Models to predict length of stay in the emergency department: a systematic literature review and appraisal.

INTRODUCTION: Prolonged Length of Stay (LOS) in ED (Emergency Department) has been associated with poor clinical outcomes. Prediction of ED LOS may help optimize resource utilization, clinical management, and benchmarking. This study aims to systematically review models for predicting ED LOS and to assess the reporting and methodological quality about these models. METHODS: The online database PubMed, Scopus, and Web of Science (10 Sep 2023) was searched for English language articles that reported prediction models of LOS in ED. Identified titles and abstracts were independently screened by two reviewers. All original papers describing either development (with or without internal validation) or external validation of a prediction model for LOS in ED were included. RESULTS: Of 12,193 uniquely identified articles, 34 studies were included (29 describe the development of new models and five describe the validation of existing models). Different statistical and machine learning methods were applied to the papers. On the 39-point reporting score and 11-point methodological quality score, the highest reporting scores for development and validation studies were 39 and 8, respectively. CONCLUSION: Various studies on prediction models for ED LOS were published but they are fairly heterogeneous and suffer from methodological and reporting issues. Model development studies were associated with a poor to a fair level of methodological quality in terms of the predictor selection approach, the sample size, reproducibility of the results, missing imputation technique, and avoiding dichotomizing continuous variables. Moreover, it is recommended that future investigators use the confirmed checklist to improve the quality of reporting.

Topic evolution before fall incidents in new fallers through natural language processing of general practitioners' clinical notes.

BACKGROUND: Falls involve dynamic risk factors that change over time, but most studies on fall-risk factors are cross-sectional and do not capture this temporal aspect. The longitudinal clinical notes within electronic health records (EHR) provide an opportunity to analyse fall risk factor trajectories through Natural Language Processing techniques, specifically dynamic topic modelling (DTM). This study aims to uncover fall-related topics for new fallers and track their evolving trends leading up to falls. METHODS: This case-cohort study utilised primary care EHR data covering information on older adults between 2016 and 2019. Cases were individuals who fell in 2019 but had no falls in the preceding three years (2016-18). The control group was randomly sampled individuals, with similar size to the cases group, who did not endure falls during the whole study follow-up period. We applied DTM on the clinical notes collected between 2016 and 2018. We compared the trend lines of the case and control groups using the slopes, which indicate direction and steepness of the change over time. RESULTS: A total of 2,384 fallers (cases) and an equal number of controls were included. We identified 25 topics that showed significant differences in trends between the case and control groups. Topics such as medications, renal care, family caregivers, hospital admission/discharge and referral/streamlining diagnostic pathways exhibited a consistent increase in steepness over time within the cases group before the occurrence of falls. CONCLUSIONS: Early recognition of health conditions demanding care is crucial for applying proactive and comprehensive multifactorial assessments that address underlying causes, ultimately reducing falls and fall-related injuries.

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.

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.

Trends in preterm birth in the Netherlands in 2011-2019: A population-based study among singletons and multiples.

INTRODUCTION: Preterm birth (PTB) is the leading cause of infant mortality and morbidity worldwide. Rates of PTB in the Netherlands are declining, possibly due to the implementation of preventive strategies. In this study we assessed the overall trend in PTB rates in the Netherlands in recent years, and in more detail in specific subgroups to investigate potential groups that require scrutiny in the near future. MATERIAL AND METHODS: Based on the national perinatal registry, we included all pregnancies without severe congenital abnormalities resulting in a birth from 24 to 42 completed weeks of gestation between 2011 and 2019 in the Netherlands. We assessed PTB rates in two different clinical subtypes (spontaneous vs. iatrogenic) and in five gestational age subgroups: 24-27+6 weeks (extreme), 28-31+6 weeks (very), 32-33+6 weeks (moderate, 34-36+6 weeks [late] and, in general, 24-36+6 weeks [overall PTB]). Trend analysis was performed using the Cochran Armitage test. We also compared PTB rates in different subgroups in the first 2 years compared to the last 2 years. Singleton and multiple gestations were analyzed separately. RESULTS: We included 1 447 689 singleton and 23 250 multiple pregnancies in our study. In singletons, we observed a significant decline in PTB from 5.5% to 5.0% (p < 0.0001), mainly due to a decrease in iatrogenic PTBs. When focusing on different gestational age subgroups, there was a decrease in all iatrogenic PTB and in moderate to late spontaneous PTB. However, in spontaneous extreme and very PTB there was an significant increase. When assessing overall PTB risk in different subgroups, the decline was only visible in women with age ≥25 years, nulliparous and primiparous women, women with a medium or high socioeconomic status and hypertensive women. In multiples, the rate of PTB remained fairly stable, from 52.3% in 2011 to 54.1% in 2019 (p = 0.57). CONCLUSIONS: In the Netherlands, between 2011 and 2019, PTB decreased, mainly due to a reduction in late PTB, and more in iatrogenic than in spontaneous PTB. Focus for the near future should be on specific subgroups in which the decline was not visible, such as women with a low socioeconomic status or a young age.

