Deep learning for quality assessment of optical coherence tomography angiography images.

Optical coherence tomography angiography (OCTA) is an emerging non-invasive technique for imaging the retinal vasculature. While there are many promising clinical applications for OCTA, determination of image quality remains a challenge. We developed a deep learning-based system using a ResNet152 neural network classifier, pretrained using ImageNet, to classify images of the superficial capillary plexus in 347 scans from 134 patients. Images were also manually graded by two independent graders as a ground truth for the supervised learning models. Because requirements for image quality may vary depending on the clinical or research setting, two models were trained-one to identify high-quality images and one to identify low-quality images. Our neural network models demonstrated outstanding area under the curve (AUC) metrics for both low quality image identification (AUC = 0.99, 95%CI 0.98-1.00, [Formula: see text] = 0.90) and high quality image identification (AUC = 0.97, 95%CI 0.96-0.99, [Formula: see text] = 0.81), significantly outperforming machine-reported signal strength (AUC = 0.82, 95%CI 0.77-0.86, [Formula: see text]= 0.52 and AUC = 0.78, 95%CI 0.73-0.83, [Formula: see text] = 0.27 respectively). Our study demonstrates that techniques from machine learning may be used to develop flexible and robust methods for quality control of OCTA images.

Treatment regimens for administration of anti-vascular endothelial growth factor agents for neovascular age-related macular degeneration.

BACKGROUND: Age-related macular degeneration (AMD) is one of the leading causes of permanent blindness worldwide. The current mainstay of treatment for neovascular AMD (nAMD) is intravitreal injection of anti-vascular endothelial growth factor (anti-VEGF) agents: aflibercept, ranibizumab, and off-label bevacizumab. Injections can be given monthly, every two or three months ('extended-fixed'), or as needed (pro re nata (PRN)). A variant of PRN is 'treat-and-extend' whereby injections are resumed if recurrence is detected and then delivered with increasing intervals. Currently, injection frequency varies among practitioners, which underscores the need to characterize an optimized approach to nAMD management. OBJECTIVES: To investigate the effects of monthly versus non-monthly intravitreous injection of an anti-VEGF agent in people with newly diagnosed nAMD. SEARCH METHODS: We searched CENTRAL, MEDLINE, Embase, LILACS, and three trials registers from 2004 to October 2019; checked references; handsearched conference abstracts; and contacted pharmaceutical companies to identify additional studies. SELECTION CRITERIA: We included randomized controlled trials (RCTs) that compared different treatment regimens for anti-VEGF agents in people with newly diagnosed nAMD. We considered standard doses only (ranibizumab 0.5 mg, bevacizumab 1.25 mg, aflibercept 2.0 mg, or a combination of these). DATA COLLECTION AND ANALYSIS: We used standard Cochrane methods for trial selection, data extraction, and analysis. MAIN RESULTS: We included 15 RCTs. The total number of participants was 7732, ranging from 37 to 2457 in each trial. The trials were conducted worldwide. Of these, six trials exclusively took place in the US, and three included centers from more than one country. Eight trials were at high risk of bias for at least one domain and all trials had at least one domain at unclear risk of bias. Seven trials (3525 participants) compared a PRN regimen with a monthly injection regimen, of which five trials delivered four to eight injections using standard PRN and three delivered nine or 10 injections using a treat-and-extend regimen in the first year. The overall mean change in best-corrected visual acuity (BCVA) at one year was +8.8 letters in the monthly injection group. Compared to the monthly injection, there was moderate-certainty evidence that the mean difference (MD) in BCVA change at one year for the standard PRN subgroup was -1.7 letters (95% confidence interval (CI) -2.8 to -0.6; 4 trials, 2299 participants), favoring monthly injections. There was low-certainty evidence of a similar BCVA change with the treat-and-extend subgroup (0.5 letters, 95% CI -3.1 to 4.2; 3 trials, 1226 participants). Compared to monthly injection, there was low-certainty evidence that fewer participants gained 15 or more lines of vision with standard PRN treatment at one year (risk ratio (RR) 0.87, 95% CI 0.76 to 0.99; 4 trials, 2299 participants) and low-certainty evidence of a similar gain with treat-and-extend versus monthly regimens (RR 1.11, 95% CI 0.91 to 1.36; 3 trials, 1169 participants). The mean change in central retinal thickness was a decrease of -166 µm in the monthly injection group; the MD compared with standard PRN was 21 µm (95% CI 6 to 32; 4 trials, 2215 participants; moderate-certainty evidence) and with treat-and extend was 22 µm (95% CI 37 to -81 µm; 2 trials, 635 participants; low-certainty evidence), in favor of monthly injection. Only one trial (498 participants) measured quality of life and reported no evidence of a difference between regimens, but data could not be extracted (low-certainty evidence). Both PRN regimens (standard and 'treat-and-extend') used fewer injections than monthly regimens (standard PRN: MD -4.6 injections, 95% CI -5.4 to -3.8; 4 trials, 2336 participants; treat-and-extend: -2.4 injections, 95% CI -2.7 to -2.1 injections; moderate-certainty evidence for both comparisons). Two trials provided cost data (1105 participants, trials conducted in the US and the UK). They found that cost differences between regimens were reduced if bevacizumab rather than aflibercept or ranibizumab were used, since bevacizumab was less costly (low-certainty evidence). PRN regimens were associated with a reduced risk of endophthalmitis compared with monthly injections (Peto odds ratio (OR) 0.13, 95% CI 0.04 to 0.46; 6 RCTs, 3175 participants; moderate-certainty evidence). Using data from all trials included in this review, we estimated the risk of endophthalmitis with monthly injections to be 8 in every 1000 people per year. The corresponding risk for people receiving PRN regimens was 1 in every 1000 people per year (95% CI 0 to 4). Three trials (1439 participants) compared an extended-fixed regimen (number of injections reported in only one large trial: 7.5 in one year) with monthly injections. There was moderate-certainty evidence that BCVA at one year was similar for extended-fixed and monthly injections (MD in BCVA change compared to extended-fixed group: -1.3 letters, 95% CI -3.9 to 1.3; RR of gaining 15 letters or more: 0.94, 95% CI 0.80 to 1.10). The change in central retinal thickness was a decrease of 137 µm in the monthly group; the MD with the extended-fixed group was 8 µm (95% CI -11 to 27; low-certainty evidence). The frequency of endophthalmitis was lower in the extended-fixed regimen compared to the monthly group, but this estimate was imprecise (RR 0.19, 95% CI 0.03 to 1.11; low-certainty evidence). If we assumed a risk of 8 cases of endophthalmitis in 1000 people receiving monthly injections over one year, then the corresponding risk with extended-fixed regimen was 2 in 1000 people (95% CI 0 to 9). Other evidence comparing different extended-fixed or PRN regimens yielded inconclusive results. AUTHORS' CONCLUSIONS: We found that, at one year, monthly regimens are probably more effective than PRN regimens using seven or eight injections in the first year, but the difference is small and clinically insignificant. Endophthalmitis is probably more common with monthly injections and differences in costs between regimens are higher if aflibercept or ranibizumab are used compared to bevacizumab. This evidence only applies to settings in which regimens are implemented as described in the trials, whereas undertreatment is likely to be common in real-world settings. There are no data from RCTs on long-term effects of different treatment regimens.

