Portland VA Research Foundation

Evidence synthesis and guideline groups have the potential to address health disparities. In June 2024, the Agency for Healthcare Research and Quality (AHRQ) and the Robert Wood Johnson Foundation (RWJF) cosponsored a summit to address racial and ethnic health equity in systematic reviews and other syntheses and guidelines, with support from Cochrane US. This article summarizes cross-cutting themes around future directions for systematic reviews and guidelines. Discussions addressed include the rationale for addressing racial health equity in systematic reviews and guidelines; representation of people with lived experience in systematic reviews and guidelines; approaches to developing and addressing equity-focused scope, including frameworks, methods, and thoughtful interpretation in systematic reviews; challenges and opportunities for guideline recommendations; need for standardized language and reporting for race and ethnicity in primary research studies, systematic reviews, and guidelines; and measures to track the progress of incorporating and addressing racial and ethnic health equity in systematic reviews and guidelines. Participants acknowledged that a one-size-fits-all approach was not possible or desired. Consensus priorities for next steps were to develop methods guidance to address equity in systematic reviews and guidelines; develop measures to track the progress of addressing racial and health equity in systematic reviews and guidelines; operationalize engaging representative interest holders in systematic reviews and guidelines; and share resources and learning for advancing health equity.

Failure of machine learning models to generalize to new data is a core problem limiting the reliability of AI systems, partly due to the lack of simple and robust methods for comparing new data to the original training dataset. We propose a standardized approach for assessing data similarity in a model-agnostic manner by constructing a supervised autoencoder for generalizability estimation (SAGE). We compare points in a low-dimensional embedded latent space, defining empirical probability measures for k-Nearest Neighbors (kNN) distance, reconstruction of inputs and task-based performance. As proof of concept for classification tasks, we use MNIST and CIFAR-10 to demonstrate how an ensemble output probability score can separate deformed images from a mixture of typical test examples, and how this SAGE score is robust to transformations of increasing severity. As further proof of concept, we extend this approach to a regression task using non-imaging data (UCI Abalone). In all cases, we show that out-of-the-box model performance increases after SAGE score filtering, even when applied to data from the model's own training and test datasets. Our out-of-distribution scoring method can be introduced during several steps of model construction and assessment, leading to future improvements in responsible deep learning implementation.

BackgroundImplementation strategies can be a vital leveraging point for enhancing the implementation and dissemination of evidence-based suicide prevention interventions and programming. However, much remains unknown about which implementation strategies are commonly used and effective for supporting suicide prevention efforts.MethodsIn light of the limited available literature, a scoping review was conducted to evaluate implementation strategies present in current suicide prevention studies. We identified studies that were published between 2013 and 2022 that focused on suicide prevention and incorporated at least one implementation strategy. Studies were coded by two independent coders who showed strong inter-rater reliability. Data were synthesized using descriptive statistics and a narrative synthesis of findings.ResultsOverall, we found that studies most commonly utilized strategies related to iterative evaluation, training, and education. The majority of studies did not include direct measurement of suicide behavior outcomes, and there were few studies that directly tested implementation strategy effectiveness.ConclusionImplementation science strategies remain an important component for improving suicide prevention and intervention implementation. Future research should consider the incorporation of more type 3 hybrid designs as well as increased systematic documentation of implementation strategies.Trial registration < de-identified >

BackgroundThere is growing interest in retained introns in a variety of disease contexts including cancer and aging. Many software tools have been developed to detect retained introns from short RNA-seq reads, but reliable detection is complicated by overlapping genes and transcripts as well as the presence of unprocessed or partially processed RNAs.ResultsWe compared introns detected by 8 tools using short RNA-seq reads with introns observed in long RNA-seq reads from the same biological specimens. We found significant disagreement among tools (Fleiss’ kappa = 0.113) such that 47.7% of all detected intron retentions were not called by more than one tool. We also observed poor performance of all tools, with none achieving an F1-score greater than 0.26, and qualitatively different behaviors between general-purpose alternative splicing detection tools and tools confined to retained intron detection.ConclusionsShort-read tools detect intron retention with poor recall and precision, calling into question the completeness and validity of a large percentage of putatively retained introns called by commonly used methods.

ABSTRACTWe compared cancer and non-cancer RNA sequencing (RNA-seq) data from The Cancer Genome Atlas (TCGA), the Genotype-Tissue Expression (GTEx) Project, and the Sequence Read Archive (SRA). We found that: 1) averaging across cancer types, 80.6% of exon-exon junctions thought to be cancer-specific based on comparison with tissue-matched samples are in fact present in other adult non-cancer tissues throughout the body; 2) 30.8% of junctions not present in any GTEx or TCGA normal tissues are shared by multiple samples within at least one cancer type cohort, and 87.4% of these distinguish between different cancer types; and 3) many of these junctions not found in GTEx or TCGA normal tissues (15.4% on average) are also found in embryological and other developmentally associated cells. This study probes the distribution of putatively cancer-specific junctions across a broad set of publicly available non-cancer human RNA-seq datasets. Overall, we identify a subset of shared cancer-specific junctions that could represent novel sources of cancer neoantigens. We further describe a framework for characterizing possible origins of these junctions, including potential developmental and embryological sources, as well as cell type-specific markers particularly related to cell types of cancer origin. These findings refine the meaning of RNA splicing event novelty, particularly with respect to the human neoepitope repertoire. Ultimately, cancer-specific exon-exon junctions may affect the anti-cancer immune response and may have a substantial causal relationship with the biology of disease.

