An evaluation of sample size requirements for developing risk prediction models with binary outcomes.

BACKGROUND: Risk prediction models are routinely used to assist in clinical decision making. A small sample size for model development can compromise model performance when the model is applied to new patients. For binary outcomes, the calibration slope (CS) and the mean absolute prediction error (MAPE) are two key measures on which sample size calculations for the development of risk models have been based. CS quantifies the degree of model overfitting while MAPE assesses the accuracy of individual predictions. METHODS: Recently, two formulae were proposed to calculate the sample size required, given anticipated features of the development data such as the outcome prevalence and c-statistic, to ensure that the expectation of the CS and MAPE (over repeated samples) in models fitted using MLE will meet prespecified target values. In this article, we use a simulation study to evaluate the performance of these formulae. RESULTS: We found that both formulae work reasonably well when the anticipated model strength is not too high (c-statistic < 0.8), regardless of the outcome prevalence. However, for higher model strengths the CS formula underestimates the sample size substantially. For example, for c-statistic = 0.85 and 0.9, the sample size needed to be increased by at least 50% and 100%, respectively, to meet the target expected CS. On the other hand, the MAPE formula tends to overestimate the sample size for high model strengths. These conclusions were more pronounced for higher prevalence than for lower prevalence. Similar results were drawn when the outcome was time to event with censoring. Given these findings, we propose a simulation-based approach, implemented in the new R package 'samplesizedev', to correctly estimate the sample size even for high model strengths. The software can also calculate the variability in CS and MAPE, thus allowing for assessment of model stability. CONCLUSIONS: The calibration and MAPE formulae suggest sample sizes that are generally appropriate for use when the model strength is not too high. However, they tend to be biased for higher model strengths, which are not uncommon in clinical risk prediction studies. On those occasions, our proposed adjustments to the sample size calculations will be relevant.

Penalized Regression Methods With Modified Cross-Validation and Bootstrap Tuning Produce Better Prediction Models.

Risk prediction models fitted using maximum likelihood estimation (MLE) are often overfitted resulting in predictions that are too extreme and a calibration slope (CS) less than 1. Penalized methods, such as Ridge and Lasso, have been suggested as a solution to this problem as they tend to shrink regression coefficients toward zero, resulting in predictions closer to the average. The amount of shrinkage is regulated by a tuning parameter, λ , $\lambda ,$ commonly selected via cross-validation ("standard tuning"). Though penalized methods have been found to improve calibration on average, they often over-shrink and exhibit large variability in the selected λ $\lambda $ and hence the CS. This is a problem, particularly for small sample sizes, but also when using sample sizes recommended to control overfitting. We consider whether these problems are partly due to selecting λ $\lambda $ using cross-validation with "training" datasets of reduced size compared to the original development sample, resulting in an over-estimation of λ $\lambda $ and, hence, excessive shrinkage. We propose a modified cross-validation tuning method ("modified tuning"), which estimates λ $\lambda $ from a pseudo-development dataset obtained via bootstrapping from the original dataset, albeit of larger size, such that the resulting cross-validation training datasets are of the same size as the original dataset. Modified tuning can be easily implemented in standard software and is closely related to bootstrap selection of the tuning parameter ("bootstrap tuning"). We evaluated modified and bootstrap tuning for Ridge and Lasso in simulated and real data using recommended sample sizes, and sizes slightly lower and higher. They substantially improved the selection of λ $\lambda $ , resulting in improved CS compared to the standard tuning method. They also improved predictions compared to MLE.

External validation of a risk model predicting failure of salvage focal ablation for prostate cancer.

OBJECTIVES: To externally validate a published model predicting failure within 2 years after salvage focal ablation in men with localised radiorecurrent prostate cancer using a prospective, UK multicentre dataset. PATIENTS AND METHODS: Patients with biopsy-confirmed ≤T3bN0M0 cancer after previous external beam radiotherapy or brachytherapy were included from the FOcal RECurrent Assessment and Salvage Treatment (FORECAST) trial (NCT01883128; 2014-2018; six centres), and from the high-intensity focussed ultrasound (HIFU) Evaluation and Assessment of Treatment (HEAT) and International Cryotherapy Evaluation (ICE) UK-based registries (2006-2022; nine centres). Eligible patients underwent either salvage focal HIFU or cryotherapy, with the choice based predominantly on anatomical factors. Per the original multivariable Cox regression model, the predicted outcome was a composite failure outcome. Model performance was assessed at 2 years post-salvage with discrimination (concordance index [C-index]), calibration (calibration curve and slope), and decision curve analysis. For the latter, two clinically-reasonable risk threshold ranges of 0.14-0.52 and 0.26-0.36 were considered, corresponding to previously published pooled 2-year recurrence-free survival rates for salvage local treatments. RESULTS: A total of 168 patients were included, of whom 84/168 (50%) experienced the primary outcome in all follow-ups, and 72/168 (43%) within 2 years. The C-index was 0.65 (95% confidence interval 0.58-0.71). On graphical inspection, there was close agreement between predicted and observed failure. The calibration slope was 1.01. In decision curve analysis, there was incremental net benefit vs a 'treat all' strategy at risk thresholds of ≥0.23. The net benefit was therefore higher across the majority of the 0.14-0.52 risk threshold range, and all of the 0.26-0.36 range. CONCLUSION: In external validation using prospective, multicentre data, this model demonstrated modest discrimination but good calibration and clinical utility for predicting failure of salvage focal ablation within 2 years. This model could be reasonably used to improve selection of appropriate treatment candidates for salvage focal ablation, and its use should be considered when discussing salvage options with patients. Further validation in larger, international cohorts with longer follow-up is recommended.