Application of Schistosomiasis Consortium for Operational Research and Evaluation Study Findings to Refine Predictive Modeling of Schistosoma mansoni and Schistosoma haematobium Control in Sub-Saharan Africa.

An essential mission of the Schistosomiasis Consortium for Operational Research and Evaluation (SCORE) was to help inform global health practices related to the control and elimination of schistosomiasis. To provide more accurate, evidence-based projections of the most likely impact of different control interventions, whether implemented alone or in combination, SCORE supported mathematical modeling teams to provide simulations of community-level Schistosoma infection outcomes in the setting of real or hypothetical programs implementing multiyear mass drug administration (MDA) for parasite control. These models were calibrated using SCORE experience with Schistosoma mansoni and Schistosoma haematobium gaining and sustaining control studies, and with data from comparable programs that used community-based or school-based praziquantel MDA in other parts of sub-Saharan Africa. From 2010 to 2019, models were developed and refined, first to project the likely SCORE control outcomes, and later to more accurately reflect impact of MDA across different transmission settings, including the role of snail ecology and the impact of seasonal rainfall on snail abundance. Starting in 2014, SCORE modeling projections were also compared with the models of colleagues in the Neglected Tropical Diseases Modelling Consortium. To explore further possible improvement to program-based control, later simulations examined the cost-effectiveness of combining MDA with environmental snail control, and the utility of early impact assessment to more quickly identify persistent hot spots of transmission. This article provides a nontechnical summary of the 11 SCORE-related modeling projects and provides links to the original open-access articles describing model development and projections relevant to schistosomiasis control policy.

Society for Perioperative Assessment and Quality Improvement (SPAQI) Consensus Statement on Perioperative Smoking Cessation.

Smokers are at increased risk for surgical complications. Despite the known benefits of smoking cessation, many perioperative health care providers do not routinely provide smoking cessation interventions. The variation in delivery of perioperative smoking cessation interventions may be due to limited high-level evidence for whether smoking cessation interventions used in the general population are effective and feasible in the surgical population, as well as the challenges and barriers to implementation of interventions. Yet smoking is a potentially modifiable risk factor for improving short- and long-term patient outcomes. The purpose of the Society for Perioperative Assessment and Quality Improvement (SPAQI) Consensus Statement on Perioperative Smoking Cessation is to present recommendations based on current scientific evidence in surgical patients. These statements address questions regarding the timing and intensity of interventions, roles of perioperative health care providers, and behavioral and pharmacological interventions. Barriers and strategies to overcome challenges surrounding implementation of interventions and future areas of research are identified. These statements are based on the current state of knowledge and its interpretation by a multidisciplinary group of experts at the time of publication.

Cost and Selection of Ophthalmic Anti-Vascular Endothelial Growth Factor Agents.

Anti-vascular endothelial growth factor (anti-VEGF) drugs - ranibizumab, aflibercept, and off-label bevacizumab - are vital to the treatment of common retinal diseases, including exudative age-related macular degeneration (AMD), diabetic macular edema (DME), and macular edema (ME) associated with retinal vein occlusion (RVO). Given the high prevalence of AMD and retinal vascular diseases, anti-VEGF agents represent a large cost burden to the United States (US) healthcare system. Although ranibizumab and aflibercept are 30-fold more expensive per injection than bevacizumab, the two more costly medications are commonly used in the US, even though all three have been shown to be effective and safe for treatment of these retinal diseases. We investigated the availability and content of professional ophthalmic guidelines on cost consideration in the selection of anti-VEGF agents. We found that current professional guidelines were limited in availability and lacked specific guidance on cost-based anti-VEGF drug selection. This represents a missed opportunity to encourage the practice of value-based medicine. [Full article available at http://rimed.org/rimedicaljournal-2016-05.asp, free with no login].