Perspectives from the 2nd International Post-Tuberculosis Symposium: mobilising advocacy and research for improved outcomes.

In 2020, it was estimated that there were 155 million survivors of TB alive, all at risk of possible post TB disability. The 2nd International Post-Tuberculosis Symposium (Stellenbosch, South Africa) was held to increase global awareness and empower TB-affected communities to play an active role in driving the agenda. We aimed to update knowledge on post-TB life and illness, identify research priorities, build research collaborations and highlight the need to embed lung health outcomes in clinical TB trials and programmatic TB care services. The symposium was a multidisciplinary meeting that included clinicians, researchers, TB survivors, funders and policy makers. Ten academic working groups set their own goals and covered the following thematic areas: 1) patient engagement and perspectives; 2) epidemiology and modelling; 3) pathogenesis of post-TB sequelae; 4) post-TB lung disease; 5) cardiovascular and pulmonary vascular complications; 6) neuromuscular & skeletal complications; 7) paediatric complications; 8) economic-social and psychological (ESP) consequences; 9) prevention, treatment and management; 10) advocacy, policy and stakeholder engagement. The working groups provided important updates for their respective fields, highlighted research priorities, and made progress towards the standardisation and alignment of post-TB outcomes and definitions.

En 2020, il est estimé qu'il y a 155 millions de survivants de la TB dans le monde, tous exposés à un risque d'invalidité post-TB. Le deuxième Symposium International Post-Tuberculose (Stellenbosch, Afrique du Sud) a été organisé dans le but de sensibiliser davantage à l'échelle mondiale et de permettre aux communautés touchées par la TB de contribuer activement à la mise en œuvre de l'agenda. De plus, nous avons entrepris de mettre à jour les connaissances sur la vie et les maladies post-TB, de déterminer les domaines de recherche prioritaires, d'établir des partenariats de recherche et de souligner l'importance d'intégrer les résultats sur la santé pulmonaire dans les essais cliniques et les services de soins de la TB. Le symposium était une réunion de travail pluridisciplinaire rassemblant des praticiens, des chercheurs, des personnes ayant survécu à la TB, des donateurs, des décideurs politiques et d'autres acteurs clés. Dix groupes de travail académiques ont établi leurs propres objectifs et ont abordé les sujets thématiques suivants : 1) engagement et perspectives des patients ; 2) épidémiologie et modélisation ; 3) pathogénie des séquelles post-TB ; 4) maladie pulmonaire post-TB (PTLD, pour l'anglais «post-TB lung disease ¼) ; 5) complications cardiovasculaires et vasculaires pulmonaires ; 6) complications neuromusculaires et squelettiques ; 7) complications pédiatriques ; 8) conséquences économiques, sociales et psychologiques (ESP, pour l'anglais «economic-social and psychological¼) ; 9) prévention, traitement et gestion ; 10) plaidoyer, politique et engagement des parties prenantes. Les groupes de travail académiques ont apporté des mises à jour significatives dans leurs domaines respectifs, ont mis en évidence les priorités de recherche et ont avancé vers la normalisation et l'harmonisation des résultats et des définitions de la post-TB.

Do NO, N2O, N2 and CO2 fluxes differ in soils sourced from cropland and varying riparian buffer vegetation? An incubation study.

Riparian buffers are expedient interventions for water quality functions in agricultural landscapes. However, the choice of vegetation and management affects soil microbial communities, which in turn affect nutrient cycling and the production and emission of gases such as nitric oxide (NO), nitrous oxide (N2O), nitrogen gas (N2) and carbon dioxide (CO2). To investigate the potential fluxes of the above-mentioned gases, soil samples were collected from a cropland and downslope grass, willow and woodland riparian buffers from a replicated plot scale experimental facility. The soils were re-packed into cores and to investigate their potential to produce the aforementioned gases via potential denitrification, a potassium nitrate (KNO3 -) and glucose (labile carbon)-containing amendment, was added prior to incubation in a specialized laboratory DENItrification System (DENIS). The resulting NO, N2O, N2 and CO2 emissions were measured simultaneously, with the most NO (2.9 ± 0.31 mg NO m-2) and N2O (1413.4 ± 448.3 mg N2O m-2) generated by the grass riparian buffer and the most N2 (698.1 ± 270.3 mg N2 m-2) and CO2 (27,558.3 ± 128.9 mg CO2 m-2) produced by the willow riparian buffer. Thus, the results show that grass riparian buffer soils have a greater NO3 - removal capacity, evidenced by their large potential denitrification rates, while the willow riparian buffers may be an effective riparian buffer as its soils potentially promote complete denitrification to N2, especially in areas with similar conditions to the current study.

