BMJ Evidence-Based Medicine最新文献

Objectives: Our objective was to develop and test prompts designed to generate balanced, evidence-based information from artificial intelligence (AI) for the development of patient decision aid (DA) content. We compared the outputs of this AI-enhanced strategy with those produced by an experienced human team using a traditional development approach.

Methods: We conducted a comparative, mixed-methods, multiple-case study, with each case being a DA. Eight DAs were randomly selected from the Ottawa Inventory, stratified by author type (commercial, academic, public institution, professional association). We then followed a systematic process involving two researchers working independently. One researcher described the topics of the selected DAs and extracted their content by listing the available options with their benefits and harms. The other researcher-blind to the DA-used the topic description to generate AI-enhanced DA content by iteratively refining the prompt structures based on the International Patient Decision Aids Standards until the generated content stabilised. Quantitative analyses compared the number of options, benefits and harms generated by the traditional and AI-enhanced strategies, while qualitative analyses examined the differences in content.

Results: The selected DAs targeted different populations (older adults, women, the general population, children) and were produced in Canada, the UK, the USA or Australia. One type of DA (n=6) focused on a specific option (eg, whether to get vaccinated against COVID-19), the other (n=2) focused on improving an outcome (eg, treating attention-deficit/hyperactivity disorder symptoms). For option-focused DAs, 66% of the benefits/harms were generated by the AI-enhanced strategy only and 6.2% by the traditional strategy only. For outcome-focused DAs, 47% of the options were generated by the AI-enhanced strategy only, and 4% by the traditional strategy only. An evidence search confirmed that the options generated only by the AI-enhanced strategy were indeed beneficial, ruling out hallucinations. However, the AI-enhanced strategy did not suggest optimal combinations. Qualitative analysis showed that AI-enhanced content was generally richer.

Conclusions: This study provides practical guidance on leveraging AI to improve the efficiency of DA development and improve their quality.

Objectives: First, investigate whether a long compared with a short abstract decreases readers' attention. Second, investigate differences regarding perceptions of informativeness, accuracy, attractiveness and conciseness.

Design: Two-arm, single-blinded, parallel-group, superiority randomised controlled trial with 1:1 allocation.

Setting/participants: Researchers worldwide who indexed any type of systematic review in PubMed with an English abstract between 1 January 2024 and 26 March 2024.

Interventions: Researchers were randomly assigned to two groups. Both groups received the same cover letter by email with a link to our survey, which was assigned to either the short (277 words) or long abstract (771 words) of the same systematic review published in two different journals.

Main outcome measures: Primary outcome was the proportion of trial participation after reading the abstract, indicating readers' attention. Secondary outcomes were researchers' perceptions of four indicators of a well-written abstract (informativeness, accuracy, attractiveness, conciseness), and general abstract characteristics.

Results: A total of 5397 authors were randomly assigned to the short (n=2691) or long abstract (n=2706). Trial participation did not differ between groups (37.8% vs 35.0%; p=0.1935). While the short abstract was considered more attractive (60.5% vs 46.6%; p=0.0034) and concise (82.3% vs 37.9%; p<0.0001), the length had no impact on its informativeness (85.5% vs 91.2%; p=0.0594) and accuracy (80.2% vs 86.3%; p=0.0868). Regarding general abstract characteristics, 76.0% preferred a maximum length of 250-300 words, nearly all a structured format and about half supported reporting funding and registration information.

Conclusions: Abstract length had no impact on readers' attention, but short abstracts were considered more attractive and concise. Guidelines like PRISMA-A should recommend a range of 250-300 words for abstracts, allowing authors to include key information while prioritising clarity and precision. With authors considering information on funding and registration as important, journals should update their author guidelines to include these by default.

Trial registration number: NCT06525805.FundingNone.

Objective: To compare the effects of early physical interventions on the prevention of intensive care unit-acquired weakness (ICUAW) and the improvement of relevant clinical outcomes in patients with critical illness.

Methods: We systematically searched the Web of Science, PubMed, Embase and the Cochrane Central Register of Controlled Trials from their inception until 20 August 2024, to identify randomised controlled trials (RCTs) enrolling patients ≥18 years old and implementing early physical intervention that commenced at any time point during mechanical ventilation (MV) use or within 7 days after intensive care unit (ICU) admission for review. We synthesised data using a random-effects model and analysed through network meta-analysis (NMA) and component network meta-analysis (CNMA).

Main outcome measures: Primary outcome is the incidence of ICUAW. Secondary outcomes included Medical Research Council sum score, length of stay in the ICU or hospital, duration of MV and mortality rates in the ICU or hospital.

Results: Our analyses included 63 RCTs involving 24 treatments and eight components. The NMA results revealed systematic early mobilisation (SEM) combined with neuromuscular electrical stimulation (NMES), SEM alone and NMES alone may lead to a moderate to large reduction in the incidence of ICUAW (odds ratios [ORs]=0.03, 0.09 and 0.12, 95% confidence intervals [CIs]=0.00 to 0.42, 0.01 to 0.97 and 0.03 to 0.44, respectively) and improved relevant clinical outcomes compared with routine care. The CNMA results further indicated that SEM (OR=0.14, 95% CI=0.02 to 0.83) and NMES (OR=0.22, 95% CI=0.09 to 0.52) effectively mitigated the ICUAW incidence.

Conclusions: SEM and NMES are optimal interventions for preventing ICUAW. Healthcare providers in ICUs should implement early mobilisation with structured protocols and patient assessments or apply NMES to specific muscle groups to prevent ICUAW in critically ill patients and improve relevant clinical outcomes.

Prospero registration number: CRD42024581173.