Development and validation of prognostic machine learning models for short- and long-term mortality among acutely admitted patients based on blood tests.

Several scores predicting mortality at the emergency department have been developed. However, all with shortcomings either simple and applicable in a clinical setting, with poor performance, or advanced, with high performance, but clinically difficult to implement. This study aimed to explore if machine learning algorithms could predict all-cause short- and long-term mortality based on the routine blood test collected at admission. METHODS: We analyzed data from a retrospective cohort study, including patients > 18 years admitted to the Emergency Department (ED) of Copenhagen University Hospital Hvidovre, Denmark between November 2013 and March 2017. The primary outcomes were 3-, 10-, 30-, and 365-day mortality after admission. PyCaret, an automated machine learning library, was used to evaluate the predictive performance of fifteen machine learning algorithms using the area under the receiver operating characteristic curve (AUC). RESULTS: Data from 48,841 admissions were analyzed, of these 34,190 (70%) were randomly divided into training data, and 14,651 (30%) were in test data. Eight machine learning algorithms achieved very good to excellent results of AUC on test data in a of range 0.85-0.93. In prediction of short-term mortality, lactate dehydrogenase (LDH), leukocyte counts and differentials, Blood urea nitrogen (BUN) and mean corpuscular hemoglobin concentration (MCHC) were the best predictors, whereas prediction of long-term mortality was favored by age, LDH, soluble urokinase plasminogen activator receptor (suPAR), albumin, and blood urea nitrogen (BUN). CONCLUSION: The findings suggest that measures of biomarkers taken from one blood sample during admission to the ED can identify patients at high risk of short-and long-term mortality following emergency admissions.

Performance of Panel-Estimated GFR Among Hospitalized Older Adults.

RATIONALE & OBJECTIVE: Older adults represent nearly half of all hospitalized patients and are vulnerable to inappropriate dosing of medications eliminated through the kidneys. However, few studies in this population have evaluated the performance of equations for estimating the glomerular filtration rate (GFR)-particularly those that incorporate multiple filtration markers. STUDY DESIGN: Cross-sectional diagnostic test substudy of a randomized clinical trial. SETTING & PARTICIPANTS: Adults≥65 years of age presenting to the emergency department of Copenhagen University Hospital Amager and Hvidovre in Hvidovre, Denmark, between October 2018 and April 2021. TESTS COMPARED: Measured GFR (mGFR) determined using 99mTc-DTPA plasma clearance compared with estimated GFR (eGFR) calculated using 6 different equations based on creatinine; 3 based on creatinine and cystatin C combined; and 2 based on panels of markers including creatinine, cystatin C, ß-trace protein (BTP) and/or ß2-microglobulin (B2M). OUTCOME: The performance of each eGFR equation compared with mGFR with respect to bias, relative bias, inaccuracy (1-P30), and root mean squared error (RMSE). RESULTS: We assessed eGFR performance for 106 patients (58% female, median age 78.3 years, median mGFR 62.9mL/min/1.73m2). Among the creatinine-based equations, the 2009 CKD-EPIcr equation yielded the smallest relative bias (+4.2%). Among the creatinine-cystatin C combination equations, the 2021 CKD-EPIcomb equation yielded the smallest relative bias (-3.4%), inaccuracy (3.8%), and RMSE (0.139). Compared with the 2021 CKD-EPIcomb, the CKD-EPIpanel equation yielded a smaller RMSE (0.136) but larger relative bias (-4.0%) and inaccuracy (5.7%). LIMITATIONS: Only White patients were included; only a subset of patients from the original clinical trial underwent GFR measurement; and filtration marker concentration can be affected by subclinical changes in volume status. CONCLUSIONS: The 2009 CKD-EPIcr, 2021 CKD-EPIcomb, and CKD-EPIpanel equations performed best and notably outperformed their respective full-age spectrum equations. The addition of cystatin C to creatinine-based equations improved performance, while the addition of BTP and/or B2M yielded minimal improvement. FUNDING: Grants from public sector industry (Amgros I/S) and government (Capital Region of Denmark). TRIAL REGISTRATION: Registered at ClinicalTrials.gov with registration number NCT03741283. PLAIN-LANGUAGE SUMMARY: Inaccurate kidney function assessment can lead to medication errors, a common cause of hospitalization and early readmission among older adults. Several novel methods have been developed to estimate kidney function based on a panel of kidney function markers that can be measured from a single blood sample. We evaluated the accuracy of these new methods (relative to a gold standard method) among 106 hospitalized older adults. We found that kidney function estimates combining 2 markers (creatinine and cystatin C) were highly accurate and noticeably more accurate than estimates based on creatinine alone. Estimates incorporating additional markers such as ß-trace protein and ß2-microglobulin did not further improve accuracy.

