Toward safer ophthalmic artificial intelligence via distributed validation on real-world data.

PURPOSE OF REVIEW: The current article provides an overview of the present approaches to algorithm validation, which are variable and largely self-determined, as well as solutions to address inadequacies. RECENT FINDINGS: In the last decade alone, numerous machine learning applications have been proposed for ophthalmic diagnosis or disease monitoring. Remarkably, of these, less than 15 have received regulatory approval for implementation into clinical practice. Although there exists a vast pool of structured and relatively clean datasets from which to develop and test algorithms in the computational 'laboratory', real-world validation remains key to allow for safe, equitable, and clinically reliable implementation. Bottlenecks in the validation process stem from a striking paucity of regulatory guidance surrounding safety and performance thresholds, lack of oversight on critical postdeployment monitoring and context-specific recalibration, and inherent complexities of heterogeneous disease states and clinical environments. Implementation of secure, third-party, unbiased, pre and postdeployment validation offers the potential to address existing shortfalls in the validation process. SUMMARY: Given the criticality of validation to the algorithm pipeline, there is an urgent need for developers, machine learning researchers, and end-user clinicians to devise a consensus approach, allowing for the rapid introduction of safe, equitable, and clinically valid machine learning implementations.

Transepithelial versus Epithelium-off Corneal Collagen Cross-linking for Corneal Ectasia: A Systematic Review and Meta-analysis.

TOPIC: To evaluate the safety and efficacy of transepithelial corneal cross-linking in comparison with the established epithelium-off technique for corneal ectasia. CLINICAL RELEVANCE: Considerable debate exists regarding whether transepithelial and epithelium-off cross-linking are comparable in their safety and efficacy. METHODS: We searched 16 electronic databases, including Medline, Embase, Web of Science, and the grey literature, current to July 8, 2020, for randomized controlled trials comparing transepithelial and epithelium-off cross-linking for corneal ectasia. We excluded studies evaluating cross-linking for nonectatic indications, as well as non-randomized controlled trials. Our primary outcome was the change in maximal keratometry (Kmax) at 12 months after cross-linking, and we considered additional topographic, visual, and safety outcomes. We summarized our analyses by calculating weighted mean differences (MDs) with associated 95% confidence intervals (CIs) for continuous outcomes and relative risks (RRs) with corresponding 95% CIs for dichotomous outcomes. We conducted trial sequential analysis to determine whether the required information size was met for each outcome. The quality of individual trials was evaluated using the Cochrane Collaboration's risk of bias assessment tool, and the evidence was assessed at an outcome level using the Grading of Recommendations Assessment, Development, and Evaluation methodology. RESULTS: Twelve studies totaling 966 eyes were eligible. A significant difference was found between transepithelial and epithelium-off cross-linking groups in the change in Kmax at 12 months (MD, 0.75; 95% CI, 0.23-1.28; P = 0.004; primary outcome) and at longest follow-up (MD, 1.20; 95% CI, 0.62-1.77; P < 0.001; secondary outcome) after treatment. No significant difference was found between the 2 groups when examining uncorrected distance visual acuity (MD, 0.04; 95% CI, -0.06 to 0.14; P = 0.386) or corrected distance visual acuity (MD, 0.01; 95% CI, -0.06 to 0.09; P = 0.732). Transepithelial cross-linking was associated with significantly fewer complications than the epithelium-off approach (RR, 0.22; 95% CI, 0.06-0.79; P = 0.020), although it was associated with an increased rate of disease progression at 12 months after treatment (RR, 4.49; 95% CI, 1.24-16.25; P = 0.022). The required information size was met for our primary outcome and trial sequential analysis supported the conventional meta-analysis. The quality of evidence was rated as moderate using the Grading of Recommendations Assessment, Development, and Evaluation methodology. DISCUSSION: The efficacy of transepithelial cross-linking remains inferior to the epithelium-off approach, although it is significantly safer.

Bedside Optic Nerve Ultrasonography for Diagnosing Increased Intracranial Pressure: A Systematic Review and Meta-analysis.

