Associations with baseline visual acuity in 12,414 eyes starting treatment for neovascular AMD.

AIMS: To determine baseline visual acuity before the start of treatment for neovascular age-related macular degeneration (AMD), compare median and visual acuity states between treatment sites and investigate the association of socio-demographic and clinical characteristics with baseline acuity. METHODS: Anonymised demographic and clinical data, collected as part of routine clinical care, were extracted from electronic medical records at treating National Health Service (NHS) Trusts. Analyses were restricted to eyes with baseline visual acuity recorded at treatment initiation. Associations with baseline acuity were investigated using multivariate linear regression. RESULTS: Analysis included 12,414 eyes of 9116 patients at 13 NHS Trusts. Median baseline acuity was LogMAR 0.46 (interquartile range = 0.26-0.80) and 34.5% of eyes had good acuity, defined as LogMAR ≤0.3. Baseline acuity was positively associated with second-treated eye status, younger age, lower socio-economic deprivation, independent living, and female sex. There was little evidence of association between baseline acuity and distance to the nearest treatment centre, systemic or ocular co-morbidity. Despite case-mix adjustments, there was evidence of significant variation of baseline visual acuity between sites. CONCLUSIONS: Despite access to publicly funded treatment within the NHS, variation in visual acuity at the start of neovascular AMD treatment persists. Identifying the characteristics associated with poor baseline acuity, targeted health awareness campaigns, professional education, and pathway re-design may help to improve baseline acuity, the first eye gap, and visual acuity outcomes.

Associations with visual acuity outcomes after 12 months of treatment in 9401 eyes with neovascular AMD.

OBJECTIVE: To record visual acuity outcomes after 12 months of treatment for neovascular age-related macular degeneration (NvAMD), investigate variation between sites and explore associations with baseline characteristics and care processes. METHODS AND ANALYSIS: Anonymised demographic and clinical data were extracted from electronic medical records at treating National Health Service (NHS) Trusts. Associations with acuity outcomes were investigated using multivariate linear and logistic regression. RESULTS: Analysis included 9401 eyes (7686 patients) treated at 13 NHS Trusts. From baseline to month 12, median acuity improved from LogMAR 0.50 (IQR 0.30-0.80) to 0.40 (0.22-0.74) and the proportion of eyes with LogMAR ≥0.3 increased from 34.5% to 39.8%. Baseline visual acuity was the strongest predictor of visual acuity outcomes. For each LogMAR 0.1 worsening of baseline acuity, the acuity at 12 months was improved by LogMAR 0.074 (95% CI 0.073 to 0.074) and the odds of a 'poor' acuity outcome was multiplied by 1.66 (95% CI 1.61 to 1.70). Younger age, independent living status, lower socioeconomic deprivation, timely loading phase completion and higher number of injections were associated with better acuity outcomes. Despite case-mix adjustments, there was evidence of significant variation in acuity outcomes between sites. CONCLUSIONS: Even after adjustment for other variables, variation in acuity outcomes after NvAMD treatment within the NHS remains. Meaningful comparison of outcomes between different providers requires adjustment for a range of baseline characteristics, not visual acuity alone. Identifying best practice at sites with better outcomes and adapting local care processes are required to tackle this health inequality.

Analysis of an adaptive lead weighted ResNet for multiclass classification of 12-lead ECGs.

Background. Twelve lead ECGs are a core diagnostic tool for cardiovascular diseases. Here, we describe and analyse an ensemble deep neural network architecture to classify 24 cardiac abnormalities from 12 lead ECGs.Method. We proposed a squeeze and excite ResNet to automatically learn deep features from 12-lead ECGs, in order to identify 24 cardiac conditions. The deep features were augmented with age and gender features in the final fully connected layers. Output thresholds for each class were set using a constrained grid search. To determine why the model made incorrect predictions, two expert clinicians independently interpreted a random set of 100 misclassified ECGs concerning left axis deviation.Results. Using the bespoke weighted accuracy metric, we achieved a 5-fold cross-validation score of 0.684, and sensitivity and specificity of 0.758 and 0.969, respectively. We scored 0.520 on the full test data, and ranked 2nd out of 41 in the official challenge rankings. On a random set of misclassified ECGs, agreement between two clinicians and training labels was poor (clinician 1:κ= -0.057, clinician 2:κ= -0.159). In contrast, agreement between the clinicians was very high (κ= 0.92).Discussion. The proposed prediction model performed well on the validation and hidden test data in comparison to models trained on the same data. We also discovered considerable inconsistency in training labels, which is likely to hinder development of more accurate models.