Private healthcare sector spine surgery: patient and surgeon profiles from a large open medical scheme in South Africa.

Background It is generally understood that the profile of spine surgery varies by setting, based on factors such as the age profile of the population, the economic context and access to healthcare. Relatively little is known about the profile of spine surgery in South Africa, although one previous report from the public healthcare sector suggested a high burden of trauma and infection-related surgery. To our knowledge, there has been no formal investigation in the private sector. A holistic understanding of spine surgery within our setting may be beneficial for applications such as resource allocation and informing the training needs of future specialists.   Objectives To provide insight into spinal surgery in the South African private healthcare sector by describing spine surgery characteristics within a large open medical scheme, including both patient and surgeon profiles.   Methods This retrospective review included adult inpatient spine surgeries funded by the largest open medical scheme in South Africa between 2008 and 2017. An anonymized dataset extracted from the scheme records included patient demographics, ICD-10 codes, procedure codes and surgeon specialization. Descriptive patient and surgery data were presented. Furthermore, the association between patient and surgery factors and surgeon specialization was investigated using univariate and multivariate analyses.   Results A total of 49,576 index spine surgeries were included. The largest proportion of surgeries involved members 40-59 years old (n=23,543, 48%), approximately half involved female members (n=25,293, 51%) and most were performed by neurosurgeons (n=35,439, 72%). At least 37,755 (76%) surgeries were for degenerative pathology, 2,100 (4%) for trauma and 242 (0.5%) for infection. Adjusted risk ratios (aRR) significantly associated with orthopaedic surgeon specialization included cervical spine region aRR = 0.49 (95% C.I. 0.39-0.61), trauma aRR = 1.50 (95% C.I. 1.20-1.88), deformity aRR = 1.77 (95% C.I. 1.33-2.35) and blood transfusion aRR = 1.46 (95% C.I. 1.12-1.91).   Conclusion Spine surgery in South Africa's largest open medical scheme was dominated by surgery for degenerative pathology in older adults and was performed largely by neurosurgeons. This constituted a stark contrast to a previous report from the public sector and highlighted a mismatch between exposure during public sector registrar training and private practice post-specialization. The findings support the need for private-public collaboration as well as the importance of spine fellowships for all specialists intending to practice spine surgery.

Incidence of spine surgery in the South African private healthcare sector: ten-year trends within a large open medical scheme.

PURPOSE: Studies from developed countries suggest a dramatic increase in the utilization of spine surgery in recent decades, however less is known about spine surgery rates in the developing world. The aim of this study was to investigate ten-year trends in the incidence of spine surgery within South Africa's largest open medical scheme. METHODS: This retrospective review included adult inpatient spine surgeries funded by the scheme between 2008 and 2017. The incidence of spine surgery was investigated by age group-overall and for degenerative pathologies, fusion and instrumentation. Surgeons per 100,000 members were determined. Trends were evaluated by linear regression and by crude 10-year change in incidence. RESULTS: A total of 49,575 spine surgeries were included. The incidence of surgery for lumbar degenerative pathology showed a significant upward trend among 60-79 year olds but declined among 40-59 year olds. The incidence of lumbar fusion and lumbar instrumentation declined significantly among 40-59 year olds with little change among 60-79 year olds. The ratio of orthopaedic spinal surgeons decreased from 10.2 to 6.3 per 100,000 members whereas the ratio of neurosurgeons decreased from 7.6 to 6.5 per 100,000. CONCLUSION: Spine surgery in the South African private healthcare sector bears some similarity to developed countries in that it is dominated by elective procedures for degenerative pathology. However, the findings did not reflect the marked increases in the utilization of spine surgery reported elsewhere. It is hypothesized that this may be partly related to differences in the supply of spinal surgery.

