Clinical Prediction Score Research Articles

ABCD2, ABCD2-I, and OTTAWA scores for stroke risk assessment: a direct retrospective comparison

Transient ischemic attack (TIA) is a neurologic emergency characterized by cerebral ischemia eliciting a temporary focal neurological deficit. Many clinical prediction scores have been proposed to assess the risk of stroke after TIA; however, studies on their clinical validity and comparisons among them are scarce. The objective is to compare the accuracy of ABCD2, ABCD2-I, and OTTAWA scores in the prediction of a stroke at 7, 90 days, and 1 year in patients presenting with TIA. Single-centre, retrospective study including patients with TIA admitted to the Emergency Department of our third-level, University Hospital, between 2018 and 2019. Five hundred three patients were included. Thirty-nine (7.7%) had a stroke within 1 year from the TIA: 9 (1.7%) and 24 (4.7%) within 7 and 90 days, respectively. ABCD2, ABCD2-I, and OTTAWA scores were significantly higher in patients who developed a stroke. AUROCs ranged from 0.66 to 0.75, without statistically significant differences at each time-point. Considering the best cut-off of each score, only ABCD2 > 3 showed a sensitivity of 100% only in the prediction of stroke within 7 days. Among clinical items of each score, duration of symptoms, previous TIA, hemiparesis, speech disturbance, gait disturbance, previous cerebral ischemic lesions, and known carotid artery disease were independent predictors of stroke. Clinical scores have moderate prognostic accuracy for stroke after TIA. Considering the independent predictors for stroke, our study indicates the need to continue research and prompts the development of new tools on predictive scores for TIA.

Early prediction of clinical scores for left ventricular reverse remodeling using extreme gradient random forest, boosting, and logistic regression algorithm representations.

ObjectiveAt present, there is no early prediction model of left ventricular reverse remodeling (LVRR) for people who are in cardiac arrest with an ejection fraction (EF) of ≤35% at first diagnosis; thus, the purpose of this article is to provide a supplement to existing research.Materials and methodsA total of 109 patients suffering from heart attack with an EF of ≤35% at first diagnosis were involved in this single-center research study. LVRR was defined as an absolute increase in left ventricular ejection fraction (LVEF) from ≥10% to a final value of >35%, with analysis features including demographic characteristics, diseases, biochemical data, echocardiography, and drug therapy. Extreme gradient boosting (XGBoost), random forest, and logistic regression algorithm models were used to distinguish between LVRR and non-LVRR cases and to obtain the most important features.ResultsThere were 47 cases (42%) of LVRR in patients suffering from heart failure with an EF of ≤35% at first diagnosis after optimal drug therapy. General statistical analysis and machine learning methods were combined to exclude a number of significant feature groups. The median duration of disease in the LVRR group was significantly lower than that in the non-LVRR group (7 vs. 48 months); the mean values of creatine kinase (CK) and MB isoenzyme of creatine kinase (CK-MB) in the LVRR group were lower than those in the non-LVRR group (80.11 vs. 94.23 U/L; 2.61 vs. 2.99 ng/ml; 27.19 vs. 28.54 mm). Moreover, AUC values for our feature combinations ranged from 97 to 94% and to 87% when using the XGBoost, random forest, and logistic regression techniques, respectively. The ablation test revealed that beats per minute (BPM) and disease duration had a greater impact on the model’s ability to accurately forecast outcomes.ConclusionShorter disease duration, slightly lower CK and CK-MB levels, slightly smaller right and left ventricular and left atrial dimensions, and lower mean heart rates were found to be most strongly predictive of LVRR development (BPM).

