.00, 80.00)	60.00 (23.00, 80.00)	2488.0	0.532
MD (dB), median (Q1, Q3)	-14.96 (-23.20, -8.50)	-15.61 (-24.55, -8.77)	2819.0	0.589
PSD (dB), median (Q1, Q3)	8.28 (5.72, 12.87)	8.82 (5.26, 11.72)	2725.0	0.410

VFI: visual field index; MD: mean deviation; PSD: pattern standard deviation

Table 5.

Comparison of baseline, final, and change in BCVA and visual field indices between the two age groups

Variable	< 50 years (n = 102)	≥ 50 years (n = 111)	Test statistic	P
BCVA (LogMAR), median (Q1, Q3 )
Baseline	0.22 (0.10, 0.52)	0.40 (0.22, 1.00)	U = 3907.5	< 0.001
Final	0.22 (0.00, 0.40)	0.40 (0.22, 1.00)	U = 3686.5	< 0.001
ΔLogMAR	-0.09 (-0.21, 0.00)	0.00 (-0.22, 0.00)	U = 5465.0	0.660
≥ 0.2 ΔLogMAR, n (%)	26 (25.49)	30 (27.03)	χ²=0.065	0.799
VFI (%), median (Q1, Q3)
Baseline	72.50 (52.00, 85.00)	62.00 (28.00, 80.00)	U = 4365.0	0.004
Final	79.00 (51.75, 90.25)	60.00 (23.00, 80.00)	U = 3962.0	< 0.001
ΔVFI	4.00 (-3.00, 9.00)	1.00 (-5.00, 7.00)	U = 4919.5	0.099
MD (dB), median (Q1, Q3)
Baseline	-11.98 (-18.21, -8.43)	-14.96 (-23.20, -8.50)	U = 4840.5	0.068
Final	-10.79 (-19.18, -6.16)	-15.61 (-24.55, -8.77)	U = 4295.5	0.002
ΔMD	1.66 (-0.77, 3.57)	0.08 (-1.65, 2.82)	U = 4731.0	0.038
PSD (dB), median (Q1, Q3)
Baseline	10.16 (6.93, 12.40)	8.28 (5.72, 12.87)	U = 5071.0	0.189
Final	8.98 (5.48, 11.44)	8.82 (5.26, 11.72)	U = 5636.5	0.957
ΔPSD	-0.59 (-3.09, 0.60)	-0.02 (-2.22, 1.08)	U = 4833.5	0.066

VFI: visual field index; MD: mean deviation; PSD: pattern standard deviation

*: Mann–Whitney U test for between-group comparisons

Δ: Amount of change

BCVA and visual field outcomes by age group

Baseline and final BCVA and visual field outcomes are summarized in Table 5. At both baseline and final follow-up, the < 50-year group demonstrated significantly lower BCVA LogMAR scores (indicating better vision) compared to the ≥ 50-year group (both P < 0.001), although the amount of change (Δ LogMAR) did not differ significantly between groups (P = 0.660), the proportion of patients achieving a Δ LogMAR improvement of ≥ 0.2 was comparable in both cohorts: 26 (25.49%) in the < 50-year group versus 30 (27.03%) in the ≥ 50-year group (χ²=0.065, P = 0.799). Regarding visual field indices, the < 50-year group had significantly higher baseline VFI (P = 0.004) and final VFI (P < 0.001) than the selected contemporary comparison group of older NAION patients; however, the change in VFI showed no significant difference between the two groups (P = 0.099). For MD, no significant difference was observed at baseline (P = 0.068); however, the < 50-year group achieved significantly better final MD (P = 0.002) and exhibited a significantly greater improvement in MD (Δ MD) compared to the ≥ 50-year group (P = 0.038). PSD showed no significant differences between groups at baseline, final follow-up, or in the amount of change (all P > 0.05). Notably, while younger patients exhibited significant within-group improvements in multiple visual function parameters over follow-up, between-group comparisons of recovery magnitude revealed a statistically significant difference only for the change in MD, with no significant differences in the magnitude of BCVA, VFI, or PSD improvement.