Outlier identification and monitoring of institutional or clinician performance: an overview of statistical methods and application to national audit data.

BACKGROUND: Institutions or clinicians (units) are often compared according to a performance indicator such as in-hospital mortality. Several approaches have been proposed for the detection of outlying units, whose performance deviates from the overall performance. METHODS: We provide an overview of three approaches commonly used to monitor institutional performances for outlier detection. These are the common-mean model, the 'Normal-Poisson' random effects model and the 'Logistic' random effects model. For the latter we also propose a visualisation technique. The common-mean model assumes that the underlying true performance of all units is equal and that any observed variation between units is due to chance. Even after applying case-mix adjustment, this assumption is often violated due to overdispersion and a post-hoc correction may need to be applied. The random effects models relax this assumption and explicitly allow the true performance to differ between units, thus offering a more flexible approach. We discuss the strengths and weaknesses of each approach and illustrate their application using audit data from England and Wales on Adult Cardiac Surgery (ACS) and Percutaneous Coronary Intervention (PCI). RESULTS: In general, the overdispersion-corrected common-mean model and the random effects approaches produced similar p-values for the detection of outliers. For the ACS dataset (41 hospitals) three outliers were identified in total but only one was identified by all methods above. For the PCI dataset (88 hospitals), seven outliers were identified in total but only two were identified by all methods. The common-mean model uncorrected for overdispersion produced several more outliers. The reason for observing similar p-values for all three approaches could be attributed to the fact that the between-hospital variance was relatively small in both datasets, resulting only in a mild violation of the common-mean assumption; in this situation, the overdispersion correction worked well. CONCLUSION: If the common-mean assumption is likely to hold, all three methods are appropriate to use for outlier detection and their results should be similar. Random effect methods may be the preferred approach when the common-mean assumption is likely to be violated.

Importance of genotype for risk stratification in arrhythmogenic right ventricular cardiomyopathy using the 2019 ARVC risk calculator.

AIMS: To study the impact of genotype on the performance of the 2019 risk model for arrhythmogenic right ventricular cardiomyopathy (ARVC). METHODS AND RESULTS: The study cohort comprised 554 patients with a definite diagnosis of ARVC and no history of sustained ventricular arrhythmia (VA). During a median follow-up of 6.0 (3.1,12.5) years, 100 patients (18%) experienced the primary VA outcome (sustained ventricular tachycardia, appropriate implantable cardioverter defibrillator intervention, aborted sudden cardiac arrest, or sudden cardiac death) corresponding to an annual event rate of 2.6% [95% confidence interval (CI) 1.9-3.3]. Risk estimates for VA using the 2019 ARVC risk model showed reasonable discriminative ability but with overestimation of risk. The ARVC risk model was compared in four gene groups: PKP2 (n = 118, 21%); desmoplakin (DSP) (n = 79, 14%); other desmosomal (n = 59, 11%); and gene elusive (n = 160, 29%). Discrimination and calibration were highest for PKP2 and lowest for the gene-elusive group. Univariable analyses revealed the variable performance of individual clinical risk markers in the different gene groups, e.g. right ventricular dimensions and systolic function are significant risk markers in PKP2 but not in DSP patients and the opposite is true for left ventricular systolic function. CONCLUSION: The 2019 ARVC risk model performs reasonably well in gene-positive ARVC (particularly for PKP2) but is more limited in gene-elusive patients. Genotype should be included in future risk models for ARVC.

Multiparametric ultrasound versus multiparametric MRI to diagnose prostate cancer (CADMUS): a prospective, multicentre, paired-cohort, confirmatory study.