Parkinson’s Disease Classification and Clinical Score Regression via United Embedding and Sparse Learning From Longitudinal Data

Parkinson's disease (PD) is known as an irreversible neurodegenerative disease that mainly affects the patient's motor system. Early classification and regression of PD are essential to slow down this degenerative process from its onset. In this article, a novel adaptive unsupervised feature selection approach is proposed by exploiting manifold learning from longitudinal multimodal data. Classification and clinical score prediction are performed jointly to facilitate early PD diagnosis. Specifically, the proposed approach performs united embedding and sparse regression, which can determine the similarity matrices and discriminative features adaptively. Meanwhile, we constrain the similarity matrix among subjects and exploit the l2,p norm to conduct sparse adaptive control for obtaining the intrinsic information of the multimodal data structure. An effective iterative optimization algorithm is proposed to solve this problem. We perform abundant experiments on the Parkinson's Progression Markers Initiative (PPMI) data set to verify the validity of the proposed approach. The results show that our approach boosts the performance on the classification and clinical score regression of longitudinal data and surpasses the state-of-the-art approaches.

Multi-Task Weakly-Supervised Attention Network for Dementia Status Estimation With Structural MRI

Accurate prediction of clinical scores (of neuropsychological tests) based on noninvasive structural magnetic resonance imaging (MRI) helps understand the pathological stage of dementia (e.g., Alzheimer's disease (AD)) and forecast its progression. Existing machine/deep learning approaches typically preselect dementia-sensitive brain locations for MRI feature extraction and model construction, potentially leading to undesired heterogeneity between different stages and degraded prediction performance. Besides, these methods usually rely on prior anatomical knowledge (e.g., brain atlas) and time-consuming nonlinear registration for the preselection of brain locations, thereby ignoring individual-specific structural changes during dementia progression because all subjects share the same preselected brain regions. In this article, we propose a multi-task weakly-supervised attention network (MWAN) for the joint regression of multiple clinical scores from baseline MRI scans. Three sequential components are included in MWAN: 1) a backbone fully convolutional network for extracting MRI features; 2) a weakly supervised dementia attention block for automatically identifying subject-specific discriminative brain locations; and 3) an attention-aware multitask regression block for jointly predicting multiple clinical scores. The proposed MWAN is an end-to-end and fully trainable deep learning model in which dementia-aware holistic feature learning and multitask regression model construction are integrated into a unified framework. Our MWAN method was evaluated on two public AD data sets for estimating clinical scores of mini-mental state examination (MMSE), clinical dementia rating sum of boxes (CDRSB), and AD assessment scale cognitive subscale (ADAS-Cog). Quantitative experimental results demonstrate that our method produces superior regression performance compared with state-of-the-art methods. Importantly, qualitative results indicate that the dementia-sensitive brain locations automatically identified by our MWAN method well retain individual specificities and are biologically meaningful.

Clinical utility of the BIWACO score for patients with atrial fibrillation after percutaneous coronary intervention

No predictive clinical risk scores for net adverse clinical events (NACE) have been developed for patients with atrial fibrillation (AF) after percutaneous coronary intervention (PCI). We evaluated NACE to develop clinically applicable risk-stratification scores in the Bleeding and thrombotic risk evaluation In patients With Atrial fibrillation under COronary intervention (BIWACO) study, a multicenter survey which has enrolled a total of 7837 patients. We also investigated the current status and time trends for the use of antithrombotic drugs. A total of 188 AF patients who had received PCI were examined. At discharge, 65% of patients were prescribed a triple therapy (TT), 6% were prescribed a dual therapy, the remaining 29% of patients received dual-antiplatelet therapy. After 4 years, the fraction of patients continuing TT decreased by 15%, whereas oral anticoagulant alone was only 2% of patients. NACE developed in 20% of patients, resulting in death in 5% of the patients, and the remaining 13% experienced bleeding events. We developed risk scores for NACE comprising the five strongest predictive items, which we designated BIWACO scores. The area under the curve was 0.774 for NACE. Our study explored the differences in treatment practices and guideline recommendations for antithrombotic therapy. We concluded that our BIWACO score is useful for predicting clinical outcomes in AF-patients after PCI.

