British Journal of Ophthalmology最新文献

Background/aims: Patients have largely been excluded from discussions on the use of their health data in developing medical artificial intelligence (AI), despite being directly affected by its integration into care. This study assessed ophthalmology patients' perspectives on AI to inform patient-aligned development and implementation.

Methods: We conducted a cross-sectional survey across ophthalmology clinics in a large academic hospital system in New York City. Consecutive patients were approached in waiting rooms by a research coordinator to maximise sociodemographic diversity and minimise bias from digital literacy or access. The survey, developed by experts in AI, ethics, ophthalmology and survey methodology, was administered via paper and Qualtrics. It addressed attitudes towards AI in clinical scenarios, willingness to share various types of personal data for AI model development and understanding of AI in ophthalmology.

Results: Among 403 respondents, 67% reported a low or no understanding of AI, and 71% expressed interest in learning more. Patients prioritised physician involvement and transparency. Comfort decreased with task complexity: highest for screening, lower for diagnosis and lowest for treatment/surgery. For model development, patients were more comfortable sharing de-identified optical coherence technology or lab data than facial images or genetic data. 90% felt consent was always necessary when using personal data to train AI models.

Conclusions: These findings highlight the need for patient education and robust data consent protocols. Implementing an opt-out system for retrospective data use may enhance trust while supporting innovation. Integrating patient perspectives into AI governance can foster trust and transparency in ophthalmology and beyond.

Background: Outcome-reporting bias (ORB), characterised by selective outcome reporting, may exaggerate intervention efficacy and safety, compromising clinical guidelines and patient care. This cross-sectional analysis investigated the prevalence and extent of ORB in seminal ophthalmic randomised controlled trials (RCTs).

Methods: Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we searched Embase, MEDLINE and CENTRAL for ophthalmic RCTs published in the 10 highest-impact ophthalmic and general medical journals from 2014 to 2023. Outcomes were incompletely reported if they lacked sufficient data for meta-analysis and unreported if specified in protocols but absent from publications. Logistic regression evaluated associations between incomplete reporting and trial or journal characteristics. The association between outcome unfavourability (ie, failure to demonstrate superiority, non-inferiority or equivalence) and incomplete reporting was calculated as OR and synthesised into a meta-analysis. Discrepancies between prespecified and published outcomes were examined.

Results: Of 9436 records, 260 RCTs met inclusion criteria. Among these, 167 (64.2%) and 124 (47.7%) trials had at least one incompletely reported outcome and unreported outcome, respectively. Unreported outcome was associated with lower 5-year impact factor (adjusted OR (aOR): 1.02, 95% CI 1.00 to 1.05, p=0.04, per 1.0 reduction), receiving only industry funding (OR: 1.98, 95% CI 1.02 to 3.85, p=0.04) or not declaring funding compared to non-industry funding (aOR: 3.52, 95% CI 1.22 to 10.18, p=0.02). Unfavourable outcomes were more likely to be incompletely reported than favourable ones (OR: 3.31, 95% CI 2.41 to 4.56, p<0.001). Outcome discrepancies occurred in 121 trials (46.5%).

Conclusions: ORB is prevalent in ophthalmic RCTs. Strengthening adherence to prespecified protocols and improving transparency in reporting are essential.

The purpose of the study was to investigate changes in aqueous humour outflow (AHO) after the Elios glaucoma procedure (excimer laser trabeculostomy, ELT) using sequential AHO angiography. AHO angiography using two different tracers was performed in six eyes of five patients undergoing cataract surgery combined with ELT. AHO was compared in regions of interest in the nasal quadrant.AHO improved significantly after ELT (p=0.03) overall, and each patient showed individual, qualitative AHO improvement. Interestingly, AHO improvement extended beyond the site of ELT application. Further studies are needed to investigate the correlation of AHO improvement and clinical efficacy.

Background/aims: To develop a risk score for moderate-risk to high-risk penetrating keratoplasty (PK) based on various risk factors and use it to stratify grafts.

Methods: We collected data on clinical features and survival status from the electronic health records of patients who underwent moderate-risk to high-risk PK between January 2012 and December 2021. Feature selection was performed using multivariate Cox regression analyses. The risk score was derived from the coefficient of variables obtained through the Cox proportional hazards model. The performance of the risk score was assessed using receiver operating characteristic curve analysis. Risk stratification was performed based on the risk scores.