BACKGROUND: Multiparametric MRI of the prostate followed by targeted biopsy is recommended for patients at risk of prostate cancer. However, multiparametric ultrasound is more readily available than multiparametric MRI. Data from paired-cohort validation studies and randomised, controlled trials support the use of multiparametric MRI, whereas the evidence for individual ultrasound methods and multiparametric ultrasound is only derived from case series. We aimed to establish the overall agreement between multiparametric ultrasound and multiparametric MRI to diagnose clinically significant prostate cancer. METHODS: We conducted a prospective, multicentre, paired-cohort, confirmatory study in seven hospitals in the UK. Patients at risk of prostate cancer, aged 18 years or older, with an elevated prostate-specific antigen concentration or abnormal findings on digital rectal examination underwent both multiparametric ultrasound and multiparametric MRI. Multiparametric ultrasound consisted of B-mode, colour Doppler, real-time elastography, and contrast-enhanced ultrasound. Multiparametric MRI included high-resolution T2-weighted images, diffusion-weighted imaging (dedicated high B 1400 s/mm2 or 2000 s/mm2 and apparent diffusion coefficient map), and dynamic contrast-enhanced axial T1-weighted images. Patients with positive findings on multiparametric ultrasound or multiparametric MRI underwent targeted biopsies but were masked to their test results. If both tests yielded positive findings, the order of targeting at biopsy was randomly assigned (1:1) using stratified (according to centre only) block randomisation with randomly varying block sizes. The co-primary endpoints were the proportion of positive lesions on, and agreement between, multiparametric MRI and multiparametric ultrasound in identifying suspicious lesions (Likert score of ≥3), and detection of clinically significant cancer (defined as a Gleason score of ≥4 + 3 in any area or a maximum cancer core length of ≥6 mm of any grade [PROMIS definition 1]) in those patients who underwent a biopsy. Adverse events were defined according to Good Clinical Practice and trial regulatory guidelines. The trial is registered on ISRCTN, 38541912, and ClinicalTrials.gov, NCT02712684, with recruitment and follow-up completed. FINDINGS: Between March 15, 2016, and Nov 7, 2019, 370 eligible patients were enrolled; 306 patients completed both multiparametric ultrasound and multiparametric MRI and 257 underwent a prostate biopsy. Multiparametric ultrasound was positive in 272 (89% [95% CI 85-92]) of 306 patients and multiparametric MRI was positive in 238 patients (78% [73-82]; difference 11·1% [95% CI 5·1-17·1]). Positive test agreement was 73·2% (95% CI 67·9-78·1; κ=0·06 [95% CI -0·56 to 0·17]). Any cancer was detected in 133 (52% [95% CI 45·5-58]) of 257 patients, with 83 (32% [26-38]) of 257 being clinically significant by PROMIS definition 1. Each test alone would result in multiparametric ultrasound detecting PROMIS definition 1 cancer in 66 (26% [95% CI 21-32]) of 257 patients who had biopsies and multiparametric MRI detecting it in 77 (30% [24-36]; difference -4·3% [95% CI -8·3% to -0·3]). Combining both tests detected 83 (32% [95% CI 27-38]) of 257 clinically significant cancers as per PROMIS definition 1; of these 83 cancers, six (7% [95% CI 3-15]) were detected exclusively with multiparametric ultrasound, and 17 (20% [12-31]) were exclusively detected by multiparametric MRI (agreement 91·1% [95% CI 86·9-94·2]; κ=0·78 [95% CI 0·69-0·86]). No serious adverse events were related to trial activity. INTERPRETATION: Multiparametric ultrasound detected 4·3% fewer clinically significant prostate cancers than multiparametric MRI, but it would lead to 11·1% more patients being referred for a biopsy. Multiparametric ultrasound could be an alternative to multiparametric MRI as a first test for patients at risk of prostate cancer, particularly if multiparametric MRI cannot be carried out. Both imaging tests missed clinically significant cancers detected by the other, so the use of both would increase the detection of clinically significant prostate cancers compared with using each test alone. FUNDING: The Jon Moulton Charity Trust, Prostate Cancer UK, and UCLH Charity and Barts Charity.

Risk prediction in multicentre studies when there is confounding by cluster or informative cluster size.

BACKGROUND: Clustered data arise in research when patients are clustered within larger units. Generalised Estimating Equations (GEE) and Generalised Linear Models (GLMM) can be used to provide marginal and cluster-specific inference and predictions, respectively. METHODS: Confounding by Cluster (CBC) and Informative cluster size (ICS) are two complications that may arise when modelling clustered data. CBC can arise when the distribution of a predictor variable (termed 'exposure'), varies between clusters causing confounding of the exposure-outcome relationship. ICS means that the cluster size conditional on covariates is not independent of the outcome. In both situations, standard GEE and GLMM may provide biased or misleading inference, and modifications have been proposed. However, both CBC and ICS are routinely overlooked in the context of risk prediction, and their impact on the predictive ability of the models has been little explored. We study the effect of CBC and ICS on the predictive ability of risk models for binary outcomes when GEE and GLMM are used. We examine whether two simple approaches to handle CBC and ICS, which involve adjusting for the cluster mean of the exposure and the cluster size, respectively, can improve the accuracy of predictions. RESULTS: Both CBC and ICS can be viewed as violations of the assumptions in the standard GLMM; the random effects are correlated with exposure for CBC and cluster size for ICS. Based on these principles, we simulated data subject to CBC/ICS. The simulation studies suggested that the predictive ability of models derived from using standard GLMM and GEE ignoring CBC/ICS was affected. Marginal predictions were found to be mis-calibrated. Adjusting for the cluster-mean of the exposure or the cluster size improved calibration, discrimination and the overall predictive accuracy of marginal predictions, by explaining part of the between cluster variability. The presence of CBC/ICS did not affect the accuracy of conditional predictions. We illustrate these concepts using real data from a multicentre study with potential CBC. CONCLUSION: Ignoring CBC and ICS when developing prediction models for clustered data can affect the accuracy of marginal predictions. Adjusting for the cluster mean of the exposure or the cluster size can improve the predictive accuracy of marginal predictions.