Advances in clinical prediction scores for prognosis of coronavirus disease-2019

Coronavirus Disease-2019 (COVID-19) has been a major public health issue all over the world, placing a significant burden on available healthcare resources. The most common types of COVID-19 are the mild and common forms. Although the proportion of the severe-critical types is smaller, the rate of death is significantly higher and the medical resources required tend to be greater. Thus, a variety of scores based on other disease and COVID-19 were used to assess the risk of poor prognosis on the COVID-19, including the common scores for community-acquired pneumonia, sepsis and viral pneumonia. Unfortunately, the above scores often lacked an adequate description of the applicable population or were at high risk of bias with unknown applicability. Therefore, the article summarized the existing scores, aiming to provide a reference for clinical prognostic risk assessment.

Axial light loss of monocytes as a readily available prognostic biomarker in patients with suspected infection at the emergency department.

To evaluate the prognostic value of the coefficient of variance of axial light loss of monocytes (cv-ALL of monocytes) for adverse clinical outcomes in patients suspected of infection in the emergency department (ED). We performed an observational, retrospective monocenter study including all medical patients ≥18 years admitted to the ED between September 2016 and June 2019 with suspected infection. Adverse clinical outcomes included 30-day mortality and ICU/MCU admission <3 days after presentation. We determined the additional value of monocyte cv-ALL and compared to frequently used clinical prediction scores (SIRS, qSOFA, MEWS). Next, we developed a clinical model with routinely available parameters at the ED, including cv-ALL of monocytes. A total of 3526 of patients were included. The OR for cv-ALL of monocytes alone was 2.21 (1.98-2.47) for 30-day mortality and 2.07 (1.86-2.29) for ICU/MCU admission <3 days after ED presentation. When cv-ALL of monocytes was combined with a clinical score, the prognostic accuracy increased significantly for all tested scores (SIRS, qSOFA, MEWS). The maximum AUC for a model with routinely available parameters at the ED was 0.81 to predict 30-day mortality and 0.81 for ICU/MCU admission. Cv-ALL of monocytes is a readily available biomarker that is useful as prognostic marker to predict 30-day mortality. Furthermore, it can be used to improve routine prediction of adverse clinical outcomes at the ED. Registered in the Dutch Trial Register (NTR) und number 6916.

Clinical Prediction Score for Early Diagnosis of Acute Pancreatitis in Emergency Departments.

BackgroundIn Thailand, most primary care hospitals cannot measure serum lipase and amylase; no 24 hours computed tomography and magnetic resonance imaging available, and no on-call gastroenterologists. Thus, acute pancreatitis cannot be diagnosed based on the established diagnostic criteria that require this information. The resultant delayed management increases morbidity and mortality. This study was performed to create a clinical prediction score for early diagnosis of acute pancreatitis in emergency departments without requiring a computed tomography scan or laboratory measurement to assist in the initial diagnosis, treatment, or referral.MethodsPatients with suspected acute pancreatitis who had available data regarding lipase and amylase measurements and visited the emergency department from June 2019 to August 2020 were retrospectively analyzed. The baseline predictive factors were compared between patients with and without acute pancreatitis according to the 2012 revised Atlanta classification. Multivariable logistic regression was used to explore potential predictive factors and develop a clinical prediction score for the diagnosis of acute pancreatitis.ResultsA total of 506 eligible patients, 84 (16%) had acute pancreatitis. The PRE-PAN score [area under the receiver operating characteristics curve, 0.88; 95% confidence interval (CI), 0.84–0.93] included six factors: alcohol drinking, epigastric pain, pain radiating to the back, persistent pain, nausea or vomiting, and the pain score. A score of >7.5 points suggested a high probability of acute pancreatitis [positive likelihood ratio, 6.80 (95% CI, 4.75–9.34; p < 0.001); sensitivity, 66.7% (95% CI, 54.6–77.3); specificity, 90.2% (95% CI, 86.6–93.1); positive predictive value, 58.5% (95% CI, 47.1–69.3);, 92.9% (95% CI, 89.6–95.4)].ConclusionA PRE-PAN risk score is a screening tool for predicting acute pancreatitis without using the lipase concentration or radiological findings. A high predictive score, especially >7.5, suggests a high probability of acute pancreatitis.