Soil CO2 emissions in cropland with fodder maize (Zea mays L.) with and without riparian buffer strips of differing vegetation.

Vegetated land areas play a significant role in determining the fate of carbon (C) in the global C cycle. Riparian buffer vegetation is primarily implemented for water quality purposes as they attenuate pollutants from immediately adjacent croplands before reaching freashwater systems. However, their prevailing conditions may sometimes promote the production and subsequent emissions of soil carbon dioxide (CO2). Despite this, the understanding of soil CO2 emissions from riparian buffer vegetation and a direct comparison with adjacent croplands they serve remain elusive. In order to quantify the extent of CO2 emissions in such an agro system, we measured CO2 emissions simultaneously with soil and environmental variables for six months in a replicated plot-scale facility comprising of maize cropping served by three vegetated riparian buffers, namely: (i) a novel grass riparian buffer; (ii) a willow riparian buffer, and; (iii) a woodland riparian buffer. These buffered treatments were compared with a no-buffer control. The woodland (322.9 ± 3.1 kg ha- 1) and grass (285 ± 2.7 kg ha- 1) riparian buffer treatments (not significant to each other) generated significantly (p = < 0.0001) the largest CO2 compared to the remainder of the treatments. Our results suggest that during maize production in general, the woodland and grass riparian buffers serving a maize crop pose a CO2 threat. The results of the current study point to the need to consider the benefits for gaseous emissions of mitigation measures conventionally implemented for improving the sustainability of water resources.

Soil N2O and CH4 emissions from fodder maize production with and without riparian buffer strips of differing vegetation.

Purpose: Nitrous oxide (N2O) and methane (CH4) are some of the most important greenhouse gases in the atmosphere of the 21st century. Vegetated riparian buffers are primarily implemented for their water quality functions in agroecosystems. Their location in agricultural landscapes allows them to intercept and process pollutants from adjacent agricultural land. They recycle organic matter, which increases soil carbon (C), intercept nitrogen (N)-rich runoff from adjacent croplands, and are seasonally anoxic. Thus processes producing environmentally harmful gases including N2O and CH4 are promoted. Against this context, the study quantified atmospheric losses between a cropland and vegetated riparian buffers that serve it. Methods: Environmental variables and simultaneous N2O and CH4 emissions were measured for a 6-month period in a replicated plot-scale facility comprising maize (Zea mays L.). A static chamber was used to measure gas emissions. The cropping was served by three vegetated riparian buffers, namely: (i) grass riparian buffer; (ii) willow riparian buffer and; (iii) woodland riparian buffer, which were compared with a no-buffer control. Results: The no-buffer control generated the largest cumulative N2O emissions of 18.9 kg ha- 1 (95% confidence interval: 0.5-63.6) whilst the maize crop upslope generated the largest cumulative CH4 emissions (5.1 ± 0.88 kg ha- 1). Soil N2O and CH4-based global warming potential (GWP) were lower in the willow (1223.5 ± 362.0 and 134.7 ± 74.0 kg CO2-eq. ha- 1 year- 1, respectively) and woodland (1771.3 ± 800.5 and 3.4 ± 35.9 kg CO2-eq. ha- 1 year- 1, respectively) riparian buffers. Conclusions: Our results suggest that in maize production and where no riparian buffer vegetation is introduced for water quality purposes (no buffer control), atmospheric CH4 and N2O concerns may result.

The bacterial community of childcare centers: potential implications for microbial dispersal and child exposure.

BACKGROUND: Bacterial communities within built environments reflect differences in sources of bacteria, building design, and environmental contexts. These communities impact the health of their occupants in many ways. Children interact with the built environment differently than do adults as a result of their unique behaviors, size, and developmental status. Consequently, understanding the broader bacterial community to which children are exposed will help inform public health efforts and contribute to our growing understanding of the bacterial community associated with childcare centers. METHODS: We sampled childcare centers to survey the variation in bacterial community composition across five surfaces found inside and outside twelve classrooms and six centers using 16S rRNA marker gene amplicon sequencing. We then correlated these bacterial community analyses of surfaces with environmental and demographic measures of illumination and classroom occupant density. RESULTS: The childcare environment was dominated by human-associated bacteria with modest input from outdoor sources. Though the bacterial communities of individual childcare centers differed, there was a greater difference in the bacterial community within a classroom than among centers. Surface habitats-fomites-within the classroom, did not differ in community composition despite differing proximity to likely sources of bacteria, and possible environmental filters, such as light. Bacterial communities did correlate with occupant density and differed significantly between high and low usage surfaces. CONCLUSIONS: Our results suggest built environments inhabited by young children are similar to functionally equivalent built environments inhabited by adults, despite the different way young children engage with their environment. Ultimately, these results will be useful when further interrogating microbial dispersal and human exposure to microorganisms in built environments that specifically cater to young children.

