Assessing the Predictive Performance of Probabilistic Caries Risk Assessment Models: The Importance of Calibration.

Probabilistic caries risk assessment models (P-CRA), such as the Cariogram, are promising tools to planning treatments in order to control and prevent caries. The usefulness of these models for informing patients and medical decision-making depends on 2 properties known as discrimination and calibration. Current common assessment of P-CRA models, however, ignores calibration, and this can be misleading. The aim of this paper was to provide tools for a proper assessment of calibration of the P-CRA models and improve calibration when lacking. A combination of standard calibration tools (calibration plot, calibration in-the-large, and calibration slope) and 3 novel measures of calibration (the Calibration Index and 2 related metrics, E50 and E90) are proposed to evaluate if a P-CRA model is well calibrated. Moreover, an approach was proposed and validated using data from a previous follow-up study performed on children evaluated by means of a reduced Cariogram model; Platt scaling and isotonic regression were applied showing a lack of calibration. The use of the Cariogram overestimates the actual risk of new caries for forecast probabilities <0.5 and underestimates the risk for forecast probabilities >0.6. Both Platt scaling and isotonic regression were able to significantly improve the calibration of the reduced Cariogram model, preserving its discrimination properties. The average specificity and sensitivity for both Platt scaling and isotonic regression using the cut-off point p= 0.5 were >83 and their sum well exceeded 160. The benefits of the proposed calibration methods are promising, but further research in this field is required.

Effects of seawater ingestion on lactate response to exercise in runners.

The aim of this study was to examine the effect of microfiltered and sterilized seawater ingestion on running performance in a hot environment. This cross-over, double-blind randomized trial included 12 experienced male runners. The subjects randomly consumed seawater (SW) or pure water (placebo) in an equivalent amount of 50 ml five minutes prior to running at 40% of their VO2 max for 95.0 ± 18.5 min, at 30°C, until they lost 3% of body weight. Every 20 minutes, a measurement of their body weight was taken and a blood lactate analysis was performed. The concentration of lactate was significantly lower after the running exercise in the SW condition compared to placebo. The results of this study provide evidence supporting the ergogenic effects of microfiltered and sterilized seawater ingestion on running performance and lactate production.

Assessing risk factors for dental caries: a statistical modeling approach.

The problem of identifying potential determinants and predictors of dental caries is of key importance in caries research and it has received considerable attention in the scientific literature. From the methodological side, a broad range of statistical models is currently available to analyze dental caries indices (DMFT, dmfs, etc.). These models have been applied in several studies to investigate the impact of different risk factors on the cumulative severity of dental caries experience. However, in most of the cases (i) these studies focus on a very specific subset of risk factors; and (ii) in the statistical modeling only few candidate models are considered and model selection is at best only marginally addressed. As a result, our understanding of the robustness of the statistical inferences with respect to the choice of the model is very limited; the richness of the set of statistical models available for analysis in only marginally exploited; and inferences could be biased due the omission of potentially important confounding variables in the model's specification. In this paper we argue that these limitations can be overcome considering a general class of candidate models and carefully exploring the model space using standard model selection criteria and measures of global fit and predictive performance of the candidate models. Strengths and limitations of the proposed approach are illustrated with a real data set. In our illustration the model space contains more than 2.6 million models, which require inferences to be adjusted for 'optimism'.

Day-to-day variability of electrocardiographic diagnosis of left ventricular hypertrophy in hypertensive patients. Influence of electrode placement.

OBJECTIVE: Although electrocardiography (ECG) is recommended in all subjects with hypertension, no information is available on the influence exerted by random changes in the placement of electrodes on the day-to-day variability of ECG criteria for diagnosis of left ventricular hypertrophy (LVH). METHODS: In a multicentre, randomized study, two standard 12-lead ECG were recorded, 24 h apart, from 276 consecutive hypertensive patients (mean age 65 +/- 12 years, 49.6% men). Overall, 142 patients were randomized to ECG with the position of electrodes marked on the skin using a dermographic pen and 134 to traditional ECG without marking the position of electrodes. Day-to-day variability of ECG criteria for LVH was compared between the two groups. RESULTS: Coefficients of variation (SD of the difference between paired voltage measurements divided by the mean value) varied consistently among subjects randomized to ECG without dermographic pen, ranging from 30% (R wave in lead I) to 81% (R wave in lead V5). Dermographic pen led to a lesser variability of ECG voltages with consequent reduction in the coefficients of variation, which ranged from 26% (R-wave amplitude in lead I) to 43% (R-wave amplitude in lead V5). The proportion of subjects who changed classification status for LVH ('reclassification rate') from the first to the second ECG session (LVH present in session 1 and absent in session 2, or vice versa) decreased for effect of dermographic pen from 11 to 4% (P = 0.040) with the Cornell voltage, from 19 to 11% (P = 0.029) with the Sokolow-Lyon voltage, and from 18 to 7% with the Romhilt-Estes criterion (P = 0.018), but not with other criteria. In particular, the typical strain and the Cornell strain were associated with the lowest reclassification rates regardless of dermographic pen. CONCLUSIONS: Random changes in the position of ECG electrodes strongly impair the day-to-day reproducibility of Cornell voltage, Sokolow-Lyon and Romhilt-Estes criteria for LVH. The typical strain and Cornell strain criteria showed a lesser spontaneous day-to-day variability.