Prolonged use of closed-loop inspired oxygen support in preterm infants: a randomised controlled trial.

OBJECTIVE: This randomised study in preterm infants on non-invasive respiratory support investigated the effectiveness of automated oxygen control (A-FiO2) in keeping the oxygen saturation (SpO2) within a target range (TR) during a 28-day period compared with manual titration (M-FiO2). DESIGN: A single-centre randomised control trial. SETTING: A level III neonatal intensive care unit. PATIENTS: Preterm infants (<28 weeks' gestation) on non-invasive respiratory support. INTERVENTIONS: A-FiO2 versus M-FiO2 control. METHODS: Main outcomes were the proportion of time spent and median area of episodes in the TR, hyperoxaemia, hypoxaemia and the trend over 28 days using a linear random intercept model. RESULTS: 23 preterm infants (median gestation 25.7 weeks; birth weight 820 g) were randomised. Compared with M-FiO2, the time spent within TR was higher in the A-FiO2 group (68.7% vs 48.0%, p<0.001). Infants in the A-FiO2 group spent less time in hyperoxaemia (13.8% vs 37.7%, p<0.001), but no difference was found in hypoxaemia. The time-based analyses showed that the A-FiO2 efficacy may differ over time, especially for hypoxaemia. Compared with the M-FiO2 group, the A-FiO2 group had a larger intercept but with an inversed slope for the daily median area below the TR (intercept 70.1 vs 36.3; estimate/day -0.70 vs 0.69, p<0.001). CONCLUSION: A-FiO2 control was superior to manual control in keeping preterm infants on non-invasive respiratory support in a prespecified TR over a period of 28 days. This improvement may come at the expense of increased time below the TR in the first days after initiating A-FiO2 control. TRIAL REGISTRATION NUMBER: NTR6731.

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.

Time-dependent association between STOPP and START criteria and gastrointestinal bleeding in older patients using routinely collected primary care data.

PURPOSE: Only few studies have assessed the preventive effect of the STOPP/START criteria on adverse events. We aim to quantify 1) the association between nonadherence to STOPP/START criteria and gastrointestinal bleedings, and 2) the association between exposure to the potentially harmful START-medications and gastrointestinal bleedings. DESIGN: A retrospective cohort study using routinely collected data of patients aged ≥ 65 years from the electronic health records (EHR) of 49 general practitioners (GPs) in 6 GP practices, from 2007 to 2014. The database is maintained in the academic research network database (AHA) of Amsterdam UMC, the Netherlands. METHODS: Gastrointestinal bleedings were identified using ICPC codes and free text inspections. Three STOPP and six START criteria pertaining to gastrointestinal bleedings were selected. Cox proportional hazards regression with time-dependent covariate analysis was performed to assess the independent association between nonadherence to the STOPP/START criteria and gastrointestinal bleedings. The analysis was performed with all criteria as a composite outcome, as well as separately for the individual criteria. RESULTS: Out of 26,576 participants, we identified 19,070 Potential Inappropriate Medications (PIM)/Potential Prescribing Omission (PPO) instances for 3,193 participants and 146 gastrointestinal bleedings in 143 participants. The hazard ratio for gastrointestinal bleedings of STOPP/STARTs, taken as composite outcome, was 5.45 (95% CI 3.62-8.21). When analysed separately, two out of nine STOPP/STARTs showed significant associations. CONCLUSION: This study demonstrates a significant positive association between nonadherence to the STOPP/START criteria and gastrointestinal bleeding. We emphasize the importance of adherence to the relevant criteria for gastrointestinal bleeding, which may be endorsed by decision support systems.