Background: Optic nerve ultrasonography (optic nerve sheath diameter sonography) has been proposed as a noninvasive, quick method for diagnosing increased intracranial pressure. Purpose: To examine the accuracy of optic nerve ultrasonography for diagnosing increased intracranial pressure in children and adults. Data Sources: 13 databases from inception through May 2019, reference lists, and meeting proceedings. Study Selection: Prospective optic nerve ultrasonography diagnostic accuracy studies, published in any language, involving any age group or reference standard. Data Extraction: 3 reviewers independently abstracted data and performed quality assessment. Data Synthesis: Of 71 eligible studies involving 4551 patients, 61 included adults, and 35 were rated as having low risk of bias. The pooled sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio of optic nerve ultrasonography in patients with traumatic brain injury were 97% (95% CI, 92% to 99%), 86% (CI, 74% to 93%), 6.93 (CI, 3.55 to 13.54), and 0.04 (CI, 0.02 to 0.10), respectively. Respective estimates in patients with nontraumatic brain injury were 92% (CI, 86% to 96%), 86% (CI, 77% to 92%), 6.39 (CI, 3.77 to 10.84), and 0.09 (CI, 0.05 to 0.17). Accuracy estimates were similar among studies stratified by patient age, operator specialty and training level, reference standard, sonographer blinding status, and cutoff value. The optimal cutoff for optic nerve sheath dilatation on ultrasonography was 5.0 mm. Limitation: Small studies, imprecise summary estimates, possible publication bias, and no evaluation of effect on clinical outcomes. Conclusion: Optic nerve ultrasonography can help diagnose increased intracranial pressure. A normal sheath diameter measurement has high sensitivity and a low negative likelihood ratio that may rule out increased intracranial pressure, whereas an elevated measurement, characterized by a high specificity and positive likelihood ratio, may indicate increased intracranial pressure and the need for additional confirmatory tests. Primary Funding Source: None. (PROSPERO: CRD42017055485).

Atraumatic versus conventional lumbar puncture needles: a systematic review and meta-analysis.

BACKGROUND: Atraumatic needles have been proposed to lower complication rates after lumbar puncture. However, several surveys indicate that clinical adoption of these needles remains poor. We did a systematic review and meta-analysis to compare patient outcomes after lumbar puncture with atraumatic needles and conventional needles. METHODS: In this systematic review and meta-analysis, we independently searched 13 databases with no language restrictions from inception to Aug 15, 2017, for randomised controlled trials comparing the use of atraumatic needles and conventional needles for any lumbar puncture indication. Randomised trials comparing atraumatic and conventional needles in which no dural puncture was done (epidural injections) or without a conventional needle control group were excluded. We screened studies and extracted data from published reports independently. The primary outcome of postdural-puncture headache incidence and additional safety and efficacy outcomes were assessed by random-effects and fixed-effects meta-analysis. This study is registered with the International Prospective Register of Systematic Reviews, number CRD42016047546. FINDINGS: We identified 20 241 reports; after exclusions, 110 trials done between 1989 and 2017 from 29 countries, including a total of 31 412 participants, were eligible for analysis. The incidence of postdural-puncture headache was significantly reduced from 11·0% (95% CI 9·1-13·3) in the conventional needle group to 4·2% (3·3-5·2) in the atraumatic group (relative risk 0·40, 95% CI 0·34-0·47, p<0·0001; I2=45·4%). Atraumatic needles were also associated with significant reductions in the need for intravenous fluid or controlled analgesia (0·44, 95% CI 0·29-0·64; p<0·0001), need for epidural blood patch (0·50, 0·33-0·75; p=0·001), any headache (0·50, 0·43-0·57; p<0·0001), mild headache (0·52, 0·38-0·70; p<0·0001), severe headache (0·41, 0·28-0·59; p<0·0001), nerve root irritation (0·71, 0·54-0·92; p=0·011), and hearing disturbance (0·25, 0·11-0·60; p=0·002). Success of lumbar puncture on first attempt, failure rate, mean number of attempts, and the incidence of traumatic tap and backache did not differ significantly between the two needle groups. Prespecified subgroup analyses of postdural-puncture headache revealed no interactions between needle type and patient age, sex, use of prophylactic intravenous fluid, needle gauge, patient position, indication for lumbar puncture, bed rest after puncture, or clinician specialty. These results were rated high-quality evidence as examined using the grading of recommendations assessment, development, and evaluation. INTERPRETATION: Among patients who had lumbar puncture, atraumatic needles were associated with a decrease in the incidence of postdural-puncture headache and in the need for patients to return to hospital for additional therapy, and had similar efficacy to conventional needles. These findings offer clinicians and stakeholders a comprehensive assessment and high-quality evidence for the safety and efficacy of atraumatic needles as a superior option for patients who require lumbar puncture. FUNDING: None.