Due to the increasing amount of available published evidence and the continual need to apply and update evidence in practice, we propose a shift in the way evidence generated by learning health systems can be integrated into more traditional evidence reviews. This paper discusses two main mechanisms to close the evidence-to-practice gap: (1) integrating Learning Health System (LHS) results with existing systematic review evidence and (2) providing this combined evidence in a standardized, computable data format. We believe these efforts will better inform practice, thereby improving individual and population health.

With more than half of individuals incarcerated having serious mental health concerns, correctional settings offer excellent opportunities for epidemiological, prevention, and intervention research. However, due to unique ethical and structural challenges, these settings create risks and vulnerabilities for participants not typically encountered in research populations. We surveyed 1,224 researchers, Institutional Review Board (IRB) members, and IRB prisoner representatives to assess their perceptions of risks and vulnerabilities associated with mental health research conducted in correctional settings. Highest ranked risks were related to privacy, stigma, and confidentiality; lowest ranked risks were related to prisoners’ loss of privileges or becoming targets of violence due to having participated in research. Cognitive impairment, mental illness, lack of autonomy, and limited access to services emerged as the greatest sources of vulnerability; being male, being female, being older than age 60, being a minority, and being pregnant were the lowest ranked sources of vulnerability. Researchers with corrections experience perceived lower risks and vulnerabilities than all other groups, raising the question whether these researchers accurately appraise risk and vulnerability based on experience, or if their lower risk and vulnerability perceptions reflect potential bias due to their vested interests. By identifying areas of particular risk and vulnerability, this study provides important information for researchers and research reviewers alike.

The recognition of an inherited component to breast cancer has led to an increase in demand for information, reassurance, and genetic testing, which has resulted in the creation of genetic clinics for familial cancer. The first step for patients referred to a cancer genetic clinic is a risk assessment. To evaluate the impact of cancer genetic risk-assessment services on patients at risk of familial breast cancer. The specialised register maintained by the Cochrane Breast Cancer Group was searched on 16th February 2005. We also searched MEDLINE, EMBASE, CINAHL, PsycLIT, CENTRAL, DARE, ASSIA, Web of Science, SIGLE and LILACS. The original searches covered the period 1985 to February 2005. We also handsearched relevant journals. For this review update the search was repeated through to April 2011. We considered trials looking at interventions for cancer genetic risk-assessment services for familial breast cancer for inclusion. Trials assessed outcomes such as understanding of risk, satisfaction and psychological well-being. We excluded studies if they concerned cancers other than breast cancer or if participants were not at risk of inherited breast cancer. We also excluded trials concerning the provision of general cancer genetic information or education as this review was concerned with the delivery of genetic risk assessment. Participants could be individuals of any age or gender, with or without a known BRCA mutation, but without a previous history of breast cancer or any other serious illness. Two review authors independently assessed trial quality and extracted data. Additional information was sought from investigators as necessary. Due to the heterogeneity of both the interventions and outcomes, we reported data descriptively. In this review update, we included five new trials, bringing the total number of included studies to eight. The included trials (pertaining to 10 papers), provided data on 1973 participants and assessed the impact of cancer genetic risk assessment on outcomes including perceived risk of inherited cancer, and psychological distress. This review suggests that cancer genetic risk-assessment services help to reduce distress, improve the accuracy of the perceived risk of breast cancer, and increase knowledge about breast cancer and genetics. The health professional delivering the risk assessment does not appear to have a significant impact on these outcomes. This review found favourable outcomes for patients after risk assessment for familial breast cancer. However, there were too few papers to make any significant conclusions about how best to deliver cancer genetic risk-assessment services. Further research is needed assessing the best means of delivering cancer risk assessment, by different health professionals, in different ways and in alternative locations.

Although human papillomavirus (HPV) may be a necessary cause of cervical cancer, most women with HPV infections do not develop this disease. This study was conducted to evaluate the possible effects of specific dietary factors on cervical carcinogenesis. Fifty hospitalized in-situ cases and 125 controls were identified from family planning or gynecologic clinics associated with Siriraj hospital in Bangkok, Thailand, and 134 hospitalized invasive cases and 384 hospitalized controls from the public wards of Siriraj Hospital were administered a food-frequency questionnaire and tested for HPV DNA in exfoliated cervical cells. Odds ratios in relation to intake of foods high in vitamin C, folate, vitamin E, vitamin A, beta-carotene, retinol, and cruciferous vegetables were estimated using logistic regression in case-control comparisons and in case-case comparisons adjusted for HPV status. High intake of foods rich in vitamin A, and particularly high-retinol foods, were associated with a reduced risk of in-situ disease and less strongly also with a reduced risk of invasive as compared to in-situ disease. No association was found between intake of cruciferous vegetables, foods high in vitamin C, folate, vitamin E, and beta-carotene and risk of either in-situ or invasive cervical cancer. Increasing intake of foods rich in total vitamin A, and particularly high-retinol foods, may reduce risk of in-situ cervical cancer, and at the highest level of intake may inhibit progression to invasion. If others confirm these results they suggest means of reducing the risk of cervical cancer that are amenable to public health action.