Bias estimation in study design: a meta-epidemiological analysis of transcatheter versus surgical aortic valve replacement.

BACKGROUND: Paucity of RCTs of non-drug technologies lead to widespread dependence on non-randomized studies. Relationship between nonrandomized study design attributes and biased estimates of treatment effects are poorly understood. Our purpose was to estimate the bias associated with specific nonrandomized study attributes among studies comparing transcatheter aortic valve implantation with surgical aortic valve replacement for the treatment of severe aortic stenosis. RESULTS: We included 6 RCTs and 87 nonrandomized studies. Surgical risk scores were similar for comparison groups in RCTs, but were higher for patients having transcatheter aortic valve implantation in nonrandomized studies. Nonrandomized studies underestimated the benefit of transcatheter aortic valve implantation compared with RCTs. For example, nonrandomized studies without adjustment estimated a higher risk of postoperative mortality for transcatheter aortic valve implantation compared with surgical aortic valve replacement (OR 1.43 [95% CI 1.26 to 1.62]) than high quality RCTs (OR 0.78 [95% CI 0.54 to 1.11). Nonrandomized studies using propensity score matching (OR 1.13 [95% CI 0.85 to 1.52]) and regression modelling (OR 0.68 [95% CI 0.57 to 0.81]) to adjust results estimated treatment effects closer to high quality RCTs. Nonrandomized studies describing losses to follow-up estimated treatment effects that were significantly closer to high quality RCT than nonrandomized studies that did not. CONCLUSION: Studies with different attributes produce different estimates of treatment effects. Study design attributes related to the completeness of follow-up may explain biased treatment estimates in nonrandomized studies, as in the case of aortic valve replacement where high-risk patients were preferentially selected for the newer (transcatheter) procedure.

Five-Year Outcomes of Graft-Free Tube Shunts and Risk Factors for Tube Exposures in Glaucoma.

PRCIS: Use of a scleral tunnel technique instead of a patch graft can be considered in most cases of tube shunt implantation. Grafts may still be considered in younger (<65 y old) East Asians. PURPOSE: (1) To assess the risk factors for tube exposure with a graft-free implantation technique and (2) to examine 5-year outcomes of graft-free tube shunt insertion. METHODS: This was a retrospective case series of 204 consecutive eyes undergoing a glaucoma tube shunt implantation with a scleral tunnel technique in lieu of a graft. Preoperative and postoperative best-corrected visual acuity, intraocular pressure, and number of glaucoma medications were compared. Failure was defined as the following: (1) intraocular pressure >21 mm Hg or ≤5 mm Hg on 2 consecutive visits after 3 mo; (2) required additional glaucoma surgery; (3) loss of light perception. Univariable and multivariable regression analyses were conducted to identify risk factors of tube exposures. RESULTS: Intraocular pressure and the number of glaucoma medications were significantly decreased at all postoperative time points ( P <0.001). Success rates were 91% at year 1, 75% at year 3, and 67% at year 5. The most common early (<3 mo) complication was tube malpositioning. The most common late (>3 mo to 5 y) complications were corneal complications and uncontrolled intraocular pressure. By year 5, 6.9% of tubes were exposed. Multivariable regression showed that age less than 65 years old (odds ratio: 3.66, P =0.04) and East Asian ethnicity (odds ratio: 3.36, P =0.04) were associated with significantly increased risk of tube exposure. CONCLUSIONS: Graft-free glaucoma tube implantation has comparable long-term outcomes and complication rates to shunts with a graft. Younger (<65 y old) East Asians are at greater risk of tube exposure without a graft.

Utility of online GCA risk models in predicting the result of temporal artery biopsy within a clinical setting: study of diagnostic and screening tests.