A huntingtin-mediated fast stress response halting endosomal trafficking is defective in Huntington's disease.

Cellular stress is a normal part of the aging process and is especially relevant in neurodegenerative disease. Canonical stress responses, such as the heat shock response, activate following exposure to stress and restore proteostasis through the action of isomerases and chaperones within the cytosol. Through live-cell imaging, we demonstrate involvement of the Huntington's disease (HD) protein, huntingtin, in a rapid cell stress response that lies temporally upstream of canonical stress responses. This response is characterized by the formation of distinct cytosolic puncta and reversible localization of huntingtin to early endosomes. The formation of these puncta, which we have termed huntingtin stress bodies (HSBs), is associated with arrest of early-to-recycling and early-to-late endosomal trafficking. The critical domains for this response have been mapped to two regions of huntingtin flanking the polyglutamine tract, and we observe polyglutamine-expanded huntingtin-expressing cells to be defective in their ability to recover from this stress response. We propose that HSB formation rapidly diverts high ATP use from vesicular trafficking during stress, thus mobilizing canonical stress responses without relying on increased energy metabolism, and that restoration from this response is defective in HD.

MD/PhD Training in Canada: Results from a national trainee and program director review.

PURPOSE: There has been limited examination of clinician scientist training in Canada, particularly regarding training integration and funding. This study assessed program structure, funding, tuition and mentorship structures available at Canadian MD/PhD programs. METHODS: Clinician Investigator Trainee Association of Canada administered an anonymous survey to current trainees and program directors that captured program structure, trainee funding, tuition and mentorship opportunities and needs across institutions. RESULTS: In June 2015, 101/228 (44%) trainees and 9/13 (69%) program directors completed the online survey. In all programs, students completed the PhD degree prior to clerkship training. Seven programs offered research training upon completion of pre-clerkship, four offered concurrent clinical and research training, and three offered alternative structures. Nine held seminars exposing students to clinical and research integration and two offered clinician scientist skills courses. Stipend funding and tuition varied, especially during clinical training years. Regarding mentorship, all programs held regular meetings, though eight programs do not have formal mentorship opportunities. Both trainees and program directors identified the need for further career planning and development support as a student priority. CONCLUSION: MD/PhD programs varied by program structure, funding, tuition and mentorship opportunities. Mechanisms to share and spread program innovations should be instated. Students may benefit from concurrent research and clinical training as well as courses specific to clinician scientist skill development. Decreasing debt burden may attract and retain trainees in this demanding path. To ensure mentorship programs align with trainee priorities, program directors should directly collaborate with students in their development and evaluation.

New meaning for NLP: the trials and tribulations of natural language processing with GPT-3 in ophthalmology.

Natural language processing (NLP) is a subfield of machine intelligence focused on the interaction of human language with computer systems. NLP has recently been discussed in the mainstream media and the literature with the advent of Generative Pre-trained Transformer 3 (GPT-3), a language model capable of producing human-like text. The release of GPT-3 has also sparked renewed interest on the applicability of NLP to contemporary healthcare problems. This article provides an overview of NLP models, with a focus on GPT-3, as well as discussion of applications specific to ophthalmology. We also outline the limitations of GPT-3 and the challenges with its integration into routine ophthalmic care.

Reinforcement learning in ophthalmology: potential applications and challenges to implementation.

Reinforcement learning is a subtype of machine learning in which a virtual agent, functioning within a set of predefined rules, aims to maximise a specified outcome or reward. This agent can consider multiple variables and many parallel actions at once to optimise its reward, thereby solving complex, sequential problems. Clinical decision making requires physicians to optimise patient outcomes within a set practice framework and, thus, presents considerable opportunity for the implementation of reinforcement learning-driven solutions. We provide an overview of reinforcement learning, and focus on potential applications within ophthalmology. We also explore the challenges associated with development and implementation of reinforcement learning solutions and discuss possible approaches to address them.

Functional characterization of variants of unknown significance in a spinocerebellar ataxia patient using an unsupervised machine learning pipeline.