Impact of Social Needs in Electronic Health Records and Claims on Health Care Utilization and Costs Risk-Adjustment Models Within Medicaid Population.

Patients enrolled in Medicaid have significantly higher social needs (SNs) than others. Using claims and electronic health records (EHRs) data, managed care organizations (MCOs) could systemically identify high-risk patients with SNs and develop population health management interventions. Impact of SNs on models predicting health care utilization and costs was assessed. This retrospective study included claims and EHRs data on 39,267 patients younger than age 65 years who were continuously enrolled during 2018-2019 in a Medicaid-managed care plan. SN marker was developed suggesting presence of International Classification of Diseases, 10th revision codes in any of the 5 SN domains. Impact of SN marker was compared across demographic and 2 diagnosis-based (ie, Charlson and Adjusted Clinical Groups risk score) prediction models of emergency department (ED) visit and hospitalizations, and total, medical, and pharmacy costs. After combining data sources, prevalence of documented SN marker increased from 11% and 13% to 18% of the study population across claims, EHRs, and both combined, respectively. SN marker improved predictions of demographic models for all utilization and total costs outcomes (area under the curve [AUC] of ED model increased from 0.57 to 0.61 and R2 of total cost model increased from 10.9 to 12.2). In both diagnosis-based models, adding SN marker marginally improved outcomes prediction (AUC of ED model increased from 0.65 to 0.66). This study demonstrated feasibility of using claims and EHRs data to systematically capture SNs and incorporate in prediction models that could enable MCOs and policy makers to adjust and develop effective population health interventions.

Non-motorised Rehabilitation Device for Performance Assessment in Upper Limb Stroke Rehabilitation: A Pilot Study

Stroke patients with upper limb disability restricted to carry out their activities of daily living. The patient needs a motivation to recover from a stroke and the patient also needs to go through a rehabilitation process at the same time. The conventional rehabilitation process scoring systems are always subjective, lack reliability and relies heavily on the ability of the trained physiotherapist that providing only rough estimates on motor function. On the other hand, robot-based assessments are objective, repeatable, and could potentially reduce the assessment time. Therefore, a simple non-motorized device was developed as a tool to objectively assess hand function of stroke patients. This study was carried out to investigate the suitability of using thedeveloped device with stroke patient populations and to evaluate the performance of clinical scores prediction of the stroke patients. A total of five patients with upper limb disability following stroke consented to take part in this study. Twelve predictive variables were investigated, relating to the total movement time, velocity, strategy, accuracy, and smoothness from three robotic assessment modules which are Draw I, Draw Diamond and Draw Circle. The hardware for measuring elbow angle hasbeen developed to measurethe shoulder movement performedby patient during the assessment process. In addition to that, the shoulder movement calculation method has been proposed and validated.The findings indicatethat the performance of prediction for all assessment modules hasbeen increased after implementing the shoulder movement calculation. It is recommended this calculation method to be used in conjunction with kinematic variables to carry out the data acquisition process in the future for improvement of effectiveness and accuracy of the robotic assessment.

Predicting the efficacy of first-line immunotherapy by combining cancer cachexia and tumor burden in advanced non-small cell lung cancer.

BackgroundCancer cachexia and tumor burden predict efficacies of programmed cell death‐1 (PD‐1)/programmed death‐ligand 1 (PD‐L1) inhibitors and chemotherapy or pembrolizumab in non‐small cell lung cancer (NSCLC). There are no predictive models that simultaneously assess cancer cachexia and tumor burden.MethodsIn the present retrospective study, we reviewed the medical records of patients with advanced NSCLC who received cancer immunotherapy as first‐line systemic therapy. Clinical immune predictive scores were defined according to multivariate analysis of progression‐free survival (PFS) and overall survival (OS).ResultsA total of 157 patients were included in the present study (75 treated with PD‐1/PD‐L1 inhibitors + chemotherapy; 82, pembrolizumab monotherapy). Multivariate analysis for PFS revealed that PD‐L1 tumor proportion scores <50%, a total target lesion diameter ≥76 mm, and cancer cachexia were independently associated with poor PFS. Multivariate analysis for OS revealed that ≥4 metastases and cancer cachexia were significantly associated with poor OS. In the immune predictive model, the median PFS was 21.7 months in the low‐risk group (N = 41); 7.6 in the medium‐risk group (N = 64); and 3.0 in the high‐risk group (N = 47). The median OS were not reached, 22.4 and 9.1 months respectively. Our immune predictive model was significantly associated with PFS (p < 0.001) and OS (p < 0.001).ConclusionWe proposed the immune predictive model, including tumor burden and cancer cachexia, which may predict the efficacy and survival outcome of first‐line immunotherapy in advanced NSCLC.