Drastic reduction of orthopaedic services at an urban tertiary hospital in South Africa during COVID-19: Lessons for the future response to the pandemic.

BACKGROUND: The COVID-19 pandemic has impacted on the global surgery landscape. OBJECTIVES: To analyse and describe the initial impact of the COVID-19 pandemic on orthopaedic surgery at Groote Schuur Hospital, a tertiary academic hospital in South Africa. METHODS: The number of orthopaedic surgical cases, emergency theatre patient waiting times, and numbers of outpatient clinic visits, ward admissions, bed occupancies and total inpatient days for January - April 2019 (pre-COVID-19) were compared with the same time frame in 2020 (COVID-19). The COVID-19 timeframe included initiation of a national 'hard lockdown' from 26 March 2020, in preparation for an increasing volume of COVID-19 cases. RESULTS: April 2020, the time of the imposed hard lockdown, was the most affected month, although the number of surgical cases had started to decrease slowly during the 3 preceding months. The total number of surgeries, outpatient visits and ward admissions decreased significantly during April 2020 (55.2%, 69.1% and 60.6%, respectively) compared with April 2019 (p<0.05). Trauma cases were reduced by 40% in April 2020. Overall emergency theatre patient waiting time was 30% lower for April 2020 compared with 2019. CONCLUSIONS: COVID-19 and the associated lockdown has heavily impacted on both orthopaedic inpatient and outpatient services. Lockdown led to a larger reduction in the orthopaedic trauma burden than in international centres, but the overall reduction in surgeries, outpatient visits and hospital admissions was less. This lesser reduction was probably due to local factors, but also to a conscious decision to avoid total collapse of our surgical services.

Development and Assessment of an Artificial Intelligence-Based Tool for Skin Condition Diagnosis by Primary Care Physicians and Nurse Practitioners in Teledermatology Practices.

Importance: Most dermatologic cases are initially evaluated by nondermatologists such as primary care physicians (PCPs) or nurse practitioners (NPs). Objective: To evaluate an artificial intelligence (AI)-based tool that assists with diagnoses of dermatologic conditions. Design, Setting, and Participants: This multiple-reader, multiple-case diagnostic study developed an AI-based tool and evaluated its utility. Primary care physicians and NPs retrospectively reviewed an enriched set of cases representing 120 different skin conditions. Randomization was used to ensure each clinician reviewed each case either with or without AI assistance; each clinician alternated between batches of 50 cases in each modality. The reviews occurred from February 21 to April 28, 2020. Data were analyzed from May 26, 2020, to January 27, 2021. Exposures: An AI-based assistive tool for interpreting clinical images and associated medical history. Main Outcomes and Measures: The primary analysis evaluated agreement with reference diagnoses provided by a panel of 3 dermatologists for PCPs and NPs. Secondary analyses included diagnostic accuracy for biopsy-confirmed cases, biopsy and referral rates, review time, and diagnostic confidence. Results: Forty board-certified clinicians, including 20 PCPs (14 women [70.0%]; mean experience, 11.3 [range, 2-32] years) and 20 NPs (18 women [90.0%]; mean experience, 13.1 [range, 2-34] years) reviewed 1048 retrospective cases (672 female [64.2%]; median age, 43 [interquartile range, 30-56] years; 41 920 total reviews) from a teledermatology practice serving 11 sites and provided 0 to 5 differential diagnoses per case (mean [SD], 1.6 [0.7]). The PCPs were located across 12 states, and the NPs practiced in primary care without physician supervision across 9 states. The NPs had a mean of 13.1 (range, 2-34) years of experience and practiced in primary care without physician supervision across 9 states. Artificial intelligence assistance was significantly associated with higher agreement with reference diagnoses. For PCPs, the increase in diagnostic agreement was 10% (95% CI, 8%-11%; P < .001), from 48% to 58%; for NPs, the increase was 12% (95% CI, 10%-14%; P < .001), from 46% to 58%. In secondary analyses, agreement with biopsy-obtained diagnosis categories of maglignant, precancerous, or benign increased by 3% (95% CI, -1% to 7%) for PCPs and by 8% (95% CI, 3%-13%) for NPs. Rates of desire for biopsies decreased by 1% (95% CI, 0-3%) for PCPs and 2% (95% CI, 1%-3%) for NPs; the rate of desire for referrals decreased by 3% (95% CI, 1%-4%) for PCPs and NPs. Diagnostic agreement on cases not indicated for a dermatologist referral increased by 10% (95% CI, 8%-12%) for PCPs and 12% (95% CI, 10%-14%) for NPs, and median review time increased slightly by 5 (95% CI, 0-8) seconds for PCPs and 7 (95% CI, 5-10) seconds for NPs per case. Conclusions and Relevance: Artificial intelligence assistance was associated with improved diagnoses by PCPs and NPs for 1 in every 8 to 10 cases, indicating potential for improving the quality of dermatologic care.

