Explore the vast range of titles available.

Evaluating the performance of a clinical prediction model is crucial to establish its predictive accuracy in the populations and settings intended for use. In this article, the first in a three part series, Collins and colleagues describe the importance of a meaningful evaluation using internal, internal-external, and external validation, as well as exploring heterogeneity, fairness, and generalisability in model performance.

[...]we may consider more direct tests for heterogeneity in predictor effects by place or time. [...]fully independent external validation with data not available at the time of prediction model development can be important (Fig. 2).

Clinical prediction models estimate an individual’s risk (probability) of a health related outcome to help guide patient counselling and clinical decision making. Most models provide a single point estimate of risk but without the associated uncertainty. Riley and colleagues argue that this needs to change, as understanding uncertainty of risk estimates helps to inform critical evaluation of a model and may impact shared decision making. Examples are provided to illustrate uncertainty in risk estimates, and key methods to quantify and present uncertainty are discussed.

Clinical prediction models aim to predict outcomes in individuals, to inform diagnosis or prognosis in healthcare. Hundreds of prediction models are published in the medical literature each year, yet many are developed using a dataset that is too small for the total number of participants or outcome events. This leads to inaccurate predictions and consequently incorrect healthcare decisions for some individuals. In this article, the authors provide guidance on how to calculate the sample size required to develop a clinical prediction model.

Pleural infections are common and drainage of the pleural space, in addition to antimicrobial therapy, is often required for adequate treatment. Guidelines suggest flushing small bore chest drains with 20-30 mL of saline every six hours, however, no randomized controlled trials (RCTs) have assessed if this practice improves outcomes for pleural space infections. As a result, flushing practice is varied, inconsistent, and confounds the interpretation of studied therapeutic modalities in pleural space infection trials. The impact of regular chest drain flushing compared to as-needed flushing on length of time to chest tube removal is unclear. Chest Drain REgular FLushing in ComplIcated Parapneumonic EFfusions and Empyemas (RELIEF) is a multi-center, open label randomized controlled trial conducted in the United States. Patients with a pleural space infection requiring chest drain placement for inpatient management will be screened for eligibility. Patients will be randomized within 24 hours of chest drain placement to a regular flushing protocol versus as-needed flushing for drain blockage. The primary outcome is time from randomization until time to chest drain removal (hours). Secondary outcomes are length of hospitalization, degree of radiographic improvement by chest X-ray, ultrasound or CT scan from time of drain placement to time of removal, need for additional procedures for the management of pleural space infection, and complications. An ordinal, multi-state transition model will be used to precisely characterize the role of flushing in longitudinal clinical outcomes in the two arms. RELIEF is a multi-center, open label randomized controlled trial that compares a regular saline flushing protocol with as-needed saline flushing of small-bore chest drains (8-20Fr) for the management of pleural space infection. This will be the first randomized controlled trial evaluating flushing protocol with patient-centered outcomes in pleural space infections. The trial was registered in ClinicalTrials.gov (NCT06427538) on 05-10-2024.

Patients with stable coronary disease were randomly assigned to an initial invasive strategy with angiography and revascularization if appropriate or to medical therapy alone. At 3.2 years, there was no significant difference between the groups with respect to the estimated rate of ischemic events. The findings were sensitive to the definition of myocardial infarction.

Abstract BACKGROUND The choice of anterior vs posterior approach for degenerative cervical myelopathy that spans multiple segments remains controversial. OBJECTIVE To compare the outcomes following the 2 approaches using multicenter prospectively collected data. METHODS Quality Outcomes Database (QOD) for patients undergoing surgery for 3 to 5 level degenerative cervical myelopathy was analyzed. The anterior group (anterior cervical discectomy [ACDF] or corpectomy [ACCF] with fusion) was compared with posterior cervical fusion. Outcomes included: patient reported outcomes (PROs): neck disability index (NDI), numeric rating scale (NRS) of neck pain and arm pain, EQ-5D, modified Japanese Orthopedic Association score for myelopathy (mJOA), and NASS satisfaction questionnaire; hospital length of stay (LOS), 90-d readmission, and return to work (RTW). Multivariable regression models were fitted for outcomes. RESULTS Of total 245 patients analyzed, 163 patients underwent anterior surgery (ACDF-116, ACCF-47) and 82 underwent posterior surgery. Patients undergoing an anterior approach had lower odds of having higher LOS (P < .001, odds ratio 0.16, 95% confidence interval 0.08-0.30). The 12-mo NDI, EQ-5D, NRS, mJOA, and satisfaction scores as well as 90-d readmission and RTW did not differ significantly between anterior and posterior groups. CONCLUSION Patients undergoing anterior approaches for 3 to 5 level degenerative cervical myelopathy had shorter hospital LOS compared to those undergoing posterior decompression and fusion. Also, patients in both groups exhibited similar long-term PROs, readmission, and RTW rates. Further investigations are needed to compare the differences in longer term reoperation rates and functional outcomes before the clinical superiority of one approach over the other can be established.

In a recent article, Wu et al.1 presented an electroencephalogram (EEG)-based prediction model for antidepressant treatment response1. Here, we point to limitations in the methods used to define response and to validate the prediction model—specifically, that change from baseline Hamilton depression rating scale (HAMD) scores needs to take into account the nonlinearity of response, and that the validation analysis transposed the predictor and the outcome. 1. ARISING FROM W. Wu et al. Nature Biotechnology. https://doi.org/10.1038/s41587-019-0397-3 (2020)

In the ISCHEMIA trial, patients with stable ischemic heart disease were randomly assigned to invasive or conservative treatment. As reported separately, the invasive strategy did not reduce clinical events. Improvements in health status were slightly greater with the invasive strategy, reflecting minimal effects in asymptomatic patients and larger effects in patients with angina symptoms at baseline.

Machine learning is increasingly being used to predict clinical outcomes. Most comparisons of different methods have been based on empirical analyses in specific datasets. We used Monte Carlo simulations to determine when machine learning methods perform better than statistical learning methods in a specific setting. We evaluated six learning methods: stochastic gradient boosting machines using trees as the base learners, random forests, artificial neural networks, the lasso, ridge regression, and linear regression estimated using ordinary least squares (OLS). Our simulations were informed by empirical analyses in patients with acute myocardial infarction (AMI) and congestive heart failure (CHF) and used six data-generating processes, each based on one of the six learning methods, to simulate continuous outcomes in the derivation and validation samples. The outcome was systolic blood pressure at hospital discharge, a continuous outcome. We applied the six learning methods in each of the simulated derivation samples and evaluated performance in the simulated validation samples. The primary observation was that neural networks tended to result in estimates with worse predictive accuracy than the other five methods in both disease samples and across all six data-generating processes. Boosted trees and OLS regression tended to perform well across a range of scenarios.