The effect of computerised decision support alerts tailored to intensive care on the administration of high-risk drug combinations, and their monitoring: a cluster randomised stepped-wedge trial.

BACKGROUND: Drug-drug interactions (DDIs) can harm patients admitted to the intensive care unit (ICU). Yet, clinical decision support systems (CDSSs) aimed at helping physicians prevent DDIs are plagued by low-yield alerts, causing alert fatigue and compromising patient safety. The aim of this multicentre study was to evaluate the effect of tailoring potential DDI alerts to the ICU setting on the frequency of administered high-risk drug combinations. METHODS: We implemented a cluster randomised stepped-wedge trial in nine ICUs in the Netherlands. Five ICUs already used potential DDI alerts. Patients aged 18 years or older admitted to the ICU with at least two drugs administered were included. Our intervention was an adapted CDSS, only providing alerts for potential DDIs considered as high risk. The intervention was delivered at the ICU level and targeted physicians. We hypothesised that showing only relevant alerts would improve CDSS effectiveness and lead to a decreased number of administered high-risk drug combinations. The order in which the intervention was implemented in the ICUs was randomised by an independent researcher. The primary outcome was the number of administered high-risk drug combinations per 1000 drug administrations per patient and was assessed in all included patients. This trial was registered in the Netherlands Trial Register (identifier NL6762) on Nov 26, 2018, and is now closed. FINDINGS: In total, 10 423 patients admitted to the ICU between Sept 1, 2018, and Sept 1, 2019, were assessed and 9887 patients were included. The mean number of administered high-risk drug combinations per 1000 drug administrations per patient was 26·2 (SD 53·4) in the intervention group (n=5534), compared with 35·6 (65·0) in the control group (n=4353). Tailoring potential DDI alerts to the ICU led to a 12% decrease (95% CI 5-18%; p=0·0008) in the number of administered high-risk drug combinations per 1000 drug administrations per patient, after adjusting for clustering and prognostic factors. INTERPRETATION: This cluster randomised stepped-wedge trial showed that tailoring potential DDI alerts to the ICU setting significantly reduced the number of administered high-risk drug combinations. Our list of high-risk drug combinations can be used in other ICUs, and our strategy of tailoring alerts based on clinical relevance could be applied to other clinical settings. FUNDING: ZonMw.

Assessing the use of HL7 FHIR for implementing the FAIR guiding principles: a case study of the MIMIC-IV Emergency Department module.

Objectives: To provide a real-world example on how and to what extent Health Level Seven Fast Healthcare Interoperability Resources (FHIR) implements the Findable, Accessible, Interoperable, and Reusable (FAIR) guiding principles for scientific data. Additionally, presents a list of FAIR implementation choices for supporting future FAIR implementations that use FHIR. Materials and methods: A case study was conducted on the Medical Information Mart for Intensive Care-IV Emergency Department (MIMIC-ED) dataset, a deidentified clinical dataset converted into FHIR. The FAIRness of this dataset was assessed using a set of common FAIR assessment indicators. Results: The FHIR distribution of MIMIC-ED, comprising an implementation guide and demo data, was more FAIR compared to the non-FHIR distribution. The FAIRness score increased from 60 to 82 out of 95 points, a relative improvement of 37%. The most notable improvements were observed in interoperability, with a score increase from 5 to 19 out of 19 points, and reusability, with a score increase from 8 to 14 out of 24 points. A total of 14 FAIR implementation choices were identified. Discussion: Our work examined how and to what extent the FHIR standard contributes to FAIR data. Challenges arose from interpreting the FAIR assessment indicators. This study stands out for providing a real-world example of a dataset that was made more FAIR using FHIR. Conclusion: To the best of our knowledge, this is the first study that formally assessed the conformance of a FHIR dataset to the FAIR principles. FHIR improved the accessibility, interoperability, and reusability of MIMIC-ED. Future research should focus on implementing FHIR in research data infrastructures.

Characteristics and outcome of COVID-19 patients admitted to the ICU: a nationwide cohort study on the comparison between the consecutive stages of the COVID-19 pandemic in the Netherlands, an update.