Diagnostic Values of Red Flags and a Clinical Prediction Score for Emergent Intracranial Lesions in Non-Traumatic Pediatric Headaches.

Introduction: Diagnosis of emergent intracranial lesions that require emergency treatment either medically or surgically in non-traumatic pediatric headaches is important. Red-flag signs and symptoms are commonly used as justification for neuroimaging; however, evidence on its diagnostic values is limited. The study aims to identify diagnostic values of red-flags and develop a clinical prediction score to help improve the diagnostic yield of neuroimaging. Methods: A retrospective review of 109 pediatric patients from 2006 to 2020 who presented with a non-traumatic headache was conducted. A clinical prediction score from red flags was developed using multivariate logistic regression. Discriminatory ability was examined using the area under the receiver operating characteristic curve. Results: A total of 51 patients were diagnosed with emergent intracranial lesions. Four potential clinical red flag predictors were identified: (1) acute onset (less than 3 months), (2) altered conscious state, (3) focal motor abnormality, and (4) and ocular/pupillary abnormality or squint. A clinical prediction score was developed with good discriminatory properties (0.84). Conclusions: Clinical predictor scores from these four red flags may play an important role in maximizing neuroimaging and proper management for pediatric patients with non-traumatic headaches. Future validation studies are needed and could guide referrals and optimize the use of neuroimaging for these patients.

Evaluation and Mitigation of Racial Bias in Clinical Machine Learning Models: Scoping Review.

BackgroundRacial bias is a key concern regarding the development, validation, and implementation of machine learning (ML) models in clinical settings. Despite the potential of bias to propagate health disparities, racial bias in clinical ML has yet to be thoroughly examined and best practices for bias mitigation remain unclear.ObjectiveOur objective was to perform a scoping review to characterize the methods by which the racial bias of ML has been assessed and describe strategies that may be used to enhance algorithmic fairness in clinical ML.MethodsA scoping review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Extension for Scoping Reviews. A literature search using PubMed, Scopus, and Embase databases, as well as Google Scholar, identified 635 records, of which 12 studies were included.ResultsApplications of ML were varied and involved diagnosis, outcome prediction, and clinical score prediction performed on data sets including images, diagnostic studies, clinical text, and clinical variables. Of the 12 studies, 1 (8%) described a model in routine clinical use, 2 (17%) examined prospectively validated clinical models, and the remaining 9 (75%) described internally validated models. In addition, 8 (67%) studies concluded that racial bias was present, 2 (17%) concluded that it was not, and 2 (17%) assessed the implementation of bias mitigation strategies without comparison to a baseline model. Fairness metrics used to assess algorithmic racial bias were inconsistent. The most commonly observed metrics were equal opportunity difference (5/12, 42%), accuracy (4/12, 25%), and disparate impact (2/12, 17%). All 8 (67%) studies that implemented methods for mitigation of racial bias successfully increased fairness, as measured by the authors’ chosen metrics. Preprocessing methods of bias mitigation were most commonly used across all studies that implemented them.ConclusionsThe broad scope of medical ML applications and potential patient harms demand an increased emphasis on evaluation and mitigation of racial bias in clinical ML. However, the adoption of algorithmic fairness principles in medicine remains inconsistent and is limited by poor data availability and ML model reporting. We recommend that researchers and journal editors emphasize standardized reporting and data availability in medical ML studies to improve transparency and facilitate evaluation for racial bias.