CAG-expanded ATXN7 has been previously defined in the pathogenesis of spinocerebellar ataxia type 7 (SCA7), a polyglutamine expansion autosomal dominant cerebellar ataxia. Pathology in SCA7 occurs as a result of a CAG triplet repeat expansion in excess of 37 in the first exon of ATXN7, which encodes ataxin-7. SCA7 presents clinically with spinocerebellar ataxia and cone-rod dystrophy. Here, we present a novel spinocerebellar ataxia variant occurring in a patient with mutations in both ATXN7 and TOP1MT, which encodes mitochondrial topoisomerase I (top1mt). Using machine-guided, unbiased microscopy image analysis, we demonstrate alterations in ataxin-7 subcellular localization, and through high-fidelity measurements of cellular respiration, bioenergetic defects in association with top1mt mutations. We identify ataxin-7 Q35P and top1mt R111W as deleterious mutations, potentially contributing to disease states. We recapitulate our mutations through Drosophila genetic models. Our work provides important insight into the cellular biology of ataxin-7 and top1mt and offers insight into the pathogenesis of spinocerebellar ataxia applicable to multiple subtypes of the illness. Moreover, our study demonstrates an effective pipeline for the characterization of previously unreported genetic variants at the level of cell biology.

Uveal melanoma incidence trends in Canada: 1992-2010 vs. 2011-2017.

Background/aims: Uveal melanoma is the most common type of non-cutaneous melanoma and the most common ocular malignancy in the adult population, especially affecting Caucasians (98% of cases). Despite its low incidence rate, we have noted increasing incidence trends in recent years. Methods: We analyzed uveal melanoma incidence data using the Canadian Cancer Registry (CCR) for 2011-2017 years. The data was examined using the International Classification of Diseases for Oncology, Third Edition, codes for all uveal melanoma subtypes. The data for 2011-2017 was then compared to previously published work by our research group for uveal melanoma incidence in Canada between 1992 and 2010 using the same methodology. Results: Between 2011 and 2017, 1,215 patients were diagnosed with uveal melanoma, 49% of whom were females. The percentage distribution of uveal melanoma between the sexes was similar between 1992-2010 and 2011-2017, whereby of the 2,215 diagnoses of uveal melanoma in 1992-2010, 47.9% were females. The change in the incidence rate for this cancer has doubled between 1992-2010 and 2011-2017, from 0.074 to 0.15 cases per million individuals per year. Our study documents that the Canadian 2011-2017 age-standardized incidence rate (ASIR) for uveal melanoma against the World Health Organization (WHO) 2000-2025 world population standard was 5.09 cases per million individuals per year (95% confidence interval, 4.73-5.44), as compared with the 1992-2010 rate of 3.34 cases per million individuals per year (95% confidence interval, CI 3.20 to 3.47). Conclusion: This work demonstrates an ongoing, steady increase in uveal melanoma incidence in Canada in recent years.

Frame-based versus frameless stereotactic brain biopsies: A systematic review and meta-analysis.

BACKGROUND: Stereotactic brain biopsy techniques have been a focus of rapid technological innovation. The recent advent of frameless stereotaxy has invited the question of whether it can provide the same diagnostic yield as frame-based techniques, without increasing risk of harm to patients. The goal of this meta-analysis was to compare each of these techniques in terms of yield and safety. METHODS: We independently searched four databases for English studies comparing frameless and frame-based stereotactic brain biopsies. Our primary outcome was biopsy diagnostic yield. Our secondary outcomes included mortality, morbidity (e.g., symptomatic postbiopsy intracranial hemorrhage, asymptomatic postbiopsy intracranial hemorrhage, new postbiopsy neurological deficit, and postbiopsy seizure), and frequency of repeat biopsy. We calculated pooled estimates and relative risks for dichotomous outcomes using Review Manager 5.3, with corresponding 95% confidence intervals. RESULTS: A total of 3256 stereotactic brain biopsies (2050 frame based and 1206 frameless), from 20 studies, were included in our final analysis. The results did not demonstrate any significant difference between the two stereotactic systems in terms of diagnostic yield (risk ratio [RR] 1.00, 95% confidence interval [CI] 0.99-1.02, P = 0.64, I2 = 0%). The only significant difference was the increased frequency of asymptomatic hemorrhages in the frameless group (RR 1.37, 95% CI 1.06-1.75, P = 0.01, I2 = 0%). Application of Grading of Recommendations Assessment, Development, and Evaluation to the results yielded very low quality of all outcomes. CONCLUSION: Based on very low-quality evidence, both frame-based and frameless stereotaxy are safe and effective for biopsy of intracranial tumors. Further study of patient preference and cost comparing analysis is required to identify if either modality should be preferred.

