Daily Anesthesiology Research Analysis
Analyzed 43 papers and selected 3 impactful papers.
Summary
Today’s top anesthesiology-related papers span ICU infection risk stratification, perioperative regional anesthesia effects on cognition, and generative AI for preanesthetic documentation. A new AURIS score outperforms the Candida score for predicting Candida auris candidaemia in colonized ICU patients, a double-blind RCT shows PENG block may reduce early POCD after THA, and an LLM-based tool shortens consultation time while maintaining clinician engagement.
Research Themes
- ICU infection risk prediction and antifungal stewardship
- Regional anesthesia and neurocognitive outcomes
- Generative AI to improve perioperative documentation workflows
Selected Articles
1. Development and validation of the AURIS score for predicting candidaemia in Candidozyma auris-colonised patients in the intensive care unit: a bicentric retrospective cohort study.
In a bicentric retrospective ICU cohort (n=422 C. auris–colonized patients), the AURIS score using four predictors (TPN, prior antifungal therapy, multifocal colonization, urinary isolation) achieved an AUC of 0.81 and outperformed the Candida score (AUC 0.75). At a 28% risk threshold, it offered high negative predictive value (0.94), supporting antifungal stewardship by identifying low-risk patients.
Impact: Provides a validated, pragmatic bedside risk tool in a high-consequence ICU pathogen where existing scores underperform, with clear stewardship implications.
Clinical Implications: Use the AURIS score to triage empiric antifungal therapy in C. auris–colonized ICU patients: de-escalate in low-risk individuals (high NPV) and prioritize diagnostics/therapy in high-risk profiles (e.g., TPN, multifocal colonization).
Key Findings
- Four independent predictors retained: TPN, prior antifungal therapy, multifocal colonization, and urinary isolation.
- Discrimination AUC 0.81; significantly outperformed the Candida score (AUC 0.75; p=0.0003).
- At 28% threshold: sensitivity 0.72, specificity 0.84, negative predictive value 0.94.
- Elastic Net refinement with internal bootstrap validation (n=5000) and external validation across two centers.
Methodological Strengths
- External validation and Elastic Net regularization with extensive bootstrap internal validation.
- Direct benchmarking against the Candida score with calibration assessment and pragmatic nomogram.
Limitations
- Retrospective design during prolonged outbreaks in two Spanish centers may limit generalizability.
- No prospective impact analysis on antifungal use, time-to-therapy, or patient-centered outcomes.
Future Directions: Prospective, multicenter impact studies across diverse C. auris clades; EHR integration with decision support; threshold optimization for stewardship and outcome benefits.
BACKGROUND: Candidozyma auris is an emerging multidrug-resistant pathogen that frequently colonises hospitalised patients and can cause invasive disease. Traditional tools, such as the Candida score, perform poorly in this setting. We aimed to externally validate and refine a clinical prediction model for C auris candidaemia among colonised patients in the intensive care unit (ICU). METHODS: We performed a retrospective analysis of prospectively and systematically collected cohort data from ICUs in two tertiary-care hospitals in Valencia, Spain, to predict candidaemia among adult C auris-colonised patients during prolonged outbreaks (October, 2017, to March, 2020). A previously derived logistic regression-based prediction model was externally validated, then refined in a bicentric cohort using Elastic Net regression. Internal validation was performed by bootstrap resampling (n=5000). Model discrimination and calibration were assessed and compared with the Candida score. FINDINGS: In the external validation cohort, 216 C auris-colonised ICU patients were included, of whom 31 (14%) developed candidaemia. After pooling this cohort with the original derivation cohort, a bicentric dataset of 422 patients was obtained, with 68 (16%) candidaemia events. Four predictors were retained: total parenteral nutrition (TPN; p<0·0001), previous antifungal therapy (p=0·027), multifocal colonisation (p=0·0020), and urinary isolation (p=0·029). These formed a simplified four-variable model (AURIS score) with a validated area under the curve of 0·81, outperforming the Candida score (0·75; p=0·0003). A graphical nomogram and point-based score for bedside risk estimation was created. At a 28% threshold, sensitivity was 0·72, specificity 0·84, and negative predictive value 0·94. INTERPRETATION: The AURIS score provides a pragmatic tool for risk stratification among C auris-colonised ICU patients, with value in identifying those at low risk of candidaemia, reducing unnecessary empirical antifungal therapy. Its predictors highlight the risk in multi-colonised carriers, the relevance of urinary colonisation, the ecological advantage from previous antifungal exposure, and the strong association with TPN. Broader validation across diverse clades and epidemiological settings is warranted before widespread implementation. FUNDING: None. TRANSLATION: For the Spanish translation of the abstract see Supplementary Materials section.
2. Effect of pericapsular nerve group block on postoperative cognitive function in older patients undergoing total hip arthroplasty.
In an 84-patient randomized, double-blind trial of elective THA under spinal anesthesia, preoperative PENG block reduced day-7 POCD (14.6% vs 37.2%), lowered early postoperative pain and opioid consumption, and improved inflammatory markers (NLR/PLR), mobilization, and length of stay.
Impact: Links a targeted regional block to preservation of early postoperative cognition, suggesting a modifiable, non-sedative strategy to mitigate POCD in high-risk older adults.
Clinical Implications: Consider PENG block as part of multimodal analgesia in older THA patients to reduce early POCD risk, opioid exposure, systemic inflammation, and facilitate early mobilization and discharge.
Key Findings
- Day-7 POCD incidence was significantly lower with PENG block (14.6% vs 37.2%).
- Lower pain scores and opioid consumption in the first 24 hours postoperatively.
- Earlier mobilization and shorter hospital stay with reduced NLR and PLR at 24–48 hours.