Multivariate analysis of visual function recovery

Factors influencing the change in BCVA are presented in Table 6. After adjustment for potential confounders, three variables remained significant: age (B = 0.009, 95% CI: 0.002 to 0.016, P = 0.013), onset-to-presentation time (B = − 0.007, 95% CI: −0.014 to − 0.001, P = 0.023), and baseline BCVA (LogMAR) (B = − 0.408, 95% CI: −0.562 to − 0.255, P < 0.001). Given that ΔLogMAR is derived from baseline measurements, these associations should be interpreted with caution due to the potential influence of mathematical coupling and regression to the mean. Specifically, older age was associated with less visual recovery (more positive ΔLogMAR), while shorter onset-to-presentation time and better baseline visual acuity were associated with greater visual recovery (more negative ΔLogMAR). For all other variables included in the multivariate model, no statistically significant associations with ΔLogMAR were observed (P > 0.05).

Table 6.

Analysis of factors influencing ΔLogMAR

Variable	Univariate B (95% CI)	P	Multivariate B (95% CI)	P
Age (years)	0.002 (-0.004, 0.008)	0.434	0.009 (0.002, 0.016)	0.013
Sex	-0.047 (-0.178, 0.084)	0.480	0.003 (-0.123, 0.129)	0.962
Onset-to-presentation time	-0.008 (-0.015, -0.001)	0.020	-0.007 (-0.014, -0.001)	0.023
RAPD positive	-0.024 (-0.154, 0.107)	0.721
Lens status	0.002 (-0.059, 0.063)	0.940	0.008 (-0.062, 0.078)	0.820
Baseline BCVA ≥ 0.5 (5 m SL)	0.289 (0.165, 0.414)	< 0.001	0.100 (-0.067, 0.268)	0.238
Baseline BCVA (LogMAR)	-0.293 (-0.390, -0.196)	< 0.001	-0.408 (-0.562, -0.255)	< 0.001
Baseline VFI	0.001 (-0.001, 0.004)	0.242	0.000 (-0.011, 0.012)	0.941
Baseline MD	-0.003 (-0.011, 0.004)	0.391	0.015 (-0.024, 0.054)	0.447
Baseline PSD	0.010 (-0.006, 0.026)	0.229	-0.010 (-0.026, 0.007)	0.240
Systemic glucocorticoid use	0.036 (-0.099, 0.170)	0.603	0.033 (-0.099, 0.166)	0.620
Diabetes mellitus	0.023 (-0.122, 0.167)	0.758	0.099 (-0.043, 0.241)	0.172
Hypertension	0.049 (-0.085, 0.183)	0.474	0.090 (-0.038, 0.217)	0.166
Hyperlipidemia	0.047 (-0.090, 0.183)	0.502
Cardiovascular disease	-0.092 (-0.332, 0.149)	0.453	-0.195 (-0.434, 0.044)	0.109
Cerebrovascular disease	-0.159 (-0.467, 0.149)	0.310
Anemia	-0.138 (-0.462, 0.186)	0.402
Smoking	-0.031 (-0.216, 0.154)	0.738
Alcohol consumption	0.116 (-0.035, 0.267)	0.132
PHOMS in affected eye	-0.019 (-0.152, 0.114)	0.782	-0.018 (-0.141, 0.105)	0.769
ODD in affected eye	-0.039 (-0.593, 0.514)	0.889	0.025 (-0.485, 0.534)	0.924

Lens status: 0=transparent, 1=mild opacity, 2=moderate opacity, 3=severe opacity, 4=IOL

Sex: 0=male, 1=female

Other binary variables: 0=negative, 1=positive

Factors influencing the change in VFI (ΔVFI) are presented in Table 7. In the multivariate regression analysis, age (B = − 0.317, 95% CI: −0.607 to − 0.027, P = 0.032) and baseline BCVA (LogMAR) (B = − 7.569, 95% CI: −13.725 to − 1.413, P = 0.016) were independently associated with ΔVFI. Specifically, younger age and lower baseline LogMAR values (indicating better vision) were significantly associated with a greater improvement in VFI. For all other variables included in the multivariate model, no statistically significant associations with ΔVFI were observed (P > 0.05).