A global review of publicly available datasets for ophthalmological imaging: barriers to access, usability, and generalisability.

Health data that are publicly available are valuable resources for digital health research. Several public datasets containing ophthalmological imaging have been frequently used in machine learning research; however, the total number of datasets containing ophthalmological health information and their respective content is unclear. This Review aimed to identify all publicly available ophthalmological imaging datasets, detail their accessibility, describe which diseases and populations are represented, and report on the completeness of the associated metadata. With the use of MEDLINE, Google's search engine, and Google Dataset Search, we identified 94 open access datasets containing 507 724 images and 125 videos from 122 364 patients. Most datasets originated from Asia, North America, and Europe. Disease populations were unevenly represented, with glaucoma, diabetic retinopathy, and age-related macular degeneration disproportionately overrepresented in comparison with other eye diseases. The reporting of basic demographic characteristics such as age, sex, and ethnicity was poor, even at the aggregate level. This Review provides greater visibility for ophthalmological datasets that are publicly available as powerful resources for research. Our paper also exposes an increasing divide in the representation of different population and disease groups in health data repositories. The improved reporting of metadata would enable researchers to access the most appropriate datasets for their needs and maximise the potential of such resources.

Nontraumatic subperiosteal orbital hemorrhage in a laboring patient with gestational immune thrombocytopenic purpura.

PURPOSE: To describe a case of nontraumatic subperiosteal orbital hemorrhage (NTSOH) in a laboring patient with gestational immune thrombocytopenic purpura. OBSERVATIONS: A 28-year-old G3P2 laboring patient was urgently evaluated by our ophthalmology unit after she developed sudden onset left eye proptosis, headache, and diplopia in the final hour of pushing. The patient's platelet count was markedly decreased at 45,000 and subsequent work-up established a diagnosis of gestational immune thrombocytopenic purpura. On examination, visual acuity was 20/25 bilaterally and there was a minus two restriction in upgaze in the left eye and a left hypotropia in primary position. Computed tomography demonstrated an elliptical, hyperdense collection at the left orbital roof consistent with NTSOH. The patient was deemed clinically stable through serial examinations and symptoms resolved with conservative management. CONCLUSIONS AND IMPORTANCE: NTSOH is a rare, potentially sight-threatening condition that requires timely ophthalmological evaluation. To our knowledge, this is the first report in the literature of NTSOH in a laboring patient with gestational immune thrombocytopenic purpura. Consideration of the possibility of NTSOH as a complication in this population may allow for appropriate diagnosis, monitoring, and treatment when indicated.

Protocol for a systematic review, meta-analysis, and trial sequential analysis of clinical outcomes following accelerated versus conventional corneal collagen cross-linking for corneal ectasia.

BACKGROUND: Collagen cross-linking (CXL) is an evolving procedure that enhances the biomechanical rigidity of the cornea and can slow or halt ectatic disease. CXL requires delivery of 5.4 J/cm2 of ultraviolet A (UVA) radiation to a cornea saturated with riboflavin in order to induce cross-link formation. The conventional CXL procedure achieves this fluence by exposing the cornea to a 3 mW/cm2 UVA lamp for 30 min; however, some surgeons have proposed accelerated protocols which achieve the same fluence in a shorter period of time by using a higher power light source. Whether accelerated protocols are as effective in arresting ectasia as the established conventional procedure remains unclear. Accordingly, this study will systematically review and synthesise the evidence on accelerated CXL and compare it to the conventional approach across an array of clinical outcomes. METHODS: We will search 16 electronic databases, including MEDLINE, Embase, and the Cochrane Library, from inception to February 1, 2019. We will include all randomised controlled trials comparing accelerated and conventional CXL for any corneal ectatic disease. Two reviewers will independently screen search results to identify eligible articles, complete data collection, and conduct quality assessment. The quality of individual trials will be assessed using the Cochrane Collaboration's Risk of Bias Assessment Tool. Our primary outcome will be the change in maximal keratometry (Kmax) at 12 months following treatment. Additional outcomes will include: incidence of disease progression, incidence of serious adverse events, as well as change in Kmax at longest follow-up, mean stromal demarcation line depth, mean uncorrected distance visual acuity, mean corrected distance visual acuity, mean Kmax, mean endothelial cell density, mean central corneal thickness, mean spherical equivalent, mean intraocular pressure, and mean corneal power, at 12 months following treatment. We will calculate relative risks and 95% confidence intervals (CIs) for dichotomous outcomes and weighted mean differences and corresponding 95% CIs for continuous outcomes. Prespecified subgroup analyses will be performed to investigate heterogeneity. We will rate the overall quality of evidence using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. DISCUSSION: This review will provide a comprehensive evaluation of the evidence on accelerated and conventional CXL approaches and serve to inform clinical practice, medical device design, and future research. Evaluating variations of the CXL protocol aimed at reducing treatment duration is of critical importance and a prerequisite to expanding treatment availability to more patients. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42018104151.