Review and evaluation of penalised regression methods for risk prediction in low-dimensional data with few events.

Risk prediction models are used to predict a clinical outcome for patients using a set of predictors. We focus on predicting low-dimensional binary outcomes typically arising in epidemiology, health services and public health research where logistic regression is commonly used. When the number of events is small compared with the number of regression coefficients, model overfitting can be a serious problem. An overfitted model tends to demonstrate poor predictive accuracy when applied to new data. We review frequentist and Bayesian shrinkage methods that may alleviate overfitting by shrinking the regression coefficients towards zero (some methods can also provide more parsimonious models by omitting some predictors). We evaluated their predictive performance in comparison with maximum likelihood estimation using real and simulated data. The simulation study showed that maximum likelihood estimation tends to produce overfitted models with poor predictive performance in scenarios with few events, and penalised methods can offer improvement. Ridge regression performed well, except in scenarios with many noise predictors. Lasso performed better than ridge in scenarios with many noise predictors and worse in the presence of correlated predictors. Elastic net, a hybrid of the two, performed well in all scenarios. Adaptive lasso and smoothly clipped absolute deviation performed best in scenarios with many noise predictors; in other scenarios, their performance was inferior to that of ridge and lasso. Bayesian approaches performed well when the hyperparameters for the priors were chosen carefully. Their use may aid variable selection, and they can be easily extended to clustered-data settings and to incorporate external information.

A note on obtaining correct marginal predictions from a random intercepts model for binary outcomes.

BACKGROUND: Clustered data with binary outcomes are often analysed using random intercepts models or generalised estimating equations (GEE) resulting in cluster-specific or 'population-average' inference, respectively. METHODS: When a random effects model is fitted to clustered data, predictions may be produced for a member of an existing cluster by using estimates of the fixed effects (regression coefficients) and the random effect for the cluster (conditional risk calculation), or for a member of a new cluster (marginal risk calculation). We focus on the second. Marginal risk calculation from a random effects model is obtained by integrating over the distribution of random effects. However, in practice marginal risks are often obtained, incorrectly, using only estimates of the fixed effects (i.e. by effectively setting the random effects to zero). We compare these two approaches to marginal risk calculation in terms of model calibration. RESULTS: In simulation studies, it has been seen that use of the incorrect marginal risk calculation from random effects models results in poorly calibrated overall marginal predictions (calibration slope <1 and calibration in the large ≠ 0) with mis-calibration becoming worse with higher degrees of clustering. We clarify that this was due to the incorrect calculation of marginal predictions from a random intercepts model and explain intuitively why this approach is incorrect. We show via simulation that the correct calculation of marginal risks from a random intercepts model results in predictions with excellent calibration. CONCLUSION: The logistic random intercepts model can be used to obtain valid marginal predictions by integrating over the distribution of random effects.

Analysis of Randomised Trials Including Multiple Births When Birth Size Is Informative.

BACKGROUND: Informative birth size occurs when the average outcome depends on the number of infants per birth. Although analysis methods have been proposed for handling informative birth size, their performance is not well understood. Our aim was to evaluate the performance of these methods and to provide recommendations for their application in randomised trials including infants from single and multiple births. METHODS: Three generalised estimating equation (GEE) approaches were considered for estimating the effect of treatment on a continuous or binary outcome: cluster weighted GEEs, which produce treatment effects with a mother-level interpretation when birth size is informative; standard GEEs with an independence working correlation structure, which produce treatment effects with an infant-level interpretation when birth size is informative; and standard GEEs with an exchangeable working correlation structure, which do not account for informative birth size. The methods were compared through simulation and analysis of an example dataset. RESULTS: Treatment effect estimates were affected by informative birth size in the simulation study when the effect of treatment in singletons differed from that in multiples (i.e. in the presence of a treatment group by multiple birth interaction). The strength of evidence supporting the effectiveness of treatment varied between methods in the example dataset. CONCLUSIONS: Informative birth size is always a possibility in randomised trials including infants from both single and multiple births, and analysis methods should be pre-specified with this in mind. We recommend estimating treatment effects using standard GEEs with an independence working correlation structure to give an infant-level interpretation.

A novel clinical risk prediction model for sudden cardiac death in hypertrophic cardiomyopathy (HCM risk-SCD).