Low 30-day mortality in South African orthopaedic patients undergoing surgery at an academic hospital during the first wave of the COVID-19 pandemic: It was safe to perform orthopaedic procedures at our hospital during the first COVID-19 peak.

BACKGROUND:  Initial local and global evidence suggests that SARS-CoV-2-infected patients who undergo surgery, and those who become infected perioperatively, have an increased mortality risk post surgery. OBJECTIVES: To analyse and describe the 30-day mortality, presurgical COVID-19 status and hospital-acquired SARS-CoV-2 infection rates of patients, both SARS-CoV-2-positive and negative, undergoing orthopaedic surgery at a tertiary academic hospital in South Africa (SA) during the first COVID-19 peak. METHODS: This single-centre, observational, prospective study included patients who underwent orthopaedic procedures from 1 April 2020 (beginning of the COVID-19 case increase in SA) to 31 July 2020 (first COVID-19 peak in SA). All patients were screened for COVID-19 and were confirmed positive if they had a positive laboratory quantitative polymerase chain reaction test for SARS-CoV-2 RNA on a nasopharyngeal or oral swab. Thirty-day mortality, presurgical COVID-19 status and hospital-acquired SARS-CoV-2 infection were assessed. RESULTS:  Overall, a total of 433 operations were performed on 346 patients during the timeframe. Of these patients, 65.9% (n=228) were male and 34.1% (n=118) were female. The mean (standard deviation) age was 42.5 (16.8) years (range 9 - 89). Of the patients, 5 (1.4%) were identified as COVID-19 patients under investigation (PUI) on admission and tested positive for SARS-CoV-2 before surgery, and 1 (0.3%) contracted SARS-CoV-2 perioperatively; all survived 30 days post surgery. Twenty-nine patients were lost to follow-up, and data were missing for 6 patients. The final analysis was performed excluding these 35 patients. Of the 311 patients included in the final 30-day mortality analysis, 303 (97%) had a follow-up observation ≥30 days after the operation. The overall 30-day mortality for these patients was 2.5% (n=8 deaths). None of the recorded deaths were of screened COVID-19 PUI. CONCLUSIONS: We report a low 30-day mortality rate of 2.5% (n=8) for patients undergoing orthopaedic surgery at our hospital during the first COVID-19 peak. None of the deaths were COVID-19 related, and all patients who tested SARS-CoV-2-positive, before or after surgery, survived. Our overall 30-day mortality rate correlates with several other reports of orthopaedic centres analysing over similar timeframes during the first peak of the COVID-19 pandemic. Regarding mortality and SARS-CoV-2 infection risk, we can conclude that with the appropriate measures taken, it was safe to undergo orthopaedic procedures at our hospital during the first peak of the COVID-19 pandemic in SA.

Secondary non-invasive prenatal screening for fetal trisomy: an effectiveness study in a public health setting.

