Machine learning to optimize literature screening in medical guideline development.

OBJECTIVES: In a time of exponential growth of new evidence supporting clinical decision-making, combined with a labor-intensive process of selecting this evidence, methods are needed to speed up current processes to keep medical guidelines up-to-date. This study evaluated the performance and feasibility of active learning to support the selection of relevant publications within medical guideline development and to study the role of noisy labels. DESIGN: We used a mixed-methods design. Two independent clinicians' manual process of literature selection was evaluated for 14 searches. This was followed by a series of simulations investigating the performance of random reading versus using screening prioritization based on active learning. We identified hard-to-find papers and checked the labels in a reflective dialogue. MAIN OUTCOME MEASURES: Inter-rater reliability was assessed using Cohen's Kappa (ĸ). To evaluate the performance of active learning, we used the Work Saved over Sampling at 95% recall (WSS@95) and percentage Relevant Records Found at reading only 10% of the total number of records (RRF@10). We used the average time to discovery (ATD) to detect records with potentially noisy labels. Finally, the accuracy of labeling was discussed in a reflective dialogue with guideline developers. RESULTS: Mean ĸ for manual title-abstract selection by clinicians was 0.50 and varied between - 0.01 and 0.87 based on 5.021 abstracts. WSS@95 ranged from 50.15% (SD = 17.7) based on selection by clinicians to 69.24% (SD = 11.5) based on the selection by research methodologist up to 75.76% (SD = 12.2) based on the final full-text inclusion. A similar pattern was seen for RRF@10, ranging from 48.31% (SD = 23.3) to 62.8% (SD = 21.20) and 65.58% (SD = 23.25). The performance of active learning deteriorates with higher noise. Compared with the final full-text selection, the selection made by clinicians or research methodologists deteriorated WSS@95 by 25.61% and 6.25%, respectively. CONCLUSION: While active machine learning tools can accelerate the process of literature screening within guideline development, they can only work as well as the input given by human raters. Noisy labels make noisy machine learning.

[How often do medical guidelines refer to articles written in Dutch?] / Hoe vaak verwijzen richtlijnen naar artikelen in hetNTvG?

OBJECTIVE: For Dutch medical guidelines, Dutch research articles published in the NTvG (NederlandsTijdschriftvoorGeneeskunde) and other medical journals are not searched systematically and are only used sporadically. Using these publications in the process of guideline development can be useful for recommendations regarding the Dutch context of care. In this research, we have investigated how often and in which parts of Dutch guidelines articles published in NTvG are used. DESIGN: We specifically investigated how often articles published in NTvG are mentioned in Dutch medical guidelines published on www.richtlijnendatabase.nl, that were developed in 2019, 2020 and 2021. METHOD: In all parts of new or revised Dutch medical guidelines published in these years on www.richtlijnendatabase.nl, we searched for references of articles published in NTvG. RESULTS: The results show that in 3% of all Dutch medical guidelines a reference to an article published in NTvG is made. These references were made in the literature summaries (21% of the references), the reflections on the literature for the Dutch context of care (48% of the references), or in other areas such as the introduction (10% of the references) or appendices (21% of the references). CONCLUSION: Articles published in NTvG may be relevant for making recommendations in Dutch medical guidelines, as these publications usually reflect the Dutch care context, and may do more so than research published in international journals. The results of this research show that the number of Dutch guidelines where these articles are used is limited. Dutch research articles may be a source of information that is yet to be tapped into.

Practice and challenges for organ donation after medical assistance in dying: A scoping review including the results of the first international roundtable in 2021.

The procedure combining medical assistance in dying (MAiD) with donations after circulatory determination of death (DCDD) is known as organ donation after euthanasia (ODE). The first international roundtable on ODE was held during the 2021 WONCA family medicine conference as part of a scoping review. It aimed to document practice and related issues to advise patients, professionals, and policymakers, aiding the development of responsible guidelines and helping to navigate the issues. This was achieved through literature searches and national and international stakeholder meetings. Up to 2021, ODE was performed 286 times in Canada, the Netherlands, Spain, and Belgium, including eight cases of ODE from home (ODEH). MAiD was provided 17,217 times (2020) in the eight countries where ODE is permitted. As of 2021, 837 patients (up to 14% of recipients of DCDD donors) had received organs from ODE. ODE raises some important ethical concerns involving patient autonomy, the link between the request for MAiD and the request to donate organs and the increased burden placed on seriously ill MAiD patients.