AIMS: Hypertrophic cardiomyopathy (HCM) is a leading cause of sudden cardiac death (SCD) in young adults. Current risk algorithms provide only a crude estimate of risk and fail to account for the different effect size of individual risk factors. The aim of this study was to develop and validate a new SCD risk prediction model that provides individualized risk estimates. METHODS AND RESULTS: The prognostic model was derived from a retrospective, multi-centre longitudinal cohort study. The model was developed from the entire data set using the Cox proportional hazards model and internally validated using bootstrapping. The cohort consisted of 3675 consecutive patients from six centres. During a follow-up period of 24 313 patient-years (median 5.7 years), 198 patients (5%) died suddenly or had an appropriate implantable cardioverter defibrillator (ICD) shock. Of eight pre-specified predictors, age, maximal left ventricular wall thickness, left atrial diameter, left ventricular outflow tract gradient, family history of SCD, non-sustained ventricular tachycardia, and unexplained syncope were associated with SCD/appropriate ICD shock at the 15% significance level. These predictors were included in the final model to estimate individual probabilities of SCD at 5 years. The calibration slope was 0.91 (95% CI: 0.74, 1.08), C-index was 0.70 (95% CI: 0.68, 0.72), and D-statistic was 1.07 (95% CI: 0.81, 1.32). For every 16 ICDs implanted in patients with ≥4% 5-year SCD risk, potentially 1 patient will be saved from SCD at 5 years. A second model with the data set split into independent development and validation cohorts had very similar estimates of coefficients and performance when externally validated. CONCLUSION: This is the first validated SCD risk prediction model for patients with HCM and provides accurate individualized estimates for the probability of SCD using readily collected clinical parameters.

Methods for observed-cluster inference when cluster size is informative: a review and clarifications.

Clustered data commonly arise in epidemiology. We assume each cluster member has an outcome Y and covariates X. When there are missing data in Y, the distribution of Y given X in all cluster members ("complete clusters") may be different from the distribution just in members with observed Y ("observed clusters"). Often the former is of interest, but when data are missing because in a fundamental sense Y does not exist (e.g., quality of life for a person who has died), the latter may be more meaningful (quality of life conditional on being alive). Weighted and doubly weighted generalized estimating equations and shared random-effects models have been proposed for observed-cluster inference when cluster size is informative, that is, the distribution of Y given X in observed clusters depends on observed cluster size. We show these methods can be seen as actually giving inference for complete clusters and may not also give observed-cluster inference. This is true even if observed clusters are complete in themselves rather than being the observed part of larger complete clusters: here methods may describe imaginary complete clusters rather than the observed clusters. We show under which conditions shared random-effects models proposed for observed-cluster inference do actually describe members with observed Y. A psoriatic arthritis dataset is used to illustrate the danger of misinterpreting estimates from shared random-effects models.

Review of methods for handling confounding by cluster and informative cluster size in clustered data.

Clustered data are common in medical research. Typically, one is interested in a regression model for the association between an outcome and covariates. Two complications that can arise when analysing clustered data are informative cluster size (ICS) and confounding by cluster (CBC). ICS and CBC mean that the outcome of a member given its covariates is associated with, respectively, the number of members in the cluster and the covariate values of other members in the cluster. Standard generalised linear mixed models for cluster-specific inference and standard generalised estimating equations for population-average inference assume, in general, the absence of ICS and CBC. Modifications of these approaches have been proposed to account for CBC or ICS. This article is a review of these methods. We express their assumptions in a common format, thus providing greater clarity about the assumptions that methods proposed for handling CBC make about ICS and vice versa, and about when different methods can be used in practice. We report relative efficiencies of methods where available, describe how methods are related, identify a previously unreported equivalence between two key methods, and propose some simple additional methods. Unnecessarily using a method that allows for ICS/CBC has an efficiency cost when ICS and CBC are absent. We review tools for identifying ICS/CBC. A strategy for analysis when CBC and ICS are suspected is demonstrated by examining the association between socio-economic deprivation and preterm neonatal death in Scotland.

Adult-onset hearing loss and incident cognitive impairment and dementia - A systematic review and meta-analysis of cohort studies.

BACKGROUND: We comprehensively summarized the cohort evidence to date on adult-onset hearing loss as risk factor for incident cognitive impairment and dementia, and examined the evidence for dose-response, risk for various dementia subtypes, and other moderators. Previous meta-analyses were less comprehensive. METHODS: We included cohort studies with participants without dementia and with hearing assessments at baseline, minimum 2 years follow-up and incident cognitive outcomes. We used random-effect models and subgroup and meta-regression on moderator analyses. RESULTS: We identified fifty studies (N=1,548,754). Hearing loss (yes/no) was associated with incident dementia risk (HR=1.35 [95% CI = 1.26 - 1.45), mild cognitive impairment (MCI HR=1.29 [95% CI = 1.11 - 1.50]), cognitive decline not specified as MCI or dementia (HR=1.29 [95% CI = 1.17 - 1.42]), and Alzheimer's disease dementia (ADD, HR=1.56 [95% CI = 1.30 - 1.87]), but not with vascular dementia (HR, 1.30 [95% CI = 0.83 - 2.05]). Each 10-decibel worsening of hearing was associated with a 16% increase in dementia risk (95% CI = 1.07 - 1.27). The effect of hearing loss did not vary across potential moderators. CONCLUSIONS: Cohort studies consistently support that adult-onset hearing loss increases the risk of incident cognitive decline, dementia, MCI, and ADD.

Nephron Sparing Treatment (NEST) for Small Renal Masses: A Feasibility Cohort-embedded Randomised Controlled Trial Comparing Percutaneous Cryoablation and Robot-assisted Partial Nephrectomy.