BACKGROUND: Previously, we reported a decreased mortality rate among patients with COVID-19 who were admitted at the ICU during the final upsurge of the second wave (February-June 2021) in the Netherlands. We examined whether this decrease persisted during the third wave and the phases with decreasing incidence of COVID-19 thereafter and brought up to date the information on patient characteristics. METHODS: Data from the National Intensive Care Evaluation (NICE)-registry of all COVID-19 patients admitted to an ICU in the Netherlands were used. Patient characteristics and rates of in-hospital mortality (the primary outcome) during the consecutive periods after the first wave (periods 2-9, May 25, 2020-January 31, 2023) were compared with those during the first wave (period 1, February-May 24, 2020). RESULTS: After adjustment for patient characteristics and ICU occupancy rate, the mortality risk during the initial upsurge of the third wave (period 6, October 5, 2021-January, 31, 2022) was similar to that of the first wave (ORadj = 1.01, 95%-CI [0.88-1.16]). The mortality rates thereafter decreased again (e.g., period 9, October 5, 2022-January, 31, 2023: ORadj = 0.52, 95%-CI [0.41-0.66]). Among the SARS-CoV-2 positive patients, there was a huge drop in the proportion of patients with COVID-19 as main reason for ICU admission: from 88.2% during the initial upsurge of the third wave to 51.7%, 37.3%, and 41.9% for the periods thereafter. Restricting the analysis to these patients did not modify the results on mortality. CONCLUSIONS: The results show variation in mortality rates among critically ill COVID-19 patients across the calendar time periods that is not explained by differences in case-mix and ICU occupancy rates or by varying proportions of patients with COVID-19 as main reason for ICU admission. The consistent increase in mortality during the initial, rising phase of each separate wave might be caused by the increased virulence of the contemporary virus strain and lacking immunity to the new strain, besides unmeasured patient-, treatment- and healthcare system characteristics.

Adverse drug events caused by three high-risk drug-drug interactions in patients admitted to intensive care units: A multicentre retrospective observational study.

AIMS: Knowledge about adverse drug events caused by drug-drug interactions (DDI-ADEs) is limited. We aimed to provide detailed insights about DDI-ADEs related to three frequent, high-risk potential DDIs (pDDIs) in the critical care setting: pDDIs with international normalized ratio increase (INR+ ) potential, pDDIs with acute kidney injury (AKI) potential, and pDDIs with QTc prolongation potential. METHODS: We extracted routinely collected retrospective data from electronic health records of intensive care units (ICUs) patients (≥18 years), admitted to ten hospitals in the Netherlands between January 2010 and September 2019. We used computerized triggers (e-triggers) to preselect patients with potential DDI-ADEs. Between September 2020 and October 2021, clinical experts conducted a retrospective manual patient chart review on a subset of preselected patients, and assessed causality, severity, preventability, and contribution to ICU length of stay of DDI-ADEs using internationally prevailing standards. RESULTS: In total 85 422 patients with ≥1 pDDI were included. Of these patients, 32 820 (38.4%) have been exposed to one of the three pDDIs. In the exposed group, 1141 (3.5%) patients were preselected using e-triggers. Of 237 patients (21%) assessed, 155 (65.4%) experienced an actual DDI-ADE; 52.9% had severity level of serious or higher, 75.5% were preventable, and 19.3% contributed to a longer ICU length of stay. The positive predictive value was the highest for DDI-INR+ e-trigger (0.76), followed by DDI-AKI e-trigger (0.57). CONCLUSION: The highly preventable nature and severity of DDI-ADEs, calls for action to optimize ICU patient safety. Use of e-triggers proved to be a promising preselection strategy.

Quality improvement of Dutch ICUs from 2009 to 2021: A registry based observational study.

PURPOSE: To investigate the development in quality of ICU care over time using the Dutch National Intensive Care Evaluation (NICE) registry. MATERIALS AND METHODS: We included data from all ICU admissions in the Netherlands from those ICUs that submitted complete data between 2009 and 2021 to the NICE registry. We determined median and interquartile range for eight quality indicators. To evaluate changes over time on the indicators, we performed multilevel regression analyses, once without and once with the COVID-19 years 2020 and 2021 included. Additionally we explored between-ICU heterogeneity by calculating intraclass correlation coefficients (ICC). RESULTS: 705,822 ICU admissions from 55 (65%) ICUs were included in the analyses. ICU length of stay (LOS), duration of mechanical ventilation (MV), readmissions, in-hospital mortality, hypoglycemia, and pressure ulcers decreased significantly between 2009 and 2019 (OR <1). After including the COVID-19 pandemic years, the significant change in MV duration, ICU LOS, and pressure ulcers disappeared. We found an ICC ≤0.07 on the quality indicators for all years, except for pressure ulcers with an ICC of 0.27 for 2009 to 2021. CONCLUSIONS: Quality of Dutch ICU care based on seven indicators significantly improved from 2009 to 2019 and between-ICU heterogeneity is medium to small, except for pressure ulcers. The COVID-19 pandemic disturbed the trend in quality improvement, but unaltered the between-ICU heterogeneity.

