Citations

619 Background: Regorafenib (at a starting dose of 160 mg/day, rego 160), regorafenib (with a weekly dose escalation, rego 80+) and TAS-102, are suggested treatment options for refractory metastatic colorectal cancer (mCRC). We aimed to evaluate the comparative effectiveness evidence supporting these 3 strategies. Methods: We searched PubMed, Embase, and Cochrane CENTRAL, for randomized controlled trials evaluating TAS 102 or regorafenib in refractory mCRC patients who progressed on/intolerant of previous oxaliplatin, irinotecan, and fluoropyrimidine. Outcomes of interest included OS and PFS. The overall effect was pooled using the DerSimonian random effects model. We conducted network meta-analysis based on White’s multivariate meta-regression to pool evidence from direct and indirect comparisons. Results: Six trials (3 of regorafenib and 3 of TAS-102) at low risk of bias (2,445 patients) were included. Direct comparisons showed that Rego 160 and TAS-102 as monotherapy were superior to BSC in terms of PFS (Rego 160: HR = 0.4, CI 0.26 to 0.63; TAS-102: HR = 0.46, CI 0.40 to 0.52) and OS (Rego 160: HR = 0.67, CI 0.48 to 0.93; TAS-102: HR = 0.67, CI 0.57 to 0.80). Network analysis showed that there was no difference in PFS or OS between Rego 160 and TAS-102. Rego 80+ was superior to BSC in terms of OS (HR = 0.44, CI 0.23 to 0.84) and PFS (HR = 0.37, CI 0.21 to 0.66). There was a numerical advantage for Rego 80+ compared to TAS-102 and Rego 160 (see table). Conclusions: Regorafenib 160 and TAS-102 appear to have similar efficacy. Rego 80+ is shown to be superior to BSC. A trend for improved OS was observed with Rego 80+ versus Rego 160 or TAS 102. [Table: see text]

Abstract Deep learning (DL) is a branch of machine learning (ML) showing increasing promise in medicine, to assist in data classification, novel disease phenotyping and complex decision making. Deep learning is a form of ML typically implemented via multi-layered neural networks. Deep learning has accelerated by recent advances in computer hardware and algorithms and is increasingly applied in e-commerce, finance, and voice and image recognition to learn and classify complex datasets. The current medical literature shows both strengths and limitations of DL. Strengths of DL include its ability to automate medical image interpretation, enhance clinical decision-making, identify novel phenotypes, and select better treatment pathways in complex diseases. Deep learning may be well-suited to cardiovascular medicine in which haemodynamic and electrophysiological indices are increasingly captured on a continuous basis by wearable devices as well as image segmentation in cardiac imaging. However, DL also has significant weaknesses including difficulties in interpreting its models (the ‘black-box’ criticism), its need for extensive adjudicated (‘labelled’) data in training, lack of standardization in design, lack of data-efficiency in training, limited applicability to clinical trials, and other factors. Thus, the optimal clinical application of DL requires careful formulation of solvable problems, selection of most appropriate DL algorithms and data, and balanced interpretation of results. This review synthesizes the current state of DL for cardiovascular clinicians and investigators, and provides technical context to appreciate the promise, pitfalls, near-term challenges, and opportunities for this exciting new area.

Objective To assess gender disparity in outcomes among hospitalised patients with acute myocardial infarction (AMI), acute decompensated heart failure (ADHF) or pneumonia. Design A retrospective cohort study. Setting A tertiary referral centre in Midwest, USA. Participants We evaluated 12 265 adult patients hospitalised with ADHF, 15 777 with AMI and 12 929 with pneumonia, from 1 January 1995 through 31 August 2015. Patients were selected using International Classification of Diseases, Ninth Revision, Clinical Modification codes. Primary and secondary outcome measures Prevalence of comorbidities, 30-day mortality and 30-day readmission. Comorbidities were chosen from the 20 chronic conditions, specified by the Office of the Assistant Secretary for Health. Logistic regression analysis was conducted adjusting for multiple confounders. Results Prevalence of comorbidities was significantly different between men and women in all three conditions. After adjusting for age, length of stay, multicomorbidities and residence, there was no significant difference in 30-day mortality between men and women in AMI or ADHF, but men with pneumonia had slightly higher 30-day mortality with an OR of 1.19 (95% CI 1.06 to 1.34). There was no significant difference in 30-day readmission between men and women with AMI or pneumonia, but women with ADHF were slightly more likely to be readmitted within 30 days with OR 0.90 (95% CI 0.82 to 0.99). Conclusion Gender differences in the distribution of comorbidities exist in patients hospitalised with AMI, ADHF and pneumonia. However, there is minimal clinically meaningful impact of these differences on outcomes. Efforts to address gender difference may need to be diverted towards targeting overall population health, reducing race/ethnicity disparity and improving access to care.