Effects of a clinical decision support system and patient portal for preventing medication-related falls in older fallers: Protocol of a cluster randomized controlled trial with embedded process and economic evaluations (ADFICE_IT).

BACKGROUND: Falls are the leading cause of injury-related mortality and hospitalization among adults aged ≥ 65 years. An important modifiable fall-risk factor is use of fall-risk increasing drugs (FRIDs). However, deprescribing is not always attempted or performed successfully. The ADFICE_IT trial evaluates the combined use of a clinical decision support system (CDSS) and a patient portal for optimizing the deprescribing of FRIDs in older fallers. The intervention aims to optimize and enhance shared decision making (SDM) and consequently prevent injurious falls and reduce healthcare-related costs. METHODS: A multicenter, cluster-randomized controlled trial with process evaluation will be conducted among hospitals in the Netherlands. We aim to include 856 individuals aged ≥ 65 years that visit the falls clinic due to a fall. The intervention comprises the combined use of a CDSS and a patient portal. The CDSS provides guideline-based advice with regard to deprescribing and an individual fall-risk estimation, as calculated by an embedded prediction model. The patient portal provides educational information and a summary of the patient's consultation. Hospitals in the control arm will provide care-as-usual. Fall-calendars will be used for measuring the time to first injurious fall (primary outcome) and secondary fall outcomes during one year. Other measurements will be conducted at baseline, 3, 6, and 12 months and include quality of life, cost-effectiveness, feasibility, and shared decision-making measures. Data will be analyzed according to the intention-to-treat principle. Difference in time to injurious fall between the intervention and control group will be analyzed using multilevel Cox regression. DISCUSSION: The findings of this study will add valuable insights about how digital health informatics tools that target physicians and older adults can optimize deprescribing and support SDM. We expect the CDSS and patient portal to aid in deprescribing of FRIDs, resulting in a reduction in falls and related injuries. TRIAL REGISTRATION: ClinicalTrials.gov NCT05449470 (7-7-2022).

Early detection of colorectal cancer by leveraging Dutch primary care consultation notes with free text embeddings.

We aimed to assess the added predictive performance that free-text Dutch consultation notes provide in detecting colorectal cancer in primary care, in comparison to currently used models. We developed, evaluated and compared three prediction models for colorectal cancer (CRC) in a large primary care database with 60,641 patients. The prediction model with both known predictive features and free-text data (with TabTxt AUROC: 0.823) performs statistically significantly better (p < 0.05) than the other two models with only tabular (as used nowadays) and text data, respectively (AUROC Tab: 0.767; Txt: 0.797). The specificity of the two models that use demographics and known CRC features (with specificity Tab: 0.321; TabTxt: 0.335) are higher than that of the model with only free-text (specificity Txt: 0.234). The Txt and, to a lesser degree, TabTxt model are well calibrated, while the Tab model shows slight underprediction at both tails. As expected with an outcome prevalence below 0.01, all models show much uncalibrated predictions in the extreme upper tail (top 1%). Free-text consultation notes show promising results to improve the predictive performance over established prediction models that only use structured features. Clinical future implications for our CRC use case include that such improvement may help lowering the number of referrals for suspected CRC to medical specialists.

Personal Continuity and Appropriate Prescribing in Primary Care.