Acute kidney injury associated with nephrotoxic drugs in critically ill patients: a multicenter cohort study using electronic health record data.

Background: Nephrotoxic drugs frequently cause acute kidney injury (AKI) in adult intensive care unit (ICU) patients. However, there is a lack of large pharmaco-epidemiological studies investigating the associations between drugs and AKI. Importantly, AKI risk factors may also be indications or contraindications for drugs and thereby confound the associations. Here, we aimed to estimate the associations between commonly administered (potentially) nephrotoxic drug groups and AKI in adult ICU patients whilst adjusting for confounding. Methods: In this multicenter retrospective observational study, we included adult ICU admissions to 13 Dutch ICUs. We measured exposure to 44 predefined (potentially) nephrotoxic drug groups. The outcome was AKI during ICU admission. The association between each drug group and AKI was estimated using etiological cause-specific Cox proportional hazard models and adjusted for confounding. To facilitate an (independent) informed assessment of residual confounding, we manually identified drug group-specific confounders using a large drug knowledge database and existing literature. Results: We included 92 616 ICU admissions, of which 13 492 developed AKI (15%). We found 14 drug groups to be associated with a higher hazard of AKI after adjustment for confounding. These groups included established (e.g. aminoglycosides), less well established (e.g. opioids) and controversial (e.g. sympathomimetics with α- and ß-effect) drugs. Conclusions: The results confirm existing insights and provide new ones regarding drug associated AKI in adult ICU patients. These insights warrant caution and extra monitoring when prescribing nephrotoxic drugs in the ICU and indicate which drug groups require further investigation.

Strain on Scarce Intensive Care Beds Drives Reduced Patient Volumes, Patient Selection, and Worse Outcome: A National Cohort Study.

OBJECTIVES: Strain on ICUs during the COVID-19 pandemic required stringent triage at the ICU to distribute resources appropriately. This could have resulted in reduced patient volumes, patient selection, and worse outcome of non-COVID-19 patients, especially during the pandemic peaks when the strain on ICUs was extreme. We analyzed this potential impact on the non-COVID-19 patients. DESIGN: A national cohort study. SETTING: Data of 71 Dutch ICUs. PARTICIPANTS: A total of 120,393 patients in the pandemic non-COVID-19 cohort (from March 1, 2020 to February 28, 2022) and 164,737 patients in the prepandemic cohort (from January 1, 2018 to December 31, 2019). INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Volume, patient characteristics, and mortality were compared between the pandemic non-COVID-19 cohort and the prepandemic cohort, focusing on the pandemic period and its peaks, with attention to strata of specific admission types, diagnoses, and severity. The number of admitted non-COVID-19 patients during the pandemic period and its peaks were, respectively, 26.9% and 34.2% lower compared with the prepandemic cohort. The pandemic non-COVID-19 cohort consisted of fewer medical patients (48.1% vs. 50.7%), fewer patients with comorbidities (36.5% vs. 40.6%), and more patients on mechanical ventilation (45.3% vs. 42.4%) and vasoactive medication (44.7% vs. 38.4%) compared with the prepandemic cohort. Case-mix adjusted mortality during the pandemic period and its peaks was higher compared with the prepandemic period, odds ratios were, respectively, 1.08 (95% CI, 1.05-1.11) and 1.10 (95% CI, 1.07-1.13). CONCLUSIONS: In non-COVID-19 patients the strain on healthcare has driven lower patient volume, selection of fewer comorbid patients who required more intensive support, and a modest increase in the case-mix adjusted mortality.