Multi-Modality Imaging for Prevention of Coronary Artery Disease and Myocardial Infarction in the General Population: Ready for Prime Time?

Cardiovascular disease (CVD) remains a leading cause of death and disability worldwide. Acute myocardial infarction (AMI) causes irreversible myocardial damage, heart failure, life-threatening arrythmias and sudden cardiac death (SCD), and is a main driver of CVD mortality and morbidity. To control the forecasted increase in CVD burden for both the individual and society, improved strategies for the prevention of AMI and SCD are required. Current prevention of AMI and SCD is directed towards risk-modifying interventions, guided by risk assessment using clinical risk prediction scores (CRPSs) and the coronary artery calcium score (CACS). Early detection of more advanced coronary artery disease (CAD), beyond risk assessment by CRPSs or CACS, is a promising strategy to allow personalized treatment for the improved prevention of AMI and SCD in the general population. We review evidence for further testing, beyond CRPSs and CACS, and therapies focusing on promising targets, including subclinical obstructive CAD, high-risk plaques, and silent myocardial ischemia. We also evaluate the potential of multi-modality imaging to enhance the conduction of adequately powered trials to provide high-quality evidence on the impact of add-on tests and therapies in the prevention of AMI and SCD in asymptomatic individuals. To conclude, we discuss the occurrence of AMI and SCD in individuals currently estimated to be at “low-risk” by the current strategy based on CRPSs, and methods to improve prevention of AMI and SCD in this “low-risk” population.

Derivation and Validation of a Clinical Prediction Score to Identify the Isolation of Pseudomonas in Pneumonia.

ABSTRACTGiven the focus of existing clinical prediction scores on identifying drug-resistant pathogens as a whole, the application to individual pathogens and other institutions may yield weaker performance. This study aimed to develop a locally derived clinical prediction model for Pseudomonas-mediated pneumonia. This retrospective study included patients ≥18 years of age who were admitted to an academic medical center between 1 July 2010 and 31 July 2020 with a CDC National Healthcare Safety Network confirmed pneumonia diagnosis and were receiving antimicrobials during the index encounter, with a positive respiratory culture. Cystic fibrosis patients were excluded. Logistic regression analysis identified risk factors associated with the isolation of Pseudomonas aeruginosa from respiratory cultures within the derivation cohort (n = 186), which were weighted to generate a prediction score that was applied to the derivation and internal validation (n = 95) cohorts. A total of 281 patients met the inclusion criteria. Five predictor variables were identified, namely, tracheostomy status (4 points), chronic obstructive pulmonary disease (5 points), enteral nutrition (9 points), chronic steroid use (11 points), and Pseudomonas aeruginosa isolation from any culture in the prior 6 months (14 points). At a score of >11, the prediction score demonstrated a sensitivity of 52.4% (95% confidence interval [CI], 36.4 to 68.0%) and a specificity of 84.9% (95% CI, 72.4 to 93.35%) in the validation cohort. Score accuracy was 70.5% (95% CI, 60.3 to 79.4%), and the area under the receiver operating characteristic curve (AUROC) was 0.77 (95% CI, 0.68 to 0.87) in the validation cohort. A prediction score for identifying Pseudomonas aeruginosa in pneumonia was derived, which may have the potential to decrease the use of broad-spectrum antibiotics. Validation with larger and external cohorts is necessary.IMPORTANCE In this study, we aimed to develop a locally derived clinical prediction model for Pseudomonas-mediated pneumonia. Utilizing a locally validated prediction score may help direct therapeutic management and be generalizable to other clinical settings and similar populations for the selection of appropriate antimicrobial coverage when data are lacking. Our study highlights a unique patient population, including immunocompromised, structural lung disease, and transplant patients. Five predictor variables were identified, namely, tracheostomy status, chronic obstructive pulmonary disease, enteral nutrition, chronic steroid use, and Pseudomonas aeruginosa isolation from any culture in the prior 6 months. A prediction score for identifying Pseudomonas aeruginosa in pneumonia was derived, which may have the potential to decrease the use of broad-spectrum antibiotics, although validation with larger and external cohorts is necessary.