Table 7.

Analysis of factors influencing ΔVFI

Variable	Univariate B (95% CI)	P	Multivariate B (95% CI)	P
Age (years)	-0.141 (-0.362, 0.080)	0.210	-0.317 (-0.607, -0.027)	0.032
Sex	0.104 (-4.788, 4.996)	0.967	0.713 (-4.317, 5.742)	0.780
Onset-to-presentation time	0.233 (-0.019, 0.484)	0.070	0.136 (-0.120, 0.391)	0.296
RAPD positive	-0.157 (-5.044, 4.729)	0.949
Lens status	0.245 (-2.032, 2.521)	0.832	1.668 (-1.122, 4.457)	0.240
Baseline BCVA ≥ 0.5 (5 m SL)	-1.427 (-6.303, 3.449)	0.565	-1.437 (-8.139, 5.265)	0.673
Baseline BCVA (LogMAR)	0.067 (-3.850, 3.984)	0.973	-7.569 (-13.725, -1.413)	0.016
Baseline VFI	-0.152 (-0.233, -0.070)	< 0.001	-0.236 (-0.708, 0.236)	0.326
Baseline MD	0.550 (0.269, 0.830)	< 0.001	0.088 (-1.468, 1.643)	0.911
Baseline PSD	-0.267 (-0.871, 0.338)	0.385	-0.617 (-1.265, 0.030)	0.061
Systemic glucocorticoid use	-1.587 (-6.620, 3.447)	0.535	-2.139 (-7.450, 3.173)	0.428
Diabetes mellitus	-0.483 (-5.879, 4.914)	0.860	-3.194 (-8.889, 2.502)	0.270
Hypertension	-3.692 (-8.693, 1.309)	0.147	-1.262 (-6.366, 3.841)	0.626
Hyperlipidemia	2.339 (-2.702, 7.380)	0.361
Cardiovascular disease	-1.478 (-10.479, 7.523)	0.747	-2.013 (-11.576, 7.550)	0.679
Cerebrovascular disease	-0.803 (-12.337, 10.730)	0.891
Anemia	-4.245 (-16.360, 7.870)	0.490
Smoking	-2.267 (-9.178, 4.645)	0.519
Alcohol consumption	-1.766 (-7.441, 3.909)	0.540
PHOMS in affected eye	-1.378 (-6.347, 3.591)	0.585	-0.249 (-5.171, 4.673)	0.921
ODD in affected eye	11.429 (-9.218, 32.075)	0.276	9.676 (-10.693, 30.044)	0.350

Lens status: 0=transparent, 1=mild opacity, 2=moderate opacity, 3=severe opacity, 4=IOL

Sex: 0=male, 1=female

Other binary variables: 0=negative, 1=positive

Factors influencing the change in MD (ΔMD) are presented in Table 8. In the multivariate linear regression analysis, no variable was independently associated with ΔMD at the conventional significance level.

Table 8.