PURPOSE: Personal continuity between patient and physician is a core value of primary care. Although previous studies suggest that personal continuity is associated with fewer potentially inappropriate prescriptions, evidence on continuity and prescribing in primary care is scarce. We aimed to determine the association between personal continuity and potentially inappropriate prescriptions, which encompasses potentially inappropriate medications (PIMs) and potential prescribing omissions (PPOs), by family physicians among older patients. METHODS: We conducted an observational cohort study using routine care data from patients enlisted in 48 Dutch family practices from 2013 to 2018. All 25,854 patients aged 65 years and older having at least 5 contacts with their practice in 6 years were included. We calculated personal continuity using 3 established measures: the usual provider of care measure, the Bice-Boxerman Index, and the Herfindahl Index. We used the Screening Tool of Older Person's Prescriptions (STOPP) and the Screening Tool to Alert doctors to Right Treatment (START) specific to the Netherlands version 2 criteria to calculate the prevalence of potentially inappropriate prescriptions. To assess associations, we conducted multilevel negative binomial regression analyses, with and without adjustment for number of chronic conditions, age, and sex. RESULTS: The patients' mean (SD) values for the usual provider of care measure, the Bice-Boxerman Continuity of Care Index, and the Herfindahl Index were 0.70 (0.19), 0.55 (0.24), and 0.59 (0.22), respectively. In our population, 72.2% and 74.3% of patients had at least 1 PIM and PPO, respectively; 30.9% and 34.2% had at least 3 PIMs and PPOs, respectively. All 3 measures of personal continuity were positively and significantly associated with fewer potentially inappropriate prescriptions. CONCLUSIONS: A higher level of personal continuity is associated with more appropriate prescribing. Increasing personal continuity may improve the quality of prescriptions and reduce harmful consequences.

Temporal validation of 30-day mortality prediction models for transcatheter aortic valve implantation using statistical process control - An observational study in a national population.

Background: Various mortality prediction models for Transcatheter Aortic Valve Implantation (TAVI) have been developed in the past years. The effect of time on the performance of such models, however, is unclear given the improvements in the procedure and changes in patient selection, potentially jeopardizing the usefulness of the prediction models in clinical practice. We aim to explore how time affects the performance and stability of different types of prediction models of 30-day mortality after TAVI. Methods: We developed both parametric (Logistic Regression) and non-parametric (XGBoost) models to predict 30-day mortality after TAVI using data from the Netherlands Heart Registration. The models were trained with data from 2013 to the beginning of 2016 and pre-control charts from Statistical Process Control were used to analyse how time affects the models' performance on independent data from the mid of 2016 to the end of 2019. The area under the Receiver Operating Characteristics curve (AUC) was used to evaluate the models in terms of discrimination and the Brier Score (BS), which is related to calibration, in terms of accuracy of the predicted probabilities. To understand the extent to which refitting the models contribute to the models' stability, we also allowed the models to be updated over time. Results: We included data from 11,291 consecutive TAVI patients from hospitals in the Netherlands. The parametric model without re-training had a median AUC of 0.64 (IQR 0.54-0.73) and BS of 0.028 (IQR 0.021-0.035). For the non-parametric model, the median AUC was 0.63 (IQR 0.48-0.68) and BS was 0.027 (IQR 0.021-0.036). Over time, the developed parametric model was stable in terms of AUC and unstable in terms of BS. The non-parametric model was considered unstable in both AUC and BS. Repeated model refitting resulted in stable models in terms of AUC and decreased the variability of BS, although BS was still unstable. The refitted parametric model had a median AUC of 0.66 (IQR 0.57-0.73) and BS of 0.027 (IQR 0.020-0.035) while the non-parametric model had a median AUC of 0.66 (IQR 0.57-0.74) and BS of 0.027 (IQR 0.023-0.035). Conclusions: The temporal validation of the TAVI 30-day mortality prediction models showed that the models refitted over time are more stable and accurate when compared to the frozen models. This highlights the importance of repeatedly refitted models over time to improve or at least maintain their performance stability. The non-parametric approach did not show improvement over the parametric approach.

Gestational age and socio-demographic factors associated with school performance at the age of 12 years, a population-based study.