Methodological Strengths
- Prospective, randomized, double-blind design with standardized assessments (T‑MMSE, NRS, NLR/PLR).
- Balanced baseline characteristics with clear perioperative endpoints including mobilization and LOS.
Limitations
- Single-center study with modest sample size; findings need replication.
- Primary cognitive benefit demonstrated at day 7; durability through day 30/90 not fully detailed.
Future Directions: Multicenter trials powered for long-term cognitive outcomes (30–90 days), mechanistic studies linking analgesia and neuroinflammation to cognition, and comparative trials versus other hip blocks.
PURPOSE: Postoperative cognitive dysfunction (POCD) is common in older patients undergoing orthopedic surgery and may hinder clinical recovery. This prospective study evaluated whether a preoperative pericapsular nerve group (PENG) block reduces POCD incidence in patients undergoing total hip arthroplasty (THA). METHODS: This prospective, randomized, double-blind study included older patients scheduled for elective THA under spinal anesthesia. Patients were randomized into 2 groups: PENG (group P) and control (group C). Group P underwent an ultrasound-guided PENG block containing 20 mL of 0.25% bupivacaine, while group C received a sham block. Cognitive performance was evaluated using the telephone version of the Mini-Mental State Examination (T-MMSE) preoperatively and on postoperative days 7, 30 and 90. We evaluated postoperative pain using the numerical rating scale (NRS) and recorded opioid consumption, time to mobilization, hospital stay duration, neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR). RESULTS: The final analysis included 84 patients (41 in group P and 43 in group C). There were no significant differences in demographic characteristics or intraoperative data between the groups (p > 0.05). The incidence of POCD on postoperative day 7 was lower in the PENG group (14.6% vs. 37.2%; p < 0.05). In the first 24h postoperatively group P reported significantly lower pain scores and opioid use (p < 0.001), were mobilized earlier and discharged sooner (p < 0.001). The NLR and PLR were lower in the PENG group at 24 and 48h postoperatively (p < 0.05). CONCLUSION: Preoperative PENG block may preserve early postoperative cognitive function in THA patients, through improved analgesia, reduced opioid consumption, attenuation of systemic inflammation, early mobilization and shorter hospital stay.
3. Generative AI in preanesthetic consultations: Effects on efficiency, documentation workload, quality, and physician-patient interaction: A simulation trial.
In a randomized, within-subject simulation of 30 anesthesiologists, an LLM-based tool reduced preanesthetic consultation duration by 18% and markedly cut screen fixation, keyboard inputs, and mouse clicks. Despite slightly higher external quality ratings for manual notes, 60% preferred AI assistance, and clinicians reported better patient engagement.
Impact: Demonstrates tangible efficiency gains and reduced digital burden in a documentation-intensive perioperative workflow, informing safe, human-in-the-loop AI adoption.
Clinical Implications: AI-drafted notes can shorten preanesthetic visits and free clinician attention for patient interaction; however, physician review is essential to ensure documentation quality and safety.
Key Findings
- Consultation duration decreased by 252 s (-18%, p < 0.0001) with AI assistance.
- Large reductions in screen fixation (-78%), refixations (-73%), keyboard input (-87%), and mouse clicks (-19%).
- External ratings favored manual documentation (+4 PDQI-9 points), yet 60% of clinicians preferred AI assistance and reported better patient engagement.
Methodological Strengths
- Randomized, within-subject design with standardized patient and eye-tracking plus validated metrics (NASA-TLX, PDQI-9).
- Granular human-computer interaction analytics capturing workflow effects beyond time alone.
Limitations
- Simulation setting with standardized patients; real-world performance and safety require clinical trials.
- Documentation quality of AI drafts rated lower by external reviewers; added review may attenuate time savings.
Future Directions: Prospective clinical implementation studies measuring documentation accuracy, safety signals, medicolegal compliance, and net time savings after physician review; bias and transparency assessments.
BACKGROUND: Clinicians spend over 30% of their workday on electronic health records, reducing patient interaction and contributing to burnout. Preanesthetic consultations demand particularly detailed documentation, making them ideal for generative artificial intelligence (AI)-driven support. OBJECTIVE: This randomized simulation study evaluated a generative AI application based on a large language model (LLM) designed to automate documentation during preanesthetic consultations. We assessed its effects on consultation efficiency, clinician workload, physician-patient interaction, documentation quality, and user experience. METHODS: Thirty anesthesiologists at University Hospital Zurich each conducted two standardized consultations with the same simulated patient, once using the AI tool Isaac (Saipient AG, Zurich) and once with conventional manual documentation. Case order was randomized. The primary outcome was consultation duration. Secondary outcomes included visual attention (eye-tracking), human-computer interaction metrics, subjective workload (NASA-TLX), documentation quality (PDQI-9), self-assessed consultation quality, and workflow preferences. RESULTS: AI-assisted documentation reduced consultation duration by an average of 252 s (-18%, p < 0.0001), screen fixation (-78%, p = 0.0002), refixations (-73%, p < 0.0001), keyboard input (-87%, p < 0.0001), and mouse clicks (-19%, p = 0.01). Clinicians reported a trend toward lower workload (-16%, p = 0.07) and better patient engagement (median rating 87 vs. 69). However, external raters judged documentation quality to be higher for manual reports (+4 PDQI-9 points; p = 0.004), and clinicians expressed less confidence in AI-generated formatting. Still, 60% preferred AI assistance overall. CONCLUSIONS: LLM-based generative AI-supported documentation significantly improved efficiency and user experience in simulated preanesthetic consultations. While real-world use will require physicians to review and approve AI-generated drafts to ensure documentation quality, the structured outputs may still help reduce typing effort and screen interaction time, although the overall time savings may be smaller in clinical practice due to this additional review step.