Analysis of factors influencing ΔMD

Variable	Univariate B (95% CI)	P	Multivariate B (95% CI)	P
Age (years)	0.055 (-0.010, 0.120)	0.096	0.071 (-0.015, 0.157)	0.104
Sex	0.071 (-1.373, 1.515)	0.923	-0.019 (-1.507, 1.469)	0.980
Onset-to-presentation time	-0.077 (-0.151, -0.002)	0.043	-0.055 (-0.131, 0.020)	0.152
RAPD positive	0.242 (-1.200, 1.685)	0.741
Lens status	0.102 (-0.570, 0.774)	0.765	-0.385 (-1.210, 0.440)	0.358
Baseline BCVA ≥ 0.5 (5 m SL)	-0.261 (-1.701, 1.180)	0.722	-0.039 (-2.021, 1.943)	0.969
Baseline BCVA (LogMAR)	0.345 (-0.811, 1.500)	0.557	1.615 (-0.206, 3.436)	0.082
Baseline VFI	0.033 (0.009, 0.057)	0.008	-0.068 (-0.208, 0.071)	0.336
Baseline MD	-0.138 (-0.222, -0.055)	0.001	-0.446 (-0.906, 0.014)	0.057
Baseline PSD	0.036 (-0.142, 0.215)	0.688	0.182 (-0.010, 0.373)	0.063
Systemic glucocorticoid use	0.486 (-1.000, 1.972)	0.520	0.663 (-0.908, 2.234)	0.406
Diabetes mellitus	0.554 (-1.038, 2.146)	0.494	1.182 (-0.503, 2.866)	0.168
Hypertension	1.389 (-0.083, 2.861)	0.064	0.661 (-0.848, 2.171)	0.389
Hyperlipidemia	-0.307 (-1.806, 1.191)	0.686
Cardiovascular disease	0.223 (-2.435, 2.881)	0.869	0.214 (-2.614, 3.043)	0.881
Cerebrovascular disease	1.315 (-2.086, 4.716)	0.447
Anemia	1.400 (-2.176, 4.976)	0.441
Smoking	0.601 (-1.440, 2.642)	0.562
Alcohol consumption	0.369 (-1.307, 2.046)	0.664
PHOMS in affected eye	0.396 (-1.072, 1.863)	0.596	-0.009 (-1.465, 1.447)	0.990
ODD in affected eye	-4.590 (-10.671, 1.492)	0.138	-3.343 (-9.368, 2.681)	0.275

Lens status: 0=transparent, 1=mild opacity, 2=moderate opacity, 3=severe opacity, 4=IOL

Sex: 0=male, 1=female

Other binary variables: 0=negative, 1=positive

Factors influencing the change in PSD (ΔPSD) are presented in Table 9. In the multivariate analysis, baseline PSD (B = − 0.331, 95% CI: −0.437 to − 0.226, P < 0.001) was independently associated with ΔPSD. This negative coefficient was consistent with higher baseline PSD being statistically associated with either a smaller increase or a larger decrease in PSD during follow-up. However, due to the non-linear behavior of PSD across disease stages, caution should be exercised when interpreting these results as predictive of long-term trends [8].

Table 9.

Analysis of factors influencing ΔPSD

Variable	Univariate B (95% CI)	P	Multivariate B (95% CI)	P
Age (years)	0.046 (0.007, 0.084)	0.021	0.013 (-0.034, 0.061)	0.577
Sex	-0.254 (-1.115, 0.607)	0.561	-0.808 (-1.632, 0.016)	0.055
Onset-to-presentation time	-0.023 (-0.068, 0.021)	0.308	-0.032 (-0.074, 0.010)	0.135
RAPD positive	0.318 (-0.542, 1.177)	0.467
Lens status	0.440 (0.044, 0.837)	0.030	0.177 (-0.280, 0.633)	0.447
Baseline BCVA ≥ 0.5 (5 m SL)	-1.358 (-2.198, -0.519)	0.002	-0.619 (-1.714, 0.476)	0.267
Baseline BCVA (LogMAR)	1.039 (0.364, 1.714)	0.003	-0.349 (-1.360, 0.661)	0.496
Baseline VFI	-0.014 (-0.029, 0.001)	0.061	-0.052 (-0.131, 0.026)	0.191
Baseline MD	0.029 (-0.022, 0.080)	0.268	-0.146 (-0.404, 0.112)	0.265
Baseline PSD	-0.346 (-0.442, -0.250)	< 0.001	-0.331 (-0.437, -0.226)	< 0.001
Systemic glucocorticoid use	0.165 (-0.722, 1.052)	0.715	0.334 (-0.543, 1.211)	0.454
Diabetes mellitus	0.572 (-0.375, 1.520)	0.235	0.199 (-0.734, 1.131)	0.675
Hypertension	0.032 (-0.853, 0.918)	0.943	-0.228 (-1.063, 0.607)	0.591
Hyperlipidemia	0.458 (-0.433, 1.349)	0.312
Cardiovascular disease	-0.272 (-1.857, 1.314)	0.736	-0.689 (-2.251, 0.872)	0.385
Cerebrovascular disease	2.018 (0.005, 4.031)	0.049	1.558 (-0.320, 3.435)	0.103
Anemia	-1.694 (-3.818, 0.430)	0.117
Smoking	0.683 (-0.532, 1.898)	0.269
Alcohol consumption	-0.255 (-1.254, 0.745)	0.616
PHOMS in affected eye	-0.061 (-0.936, 0.815)	0.892	-0.008 (-0.813, 0.798)	0.985
ODD in affected eye	-2.616 (-6.245, 1.013)	0.157	-1.281 (-4.606, 2.044)	0.448