BACKGROUND: Gestational age is positively associated with cognitive development, but socio-demographic factors also influence school performance. Previous studies suggested possible interaction, putting children with low socio-economic status (SES) at increased risk of the negative effects of prematurity. OBJECTIVES: To investigate the association between gestational age in weeks, socio-demographic characteristics, and school performance at the age of 12 years among children in regular primary education. METHODS: Population-based cohort study among liveborn singletons (N = 860,332) born in the Netherlands in 1999-2006 at 25-42 weeks' gestation, with school performance from 2011 to 2019. Regression analyses were conducted investigating the association of gestational age and sociodemographic factors with school performance and possible interaction. RESULTS: School performance increased with gestational age up to 40 weeks. This pattern was evident across socio-demographic strata. Children born at 25 weeks had -0.57 SD (95% confidence interval -0.79, -0.35) lower school performance z-scores and lower secondary school level compared to 40 weeks. Low maternal education, low maternal age, and non-European origin were strongly associated with lower school performance. Being born third or later and low socioeconomic status (SES) were also associated with lower school performance, but differences were smaller than among other factors. When born preterm, children from mothers with low education level, low or high age, low SES or children born third or later were at higher risk for lower school performance compared to children of mothers with intermediate education level, aged 25-29 years, with intermediate SES or first borns (evidence of interaction). CONCLUSIONS: Higher gestational age is associated with better school performance at the age of 12 years along the entire spectrum of gestational age, beyond the cut-off of preterm birth and across socio-demographic differences. Children in socially or economically disadvantaged situations might be more vulnerable to the negative impact of preterm birth. Other important factors in school performance are maternal education, maternal age, ethnicity, birth order and SES. Results should be interpreted with caution due to differential loss to follow-up.

Validation of the ADFICE_IT Models for Predicting Falls and Recurrent Falls in Geriatric Outpatients.

OBJECTIVES: Before being used in clinical practice, a prediction model should be tested in patients whose data were not used in model development. Previously, we developed the ADFICE_IT models for predicting any fall and recurrent falls, referred as Any_fall and Recur_fall. In this study, we externally validated the models and compared their clinical value to a practical screening strategy where patients are screened for falls history alone. DESIGN: Retrospective, combined analysis of 2 prospective cohorts. SETTING AND PARTICIPANTS: Data were included of 1125 patients (aged ≥65 years) who visited the geriatrics department or the emergency department. METHODS: We evaluated the models' discrimination using the C-statistic. Models were updated using logistic regression if calibration intercept or slope values deviated significantly from their ideal values. Decision curve analysis was applied to compare the models' clinical value (ie, net benefit) against that of falls history for different decision thresholds. RESULTS: During the 1-year follow-up, 428 participants (42.7%) endured 1 or more falls, and 224 participants (23.1%) endured a recurrent fall (≥2 falls). C-statistic values were 0.66 (95% CI 0.63-0.69) and 0.69 (95% CI 0.65-0.72) for the Any_fall and Recur_fall models, respectively. Any_fall overestimated the fall risk and we therefore updated only its intercept whereas Recur_fall showed good calibration and required no update. Compared with falls history, Any_fall and Recur_fall showed greater net benefit for decision thresholds of 35% to 60% and 15% to 45%, respectively. CONCLUSIONS AND IMPLICATIONS: The models performed similarly in this data set of geriatric outpatients as in the development sample. This suggests that fall-risk assessment tools that were developed in community-dwelling older adults may perform well in geriatric outpatients. We found that in geriatric outpatients the models have greater clinical value across a wide range of decision thresholds compared with screening for falls history alone.

Leveraging Multi-Word Concepts to Predict Acute Kidney Injury in Intensive Care.

Acute kidney injury (AKI) is an abrupt decrease in kidney function widespread in intensive care. Many AKI prediction models have been proposed, but only few exploit clinical notes and medical terminologies. Previously, we developed and internally validated a model to predict AKI using clinical notes enriched with single-word concepts from medical knowledge graphs. However, an analysis of the impact of using multi-word concepts is lacking. In this study, we compare the use of only the clinical notes as input to prediction to the use of clinical notes retrofitted with both single-word and multi-word concepts. Our results show that 1) retrofitting single-word concepts improved word representations and improved the performance of the prediction model; 2) retrofitting multi-word concepts further improves both results, albeit slightly. Although the improvement with multi-word concepts was small, due to the small number of multi-word concepts that could be annotated, multi-word concepts have proven to be beneficial.