There is a paucity of high-level evidence on small renal mass (SRM) management, as previous classical randomised controlled trials (RCTs) failed to meet accrual targets. Our objective was to assess the feasibility of recruitment to a cohort-embedded RCT comparing cryoablation (CRA) to robotic partial nephrectomy (RPN). A total of 200 participants were recruited to the cohort, of whom 50 were enrolled in the RCT. In the CRA intervention arm, 84% consented (95% confidence interval [CI] 64-95%) and 76% (95% CI 55-91%) received CRA; 100% (95% CI 86-100%) of the control arm underwent RPN. The retention rate was 90% (95% CI 79-96%) at 6 mo. In the RPN group 2/25 (8%) were converted intra-operative to radical nephrectomy. Postoperative complications (Clavien-Dindo grade 1-2) occurred in 12% of the CRA group and 29% of the RPN group. The median length of hospital stay was shorter for CRA (1 vs 2 d; p = 0.019). At 6 mo, the mean change in renal function was -5.0 ml/min/1.73 m2 after CRA and -5.8 ml/min/1.73 m2 after RPN. This study demonstrates the feasibility of a cohort-embedded RCT comparing CRA and RPN. These data can be used to inform multicentre trials on SRM management. PATIENT SUMMARY: We assessed whether patients with a small kidney tumour would consent to a trial comparing two different treatments: cryoablation (passing small needles through the skin to freeze the kidney tumour) and surgery to remove part of the kidney. We found that most patients agreed and a full trial would therefore be feasible.

The Role of Multiparametric MRI and MRI-targeted Biopsy in the Diagnosis of Radiorecurrent Prostate Cancer: An Analysis from the FORECAST Trial.

BACKGROUND: The role of multiparametric magnetic resonance imaging (MRI) for detecting recurrent prostate cancer after radiotherapy is unclear. OBJECTIVE: To evaluate MRI and MRI-targeted biopsies for detecting intraprostatic cancer recurrence and planning for salvage focal ablation. DESIGN, SETTING, AND PARTICIPANTS: FOcal RECurrent Assessment and Salvage Treatment (FORECAST; NCT01883128) was a prospective cohort diagnostic study that recruited 181 patients with suspected radiorecurrence at six UK centres (2014 to 2018); 144 were included here. INTERVENTION: All patients underwent MRI with 5 mm transperineal template mapping biopsies; 84 had additional MRI-targeted biopsies. MRI scans with Likert scores of 3 to 5 were deemed suspicious. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: First, the diagnostic accuracy of MRI was calculated. Second, the pathological characteristics of MRI-detected and MRI-undetected tumours were compared using the Wilcoxon rank sum test and chi-square test for trend. Third, four biopsy strategies involving an MRI-targeted biopsy alone and with systematic biopsies of one to two other quadrants were studied. Fisher's exact test was used to compare MRI-targeted biopsy alone with the best other strategy for the number of patients with missed cancer and the number of patients with cancer harbouring additional tumours in unsampled quadrants. Analyses focused primarily on detecting cancer of any grade or length. Last, eligibility for focal therapy was evaluated for men with localised (≤T3bN0M0) radiorecurrent disease. RESULTS AND LIMITATIONS: Of 144 patients, 111 (77%) had cancer detected on biopsy. MRI sensitivity and specificity at the patient level were 0.95 (95% confidence interval [CI] 0.92 to 0.99) and 0.21 (95% CI 0.07 to 0.35), respectively. At the prostate quadrant level, 258/576 (45%) quadrants had cancer detected on biopsy. Sensitivity and specificity were 0.66 (95% CI 0.59 to 0.73) and 0.54 (95% CI 0.46 to 0.62), respectively. At the quadrant level, compared with MRI-undetected tumours, MRI-detected tumours had longer maximum cancer core length (median difference 3 mm [7 vs 4 mm]; 95% CI 1 to 4 mm, p < 0.001) and a higher grade group (p = 0.002). Of the 84 men who also underwent an MRI-targeted biopsy, 73 (87%) had recurrent cancer diagnosed. Performing an MRI-targeted biopsy alone missed cancer in 5/73 patients (7%; 95% CI 3 to 15%); with additional systematic sampling of the other ipsilateral and contralateral posterior quadrants (strategy 4), 2/73 patients (3%; 95% CI 0 to 10%) would have had cancer missed (difference 4%; 95% CI -3 to 11%, p = 0.4). If an MRI-targeted biopsy alone was performed, 43/73 (59%; 95% CI 47 to 69%) patients with cancer would have harboured undetected additional tumours in unsampled quadrants. This reduced but only to 7/73 patients (10%; 95% CI 4 to 19%) with strategy 4 (difference 49%; 95% CI 36 to 62%, p < 0.0001). Of 73 patients, 43 (59%; 95% CI 47 to 69%) had localised radiorecurrent cancer suitable for a form of focal ablation. CONCLUSIONS: For patients with recurrent prostate cancer after radiotherapy, MRI and MRI-targeted biopsy, with or without perilesional sampling, will diagnose cancer in the majority where present. MRI-undetected cancers, defined as Likert scores of 1 to 2, were found to be smaller and of lower grade. However, if salvage focal ablation is planned, an MRI-targeted biopsy alone is insufficient for prostate mapping; approximately three of five patients with recurrent cancer found on an MRI-targeted biopsy alone harboured further tumours in unsampled quadrants. Systematic sampling of the whole gland should be considered in addition to an MRI-targeted biopsy to capture both MRI-detected and MRI-undetected disease. PATIENT SUMMARY: After radiotherapy, magnetic resonance imaging (MRI) is accurate for detecting recurrent prostate cancer, with missed cancer being smaller and of lower grade. Targeting a biopsy to suspicious areas on MRI results in a diagnosis of cancer in most patients. However, for every five men who have recurrent cancer, this targeted approach would miss cancers elsewhere in the prostate in three of these men. If further focal treatment of the prostate is planned, random biopsies covering the whole prostate in addition to targeted biopsies should be considered so that tumours are not missed.