Results: The study cohort comprised 2297 eyes (2166 patients) subjected to moderate-risk to high-risk PK. The risk score was derived based on six variables. Each variable was assigned a score of 0, 1, 2 or 3 based on its coefficient. The discriminative ability of this score was robust, achieving an area under the receiver operating characteristic curve of 0.747 in the external validation cohort, which surpassed the predictive performance of any single variable alone. According to the established risk score, patients were stratified into three distinct prognostic subgroups: moderate-risk (0-2 points), high-risk (3-5 points) and extremely high-risk groups (≥6 points). The survival curves across these stratified groups differed significantly (log-rank test, p<0.001).

Conclusion: The risk score demonstrated good discriminatory ability for moderate-risk to high-risk PK and can help to further stratify these grafts.

Aims: To assess the long-term anatomical and visual outcomes of primary scleral buckle (SB) surgery for rhegmatogenous retinal detachment (RRD) and identify predictors of surgical and functional success.

Methods: All primary SB procedures performed at Manchester Royal Eye Hospital between January 2008 and December 2023 were reviewed retrospectively. Preoperative, intraoperative and postoperative data were extracted from electronic records. Primary end-points were single-surgery anatomical success (SSAS) and final best-corrected visual acuity (BCVA); univariate and multivariate regression analyses examined predictive variables and complications.

Results: 608 eyes were included. SSAS was achieved in 515 eyes (85%) and final anatomical success in 602 eyes (99%). Mean BCVA improved from 0.62±0.85 to 0.32±0.49 logarithm of the minimum angle of resolution (logMAR) (p<0.01). On multivariate analysis, age >40 years (OR 0.55, 95% CI 0.37 to 0.83; p=0.004), macula-off status (OR 0.65, 95% CI 0.44 to 0.97; p=0.034) and ocular trauma (OR 0.40, 95% CI 0.19 to 0.82; p=0.012) independently reduced SSAS. For visual outcomes, macula-off detachment (β=+0.36; p<0.001) and ocular trauma (β=+0.42; p<0.001) were independent predictors of worse postoperative BCVA. The most common complications were subretinal haemorrhage (4.6%), inadvertent deep sutures (3.9%) and postoperative ocular hypertension (7.4%).

Conclusions: Primary SB provided high anatomical success, significant visual improvement and a favourable safety profile in this large single-centre study. Age above 40 years, macula-off status and ocular trauma predicted poorer SSAS; trauma and macula-off status were also associated with worse postoperative BCVA on multivariate logistic regression.

Background/aims: To characterise the temporal dynamics of choroidal changes relative to myopia onset and determine their associations with myopia onset in school children.

Methods: The 3-year prospective cohort study consisted of 180 emmetropic school-children. Cycloplegic auto-refractions were performed annually. Choroidal images acquired via swept-source optical coherence tomography until myopia onset were analysed. Choroidal architecture was characterised based on submacular choroidal thickness (ChT), choroidal vascularity index (CVI), luminal area (LA), stromal area (SA) and total choroidal area (TCA) measurements. Time to myopia onset was defined as the first annual follow-up visit with spherical equivalent ≤-0.50 dioptres.

Results: A total of 165 participants (330 eyes, 91.7%) who completed at least one annual follow-up were included. During 36 months, 168 eyes of 100 participants developed myopia. Refractive shift and axial elongation showed significant non-linear trends in which they accelerated dramatically approximately 2 years prior to myopia onset. Notably, ChT, as well as CVI, LA, SA and TCA, exhibited significant linear declines throughout the 3 years when these eyes progressed to myopia, whereas they remained unchanged in persistent non-myopic eyes. Multivariable accelerated failure time models showed that a greater decline rate of these choroidal parameters (per SD) shortened the time to myopia onset, with time ratios ranging from 1.26 to 1.76. The area under receiver-operating characteristic curves for annual choroidal changes effectively distinguished incident myopia (0.83-0.89).

Conclusions: Alterations in choroidal architecture are an early indicator for the disruption of normal emmetropisation and the onset of myopia.

Background: Infectious keratitis (IK) is a leading cause of corneal blindness globally. While treatment can limit infection-related complications, the timeline and predictors of visual recovery in IK remain poorly characterised.

Methods: We conducted a retrospective cohort study of 966 patients (984 eyes) with non-viral IK at a tertiary eye care centre (2018-2023). Electronic health records were reviewed for microbiologic data, and best-corrected visual acuity (BCVA) measured at presentation and up to 90 days. Outcomes included change in BCVA evaluated across time intervals. Spearman correlation assessed associations between BCVA change and number of follow-up visits.