Palliative care interventions in intensive care unit patients.

PURPOSE: The integration of palliative care into intensive care units (ICUs) is advocated to mitigate physical and psychological burdens for patients and their families, and to improve end-of-life care. The most efficacious palliative care interventions, the optimal model of their delivery and the most appropriate outcome measures in ICU are not clear. METHODS: We conducted a systematic review of randomised clinical trials and observational studies to evaluate the number and types of palliative care interventions implemented within the ICU setting, to assess their impact on ICU practice and to evaluate differences in palliative care approaches across different countries. RESULTS: Fifty-eight full articles were identified, including 9 randomised trials and 49 cohort studies; all but 4 were conducted within North America. Interventions were categorised into five themes: communication (14, 24.6%), ethics consultations (5, 8.8%), educational (18, 31.6%), involvement of a palliative care team (28, 49.1%) and advance care planning or goals-of-care discussions (7, 12.3%). Thirty studies (51.7%) proposed an integrative model, whilst 28 (48.3%) reported a consultative one. The most frequently reported outcomes were ICU or hospital length of stay (33/55, 60%), limitation of life-sustaining treatment decisions (22/55, 40%) and mortality (15/55, 27.2%). Quantitative assessment of pooled data was not performed due to heterogeneity in interventions and outcomes between studies. CONCLUSION: Beneficial effects on the most common outcomes were associated with strategies to enhance palliative care involvement, either with an integrative or a consultative approach. Few studies reported functional outcomes for ICU patients. Almost all studies were from North America, limiting the generalisability to other healthcare systems.

Palliative care interventions in intensive care unit patients - a systematic review protocol.

BACKGROUND: Even though data suggest that palliative care (PC) improves patient quality of life, caregiver burden, cost, and intensive care unit (ICU) length of stay, integration of PC in the ICU is far from being universally accepted. Poor understanding of what PC provides is one of the barriers to the widespread implementation of their services in ICU. Evidence suggests that the availability of specialist PC is lacking in most European countries and provided differently depending on geographical location. The aim of this systematic review is to compare the numbers and types of PC interventions and gauge their impact on stakeholder outcomes and ICU resource utilisation. METHODS: We will undertake a systematic review of the published peer-reviewed journal articles; our search will be carried out MEDLINE, Embase, Cochrane, CINAHL, and PsycINFO. The search strategy will include variations in the term 'palliative care' and 'intensive care'. All studies with patient populations undergoing palliative care interventions will be selected. Only full-text articles will be considered, and conference abstracts excluded. There will be no date restrictions on the year of publications or on language. The primary aim of the present study is to compare the numbers and types of PC interventions in ICU and their impact on stakeholder (patient, family, clinician, other) outcomes. Reporting of findings will follow the Preferred Reporting Items of Systematic Reviews and Meta-Analyses (PRISMA) guidelines. DISCUSSION: This review will provide insight into the implementation of palliative care in ICU, elucidate differences between countries and health systems, reveal most effective models, and contribute to identifying research priorities to improve outcomes. SYSTEMATIC REVIEW REGISTRATION: International Prospective Register of Systematic reviews PROSPERO ( CRD42018094315 ).

Nutritional interventions in community-dwelling Alzheimer patients with (risk of) undernutrition: a systematic review.

ABSTRACT Background: Weight loss and undernutrition are common in patients with Alzheimer's disease (AD) and associated with negative health outcomes. In the current guidelines on diagnosis and treatment of AD, no recommendations for treatment of (risk of) undernutrition in community-dwelling AD patients are given. Methods: We conducted a systematic review on the effect of nutritional interventions in community-dwelling AD patients with (risk of) undernutrition, according to the methods outlined by the Cochrane Collaboration. Three electronic databases and three trial registers were searched from inception till April 2013. Results: Literature search in the electronic databases yielded 546 records of which one was relevant for this review. This study, with a high risk of bias, demonstrated that oral nutritional supplements improved nutritional outcomes without effect on clinical and biochemical outcomes. The search in the trial registers yielded 369 records of which two were relevant. One trial was terminated because of failing inclusion, the other is ongoing. Conclusions: This systematic review on the effect of nutritional interventions in community-dwelling AD patients with (risk of) undernutrition, reveals a serious lack of evidence. Therefore, it is not possible to state what the best approach is.