The Sexunzipped trial: optimizing the design of online randomized controlled trials.

BACKGROUND: Sexual health problems such as unwanted pregnancy and sexually transmitted infection are important public health concerns and there is huge potential for health promotion using digital interventions. Evaluations of digital interventions are increasingly conducted online. Trial administration and data collection online offers many advantages, but concerns remain over fraudulent registration to obtain compensation, the quality of self-reported data, and high attrition. OBJECTIVE: This study addresses the feasibility of several dimensions of online trial design-recruitment, online consent, participant identity verification, randomization and concealment of allocation, online data collection, data quality, and retention at 3-month follow-up. METHODS: Young people aged 16 to 20 years and resident in the United Kingdom were recruited to the "Sexunzipped" online trial between November 2010 and March 2011 (n=2036). Participants filled in baseline demographic and sexual health questionnaires online and were randomized to the Sexunzipped interactive intervention website or to an information-only control website. Participants were also randomly allocated to a postal request (or no request) for a urine sample for genital chlamydia testing and receipt of a lower (£10/US$16) or higher (£20/US$32) value shopping voucher compensation for 3-month outcome data. RESULTS: The majority of the 2006 valid participants (90.98%, 1825/2006) were aged between 18 and 20 years at enrolment, from all four countries in the United Kingdom. Most were white (89.98%, 1805/2006), most were in school or training (77.48%, 1545/1994), and 62.81% (1260/2006) of the sample were female. In total, 3.88% (79/2036) of registrations appeared to be invalid and another 4.00% (81/2006) of participants gave inconsistent responses within the questionnaire. The higher value compensation (£20/US$32) increased response rates by 6-10%, boosting retention at 3 months to 77.2% (166/215) for submission of online self-reported sexual health outcomes and 47.4% (118/249) for return of chlamydia urine samples by post. CONCLUSIONS: It was quick and efficient to recruit young people to this online trial. Our procedures for obtaining online consent, verifying participant identity, automated randomization, and concealment of allocation worked well. The optimal response rate for the online sexual health outcome measurement was comparable to face-to-face trials. Multiple methods of participant contact, requesting online data only, and higher value compensation increased trial retention at 3-month follow-up. TRIAL REGISTRATION: International Standard Randomized Controlled Trial Number (ISRCTN): 55651027; http://www.controlled-trials.com/ISRCTN55651027 (Archived by WebCite at http://www.webcitation.org/6LbkxdPKf).

A cost-effectiveness analysis of hypertrophic cardiomyopathy sudden cardiac death risk algorithms for implantable cardioverter defibrillator decision-making.

AIMS: To conduct a contemporary cost-effectiveness analysis examining the use of implantable cardioverter defibrillators (ICD) for primary prevention in patients with hypertrophic cardiomyopathy (HCM). METHODS: A discrete-time Markov model was used to determine the cost-effectiveness of different ICD decision-making rules for implantation. Several scenarios were investigated including the reference scenario of implantation rates according to observed real world practice. A 12-year time horizon with an annual cycle length was used. Transition probabilities used in the model were obtained using Bayesian analysis. The study has been reported according to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist. RESULTS: Using a 5-year SCD risk threshold of 6% was cheaper than current practice and has marginally better total quality adjusted life years (QALYs). This is the most cost-effective of the options considered, with an incremental cost effectiveness ratio of £834 per QALY. Sensitivity analyses highlighted that this decision is largely driven by what health related quality of life (HRQL) is attributed to ICD patients and time horizon. CONCLUSION: We present a timely new perspective on HCM ICD cost-effectiveness, using methods reflecting real-world practice. While we have shown that a 6% 5-year SCD risk cut-off provides the best cohort stratification to aid ICD decision-making, this will also be influenced by the particular values of costs and HRQL for subgroups or at a local level. The process of explicitly demonstrating the main factors which drive conclusions from such an analysis will help to inform shared decision-making in this complex area for all stakeholders concerned.

Health-Related Quality of Life in Adults With Classical Infratentorial Superficial Siderosis: A Cross-sectional Study.