Correctly structured problem lists lead to better and faster clinical decision-making in electronic health records compared to non-curated problem lists: A single-blinded crossover randomized controlled trial.

BACKGROUND: Correctly structured problem lists in electronic health records (EHRs) offer major benefits to patient care. Without structured lists, diagnosis information is often scatteredly documented in free text, which may contribute to errors and inefficient information retrieval. This study aims to assess whether EHRs with correctly structured problem lists result in better and faster clinical decision-making compared to non-curated problem lists. METHODS: Two versions of two patient records (A and B) were created in an EHR training environment: one version included diagnosis information structured and coded on the problem list ("correctly structured problem list"), the other version had missing problem list diagnoses and diagnosis information partly documented in free text ("non-curated problem list"). In this single-blinded crossover randomized controlled trial, healthcare providers, who can prescribe medications, from two Dutch university medical center locations first evaluated a randomized version of patient A, then B. Participants were asked to motivate their answer to two medication prescription questions. One (test) question required information similarly presented in both record versions. The second (comparison) question required information documented on problem lists and/or in notes. The primary outcome measure was the correctness of the motivated answer to the comparison question. Secondary outcome measure was the time to answer and motivate both questions correctly. RESULTS: As planned, 160 participants enrolled. Two were excluded for not meeting inclusion criteria. Correctly structured problem lists increased providers' ability to answer the comparison question correctly (56.3 % versus 33.5 %, McNemar odds ratio 2.80 (1.65-4.93) 95 %-CI). Median time to answer both questions correctly was significantly lower for EHRs with correctly structured problem lists (Wilcoxon-signed-rank test p = 0.00002, with incorrect answers coded equally at slowest time). CONCLUSIONS: Correctly structured problem lists lead to better and faster clinical decision-making. Increased structured problem lists usage may be warranted for which implementation policies should be developed.

Determining and assessing characteristics of data element names impacting the performance of annotation using Usagi.

INTRODUCTION: Hospitals generate large amounts of data and this data is generally modeled and labeled in a proprietary way, hampering its exchange and integration. Manually annotating data element names to internationally standardized data element identifiers is a time-consuming effort. Tools can support performing this task automatically. This study aimed to determine what factors influence the quality of automatic annotations. METHODS: Data element names were used from the Dutch COVID-19 ICU Data Warehouse containing data on intensive care patients with COVID-19 from 25 hospitals in the Netherlands. In this data warehouse, the data had been merged using a proprietary terminology system while also storing the original hospital labels (synonymous names). Usagi, an OHDSI annotation tool, was used to perform the annotation for the data. A gold standard was used to determine if Usagi made correct annotations. Logistic regression was used to determine if the number of characters, number of words, match score (Usagi's certainty) and hospital label origin influenced Usagi's performance to annotate correctly. RESULTS: Usagi automatically annotated 30.5% of the data element names correctly and 5.5% of the synonymous names. The match score is the best predictor for Usagi finding the correct annotation. It was determined that the AUC of data element names was 0.651 and 0.752 for the synonymous names respectively. The AUC for the individual hospital label origins varied between 0.460 to 0.905. DISCUSSION: The results show that Usagi performed better to annotate the data element names than the synonymous names. The hospital origin in the synonymous names dataset was associated with the amount of correctly annotated concepts. Hospitals that performed better had shorter synonymous names and fewer words. Using shorter data element names or synonymous names should be considered to optimize the automatic annotating process. Overall, the performance of Usagi is too poor to completely rely on for automatic annotation.

The need to strengthen the evaluation of the impact of Artificial Intelligence-based decision support systems on healthcare provision.

Despite the renewed interest in Artificial Intelligence-based clinical decision support systems (AI-CDS), there is still a lack of empirical evidence supporting their effectiveness. This underscores the need for rigorous and continuous evaluation and monitoring of processes and outcomes associated with the introduction of health information technology. We illustrate how the emergence of AI-CDS has helped to bring to the fore the critical importance of evaluation principles and action regarding all health information technology applications, as these hitherto have received limited attention. Key aspects include assessment of design, implementation and adoption contexts; ensuring systems support and optimise human performance (which in turn requires understanding clinical and system logics); and ensuring that design of systems prioritises ethics, equity, effectiveness, and outcomes. Going forward, information technology strategy, implementation and assessment need to actively incorporate these dimensions. International policy makers, regulators and strategic decision makers in implementing organisations therefore need to be cognisant of these aspects and incorporate them in decision-making and in prioritising investment. In particular, the emphasis needs to be on stronger and more evidence-based evaluation surrounding system limitations and risks as well as optimisation of outcomes, whilst ensuring learning and contextual review. Otherwise, there is a risk that applications will be sub-optimally embodied in health systems with unintended consequences and without yielding intended benefits.