BACKGROUND: Temporal artery biopsy (TAB) is the gold standard for the diagnosis of giant cell arteritis (GCA) but has many limitations. The Ing model, González-López model, and Weis model are tools to estimate a patient's likelihood of GCA. This study investigates the utility of these prediction models in triaging patients referred for TAB. METHODS: This study is a retrospective examination of patients who underwent TAB by a neuro-ophthalmologist over a 5-year period. Data collected through chart review were inputted into prediction models to evaluate GCA risk and compared with TAB results and clinical diagnosis. Cut-off values for 100% sensitivity and specificity for TAB result were used to determine whether TAB could be avoided where there was preoperative certainty of the result. RESULTS: Among 155 eligible patients, mean age was 73 years, and 78.1% were female. TAB was negative in 103 patients (66.5%) and positive in 42 patients (27.1%). Twenty-three patients (22.3%) were diagnosed clinically and treated for biopsy-negative GCA. The Ing model had no positive biopsies below 10.59% and no negative biopsies above 68.44%. The González-López model had no positive biopsies below 0.27% and no negative biopsies above 98.08%. The Weis model had no positive biopsies with a score less than zero. CONCLUSION: Forty-one biopsies (28.9%) could have been avoided using the Ing model, 9 (6.34%) using the González-López model, and 28 (19.7%) using the Weis model. The findings suggest that the Ing and Weis models are useful screening tools for GCA with the potential to improve the effective use of health care resources.

Patterns and utility of myelin oligodendrocyte glycoprotein (MOG) antibody testing in cerebrospinal fluid.

BACKGROUND: Myelin oligodendrocyte glycoprotein (MOG) antibody-associated disease (MOGAD) is an idiopathic central nervous system (CNS) demyelinating disease gaining recognition with wider availability of cell-based assay (CBA) testing and recently published diagnostic criteria. However, uncertainty remains regarding the interpretation of antibody titers, particularly cerebrospinal fluid (CSF) MOG antibody titers. METHODS: All MOG IgG CBA results performed by the provincial MitogenDx laboratory in Alberta from July 2017 to July 2023 were retrieved. Chart review was performed in patients with both serum and CSF testing and ≥ 1 positive MOG antibody result. Demographics, antibody titers, clinical and imaging features, treatment, and diagnosis were analyzed based on serum/CSF status. RESULTS: Among 4494 MOG CBA assays, there were 413 CSF samples in 402 patients, and 268 patients had at least one associated serum sample. Mean time between CSF and serum testing was 20.9 days (range 0-870 days), most with testing within 30 days. Five of the 268 patients had serum positive/CSF positive MOG antibodies, 4 with acute disseminated encephalomyelitis and 1 with longitudinally extensive transverse myelitis. Twenty-three patients had serum positive/CSF negative MOG and 13/23 with optic neuritis. CSF MOG antibody positive patients were younger, and more likely to remain MOG seropositive versus CSF negative patients. No seronegative patient had MOG antibodies in CSF. CONCLUSIONS: In province-wide testing, CSF MOG antibodies were rare, only in MOG seropositive patients and none with optic neuritis. Our study does not support a clear role for CSF MOG antibody testing in the majority of patients, although further study is required.

A Comparison of Diagnostic Accuracy of Imaging Modalities to Detect Optic Disc Drusen: The Age of Enhanced Depth Imaging Optical Coherence Tomography.

PURPOSE: To identify the most accurate diagnostic imaging modality to detect optic disc drusen (ODD) between B-scan ultrasonography (US), fundus photography, fundus autofluorescence (FAF), and enhanced depth imaging optical coherence tomography (EDI-OCT). DESIGN: Comparative diagnostic analysis. METHODS: Two hundred five eyes of 105 patients referred to 2 tertiary care neuro-ophthalmology clinics for suspected ODD were recruited: 108 eyes had ODD and 97 did not have ODD. All eyes received a full in-person ophthalmic exam with 3D view of the optic nerve and all 4 imaging modalities. Images were independently reviewed by 3 masked neuro-ophthalmologists to determine the presence or absence of ODD. Final interpretation was made through consensus. The reference standard was defined as the attending ophthalmologist's clinical judgement based on open chart review, with access to all image modalities and clinical information, including disease course. Main outcome measures were sensitivity, specificity, accuracy, and precision for each imaging modality. Examiner confidence was quantified as the proportion of eyes in which the reviewers were certain of their decision. RESULTS: The EDI-OCT had the highest sensitivity and accuracy (95%, 97%) to detect ODD, compared with FAF (84%, 92%), US (74%, 86%), and fundus photography (38%, 66%), respectively. All image modalities had high specificity (> 97%) and precision (> 93%). The EDI-OCT also had highest examiner confidence (96%) compared with all others (88%). CONCLUSIONS: Among all modalities, EDI-OCT was the imaging modality with the highest diagnostic utility for the detection of ODD and should be considered as the preferred initial diagnostic modality.