BACKGROUND AND OBJECTIVES: Infratentorial superficial siderosis (iSS) is a rare but disabling neurological condition characterised by progressive hearing loss, balance and mobility problems. The functional decline in these neurological domains with iSS progression is likely to adversely impact health-related quality of life (HRQoL). We studied HRQoL of adults with iSS using two common generic HRQoL measures (Health Utilities Index Mark III (HUI3) and EuroQoL EQ5D (5 Level) to determine the most impacted domains and evaluate the association between HRQoL scores and disease duration. METHODS: This observational study was an anonymous online survey. Following institutional Research Ethics Committee approval, we contacted dedicated international organisations, charities and patient-groups identified through online searches, social media and collaborative networks, to distribute the study information and study link, inviting their members diagnosed with iSS to participate. Participation required access to a digital device connected to the internet, confirmation of eligibility (aged ≥18 years and previously diagnosed with iSS) and informed consent to participate in the survey, which included study-specific questions (demographics, iSS, hearing) and HRQoL questionnaires. Survey responses were captured by the Research Electronic Data Capture (REDCap) survey software and analysed using the SPSS statistical package. Linear regression analysis was performed to investigate the association between HRQoL scores and disease duration. RESULTS: Of fifty participants,60% were male; the median (interquartile range, IQR) age was 60 (15) years. The median (IQR) multi-attribute scores for HUI3 and EQ5D were 0.36 (0.53) and 0.64 (0.33), respectively. The most frequently affected domains (moderate or worse category) were hearing (64%), and pain (48%) for HUI3, and mobility (54%) and pain (50%) for EQ5D. There was a weak association between disease duration and multi-attribute scores for HUI3 (R=0.353; adjusted R2=0.096; b=-0.008; p=0.047) but not EQ5D. DISCUSSION: Our findings demonstrate low HRQoL scores which capture low functional status in several domains typically affected in iSS, suggesting that iSS has a major adverse impact on quality of life in multiple functional domains. Measures of HRQoL in iSS should be included in clinical and research settings, including treatment trials.

Magnetic Resonance Imaging and Targeted Biopsies Compared to Transperineal Mapping Biopsies Before Focal Ablation in Localised and Metastatic Recurrent Prostate Cancer After Radiotherapy.

BACKGROUND: Recurrent prostate cancer after radiotherapy occurs in one in five patients. The efficacy of prostate magnetic resonance imaging (MRI) in recurrent cancer has not been established. Furthermore, high-quality data on new minimally invasive salvage focal ablative treatments are needed. OBJECTIVE: To evaluate the role of prostate MRI in detection of prostate cancer recurring after radiotherapy and the role of salvage focal ablation in treating recurrent disease. DESIGN, SETTING, AND PARTICIPANTS: The FORECAST trial was both a paired-cohort diagnostic study evaluating prostate multiparametric MRI (mpMRI) and MRI-targeted biopsies in the detection of recurrent cancer and a cohort study evaluating focal ablation at six UK centres. A total of 181 patients were recruited, with 155 included in the MRI analysis and 93 in the focal ablation analysis. INTERVENTION: Patients underwent choline positron emission tomography/computed tomography and a bone scan, followed by prostate mpMRI and MRI-targeted and transperineal template-mapping (TTPM) biopsies. MRI was reported blind to other tests. Those eligible underwent subsequent focal ablation. An amendment in December 2014 permitted focal ablation in patients with metastases. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Primary outcomes were the sensitivity of MRI and MRI-targeted biopsies for cancer detection, and urinary incontinence after focal ablation. A key secondary outcome was progression-free survival (PFS). RESULTS AND LIMITATIONS: Staging whole-body imaging revealed localised cancer in 128 patients (71%), with involvement of pelvic nodes only in 13 (7%) and metastases in 38 (21%). The sensitivity of MRI-targeted biopsy was 92% (95% confidence interval [CI] 83-97%). The specificity and positive and negative predictive values were 75% (95% CI 45-92%), 94% (95% CI 86-98%), and 65% (95% CI 38-86%), respectively. Four cancer (6%) were missed by TTPM biopsy and six (8%) were missed by MRI-targeted biopsy. The overall MRI sensitivity for detection of any cancer was 94% (95% CI 88-98%). The specificity and positive and negative predictive values were 18% (95% CI 7-35%), 80% (95% CI 73-87%), and 46% (95% CI 19-75%), respectively. Among 93 patients undergoing focal ablation, urinary incontinence occurred in 15 (16%) and five (5%) had a grade ≥3 adverse event, with no rectal injuries. Median follow-up was 27 mo (interquartile range 18-36); overall PFS was 66% (interquartile range 54-75%) at 24 mo. CONCLUSIONS: Patients should undergo prostate MRI with both systematic and targeted biopsies to optimise cancer detection. Focal ablation for areas of intraprostatic recurrence preserves continence in the majority, with good early cancer control. PATIENT SUMMARY: We investigated the role of magnetic resonance imaging (MRI) scans of the prostate and MRI-targeted biopsies in outcomes after cancer-targeted high-intensity ultrasound or cryotherapy in patients with recurrent cancer after radiotherapy. Our findings show that these patients should undergo prostate MRI with both systematic and targeted biopsies and then ablative treatment focused on areas of recurrent cancer to preserve their quality of life. This trial is registered at ClinicalTrials.gov as NCT01883128.