Risk Assessment Models for Venous Thromboembolism in Medical Inpatients.

Importance: Thromboprophylaxis is recommended for medical inpatients at risk of venous thromboembolism (VTE). Risk assessment models (RAMs) have been developed to stratify VTE risk, but a prospective head-to-head comparison of validated RAMs is lacking. Objectives: To prospectively validate an easy-to-use RAM, the simplified Geneva score, and compare its prognostic performance with previously validated RAMs. Design, Setting, and Participants: This prospective cohort study was conducted from June 18, 2020, to January 4, 2022, with a 90-day follow-up. A total of 4205 consecutive adults admitted to the general internal medicine departments of 3 Swiss university hospitals for hospitalization for more than 24 hours due to acute illness were screened for eligibility; 1352 without therapeutic anticoagulation were included. Exposures: At admission, items of 4 RAMs (ie, the simplified and original Geneva score, the Padua score, and the IMPROVE [International Medical Prevention Registry on Venous Thromboembolism] score) were collected. Patients were stratified into high and low VTE risk groups according to each RAM. Main Outcomes and Measures: Symptomatic VTE within 90 days. Results: Of 1352 medical inpatients (median age, 67 years [IQR, 54-77 years]; 762 men [55.4%]), 28 (2.1%) experienced VTE. Based on the simplified Geneva score, 854 patients (63.2%) were classified as high risk, with a 90-day VTE risk of 2.6% (n = 22; 95% CI, 1.7%-3.9%), and 498 patients (36.8%) were classified as low risk, with a 90-day VTE risk of 1.2% (n = 6; 95% CI, 0.6%-2.6%). Sensitivity of the simplified Geneva score was 78.6% (95% CI, 60.5%-89.8%) and specificity was 37.2% (95% CI, 34.6%-39.8%); the positive likelihood ratio of the simplified Geneva score was 1.25 (95% CI, 1.03-1.52) and the negative likelihood ratio was 0.58 (95% CI, 0.28-1.18). In head-to-head comparisons, sensitivity was highest for the original Geneva score (82.1%; 95% CI, 64.4%-92.1%), while specificity was highest for the IMPROVE score (70.4%; 95% CI, 67.9%-72.8%). After adjusting the VTE risk for thromboprophylaxis use and site, there was no significant difference between the high-risk and low-risk groups based on the simplified Geneva score (subhazard ratio, 2.04 [95% CI, 0.83-5.05]; P = .12) and other RAMs. Discriminative performance was poor for all RAMs, with an area under the receiver operating characteristic curve ranging from 53.8% (95% CI, 51.1%-56.5%) for the original Geneva score to 58.1% (95% CI, 55.4%-60.7%) for the simplified Geneva score. Conclusions and Relevance: This head-to-head comparison of validated RAMs found suboptimal accuracy and prognostic performance of the simplified Geneva score and other RAMs to predict hospital-acquired VTE in medical inpatients. Clinical usefulness of existing RAMs is questionable, highlighting the need for more accurate VTE prediction strategies.

Development and validation of a prognostic COVID-19 severity assessment (COSA) score and machine learning models for patient triage at a tertiary hospital.

BACKGROUND: Clinical risk scores and machine learning models based on routine laboratory values could assist in automated early identification of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) patients at risk for severe clinical outcomes. They can guide patient triage, inform allocation of health care resources, and contribute to the improvement of clinical outcomes. METHODS: In- and out-patients tested positive for SARS-CoV-2 at the Insel Hospital Group Bern, Switzerland, between February 1st and August 31st ('first wave', n = 198) and September 1st through November 16th 2020 ('second wave', n = 459) were used as training and prospective validation cohort, respectively. A clinical risk stratification score and machine learning (ML) models were developed using demographic data, medical history, and laboratory values taken up to 3 days before, or 1 day after, positive testing to predict severe outcomes of hospitalization (a composite endpoint of admission to intensive care, or death from any cause). Test accuracy was assessed using the area under the receiver operating characteristic curve (AUROC). RESULTS: Sex, C-reactive protein, sodium, hemoglobin, glomerular filtration rate, glucose, and leucocytes around the time of first positive testing (- 3 to + 1 days) were the most predictive parameters. AUROC of the risk stratification score on training data (AUROC = 0.94, positive predictive value (PPV) = 0.97, negative predictive value (NPV) = 0.80) were comparable to the prospective validation cohort (AUROC = 0.85, PPV = 0.91, NPV = 0.81). The most successful ML algorithm with respect to AUROC was support vector machines (median = 0.96, interquartile range = 0.85-0.99, PPV = 0.90, NPV = 0.58). CONCLUSION: With a small set of easily obtainable parameters, both the clinical risk stratification score and the ML models were predictive for severe outcomes at our tertiary hospital center, and performed well in prospective validation.

Prevalence of potentially inappropriate prescribing in a subpopulation of older European clinical trial participants: a cross-sectional study.

OBJECTIVES: To estimate and compare the prevalence and type of potentially inappropriate prescribing (PIP) and potential prescribing omissions (PPOs) among community-dwelling older adults (≥65 years) enrolled to a clinical trial in three European countries. DESIGN: A secondary analysis of the Thyroid Hormone Replacement for Subclinical Hypothyroidism Trial dataset. PARTICIPANTS: A subset of 48/80 PIP and 22/34 PPOs indicators from the Screening Tool of Older Persons Prescriptions/Screening Tool to Alert doctors to Right Treatment (STOPP/START) V2 criteria were applied to prescribed medication data for 532/737 trial participants in Ireland, Switzerland and the Netherlands. RESULTS: The overall prevalence of PIP was lower in the Irish participants (8.7%) compared with the Swiss (16.7%) and Dutch (12.5%) participants (P=0.15) and was not statistically significant. The overall prevalence of PPOs was approximately one-quarter in the Swiss (25.3%) and Dutch (24%) participants and lower in the Irish (14%) participants (P=0.04) and the difference was statistically significant. The hypnotic Z-drugs were the most frequent PIP in Irish participants, (3.5%, n=4), while it was non-steroidal anti-inflammatory drug and oral anticoagulant combination, sulfonylureas with a long duration of action, and benzodiazepines (all 4.3%, n=7) in Swiss, and benzodiazepines (7.1%, n=18) in Dutch participants. The most frequent PPOs in Irish participants were vitamin D and calcium in osteoporosis (3.5%, n=4). In the Swiss and Dutch participants, they were bone antiresorptive/anabolic therapy in osteoporosis (9.9%, n=16, 8.6%, n=22) respectively. The odds of any PIP after adjusting for age, sex, multimorbidity and polypharmacy were (adjusted OR (aOR)) 3.04 (95% CI 1.33 to 6.95, P<0.01) for Swiss participants and aOR 1.74 (95% CI 0.79 to 3.85, P=0.17) for Dutch participants compared with Irish participants. The odds of any PPOs were aOR 2.48 (95% CI 1.27 to 4.85, P<0.01) for Swiss participants and aOR 2.10 (95% CI 1.11 to 3.96, P=0.02) for Dutch participants compared with Irish participants. CONCLUSIONS: This study has estimated and compared the prevalence and type of PIP and PPOs among this cohort of community-dwelling older people. It demonstrated a significant difference in the prevalence of PPOs between the three populations. Further research is urgently needed into the impact of system level factors as this has important implications for patient safety, healthcare provision and economic costs.