Lens status: 0=transparent, 1=mild opacity, 2=moderate opacity, 3=severe opacity, 4=IOL

Sex: 0=male, 1=female

Other binary variables: 0=negative, 1=positive

Discussion

Although NAION is typically considered a disease of middle-aged and elderly individuals, it is not uncommon in patients under 50 years of age. Previous studies indicate that this younger demographic accounts for approximately 7.7%–36.6% of all NAION cases [9, 10]. Consistent with this range, our cohort included 102 patients aged under 50 years, representing 18.55% of the total. Notably, the majority of these young patients were in their forties, with a median age at onset of 45 years (interquartile range: 42–48). This age distribution differs from findings in a retrospective study in Hong Kong, which reported that adult optic neuritis predominantly affects the 23–35-year group [11]. Given the frequent misdiagnosis of young NAION patients as having optic neuritis, accurate differentiation based on these age-specific profiles is warranted.

This study delineates distinct age-related profiles in NAION. Patients aged < 50 years exhibited a significant male predominance (62.75% vs. 45.05%, P = 0.010). Within this younger cohort, a higher prevalence of smoking was observed compared to the older group. Conversely, patients aged ≥ 50 years demonstrated a markedly higher prevalence of systemic vascular comorbidities, specifically hypertension and cardiovascular disease. These findings suggest that the relative contribution of risk factors varies by age, with behavioral factors being more frequently observed in younger adults and vascular comorbidities characterizing the risk profile of older patients within this specific comparison cohort. It should be noted that these observations regarding older adults are specific to this consecutively enrolled cohort from a recent time window and may not be generalizable to all older NAION patients.

The pathophysiology of NAION remains incompletely understood but is predominantly attributed to hypoperfusion of the short posterior ciliary arteries, potentially exacerbated by autoregulatory dysfunction, or to venous congestion causing compartment syndrome within the optic nerve head [12, 13]. In our cohort, the prevalence of several vascular risk factors—including diabetes, hyperlipidemia, cerebrovascular disease, anemia, and alcohol consumption—did not differ significantly between age groups, suggesting these are common denominators across the lifespan. While diabetes remains a consistent risk factor, its impact may vary by age [9]. Furthermore, our observation that smoking was more prevalent in the younger cohort, whereas hypertension and cardiovascular disease were more prevalent in the older comparison group, aligns with established age-stratified patterns [3, 10, 14]. The higher prevalence of hypertension and cardiovascular risk factors in this selected contemporary comparison group is expected; however, whether these findings reflect age itself or the burden of age-associated comorbidities has not been fully disentangled.