OBJECTIVE: To evaluate the effectiveness of secondary screening using non-invasive prenatal testing (NIPT) in a routine NHS setting including test performance, turn-around times (TATs) and no-call (failure to obtain result) rates. To examine the influence of maternal and fetal characteristics on test performance. DESIGN: Retrospective cohort. SETTING: London teaching hospital. SAMPLE: A total of 8651 pregnancies undergoing screening for fetal trisomy using NIPT provided by an NHS cell-free DNA screening laboratory - the SAFE laboratory. METHODS: Screening test evaluation and TATs. Univariate and multivariate logistic regression analysis to identify significant predictors of no-call results and reported by low fetal fraction (<2%), very high fetal fraction (>40%) and processing failure. MAIN OUTCOME MEASURES: Test performance, TATs and no-call rates, factors affecting no-call results. RESULTS: Average TAT was 4.0 days (95% CI 4.0-4.2 days). Test sensitivities for trisomies 21 and 13/18 were 98.9% (95% CI 95.9-99.9%) and 90.4% (95% CI 80.0-96.8%), respectively. The overall no-call rate was 32/8651 (0.37%, 95% CI 0.26-0.52%). The overall risk of a no-call result was influenced by gestational age, dichorionic twin pregnancy, history of malignancy and pregnancies affected by trisomy 13/18, but not by maternal weight or use of low-molecular-weight heparin. CONCLUSIONS: High-throughput NIPT can be effectively embedded into a public health NHS setting. TATs of 4 days and no-calls of <0.5% were well within clinically desirable tolerances. Gestational age, maternal weight, assisted reproductive techniques, use of low-molecular-weight heparin and past history of malignancy did not have major impacts on test no-call rates and should not constitute reasons for withholding the option of NIPT from women. TWEETABLE ABSTRACT: Turn-around times of 4 days, no-call (test failure) rates of 0.37% and highly accurate NIPT can be successfully embedded in the NHS.

Early Care and Education Professionals' Breastfeeding Knowledge and Practices Before and After an E-Learning Program.

Objective: To assess early care and education professionals' breastfeeding knowledge and practices before and after an e-learning program. Participants: Early care and education professionals from New Hampshire (U.S.A.) licensed child care programs were invited to complete a pre-assessment followed by a 90-minute e-learning breastfeeding program. Three months post-training, participants were invited to complete the post-assessment. Analysis: McNemar tests were used to assess changes from pre-post-assessment for dichotomous variables. McNemar-Bowker tests were used to determine differences from pre-post for variables with more than two categories. When the McNemar-Bowker test was significant, a multiple comparison correction (Bonferroni) was used. Results: 114 participants completed the e-learning program and pre-post assessment. Results showed significant improvement from pre-post in 10 of 15 breastfeeding knowledge questions related to health of baby, mother and child care centers, economics, and environmental impact. There were significant changes from pre-post in 24 of 50 breastfeeding practice questions in handling breast milk, promoting breastfeeding, and supporting mothers. Conclusions and Implications: This study indicates improvement in early care and education professionals' breastfeeding knowledge and practices; however, opportunities exist to design targeted initiatives to further strengthen practices that support breastfeeding families in the child care environment.

Viewership footprint for a low-resource, student-centred collaborative video platform to teach orthopaedics in southern Africa.

BACKGROUND: Institutions are increasingly using technology to augment the class learning experience of medical students. Especially in Africa, local content is key to allow insights and knowledge to emerge and build transformative capacity for students and patients. There is currently no peer-reviewed video content produced by students with the aim of providing education on orthopaedic topics for medical students and patients in this region. OBJECTIVES: To evaluate the demographic and geographical viewership as well as video-specific statistics of orthopaedic teaching videos for medical students on a YouTube channel, with the expressed aim of informing future content production. METHODS: Videos were produced by South African (SA) medical students as a problem-based collaborative project. Student-owned smartphones and various types of free video editing software were used to produce these videos, which were then assessed by a group of orthopaedic specialists and uploaded onto a YouTube channel (UCTeach). The analytical reports of this channel generated by Google and YouTube were analysed regarding watch time per day (minutes), average view duration (minutes), most watched videos, top geographies, age and gender. RESULTS: A total of 83 videos were uploaded to the UCTeach Ortho channel during a 2-year period, with a total watch time of 857 062 minutes and 337 983 views. The majority of viewers were between the ages of 18 and 34 years (85%). India had the most views (n=69 089), followed by the USA (n=66 257) and SA (n=21 882). Most of the videos were watched on mobile phones (n=183 299) and computers (n=128 228). The most watched video, produced in April 2016, was on physiological and pathological gait, with 51 314 views. CONCLUSIONS: Our study provides proof of concept for a new educational material creation and dissemination strategy. A low-cost local collaborative orthopaedic video project by medical students for medical students can lead to high view counts and watch time on YouTube. It is accessible to audiences in low-, middle- and high-income countries. The students' educational videos also reached a global audience consistently over a 3-year period.