A five-parameter score for predicting saphenous vein graft degenerative and/or occlusive disease in recurring ischemic symptoms after one year post coronary artery bypass grafting.

The recurrence rate of ischemic symptoms after coronary artery bypass grafting (CABG) is increasing in recent years. How to prevent and treat saphenous vein graft disease (SVGD [symptomatic ⩾50% stenosis in at least one Saphenous vein graft]) has been a clinical challenge to date. Different pathogenesis may exist in SVGD of different periods. There are currently few available scores for estimating the risk of SVGD after one year post CABG. We sought to develop and validate a simple predictive clinical risk score for SVGD with recurring ischemia after one year post CABG. This was a cross-sectional study and the results were validated using bootstrap resampling on a separate cohort. A nomogram and risk scoring system were developed based on retrospective data from a training cohort of 606 consecutive patients with recurring ischemia >1 year after CABG. Logistic regression model was used to find the predictive factors and to build a nomogram. To assess the generalization, models were validated using bootstrap resampling and an external cross-sectional study of 187 consecutive patients in four other hospitals. In multivariable analysis of the primary cohort, native lesion vessel number, SVG age, recurring ischemia type, very low-density lipoprotein level, and left ventricular end-diastolic diameter were independent predictors. A summary risk score was derived from nomogram, with a cut-off value of 15. In internal and external validation, the C-index was 0.86 and 0.82, indicating good discrimination. The calibration curve for probability of SVGD showed optimal agreement between actual observations and risk score prediction. A simple-to-use risk scoring system based on five easily variables was developed and validated to predict the risk of SVGD among patients who recurring ischemia after one year post CABG. This score may be useful for providing patients with individualized estimates of SVGD risk.

Validating a clinical prediction score for Legionella-related community acquired pneumonia

BackgroundLegionella-related community acquired pneumonia (CAP) is a disease with an increasing incidence and a high mortality rate, especially if empirical antibiotic therapy is inadequate. Antibiotic treatment highly relies on clinical symptoms, although proven non-specific, because currently available diagnostic techniques provide insufficient accuracy for detecting Legionella CAP on admission. This study validates a diagnostic scoring system for detection of Legionella-related CAP, based on six items on admission (Legionella prediction score).MethodsWe included patients with Legionella-related CAP admitted to five large Dutch hospitals between 2006 and 2016. Controls were non-Legionella-related CAP patients. The following six conditions were rewarded one point if present: fever > 39.4 °C; dry cough; hyponatremia (sodium) < 133 mmol/L; lactate dehydrogenase (LDH) > 225 mmol/L; C-reactive protein (CRP) > 187 mg/L and platelet count < 171 × 109/L. The accuracy of the prediction score was assessed by calculating the area under the curve (AUC) through logistic regression analysis.ResultsWe included 131 cases and 160 controls. A score of 0 occurred in non-Legionella-related CAP patients only, a score of 5 and 6 in Legionella-related CAP patients only. A cut-off ≥ 4 resulted in a sensitivity of 58.8% and a specificity of 93.1%. The AUC was 0.89 (95% CI 0.86–0.93). The strongest predictors were elevated LDH, elevated CRP and hyponatremia.ConclusionsThis multi-centre study validates the Legionella prediction score, an easily applicable diagnostic scoring system, in a large group of patients and finds high diagnostic accuracy. The score shows promise for future prospective validation and could contribute to targeted antibiotic treatment of suspected Legionella CAP.