Evaluation of Multi-Frequency Bioelectrical Impedance Analysis against Dual-Energy X-ray Absorptiometry for Estimation of Low Muscle Mass in Older Hospitalized Patients.

The accuracy of multi-frequency (MF) bioelectrical impedance analysis (BIA) to estimate low muscle mass in older hospitalized patients remains unclear. This study aimed to describe the ability of MF-BIA to identify low muscle mass as proposed by The Global Leadership Initiative on Malnutrition (GLIM) and The European Working Group on Sarcopenia in Older People (EWGSOP-2) and examine the association between muscle mass, dehydration, malnutrition, and poor appetite in older hospitalized patients. In this prospective exploratory cohort study, low muscle mass was estimated with MF-BIA against dual-energy X-ray absorptiometry (DXA) in 42 older hospitalized adults (≥65 years). The primary variable for muscle mass was appendicular skeletal muscle mass (ASM), and secondary variables were appendicular skeletal muscle mass index (ASMI) and fat-free mass index (FFMI). Cut-off values for low muscle mass were based on recommendations by GLIM and EWGSOP-2. MF-BIA was evaluated against DXA on the ability to estimate absolute values of muscle mass by mean bias, limits of agreement (LOA), and accuracy (5% and 10% levels). Agreement between MF-BIA and DXA to identify low muscle mass was evaluated with sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV). The association between muscle mass, dehydration, malnutrition, and poor appetite was visually examined with boxplots. MF-BIA overestimated absolute values of ASM with a mean bias of 0.63 kg (CI: -0.20:1.46, LOA: -4.61:5.87). Agreement between MF-BIA and DXA measures of ASM showed a sensitivity of 86%, specificity of 94%, PPV of 75% and NPV of 97%. Boxplots indicate that ASM is lower in patients with malnutrition. This was not observed in patients with poor appetite. We observed a tendency toward higher ASM in patients with dehydration. Estimation of absolute ASM values with MF-BIA should be interpreted with caution, but MF-BIA might identify low muscle mass in older hospitalized patients.

Variations in physical activity and sedentary behavior during and after hospitalization in acutely admitted older medical patients: a longitudinal study.

BACKGROUND: Inactivity is frequent among older patients during hospitalization. It is unknown how patients' daily activity pattern (diurnal profile) vary between hospitalization and after discharge. This study aims to describe and compare the distribution of physical activity and sedentary behavior in acutely hospitalized older patients during hospitalization and after discharge. METHODS: We included data on 80 patients (+65 years) admitted with acute medical illness from the STAND-Cph trial. Physical activity and sedentary behavior were measured as daily number of steps, uptime (walking/standing) and sedentary behavior (lying/sitting) with an activity monitor (activPAL3, PAL Technologies Ltd). The patients wore the monitor for three periods of one week: during hospitalization, after discharge, and four weeks after discharge. RESULTS: The patients' median age was 80 years [IQR: 75;88], 68% were female and the median De Morton Mobility Index (DEMMI) was 57 [IQR: 48;67]. The daily median uptime was 1.7 h [IQR: 1;2.8] during hospitalization, 4.0 h [IQR: 2.7;5.4] after discharge and 4.0 h [IQR: 2.8;5.8] four weeks after discharge. The daily median number of steps was 728 [IQR: 176;2089], 2207 [IQR: 1433;3148], and 2622 [IQR: 1714;3865], respectively, and median daily sedentary behavior was 21.4 h (IQR: 20.7;22.4), 19.5 h (IQR: 18.1;21.0) and 19.6 h (IQR: 18.0;20.8), respectively. During hospitalization, a small activity peak was observed between 9-11 AM without any notable variation after. At discharge and four weeks after discharge, a peak in physical activity was seen between 9-12 AM and at 5 PM. CONCLUSION: Older hospitalized patients spend most of their time being sedentary with their highest activity between 9-11 AM. Daily activity doubles after discharge with one extra peak in the afternoon. Daily routines might be disrupted, and older patients have the potential to be more physically active during hospitalization. Interventions that encourage physical activity during hospitalization are warranted.