Notably, we observed a significantly higher rate of RAPD positivity in the selected contemporary comparison group of older patients compared to the younger cohort (59.46% vs. 45.10%, P = 0.036), suggesting more severe afferent pathway damage within this specific group. ODD, calcific deposits that cause disc crowding [15, 16], were observed exclusively in patients aged < 50 years in our cohort, although this difference did not reach statistical significance (P = 0.108), likely attributable to the limited sample size or undetected buried drusen. The exclusive occurrence of ODD in younger adults suggests a potential association with the pathogenesis of early-onset NAION, which differs from the profile of the older comparison group where systemic vascular comorbidities are more prevalent. Given that bilateral ODD significantly elevates the risk of fellow eye involvement in this demographic [17], thorough evaluation using advanced imaging modalities, such as enhanced depth imaging optical coherence tomography (EDI-OCT), is warranted for young NAION patients. Conversely, a higher proportion of eyes with PHOMS was observed in the selected contemporary comparison group of older patients. We hypothesize that this discrepancy reflects differences in disease stage rather than a direct age-related effect. Our cohort comprised eyes with acute optic disc edema, whereas previous studies [18]included patients with varying disease durations. In the context of this selected contemporary comparison group of older patients, the higher proportion of eyes with PHOMS could reflect a greater representation of longer-standing or more chronic optic nerve head conditions, allowing more time for PHOMS to develop or persist. Since PHOMS are dynamic structures that may shrink or resolve over time, the observed age-related difference may be attributed to a higher prevalence of chronic phases in this cohort, rather than age serving as a direct determinant of PHOMS formation or persistence.

NAION typically presents unilaterally, with a 5-year fellow-eye involvement risk of 14%–18% [19–21]. Although existing literature remains conflicting regarding the influence of age on bilateral disease risk [21, 22], our cohort showed higher bilateral involvement in patients ≥ 50 years (28.83% vs. 14.71%, P < 0.05). In the context of this selected contemporary comparison group of older patients, the observed associations—including the higher proportion of eyes with PHOMS and increased bilateral involvement—should be interpreted as unadjusted observational findings rather than evidence of age as an independent risk factor. Given our fixed 6-month follow-up design without adjustment for follow-up duration, time since initial attack, systemic risk burden, or disc parameters, the elevated rates in older patients may correspond to the cumulative burden of vascular risk factors or a higher prevalence of chronic phases in this cohort, allowing more time for PHOMS to develop or persist. Since PHOMS are dynamic structures that may shrink or resolve over time, these findings apply specifically to this selected contemporary comparison group and do not establish age as a direct determinant of PHOMS formation or persistence across all older NAION patients.

Distinct distribution patterns of visual field defects were observed in NAION patients stratified by age: inferior hemifield defects were more common in patients < 50 years (50.98%), whereas diffuse defects were more frequent in those ≥ 50 years (40.54%). We selected 50 years as the stratification cutoff to align with prior studies defining young-onset NAION [14], while acknowledging that the overall risk of NAION increases with age and the mean age at onset is approximately 64 years [23].

Distinct age-related profiles in visual function were observed among patients with NAION. The young patient group demonstrated significantly better baseline and final follow-up BCVA and VFI, as well as a greater magnitude of mean defect improvement (ΔMD) (P = 0.038). These findings suggest that while younger patients present with more favorable initial functional status, the observed between-group difference appears to be largely related to a greater improvement in MD rather than an overall superior recovery capacity across all visual function metrics. Notably, comparable systemic glucocorticoid usage rates between groups suggest that treatment allocation reflected individualized clinical assessment rather than age-based preferences. In comparison with previous studies, specifically the work by Behbehani et al. [14], it was reported that young NAION patients achieved better final LogMAR BCVA (0.55 ± 0.57 vs. 0.90 ± 0.73, P = 0.03) and a higher proportion reached Snellen 20/40 or better (53% vs. 28.3%, P = 0.03). However, in that study, the difference in baseline BCVA did not reach statistical significance (P = 0.17), and the magnitude of MD improvement was similar between groups (P = 0.4). In contrast, our study is consistent with the final BCVA advantage in younger patients and further identifies a significant difference specifically in the magnitude of MD change. This discrepancy suggests that variations in enrollment time windows, follow-up duration, and population composition across different studies may account for the heterogeneity in reported recovery magnitudes [14, 24].