Results: The average presenting BCVA was 1.04±0.95 logMAR (20/200), improving to 0.75±0.88 logMAR (20/112) at 90 days (-0.29 logMAR; p<0.001). The average number of follow-up visits was 4.18 days within the 90-day follow-up. 27.2% had culture-proven infections, predominantly bacterial (65.6%). Culture-positive cases had poorer initial vision (1.51±0.94 logMAR) but improved significantly (-0.43 logMAR; p<0.001). Most visual recovery occurred after 4 weeks of treatment. Poorer presenting BCVA correlated with higher follow-up visit frequency (ρ=0.37, p<0.001). A significant correlation was observed between BCVA improvement and number of visits (ρ=0.14, p<0.001). Bacterial keratitis showing better recovery than fungal, Acanthamoeba and polymicrobial infections. Surgical intervention was required in 3.5% of eyes, primarily in fungal, Acanthamoeba and polymicrobial cases.

Conclusions: Most visual improvement after IK occurs after 4 weeks, emphasising the need for continued follow-up and patient counselling regarding expected recovery timelines. Outcomes vary by initial severity and causative organism, highlighting opportunities for stratified management in IK.

Aims: To quantify nationwide incidence and contemporary treatment patterns of acute retinal necrosis (ARN).

Methods: Retrospective cohort using Japan's National Database of Health Insurance Claims and Specific Health Checkups, which covers >95% of claims. New-onset ARN (2016-2022) was identified by ARN diagnostic codes plus antiviral therapy. Annual, age-specific and sex-specific incidence rates used governmental denominators; age-standardised rates used the WHO world standard. Surgical management (vitreoretinal surgery, silicone oil (SO)/perfluorocarbon (PFC) use, glaucoma surgery) was assessed up to 2 years after diagnosis.

Results: Among 586 incident cases (313 men, 273 women), crude incidence was 0.066 (95% CI 0.061 to 0.072) per 100 000 person-years; the age-standardised rate was 0.040 per 100 000 person-years. Incidence rose with age, peaking at 70-74 years, and was consistently higher in men than in women. Vitreoretinal surgery occurred in 42.4% by 30 days and 59.4% by 365 days after diagnosis; 25.1% underwent ≥2 procedures within 1 year. SO or PFC was used in 40.8% by 1 year. Glaucoma surgery was uncommon (<2% by 2 years).

Conclusion: In the largest population-based study of ARN to date, nationwide incidence in Japan closely matched prior UK estimates, suggesting cross-system consistency in ARN epidemiology. Management was characterised by early, intensive surgery with frequent tamponade use and substantial reoperation, informing clinical triage and resource planning for this rare, vision-threatening condition.

Background/aims: To evaluate psychiatric outcomes following enucleation or evisceration (EE) and determine whether these outcomes differ by the underlying aetiology related to surgery, including trauma, blind painful eye, infection and malignancy.

Methods: This retrospective cohort study utilised the TriNetX Research Network, analysing deidentified records of 20 109 patients who underwent EE. Psychiatric outcomes were compared with 325 108 patients who underwent other major ocular surgeries. Patients were propensity-score matched, and EE cases were stratified into trauma, blind painful eye, infection or malignancy cohorts. Outcomes included new diagnoses of depression, anxiety, stress and adjustment disorders, substance use disorders and new prescriptions of antidepressants or anxiolytics within 1 year. Predictors of psychiatric outcomes were evaluated using the Cox proportional hazards models at 3 years postindex.

Results: Compared to matched controls, the EE cohort had higher rates of antidepressant use (risk ratio, RR 1.14; 95% CI, 1.04 to 1.25; p<0.01) and anxiolytic use (RR 1.15; CI, 1.07 to 1.25; p<0.01), despite similar diagnostic rates of depression. Among subgroups, trauma-related EE was uniquely associated with elevated risk of depression (RR 1.85; CI, 1.06 to 3.21; p=0.03), antidepressant use (RR 2.18; CI, 1.45 to 3.26; p<0.01), anxiolytic use (RR 1.63; CI, 1.13 to 2.33; p<0.01) and substance use disorders (RR 2.04; CI, 1.10 to 3.79; p=0.02). Trauma remained an independent predictor of depression after adjusting for psychiatric history and covariates (HR 1.34; CI, 1.11 to 1.62; p<0.01).

Conclusion: Psychiatric morbidity following eye removal is influenced by related surgical aetiology. Trauma-related EE carries a distinctly higher psychiatric burden, supporting the need for targeted mental health screening and intervention in this vulnerable subgroup.