Surgical Management for Olecranon Fractures in Adults: A Systematic Review and Meta-analysis.

Olecranon fractures are common upper-extremity fractures amenable to various surgical techniques. The authors conducted a systematic review and meta-analysis to evaluate surgical management strategies for olecranon fractures in adult patients. Two investigators independently screened articles from records obtained from MEDLINE, Embase, SPORTDiscus, CINAHL, and the Cochrane Library databases. The authors compared tension-band wiring (TBW) with plate fixation (PF) for the outcomes of complications and hardware removal. Subgroup analyses were conducted comparing randomized controlled trials with observational studies for each outcome. In addition, the authors evaluated novel fixation techniques other than TBW and PF. Twenty-four studies were included in the systematic review and 10 studies were included in the meta-analysis of TBW vs PF. Compared with TBW (n=270), patients who underwent PF (n=369) had a significantly lower complication rate (relative risk, 0.48; 95% confidence interval, 0.36-0.64; P<.01; I2=16%). Plate fixation (41 of 332) also had less hardware removal compared with TBW (79 of 236; relative risk, 0.36; 95% confidence interval, 0.25-0.50; P<.01; I2=0%). Subgroup analyses showed no difference when comparing randomized controlled trials with observational studies for the outcomes of complication rate (P=.45) and hardware removal (P=.54). Qualitative analysis revealed novel methods of olecranon fixation: a modified cable pin system, tension banding through 2 cannulated screws, olecranon memory connector fixation, and cancellous screw TBW. This meta-analysis revealed significantly lower complication and hardware removal rates for PF compared with TBW. Several novel methods of olecranon fixation may represent viable alternatives. [Orthopedics. 2019; 42(2):75-82.].

Electroretinography in psychiatry: A systematic literature review.

This review aims to consolidate the available information on use of electroretinography as a diagnostic tool in psychiatry. The electroretinogram (ERG) has been found to have diagnostic utility in cocaine withdrawal (reduced light-adapted b-wave response), major depressive disorder (reduced contrast gain in pattern ERG), and schizophrenia (reduced a- and b-wave amplitudes). This review examines these findings as well as the applicability of ERG to substance use disorder, Alzheimer's disease, autism spectrum disorder, panic disorder, eating disorders, attention deficit hyperactivity disorder, and medication use. While there have been promising results, current research suffers from a lack of specificity. Further research that quantifies anomalies in ERG present in psychiatric illness is needed.

Transepithelial versus epithelium-off corneal collagen cross-linking for corneal ectasia: protocol for a systematic review, meta-analysis and trial sequential analysis of randomised controlled trials.