After adjusting for potential confounders, multivariate analysis identified age as an independent predictor of BCVA recovery (B = 0.009, 95% CI: 0.002 to 0.016, P = 0.013) and VFI improvement (B = − 0.317, 95% CI: −0.607 to − 0.027, P = 0.032). This finding is consistent with previous reports by Behbehani et al., who observed that younger patients had a significantly higher likelihood of achieving favorable final BCVA (OR = 2.87, 95% CI: 1.12 to 7.40) [14]. From a pathophysiological perspective, the greater magnitude of MD improvement observed in the younger group might reflect enhanced neuroplasticity and axonal repair capacity, along with a reduced systemic comorbidity burden, rather than indicating a universally superior recovery capacity across all visual function metrics [14, 25, 26]. However, after adjusting for potential confounders, age did not remain an independent predictor of ΔMD, suggesting that the observed association may be influenced by other baseline factors such as initial VFI. During the follow-up period, significant improvements in all four visual function indices, including BCVA, were observed primarily in the young adult group. While this pattern suggests favorable within-group trends, caution should be exercised when drawing broad conclusions regarding overall recovery capacity solely based on these intra-group improvements.

The negative regression coefficient for onset-to-presentation time (B = − 0.007) indicates that, after adjusting for age and baseline BCVA, a longer interval between symptom onset and presentation is associated with a smaller improvement in BCVA (i.e., less favorable visual recovery). This observation is consistent with findings in several ophthalmic conditions where symptom duration has been linked to visual outcomes [27, 28]. However, the relationship between timing and outcomes may vary across diseases and study designs.

The independent associations between baseline BCVA and both ΔLogMAR and ΔVFI observed in the present study warrant cautious interpretation due to the potential influence of mathematical coupling and regression to the mean. Since ΔLogMAR is mathematically derived from baseline measurements, an intrinsic statistical correlation exists between baseline values and the resulting change scores. Specifically, eyes with poorer initial visual acuity (higher LogMAR values) possess greater theoretical capacity for numerical improvement, whereas those with better baseline vision have limited room for gain. Consequently, the negative coefficient for baseline BCVA in the ΔLogMAR model may be partially attributable to these statistical constraints, although a genuine biological association with visual recovery capacity cannot be entirely excluded. Similarly, the association between baseline BCVA and ΔVFI may reflect inherent correlations between baseline functional status and subsequent functional change. These findings align with prior literature advocating prudence when interpreting regression models based on change scores, as such associations may partly stem from measurement-related statistical artifacts and should not be interpreted as definitive independent predictors of recovery magnitude without accounting for these methodological considerations [29].

Furthermore, the observed association between baseline PSD and ΔPSD in the multivariate model—whereby higher baseline PSD corresponded to either a smaller increase or a larger decrease in PSD during follow-up—also requires careful interpretation. Beyond the statistical coupling inherent to change-score analyses, PSD exhibits non-linear behavior across disease stages: it rises with progressive visual field loss in early-stage disease but peaks and declines in moderate-to-advanced stages. As highlighted in the European Glaucoma Society guidelines, PSD is not suitable for trend analysis owing to this stage-dependent non-linearity. Accordingly, the association between baseline PSD and ΔPSD may represent a composite effect of statistical coupling and disease-stage characteristics and should not be extrapolated as a definitive independent predictor of long-term progression patterns [8]. Notably, although a significant difference in the change of MD was observed between the two groups, no significant difference was found in the final PSD values (P = 0.957). Given that PSD reflects the degree of localized irregularity within the visual field, this age-independent terminal presentation might be explained by two potential mechanisms: first, diffuse retinal nerve fiber layer damage or media opacities might partially obscure genuine changes in localized sensitivity; second, patients across different age groups might converge toward a similar stable pattern of visual field defects in the late stages of the disease course, thereby masking intergroup differences in PSD [14].