Conservative versus Liberal Oxygenation Targets in Intensive Care Unit Patients (ICONIC): A Randomized Clinical Trial.

Rationale: Supplemental oxygen is widely administered to ICU patients, but appropriate oxygenation targets remain unclear. Objectives: This study aimed to determine whether a low-oxygenation strategy would lower 28-day mortality compared with a high-oxygenation strategy. Methods: This randomized multicenter trial included mechanically ventilated ICU patients with an expected ventilation duration of at least 24 hours. Patients were randomized 1:1 to a low-oxygenation (PaO2, 55-80 mm Hg; or oxygen saturation as measured by pulse oximetry, 91-94%) or high-oxygenation (PaO2, 110-150 mm Hg; or oxygen saturation as measured by pulse oximetry, 96-100%) target until ICU discharge or 28 days after randomization, whichever came first. The primary outcome was 28-day mortality. The study was stopped prematurely because of the COVID-19 pandemic when 664 of the planned 1,512 patients were included. Measurements and Main Results: Between November 2018 and November 2021, a total of 664 patients were included in the trial: 335 in the low-oxygenation group and 329 in the high-oxygenation group. The median achieved PaO2 was 75 mm Hg (interquartile range, 70-84) and 115 mm Hg (interquartile range, 100-129) in the low- and high-oxygenation groups, respectively. At Day 28, 129 (38.5%) and 114 (34.7%) patients had died in the low- and high-oxygenation groups, respectively (risk ratio, 1.11; 95% confidence interval, 0.9-1.4; P = 0.30). At least one serious adverse event was reported in 12 (3.6%) and 17 (5.2%) patients in the low- and high-oxygenation groups, respectively. Conclusions: Among mechanically ventilated ICU patients with an expected mechanical ventilation duration of at least 24 hours, using a low-oxygenation strategy did not result in a reduction of 28-day mortality compared with a high-oxygenation strategy. Clinical trial registered with the National Trial Register and the International Clinical Trials Registry Platform (NTR7376).

Evaluation of Patient-Friendly Diagnosis Clarifications in a Hospital Patient Portal.

BACKGROUND: Medical data can be difficult to comprehend for patients, but only a limited number of patient-friendly terms and definitions are available to clarify medical concepts. Therefore, we developed an algorithm that generalizes diagnoses to more general concepts that do have patient-friendly terms and definitions in SNOMED CT. We implemented the generalizations, and diagnosis clarifications with synonyms and definitions that were already available, in the problem list of a hospital patient portal. OBJECTIVE: We aimed to assess the extent to which the clarifications cover the diagnoses in the problem list, the extent to which clarifications are used and appreciated by patient portal users, and to explore differences in viewing problems and clarifications between subgroups of users and diagnoses. METHODS: We measured the coverage of diagnoses by clarifications, usage of the problem list and the clarifications, and user, patient and diagnosis characteristics with aggregated, routinely available electronic health record and log file data. Additionally, patient portal users provided quantitative and qualitative feedback about the clarification quality. RESULTS: Of all patient portal users who viewed diagnoses on their problem list (n = 2,660), 89% had one or more diagnoses with clarifications. In addition, 55% of patient portal users viewed the clarifications. Users who rated the clarifications (n = 108) considered the clarifications to be of good quality on average, with a median rating per patient of 6 (interquartile range: 4-7; from 1 very bad to 7 very good). Users commented that they found clarifications to be clear and recognized the clarifications from their own experience, but sometimes also found the clarifications incomplete or disagreed with the diagnosis itself. CONCLUSION: This study shows that the clarifications are used and appreciated by patient portal users. Further research and development will be dedicated to the maintenance and further quality improvement of the clarifications.

Drug-related causes attributed to acute kidney injury and their documentation in intensive care patients.