Deep Ensemble Learning for Retinal Image Classification.

Purpose: Vision impairment affects 2.2 billion people worldwide, half of which is preventable with early detection and treatment. Currently, automatic screening of ocular pathologies using convolutional neural networks (CNNs) on retinal fundus photographs is limited to a few pathologies. Simultaneous detection of multiple ophthalmic pathologies would increase clinical usability and uptake. Methods: Two thousand five hundred sixty images were used from the Retinal Fundus Multi-Disease Image Dataset (RFMiD). Models were trained (n = 1920) and validated (n = 640). Five selected CNN architectures were trained to predict the presence of any pathology and categorize the 28 pathologies. All models were trained to minimize asymmetric loss, a modified form of binary cross-entropy. Individual model predictions were averaged to obtain a final ensembled model and assessed for mean area under the receiver-operator characteristic curve (AUROC) for disease screening (healthy versus pathologic image) and classification (AUROC for each class). Results: The ensemble network achieved a disease screening (healthy versus pathologic) AUROC score of 0.9613. The highest single network score was 0.9586 using the SE-ResNeXt architecture. For individual disease classification, the average AUROC score for each class was 0.9295. Conclusions: Retinal fundus images analyzed by an ensemble of CNNs trained to minimize asymmetric loss were effective in detection and classification of ocular pathologies than individual models. External validation is needed to translate machine learning models to diverse clinical contexts. Translational Relevance: This study demonstrates the potential benefit of ensemble-based deep learning methods on improving automatic screening and diagnosis of multiple ocular pathologies from fundoscopy imaging.

A cost-utility analysis of transcatheter versus surgical aortic valve replacement for the treatment of aortic stenosis in the population with intermediate surgical risk.

OBJECTIVE: Although transcatheter aortic valve implantation has been shown to be noninferior to surgical aortic valve replacement in patients with severe aortic stenosis at intermediate surgical risk, the cost-effectiveness of this strategy in this population is unknown. Our objective was to conduct a cost-utility analysis comparing transcatheter aortic valve implantation with surgical aortic valve replacement in the population with intermediate risk severe aortic stenosis. METHODS: A fully probabilistic Markov model with 30-day cycles was constructed from the Canadian third-party payer's perspective to estimate the difference in cost and effectiveness (measured as quality-adjusted life years) of transcatheter aortic valve implantation versus surgical aortic valve replacement for intermediate-risk patients over a lifetime time horizon, discounted at 1.5% per annum. Clinical trial data from The Placement of Aortic Transcatheter Valve 2 informed the efficacy inputs. Costs (adjusted to 2016 Canadian dollars) were obtained from the Canadian Institute of Health Information and the Ontario Schedule of Benefits. Incremental cost-effectiveness ratios were calculated. RESULTS: In the base-case analysis, total lifetime costs for transcatheter aortic valve implantation were $10,548 higher than surgical aortic valve replacement but added 0.23 quality-adjusted life years, for an incremental cost-effectiveness ratio of $46,083/quality-adjusted life-years gained. Deterministic 1-way analyses showed that the incremental cost-effectiveness ratio was sensitive to rates of complications and cost of the transcatheter aortic valve implantation prosthesis. There was moderate-to-high parameter uncertainty; transcatheter aortic valve implantation was the preferred option in only 52.7% and 55.4% of the simulations at a $50,000 and $100,000 per quality-adjusted life years willingness-to-pay thresholds, respectively. CONCLUSIONS: On the basis of current evidence, transcatheter aortic valve implantation may be cost-effective for the treatment of severe aortic stenosis in patients with intermediate surgical risk. There remains moderate-to-high uncertainty surrounding the base-case incremental cost-effectiveness ratio.