A deep learning system for differential diagnosis of skin diseases.

Skin conditions affect 1.9 billion people. Because of a shortage of dermatologists, most cases are seen instead by general practitioners with lower diagnostic accuracy. We present a deep learning system (DLS) to provide a differential diagnosis of skin conditions using 16,114 de-identified cases (photographs and clinical data) from a teledermatology practice serving 17 sites. The DLS distinguishes between 26 common skin conditions, representing 80% of cases seen in primary care, while also providing a secondary prediction covering 419 skin conditions. On 963 validation cases, where a rotating panel of three board-certified dermatologists defined the reference standard, the DLS was non-inferior to six other dermatologists and superior to six primary care physicians (PCPs) and six nurse practitioners (NPs) (top-1 accuracy: 0.66 DLS, 0.63 dermatologists, 0.44 PCPs and 0.40 NPs). These results highlight the potential of the DLS to assist general practitioners in diagnosing skin conditions.

Risk scoring for time to end-stage knee osteoarthritis: data from the Osteoarthritis Initiative.

OBJECTIVE: This study constructs a risk score for patients' progression to end-stage knee osteoarthritis (OA) within 4 years. DESIGN: The Osteoarthritis Initiative (OAI) was a longitudinal study of the onset and progression of knee OA. Using a recent definition of end-stage knee OA, we implement interval-censored survival forests to select predictors of this endpoint. We fit an interval-censored Cox model for time to end-stage knee OA, using the selected predictors. The risk score is the Cox model's fitted linear combination of the nine selected baseline structural and symptomatic knee OA variables. RESULTS: We fit our models on a training set of 2,701 patients, and we evaluate on an independent test set of 1,436 patients. On the test sample, we observe a concordance index of 0.86 between risk score and time to end-stage, AUC of 0.87 for predicting end-stage within 24, 36, and 48 months, and positive predictive values that increase with the risk score. This risk stratification algorithm could enrich clinical trial patient enrollment. By enrolling test sample patients with scores above a threshold, a trial could have included 91% of test set patients who reach end-stage within 4 years while only enrolling 45% of the test sample. CONCLUSION: Using statistical methods, we construct and validate an interpretable risk score for time to end-stage knee OA. This score can help disease-modifying OA treatment developers to select candidates with the highest risk of fast-progressing knee OA.

A multicenter evaluation of a sample to answer real-time PCR assay for toxigenic C. difficile in symptomatic subjects.

We evaluated the performance of the Luminex ARIES® C. difficile Assay on 984 stool specimens prospectively collected from patients being tested for CDI at 4 clinical laboratories in the United States. Results were compared to direct and enriched toxigenic culture. Positive percent agreement (PPA) of the ARIES® C. difficile Assay was 98.1% versus direct toxigenic culture, and sensitivity versus direct plus enriched toxigenic culture was 90.5%. Negative percent agreement (NPA) of the ARIES® C. difficile Assay against direct culture was 92.6%, and specificity versus direct plus enriched toxigenic culture was 95.8%. The ARIES® C. difficile Assay was also compared to the results of routine (molecular, antigen, and/or toxin) methods for C. difficile testing used at each institution. The PPA of the ARIES® C. difficile Assay ranged from 82.9% to 100%. NPA values against these commercial assays ranged from 94.5% to 100%.

Impaired neural differentiation and glymphatic CSF flow in the Ccdc39 rat model of neonatal hydrocephalus: genetic interaction with L1cam.