Comparing In-Hospital Mortality Prediction by Senior Emergency Resident's Judgment and Prognostic Models in the Emergency Department.

Background: A comparison of emergency residents' judgments and two derivatives of the Sequential Organ Failure Assessment (SOFA), namely, the mSOFA and the qSOFA, was conducted to determine the accuracy of predicting in-hospital mortality among critically ill patients in the emergency department (ED). Methods: A prospective cohort research was performed on patients over 18 years of age presented to the ED. We used logistic regression to develop a model for predicting in-hospital mortality by using qSOFA, mSOFA, and residents' judgment scores. We compared the accuracy of prognostic models and residents' judgment in terms of the overall accuracy of the predicted probabilities (Brier score), discrimination (area under the ROC curve), and calibration (calibration graph). Analyses were carried out using R software version R-4.2.0. Results: In the study, 2,205 patients with median age of 64 (IQR: 50-77) years were included. There were no significant differences between the qSOFA (AUC 0.70; 95% CI: 0.67-0.73) and physician's judgment (AUC 0.68; 0.65-0.71). Despite this, the discrimination of mSOFA (AUC 0.74; 0.71-0.77) was significantly higher than that of the qSOFA and residents' judgments. Additionally, the AUC-PR of mSOFA, qSOFA, and emergency resident's judgments was 0.45 (0.43-0.47), 0.38 (0.36-0.40), and 0.35 (0.33-0.37), respectively. The mSOFA appears stronger in terms of overall performance: 0.13 vs. 0.14 and 0.15. All three models showed good calibration. Conclusion: The performance of emergency residents' judgment and the qSOFA was the same in predicting in-hospital mortality. However, the mSOFA predicted better-calibrated mortality risk. Large-scale studies should be conducted to determine the utility of these models.

Development and Internal Validation of a Prediction Model for Falls Using Electronic Health Records in a Hospital Setting.

OBJECTIVE: Fall prevention is important in many hospitals. Current fall-risk-screening tools have limited predictive accuracy specifically for older inpatients. Their administration can be time-consuming. A reliable and easy-to-administer tool is desirable to identify older inpatients at higher fall risk. We aimed to develop and internally validate a prognostic prediction model for inpatient falls for older patients. DESIGN: Retrospective analysis of a large cohort drawn from hospital electronic health record data. SETTING AND PARTICIPANTS: Older patients (≥70 years) admitted to a university medical center (2016 until 2021). METHODS: The outcome was an inpatient fall (≥24 hours of admission). Two prediction models were developed using regularized logistic regression in 5 imputed data sets: one model without predictors indicating missing values (Model-without) and one model with these additional predictors indicating missing values (Model-with). We internally validated our whole model development strategy using 10-fold stratified cross-validation. The models were evaluated using discrimination (area under the receiver operating characteristic curve) and calibration (plot assessment). We determined whether the areas under the receiver operating characteristic curves (AUCs) of the models were significantly different using DeLong test. RESULTS: Our data set included 21,286 admissions. In total, 470 (2.2%) had a fall after 24 hours of admission. The Model-without had 12 predictors and Model-with 13, of which 4 were indicators of missing values. The AUCs of the Model-without and Model-with were 0.676 (95% CI 0.646-0.707) and 0.695 (95% CI 0.667-0.724). The AUCs between both models were significantly different (P = .013). Calibration was good for both models. CONCLUSIONS AND IMPLICATIONS: Both the Model-with and Model-without indicators of missing values showed good calibration and fair discrimination, where the Model-with performed better. Our models showed competitive performance to well-established fall-risk-screening tools, and they have the advantage of being based on routinely collected data. This may substantially reduce the burden on nurses, compared with nonautomatic fall-risk-screening tools.

Predicting future falls in older people using natural language processing of general practitioners' clinical notes.