INTRODUCTION: Corneal ectasias are progressive, degenerative ocular diseases defined by abnormal structural changes in the cornea, leading to distortion of vision and substantial reduction in quality of life. Corneal collagen cross-linking (CXL) increases the biomechanical rigidity of the cornea and has been shown to halt ectatic processes. The established CXL protocol requires removal of the corneal epithelium. However, some surgeons have proposed transepithelial approaches to enhance patient recovery and minimise adverse events. Whether novel transepithelial approaches are as effective in arresting ectasia as the established epithelium-off protocol remains unclear. This study will systematically review the evidence on transepithelial CXL approaches and compare it to the epithelium-off protocol. METHODS AND ANALYSIS: We will include randomised controlled trials (RCTs) comparing transepithelial and epithelium-off CXL for any corneal ectasia. We will search 16 electronic databases including MEDLINE and Embase, as well as the grey literature. Two reviewers will independently screen search results to identify eligible studies, complete data abstraction and conduct quality assessment. We will assess the quality of individual RCTs using the Cochrane risk of bias assessment tool. Our primary outcome will be the change in maximal keratometry at 12 months after treatment, and we will examine 11 additional outcomes. We will summarise our analyses by measures of association (relative risk or odds ratio) and corresponding 95% confidence intervals (CIs) for dichotomous outcomes and weighted mean differences with 95% CIs for continuous outcomes. Prespecified subgroup analyses will be conducted to explore heterogeneity. The overall quality of evidence will be rated using the Grading of Recommendations Assessment, Development and Evaluation approach. ETHICS AND DISSEMINATION: Ethics approval is not required for this systematic review as it draws from previously published data. Results of the study will be submitted to a peer-reviewed journal for publication and discussed at conferences and seminars. PROSPERO REGISTRATION NUMBER: CRD42018102069.

The Integration of Clinical and Research Training: How and Why MD-PhD Programs Work.

For over 60 years, MD-PhD programs have provided integrated clinical and research training to produce graduates primed for physician-scientist careers. Yet the nature of this integrated training is poorly characterized, with no program theory of MD-PhD training to guide program development or evaluation. The authors address this gap by proposing a program theory of integrated MD-PhD training applying constructs from cognitive psychology and medical education. The authors argue that integrated physician-scientist training requires development of at least three elements in trainees: cognitive synergy, sense of self, and professional capacity. First, integrated programs need to foster the cognitive ability to synergize and transfer knowledge between the clinical and research realms. Second, integrated programs need to facilitate development of a unique and emergent identity as a physician-scientist that is more than the sum of the individual roles of physician and scientist. Third, integrated programs should develop core competencies unique to physician-scientists in addition to those required of each independently. The authors describe how programs can promote development of these elements in trainees, summarized in a logic model. Activities and process measures are provided to assist institutions in enhancing integration. Specifically, programs can enact the proposed theory by providing tailored MD-PhD curricula, personal development planning, and a supportive community of practice. It is high time to establish a theory behind integrated MD-PhD training as the basis for designing interventions and evaluations to develop the foundations of physician-scientist expertise.

Biopsy Versus Subtotal Versus Gross Total Resection in Patients with Low-Grade Glioma: A Systematic Review and Meta-Analysis.

BACKGROUND: The role of the extent of surgical resection (EOR) in clinical outcomes for patients with low-grade glioma requires further examination. The goal of the present study was to evaluate the association between variable degrees of EOR and clinical outcomes for patients with low-grade glioma. METHODS: We conducted a systematic review and meta-analysis and searched databases for reports of low-grade glioma EOR. Eligible studies compared patient outcomes, including ≥2 categories of EOR (biopsy, resection of any extent, subtotal resection [STR], or gross total resection [GTR]). Treatment effects were evaluated using pooled estimates, mean differences, or risk ratios (RRs) with corresponding 95% confidence intervals (CIs) using random effects modeling. RESULTS: Our literature search yielded 60 studies with 13,289 patients. Pooled estimates of overall survival (OS) showed an increase from 3.79 years in the biopsy group to 6.68 years in STR to 10.65 years in GTR. OS was favorable with resection of any extent compared with (mean difference, 3.24; 95% CI, 0.64-5.84; P = 0.015). Pooled estimates of seizure control showed an improvement from 47.8% with biopsy to 54.2% with STR and 81.0% with GTR. Compared with STR, GTR delayed malignant transformation (RR, 0.43; 95% CI, 0.20-0.93; P = 0.032), without increasing postoperative mortality (RR, 0.38; 95% CI, 0.07-1.97; P = 0.250) or morbidity (RR, 1.22; 95% CI, 0.65-2.28; P = 0.540). CONCLUSION: Among patients with low-grade gliomas, greater degrees of safe EOR were associated with longer OS and progression-free survival, better seizure control, and delayed malignant transformation, without increasing mortality or morbidity.