The study’s retrospective, single-center design with a limited sample size may introduce selection bias, result in incomplete systemic data, and limit the ability to draw causal inferences or ensure broad generalizability. Although follow-up was extended to 6 months, it remains insufficient for assessing long-term outcomes and is susceptible to patient attrition. Regarding statistical analysis, although baseline visual function metrics were adjusted for as covariates to mitigate potential biases, we acknowledge that the inherent mathematical coupling between baseline measurements and derived change scores, along with regression to the mean, may not be fully accounted for. Consequently, associations between baseline status and functional improvement should be interpreted cautiously, avoiding overly strong prognostic claims. Furthermore, OSA was not systematically evaluated; therefore, the absence of documented OSA diagnoses, particularly in younger patients, limits our ability to assess its impact on prognosis and recurrence. Notably, while all eligible young patients (< 50 years) throughout the entire study period were included, the older subgroup (≥ 50 years) comprised patients consecutively enrolled during a specific recent window (October 2024 to February 2026). This temporal discrepancy was necessitated by the substantial volume of records and resource constraints that limited comprehensive chart abstraction across the full study duration. Although this strategy ensures no selective sampling within the defined timeframe, we acknowledge that it may affect the temporal representativeness of the older cohort. Consequently, our findings regarding age-related differences must be framed strictly as applying to this selected contemporary comparison group rather than being generalized to all older NAION patients.

Additionally, the relatively high number of covariates compared with the sample size may lead to model overfitting. Since this study is exploratory and lacks an independent validation cohort, formal model diagnostics and sensitivity analyses (such as bootstrap resampling) were not conducted. Accordingly, the calculated associations should be interpreted with caution, especially regarding their stability and generalizability. Despite these limitations, key covariates including lens status (phakic/pseudophakia) and baseline disease severity were incorporated into multivariate models to mitigate confounding. Moreover, various other relevant variables were assessed; some yielded negative results (e.g., systemic corticosteroid use), which may reflect the influence of retrospective design, indication bias, and sample heterogeneity.

Conclusion

This study identifies distinct risk factor profiles and visual recovery patterns between younger patients and the older comparison cohort. Younger patients had higher smoking rates and predominantly inferior hemifield defects, whereas the older comparison cohort exhibited higher burdens of hypertension, cardiovascular disease and PHOMS, with diffuse visual field defects more common in this group. Regarding visual outcomes, while younger patients demonstrated significant within-group improvements in BCVA and visual field indices, only the change in MD differed significantly between groups. Therefore, claims of superior overall recovery capacity in younger patients should be moderated. Although multivariate analysis suggests an association between age and both visual (ΔLogMAR) and visual field (ΔVFI) recovery, this finding requires cautious interpretation given that ΔLogMAR and ΔVFI are susceptible to baseline values, and only ΔMD showed significant intergroup differences; thus, these results do not support the conclusion that younger patients have superior overall recovery capacity. In addition to age-related differences, we observed that shorter onset-to-presentation time and better baseline visual acuity were associated with more substantial visual gains, while more severe baseline visual field defects (lower MD) were associated with greater improvements in MD. However, these associations do not establish a causal benefit of earlier presentation and may reflect regression toward the mean or other statistical phenomena related to baseline severity. Consequently, these results apply strictly to this selected contemporary comparison group and should not be generalized to all older NAION patients without prospective validation.

Author contributions

C.Z. and J.L. wrote the main manuscript text. B.H., W.X., Q.W. and G.Y. collected the data and constructed the tables. All authors reviewed and approved the final manuscript.

Funding

Not applicable.

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

This retrospective study was approved by the Ethics Committee of Jinan Mingshui Eye Hospital and adheres to the Declaration of Helsinki. Due to the retrospective nature of the study with anonymous patient information, the ethics committee waived the requirement for informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.