BACKGROUND: Falls in older people are common and morbid. Prediction models can help identifying individuals at higher fall risk. Electronic health records (EHR) offer an opportunity to develop automated prediction tools that may help to identify fall-prone individuals and lower clinical workload. However, existing models primarily utilise structured EHR data and neglect information in unstructured data. Using machine learning and natural language processing (NLP), we aimed to examine the predictive performance provided by unstructured clinical notes, and their incremental performance over structured data to predict falls. METHODS: We used primary care EHR data of people aged 65 or over. We developed three logistic regression models using the least absolute shrinkage and selection operator: one using structured clinical variables (Baseline), one with topics extracted from unstructured clinical notes (Topic-based) and one by adding clinical variables to the extracted topics (Combi). Model performance was assessed in terms of discrimination using the area under the receiver operating characteristic curve (AUC), and calibration by calibration plots. We used 10-fold cross-validation to validate the approach. RESULTS: Data of 35,357 individuals were analysed, of which 4,734 experienced falls. Our NLP topic modelling technique discovered 151 topics from the unstructured clinical notes. AUCs and 95% confidence intervals of the Baseline, Topic-based and Combi models were 0.709 (0.700-0.719), 0.685 (0.676-0.694) and 0.718 (0.708-0.727), respectively. All the models showed good calibration. CONCLUSIONS: Unstructured clinical notes are an additional viable data source to develop and improve prediction models for falls compared to traditional prediction models, but the clinical relevance remains limited.

Does induction of labor at 41 weeks (early, mid or late) improve birth outcomes in low-risk pregnancy? A nationwide propensity score-matched study.

INTRODUCTION: This study aimed to assess whether induction of labor at 41 weeks of gestation improved perinatal outcomes in a low-risk pregnancy compared with expectant management. MATERIAL AND METHODS: Registry-based national cohort study in The Netherlands. The study population comprised 239 971 low-risk singleton pregnancies from 2010 to 2019, with birth occurring from 41+0 to 42+0 weeks. We used propensity score matching to compare induction of labor in three 2-day groups to expectant management, and further conducted separate analyses by parity. The main outcome measures were stillbirth, perinatal mortality, 5-min Apgar <4 and <7, neonatal intensive care unit (NICU) admissions ≥24 h, and emergency cesarean section rate. RESULTS: Compared with expectant management, induction of labor at 41+0 to 41+1 weeks resulted in reduced stillbirths (adjusted odds ratio [aOR] 0.15, 95% confidence interval [CI] 0.05-0.51) in both nulliparous and multiparous women. Induction of labor increased 5-min Apgar score <7 (aOR 1.30, 95% CI 1.09-1.55) and NICU admissions ≥24 h (aOR 2.12, 95% CI 1.53-2.92), particularly in nulliparous women, and increased the cesarean section rate (aOR 1.42, 95% CI 1.34-1.51). At 41+2-41+3 weeks, induction of labor reduced perinatal mortality (aOR 0.13, 95% CI 0.04-0.43) in both nulliparous and multiparous women. The rate of 5-min Apgar score <7 was increased (aOR 1.26, 95% CI 1.06-1.50), reaching significance in multiparous women. The cesarean section rate increased (aOR 1.57, 95% CI 1.48-1.67) in both nulliparous and multiparous women. Induction of labor at 41+4 to 41+5 weeks reduced stillbirths (aOR 0.30, 95% CI 0.10-0.93). Induction of labor increased rates of 5-min Apgar score <4 (aOR 1.61, 95% CI 1.01-2.56) and NICU admissions ≥24 h (aOR 1.52, 95% CI 1.08-2.13) in nulliparous women. Cesarean section rate was increased (aOR 1.47, 95% CI 1.38-1.57) in nulliparous and multiparous women. CONCLUSIONS: At 41+2 to 41+3 weeks, induction of labor reduced perinatal mortality, and in all 2-day groups at 41 weeks, it reduced stillbirths, compared with expectant management. Low 5-min Apgar score (<7 and <4) and NICU admissions ≥24 h occurred more often with induction of labor, especially in nulliparous women. Induction of labor in all 2-day groups coincided with elevated cesarean section rates in nulliparous and multiparous women. These findings pertaining to the choice of induction of labor vs expectant management should be discussed when counseling women at 41 weeks of gestation.