Objective There exists a paucity of work in the development and testing of theoretical models specific to childhood health disparities even though they have been linked to the prevalence of adult health disparities including high rates of chronic disease. We conducted a systematic review and thematic analysis of existing models of health disparities specific to children to inform development of a unified conceptual framework. Methods We systematically reviewed articles reporting theoretical or explanatory models of disparities on a range of outcomes related to child health. We searched Ovid Medline In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid Embase, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus (database inception to 9 July 2015). A metanarrative approach guided the analysis process. Results A total of 48 studies presenting 48 models were included. This systematic review found multiple models but no consensus on one approach. However, we did discover a fair amount of overlap, such that the 48 models reviewed converged into the unified conceptual framework. The majority of models included factors in three domains: individual characteristics and behaviours (88%), healthcare providers and systems (63%), and environment/community (56%), . Only 38% of models included factors in the health and public policies domain. Conclusions A disease-agnostic unified conceptual framework may inform integration of existing knowledge of child health disparities and guide future research. This multilevel framework can focus attention among clinical, basic and social science research on the relationships between policy, social factors, health systems and the physical environment that impact children's health outcomes.

We aimed to develop tools that can facilitate uptake of evidence summarised in systematic reviews by clinical decision makers in health systems. After conducting a systematic review on the management of anxiety in children, we interviewed health system representatives, clinicians and patients to ask about additional information needed for decision-making. Using stakeholders’ feedback and literature searches for contextual and implementation information, we developed two tools (decision aids (DAs)), one for the health system and the second for the clinical encounter. This information mapped to factors of the GRADE (Grading of Recommendations, Assessment, Development and Evaluation) Evidence to Decision Framework. The health system DAs provided information on which patients are candidate for treatment, values and preferences, costs and resources, acceptability, impact on health equity, feasibility, drug dosing, alternative therapies, remission rates and prognosis. Health system stakeholders found the DA useful for clinical decision-making and generalisable to other conditions. The encounter DA was produced as cards containing information on issues that drive treatment decisions (effect on symptoms, effect on function, treatment burden, side effects and cost). Patients and parents prioritised the cards and chose the order in which these issues were discussed with clinician. The encounter DA was found to be helpful by patients, parents and clinicians. We conclude that the uptake of evidence summaries by health systems can be enhanced by developing tools that provide contextual and implementation information about clinical care. A dual approach addressing health system stakeholders as well as clinicians and patients is likely feasible and helpful.

e15779 Background: There is paucity of data regarding the best available second-line treatment following progression on gemcitabine-based regimens in metastatic pancreatic ductal adenocarcinoma (PDAC). While a Nanoliposomal formulation of irinotecan (MM398) is considered a standard of care, there is conflicting data relating to the use of oxaliplatin in this setting. We performed a meta-analysis to determine the effectiveness of adding oxaliplatin (OX) or various irinotecan (IRI) formulations to a fluoropyrimidine (FP) as a second-line in PDAC patients. Methods: We searched different databases, including PubMed, Embase and Cochrane, to identify randomized controlled trials comparing FP monotherapy to FP combination therapy that includes either oxaliplatin (FPOX) or various irinotecan formulations (FPIRI) in PDAC patients who progressed after first-line treatment. Secondary analyses were planned to assess the effectiveness of FPOX and FPIRI compared to FP. Outcomes of interest included overall survival (OS) and progression-free survival (PFS). The overall effect was pooled using the DerSimonian and Laird random effects models. Results: Five studies (2 with FPIRI and 3 with FPOX) with 895 patients were identified. Patients randomized to FPIRI/FPOX had a significantly improved PFS (HR = 0.74, CI 0.62 to 0.89) and a trend towards an improved OS compared to FP monotherapy (HR = 0.88, CI 0.65 to 1.19). When comparing FPIRI to FP, there was an improvement in both PFS (HR = 0.64, CI 0.47 to 0.87) and OS (HR = 0.70, CI 0.55 to 0.89) in patients treated with the combination. Conversely, FPOX showed only a modest improvement in PFS (HR = 0.81, CI 0.67, 0.97) with no improvement in OS (HR = 1.03, CI 0.64 to 1.67). Conclusions: Combination chemotherapy with oxaliplatin or various irinotecan formulations seem to improve PFS vs. single agent FP. FPIRI, but not FPOX seem to confer an OS advantage. Oxaliplatin with FP following gemcitabine failure may need further confirmatory studies to establish its role in refractory pancreas cancer.