PURPOSE: To investigate drug-related causes attributed to acute kidney injury (DAKI) and their documentation in patients admitted to the Intensive Care Unit (ICU). METHODS: This study was conducted in an academic hospital in the Netherlands by reusing electronic health record (EHR) data of adult ICU admissions between November 2015 to January 2020. First, ICU admissions with acute kidney injury (AKI) stage 2 or 3 were identified. Subsequently, three modes of DAKI documentation in EHR were examined: diagnosis codes (structured data), allergy module (semi-structured data), and clinical notes (unstructured data). RESULTS: n total 8124 ICU admissions were included, with 542 (6.7%) ICU admissions experiencing AKI stage 2 or 3. The ICU physicians deemed 102 of these AKI cases (18.8%) to be drug-related. These DAKI cases were all documented in the clinical notes (100%), one in allergy module (1%) and none via diagnosis codes. The clinical notes required the highest time investment to analyze. CONCLUSIONS: Drug-related causes comprise a substantial part of AKI in the ICU patients. However, current unstructured DAKI documentation practice via clinical notes hampers our ability to gain better insights about DAKI occurrence. Therefore, both automating DAKI identification from the clinical notes and increasing structured DAKI documentation should be encouraged.

Increased mortality in ICU patients ≥70 years old with COVID-19 compared to patients with other pneumonias.

BACKGROUND: Patients over 70 years old represent a substantial proportion of the COVID-19 ICU population and their mortality rates are high. The aim of this study is to describe the outcomes of patients ≥70 years old admitted to Dutch ICUs with COVID-19, compared to patients ≥70 years old admitted to the ICU for bacterial and other viral pneumonias, with adjustments for age, comorbidities, severity of illness, and ICU occupancy rate. METHODS: Retrospective cohort study including patients ≥70 years old admitted to Dutch ICUs, comparing patients admitted with COVID-19 from March 1st 2020 to January 1st 2022 with patients ≥70 years old admitted because of a bacterial and other viral pneumonia, both divided in a historical (i.e., January 1st 2017 to January 1st 2020) and current cohort (i.e., March 1st 2020 to January 1st 2022). Primary outcome is hospital mortality. RESULTS: 11,525 unique patients ≥70 years old admitted to Dutch ICUs were included; 5094 with COVID-19, 5334 with a bacterial pneumonia, and 1312 with another viral pneumonia. ICU-mortality and in-hospital mortality rates of the patients ≥70 years old admitted with COVID-19 were 39.7% and 47.6% respectively. ICU- and hospital mortality rates of the patients who were admitted in the same or in an historical time period with a bacterial pneumonia or other viral pneumonias were considerably lower (19.5% and 28.6% for patients with a bacterial pneumonia in the historical cohort and 19.1% and 28.8% in the same period, for the patients with other viral pneumonias 20.7% and 28.9%, and 22.7% and 31.8% respectively, all p < 0.001). Differences persisted after correction for several clinical characteristics and ICU occupancy rate. CONCLUSIONS: In ICU-patients ≥70 years old, COVID-19 is more severe compared to bacterial or viral pneumonia.

Electronic health record-based prediction models for in-hospital adverse drug event diagnosis or prognosis: a systematic review.

OBJECTIVE: We conducted a systematic review to characterize and critically appraise developed prediction models based on structured electronic health record (EHR) data for adverse drug event (ADE) diagnosis and prognosis in adult hospitalized patients. MATERIALS AND METHODS: We searched the Embase and Medline databases (from January 1, 1999, to July 4, 2022) for articles utilizing structured EHR data to develop ADE prediction models for adult inpatients. For our systematic evidence synthesis and critical appraisal, we applied the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS). RESULTS: Twenty-five articles were included. Studies often did not report crucial information such as patient characteristics or the method for handling missing data. In addition, studies frequently applied inappropriate methods, such as univariable screening for predictor selection. Furthermore, the majority of the studies utilized ADE labels that only described an adverse symptom while not assessing causality or utilizing a causal model. None of the models were externally validated. CONCLUSIONS: Several challenges should be addressed before the models can be widely implemented, including the adherence to reporting standards and the adoption of best practice methods for model development and validation. In addition, we propose a reorientation of the ADE prediction modeling domain to include causality as a fundamental challenge that needs to be addressed in future studies, either through acquiring ADE labels via formal causality assessments or the usage of adverse event labels in combination with causal prediction modeling.