Utility of suPAR and NGAL for AKI Risk Stratification and Early Optimization of Renal Risk Medications among Older Patients in the Emergency Department.

Diagnosis of acute kidney injury (AKI) based on plasma creatinine often lags behind actual changes in renal function. Here, we investigated early detection of AKI using the plasma soluble urokinase plasminogen activator receptor (suPAR) and neutrophil gelatinase-sssociated lipocalin (NGAL) and observed the impact of early detection on prescribing recommendations for renally-eliminated medications. This study is a secondary analysis of data from the DISABLMENT cohort on acutely admitted older (≥65 years) medical patients (n = 339). Presence of AKI according to kidney disease: improving global outcomes (KDIGO) criteria was identified from inclusion to 48 h after inclusion. Discriminatory power of suPAR and NGAL was determined by receiver-operating characteristic (ROC). Selected medications that are contraindicated in AKI were identified in Renbase®. A total of 33 (9.7%) patients developed AKI. Discriminatory power for suPAR and NGAL was 0.69 and 0.78, respectively, at a cutoff of 4.26 ng/mL and 139.5 ng/mL, respectively. The interaction of suPAR and NGAL yielded a discriminatory power of 0.80, which was significantly higher than for suPAR alone (p = 0.0059). Among patients with AKI, 22 (60.6%) used at least one medication that should be avoided in AKI. Overall, suPAR and NGAL levels were independently associated with incident AKI and their combination yielded excellent discriminatory power for risk determination of AKI.

The effect of circadian-adjusted LED-based lighting on sleep, daytime sleepiness and biomarkers of inflammation in a randomized controlled cross-over trial by pragmatic design in elderly care home dwellers.

AIM: Elderly multimorbid care home dwellers are a heterogenic group of frail individuals that exhibit sleep disturbances and a range of co-morbidities. The project aimed to study the possible effect of indoor circadian-adjusted LED-lighting (CaLED) in the elderly residents' care home on their sleeping patterns and systemic biomarkers associated with inflammation. METHODS: A 16-week trial study was performed to follow the intervention and control groups using the Pittsburgh Sleep Quality Index (PSQI) and Epworth Sleepiness Scale (ESS) to monitor sleep and daytime sleepiness, and biomarkers IL-6, TNF-α and suPAR, to estimate the levels of inflammation. RESULTS: There was no significant impact on sleep improvement after the short intervention time when analyzing the PSQI and ESS results. However, we found several challenges using these tools for this specific group of individuals. Thus, important knowledge was gained for future studies in elderly care home dwellers. The inflammation state throughout the entire study period was stable for most of the elderly and no significant change was detected from before to after the intervention. This study represents a first-to-date attempt to ameliorate the adverse effects of sleep disturbances that characterize a randomly chosen group of elderly multimorbid subjects, by using circadian-adjusted LED-lighting in a natural care home environment. CONCLUSION: In this pragmatic randomized study of home dwelling individuals we were not able to demonstrate an improved sleep pattern as judged by PSQI, ESS or a change in inflammatory state.