Neonatal hydrocephalus affects about one child per 1000 births and is a major congenital brain abnormality. We previously discovered a gene mutation within the coiled-coil domain-containing 39 (Ccdc39) gene, which causes the progressive hydrocephalus (prh) phenotype in mice due to lack of ependymal-cilia-mediated cerebrospinal fluid (CSF) flow. In this study, we used CRISPR/Cas9 to introduce the Ccdc39 gene mutation into rats, which are more suitable for imaging and surgical experiments. The Ccdc39prh/prh mutants exhibited mild ventriculomegaly at postnatal day (P)5 that progressed into severe hydrocephalus by P11 (P<0.001). After P11, macrophage and neutrophil invasion along with subarachnoid hemorrhage were observed in mutant brains showing reduced neurofilament density, hypomyelination and increased cell death signals compared with wild-type brains. Significantly more macrophages entered the brain parenchyma at P5 before hemorrhaging was noted and increased expression of a pro-inflammatory factor (monocyte chemoattractant protein-1) was found in the cortical neural and endothelial cells in the mutant brains at P11. Glymphatic-mediated CSF circulation was progressively impaired along the middle cerebral artery from P11 as mutants developed severe hydrocephalus (P<0.001). In addition, Ccdc39prh/prh mutants with L1 cell adhesion molecule (L1cam) gene mutation, which causes X-linked human congenital hydrocephalus, showed an accelerated early hydrocephalus phenotype (P<0.05-0.01). Our findings in Ccdc39prh/prh mutant rats demonstrate a possible causal role of neuroinflammation in neonatal hydrocephalus development, which involves impaired cortical development and glymphatic CSF flow. Improved understanding of inflammatory responses and the glymphatic system in neonatal hydrocephalus could lead to new therapeutic strategies for this condition.This article has an associated First Person interview with the joint first authors of the paper.

Deep Learning and Glaucoma Specialists: The Relative Importance of Optic Disc Features to Predict Glaucoma Referral in Fundus Photographs.

PURPOSE: To develop and validate a deep learning (DL) algorithm that predicts referable glaucomatous optic neuropathy (GON) and optic nerve head (ONH) features from color fundus images, to determine the relative importance of these features in referral decisions by glaucoma specialists (GSs) and the algorithm, and to compare the performance of the algorithm with eye care providers. DESIGN: Development and validation of an algorithm. PARTICIPANTS: Fundus images from screening programs, studies, and a glaucoma clinic. METHODS: A DL algorithm was trained using a retrospective dataset of 86 618 images, assessed for glaucomatous ONH features and referable GON (defined as ONH appearance worrisome enough to justify referral for comprehensive examination) by 43 graders. The algorithm was validated using 3 datasets: dataset A (1205 images, 1 image/patient; 18.1% referable), images adjudicated by panels of GSs; dataset B (9642 images, 1 image/patient; 9.2% referable), images from a diabetic teleretinal screening program; and dataset C (346 images, 1 image/patient; 81.7% referable), images from a glaucoma clinic. MAIN OUTCOME MEASURES: The algorithm was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity for referable GON and glaucomatous ONH features. RESULTS: The algorithm's AUC for referable GON was 0.945 (95% confidence interval [CI], 0.929-0.960) in dataset A, 0.855 (95% CI, 0.841-0.870) in dataset B, and 0.881 (95% CI, 0.838-0.918) in dataset C. Algorithm AUCs ranged between 0.661 and 0.973 for glaucomatous ONH features. The algorithm showed significantly higher sensitivity than 7 of 10 graders not involved in determining the reference standard, including 2 of 3 GSs, and showed higher specificity than 3 graders (including 1 GS), while remaining comparable to others. For both GSs and the algorithm, the most crucial features related to referable GON were: presence of vertical cup-to-disc ratio of 0.7 or more, neuroretinal rim notching, retinal nerve fiber layer defect, and bared circumlinear vessels. CONCLUSIONS: A DL algorithm trained on fundus images alone can detect referable GON with higher sensitivity than and comparable specificity to eye care providers. The algorithm maintained good performance on an independent dataset with diagnoses based on a full glaucoma workup.