Influenza Season and Outcome After Elective Cardiac Surgery: An Observational Cohort Study.

BACKGROUND: An asymptomatic respiratory viral infection during cardiac surgery could lead to pulmonary complications and increased mortality. For elective surgery, testing for respiratory viral infection before surgery or vaccination could reduce the number of these pulmonary complications. The aim of this study was to investigate the association between influenzalike illness (ILI) seasons and prolonged mechanical ventilation and inhospital mortality in a Dutch cohort of adult elective cardiac surgery patients. METHODS: Cardiac surgery patients who were admitted to the intensive care unit between January 1, 2014, and February 1, 2020, were included. The primary endpoint was the duration of invasive mechanical ventilation in the ILI season compared with baseline season. Secondary endpoints were the median Pao2 to fraction of inspired oxygen ratio on days 1, 3, and 7 and postoperative inhospital mortality. RESULTS: A total of 42,277 patients underwent cardiac surgery, 12,994 (30.7%) in the ILI season, 15,843 (37.5%) in the intermediate season, and 13,440 (31.8%) in the baseline season. No hazard rates indicative of a longer duration of invasive mechanical ventilation during the ILI season were found. No differences were found for the median Pao2 to fraction of inspired oxygen ratio between seasons. However, inhospital mortality was higher in the ILI season compared with baseline season (odds ratio 1.67; 95% CI, 1.14-2.46). CONCLUSIONS: Patients undergoing cardiac surgery during the ILI season were at increased risk of inhospital mortality compared with patients in the baseline season. No evidence was found that this difference is caused by direct postoperative pulmonary complications.

Age Moderates the Effect of Obesity on Mortality Risk in Critically Ill Patients With COVID-19: A Nationwide Observational Cohort Study.

OBJECTIVES: A high body mass index (BMI) is associated with an unfavorable disease course in COVID-19, but not among those who require admission to the ICU. This has not been examined across different age groups. We examined whether age modifies the association between BMI and mortality among critically ill COVID-19 patients. DESIGN: An observational cohort study. SETTING: A nationwide registry analysis of critically ill patients with COVID-19 registered in the National Intensive Care Evaluation registry. PATIENTS: We included 15,701 critically ill patients with COVID-19 (10,768 males [68.6%] with median [interquartile range] age 64 yr [55-71 yr]), of whom 1,402 (8.9%) patients were less than 45 years. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: In the total sample and after adjustment for age, gender, Acute Physiology and Chronic Health Evaluation IV, mechanical ventilation, and use of vasoactive drugs, we found that a BMI greater than or equal to 30 kg/m 2 does not affect hospital mortality (adjusted odds ratio [OR adj ] = 0.98; 95% CI, 0.90-1.06; p = 0.62). For patients less than 45 years old, but not for those greater than or equal to 45 years old, a BMI greater than or equal to 30 kg/m 2 was associated with a lower hospital mortality (OR adj = 0.59; 95% CI, 0.36-0.96; p = 0.03). CONCLUSIONS: A higher BMI may be favorably associated with a lower mortality among those less than 45 years old. This is in line with the so-called "obesity paradox" that was established for other groups of critically ill patients in broad age ranges. Further research is needed to understand this favorable association in young critically ill patients with COVID-19.

Adverse drug event detection using natural language processing: A scoping review of supervised learning methods.

To reduce adverse drug events (ADEs), hospitals need a system to support them in monitoring ADE occurrence routinely, rapidly, and at scale. Natural language processing (NLP), a computerized approach to analyze text data, has shown promising results for the purpose of ADE detection in the context of pharmacovigilance. However, a detailed qualitative assessment and critical appraisal of NLP methods for ADE detection in the context of ADE monitoring in hospitals is lacking. Therefore, we have conducted a scoping review to close this knowledge gap, and to provide directions for future research and practice. We included articles where NLP was applied to detect ADEs in clinical narratives within electronic health records of inpatients. Quantitative and qualitative data items relating to NLP methods were extracted and critically appraised. Out of 1,065 articles screened for eligibility, 29 articles met the inclusion criteria. Most frequent tasks included named entity recognition (n = 17; 58.6%) and relation extraction/classification (n = 15; 51.7%). Clinical involvement was reported in nine studies (31%). Multiple NLP modelling approaches seem suitable, with Long Short Term Memory and Conditional Random Field methods most commonly used. Although reported overall performance of the systems was high, it provides an inflated impression given a steep drop in performance when predicting the ADE entity or ADE relation class. When annotating corpora, treating an ADE as a relation between a drug and non-drug entity seems the best practice. Future research should focus on semi-automated methods to reduce the manual annotation effort, and examine implementation of the NLP methods in practice.