Predictive preoperative clinical score for patients with liver-only oligometastatic colorectal cancer

BackgroundResection of liver metastases from colorectal cancer (CRC) in the oligometastatic stage improves survival and is a potentially curative treatment. Thus, predictive scores that reliably identify those patients who especially benefit from surgery are essential.Patients and methodsIn this multicenter analysis, 512 patients had undergone surgery for liver metastases from CRC. We investigated distinct cancer-specific risk factors that are routinely available in clinical practice and developed a predictive preoperative score using a training cohort (TC), which was thereafter tested in a validation cohort (VC).ResultsInflammatory response to the tumor, a right-sided primary tumor, multiple liver metastases, and node-positive primary tumor were significant adverse variables for overall survival (OS). Patients were stratified in five groups according to the cumulative score given by the presence of these risk factors. Median OS for patients without risk factors was 133.8 months [95% confidence interval (CI) 81.2-not reached (nr)] in the TC and was not reached in the VC. OS decreased significantly for each subsequent group with increasing number of risk factors. Median OS was significantly shorter (P < 0.0001) for patients presenting all four risk factors: 14.3 months (95% CI 10.5 months-nr) in the TC and 16.6 months (95% CI 14.6 months-nr) in the VC.ConclusionsIncluding easily obtainable variables, this preoperative score identifies oligometastatic CRC patients with prolonged survival rates that may be cured, and harbors potential to be implemented in daily clinical practice.

Clinical prediction scores and the utility of time to blood culture positivity in stratifying the risk of infective endocarditis in Staphylococcus aureus bacteraemia.

Infective endocarditis (IE) complicates up to a quarter of Staphylococcus aureus bacteraemia (SAB) cases. Risk scores predict IE complicating SAB but have undergone limited external validation, especially in community-acquired infections and those who use IV drugs. Addition of the time to positive culture (TTP) may provide incremental risk prognostication. To externally validate risk scores for predicting IE in SAB and assess the incremental value of TTP. The modified Duke score was calculated for adults hospitalized with SAB at a major tertiary institution. All patients underwent echocardiography. Sensitivity and specificity of the risk scores for predicting IE were calculated, and the incremental value of TTP was assessed. One hundred and six cases were analysed and 18 (17%) met definite IE criteria. The optimal TTP to predict IE was 11.5 h (sensitivity 88.9%; specificity 71.6%). The sensitivity of VIRSTA and PREDICT (Predicting risk of endocarditis using a clinical tool) were similar (94.4% for both) and higher than POSITIVE (Prediction Of Staphylococcus aureus Infective endocarditis Time to positivity, IV drug use, Vascular phenomena, pre-Existing heart condition; 77.8%). The receiver-operator characteristic AUCs were VIRSTA 0.83, PREDICT 0.75, POSITIVE 0.89 and TTP 0.85. Adding TTP to VIRSTA (i.e. VIRSTA+) resulted in the highest AUC (0.90), sensitivity (100%) and negative predictive value (100%), albeit with a low specificity (33%). The VIRSTA and POSITIVE scores were the strongest predictors for IE complicating SAB. The addition of TTP to VIRSTA (VIRSTA+) significantly improved discriminatory value and may be safely used to rationalize echocardiography strategies.

Composite Outcomes in Clinical Prediction Modeling: Are We Trying to Predict Apples and Oranges?

Composite outcomes are widely used in clinical research. Existing literature has considered the pros and cons of composite outcomes in clinical trials, but their extensive use in clinical prediction has received much less attention. Clinical prediction assists decision-making by directing patients with higher risks of adverse outcomes toward interventions that provide the greatest benefits to those at the greatest risk. In this article, we summarize our existing understanding of the advantages and disadvantages of composite outcomes, consider how these relate to clinical prediction, and highlight the problem of key predictors having markedly different associations with individual components of the composite outcome. We suggest that a "composite outcome fallacy" may occur when a clinical prediction model is based on strong associations between key predictors and one component of a composite outcome (such as mortality) and used to direct patients toward intervention when these predictors actually have an inverse association with a more relevant component of the composite outcome (such as the use of a lifesaving intervention). We propose that clinical prediction scores using composite outcomes should report their accuracy for key components of the composite outcome and examine for inconsistencies among predictor variables.