Real-world Evidence of the Effects of Novel Treatments for COVID-19 on Mortality: A Nationwide Comparative Cohort Study of Hospitalized Patients in the First, Second, Third, and Fourth Waves in the Netherlands.

Background: Large clinical trials on drugs for hospitalized coronavirus disease 2019 (COVID-19) patients have shown significant effects on mortality. There may be a discrepancy with the observed real-world effect. We describe the clinical characteristics and outcomes of hospitalized COVID-19 patients in the Netherlands during 4 pandemic waves and analyze the association of the newly introduced treatments with mortality, intensive care unit (ICU) admission, and discharge alive. Methods: We conducted a nationwide retrospective analysis of hospitalized COVID-19 patients between February 27, 2020, and December 31, 2021. Patients were categorized into waves and into treatment groups (hydroxychloroquine, remdesivir, neutralizing severe acute respiratory syndrome coronavirus 2 monoclonal antibodies, corticosteroids, and interleukin [IL]-6 antagonists). Four types of Cox regression analyses were used: unadjusted, adjusted, propensity matched, and propensity weighted. Results: Among 5643 patients from 11 hospitals, we observed a changing epidemiology during 4 pandemic waves, with a decrease in median age (67-64 years; P < .001), in in-hospital mortality on the ward (21%-15%; P < .001), and a trend in the ICU (24%-16%; P = .148). In ward patients, hydroxychloroquine was associated with increased mortality (1.54; 95% CI, 1.22-1.96), and remdesivir was associated with a higher rate of discharge alive within 29 days (1.16; 95% CI, 1.03-1.31). Corticosteroids were associated with a decrease in mortality (0.82; 95% CI, 0.69-0.96); the results of IL-6 antagonists were inconclusive. In patients directly admitted to the ICU, hydroxychloroquine, corticosteroids, and IL-6 antagonists were not associated with decreased mortality. Conclusions: Both remdesivir and corticosteroids were associated with better outcomes in ward patients with COVID-19. Continuous evaluation of real-world treatment effects is needed.

Assess and validate predictive performance of models for in-hospital mortality in COVID-19 patients: A retrospective cohort study in the Netherlands comparing the value of registry data with high-granular electronic health records.

PURPOSE: To assess, validate and compare the predictive performance of models for in-hospital mortality of COVID-19 patients admitted to the intensive care unit (ICU) over two different waves of infections. Our models were built with high-granular Electronic Health Records (EHR) data versus less-granular registry data. METHODS: Observational study of all COVID-19 patients admitted to 19 Dutch ICUs participating in both the national quality registry National Intensive Care Evaluation (NICE) and the EHR-based Dutch Data Warehouse (hereafter EHR). Multiple models were developed on data from the first 24 h of ICU admissions from February to June 2020 (first COVID-19 wave) and validated on prospective patients admitted to the same ICUs between July and December 2020 (second COVID-19 wave). We assessed model discrimination, calibration, and the degree of relatedness between development and validation population. Coefficients were used to identify relevant risk factors. RESULTS: A total of 1533 patients from the EHR and 1563 from the registry were included. With high granular EHR data, the average AUROC was 0.69 (standard deviation of 0.05) for the internal validation, and the AUROC was 0.75 for the temporal validation. The registry model achieved an average AUROC of 0.76 (standard deviation of 0.05) in the internal validation and 0.77 in the temporal validation. In the EHR data, age, and respiratory-system related variables were the most important risk factors identified. In the NICE registry data, age and chronic respiratory insufficiency were the most important risk factors. CONCLUSION: In our study, prognostic models built on less-granular but readily-available registry data had similar performance to models built on high-granular EHR data and showed similar transportability to a prospective COVID-19 population. Future research is needed to verify whether this finding can be confirmed for upcoming waves.