Search

Objective: To compare how different imputation methods affect the estimates and performance of a prediction model for premature mortality. Study Design and Setting: Sex-specific Weibull accelerated failure time survival models were run on four separate datasets using complete case, mode, single and multiple imputation to impute missing values. Six performance measures were compared to access predictive accuracy (Nagelkerke R2, integrated brier score), discrimination (Harrell's c-index, discrimination slope) and calibration (calibration in the large, calibration slope). Results: The highest proportion of missingness for a single variable was 10.86% for the female model and 8.24% for the male model. Comparing the performance measures for complete case, mode, single and multiple imputation: the Nagelkerke R2 values for the female model was 0.1084, 0.1116, 0.1120 and 0.111–0.1120 with the male model exhibited similar variation of 0.1050, 0.1078, 0.1078 and 0.1078–0.1081. Harrell's c-index also demonstrated small variation with values of 0.8666, 0.8719, 0.8719 and 0.8711–0.8719 for the female model and 0.8549, 0.8548, 0.8550 and 0.8550–0.8553 for the male model. Conclusion: In the scenarios examined in this study, mode imputation performed well when using a population health survey compared to single and multiple imputation when predictive performance measures is the main model goal. To generate unbiased hazard ratios, multiple imputation methods were superior. This study shows the need to consider the best imputation approach for a predictive model development given the conditions of missing data and the goals of the analysis. [ABSTRACT FROM AUTHOR]

To identify profiles of aging by combining psychological distress, cognition and functional disability, and their associated factors. Data were drawn from the Étude sur la Santé des Aînés-Services study and included 1585 older adults. Sociodemographic, psychosocial, lifestyle and health factors were informed from structured interviews. Group-based multi-trajectory modeling and multinomial logistic regression were used to identify aging profiles and correlates. Sampling weights were applied to account for the sampling plan. The weighted sample size was 1591. Three trajectories were identified: a favorable (79.0%), intermediate (14.5%), and severe scenario (6.5%). Factors associated with the severe scenario were older age, male gender, lower education, the presence of anxiety disorders, low physical activity, and smoking. Membership in the intermediate scenario was associated with daily hassles, physical disorders, anxiety and depression, antidepressant/psychotherapy use, low physical activity, and no alcohol use. High social support was protective against less favorable profiles. Symptoms of anxiety and depression and high burden of physical disorders were associated with less favorable trajectories. Modifiable lifestyle factors have a significant effect on healthy aging. Assessment and management of anxio-depressive symptoms are important in older adults. Clinical interventions including access to psychotherapy and promotion of healthier lifestyles should be considered. [ABSTRACT FROM AUTHOR]

Purpose: Socially disadvantaged patients may lack self-efficacy to navigate a complex health system making them vulnerable to healthcare inequity. We aimed to develop an Index of social vulnerability that predicts increased risk of negative healthcare events (e.g. emergency hospitalization), independent of chronic disease burden. The analysis illustrates the conceptual and practical steps leading to the development of a pragmatic Index of social vulnerability to limited healthcare self-efficacy. Methods: Using data from a 3-year cohort of 2507 adult primary care patients in Québec (Canada), we applied two complementary structural equation modelling approaches—Partial Least Squares Path Modelling (PLS-PM) and Multiple indicators and Multiple Causes (MIMIC) modelling—to identify a minimal set of social characteristics that could be summed into an Index related to limited healthcare self-efficacy. We then used logistic regression to determine if the Index predicted: hospital emergency department use; hospital admissions; unmet need for care, and others. We privileged parsimony over explanatory capacity in our analytic decisions to make the Index pragmatic for epidemiologic and clinical use. Results: The Individual Social Vulnerability Index is the weighted sum of five indicators: two indicators of social support; educational achievement; financial status; limited language proficiency. The Index predicts increased likelihood of all negative healthcare outcomes except unmet need, with a clear threshold at Index ≥ 2. The effect is independent of chronic disease burden. Conclusion: When social deficits outweigh social assets by two or more (Index ≥ 2), there is an increased risk of negative healthcare events beyond the risk attributable to poor health. The Index is a pragmatic tool to identify a minority of patients who will require additional support to receive equitable healthcare. [ABSTRACT FROM AUTHOR]

Simple Summary: Diagnosing most cancers at an earlier stage improves outcomes because treatments are more effective and often less invasive. This study looked at the health patterns of adults enrolled in Alberta's Tomorrow Project before they were diagnosed with cancer to identify factors related to cancers that are caught early or late. These identified factors were then tested in a similar group of adults from the British Columbia Generations Project who also subsequently developed cancer, to see how well they could predict stage at diagnosis. The confirmed health patterns can be used to improve cancer prevention programs and activities to find cancer earlier in people who are at most risk of being diagnosed with late-stage cancer. Risk prediction models for cancer stage at diagnosis may identify individuals at higher risk of late-stage cancer diagnoses. Partial proportional odds risk prediction models for cancer stage at diagnosis for males and females were developed using data from Alberta's Tomorrow Project (ATP). Prediction models were validated on the British Columbia Generations Project (BCGP) cohort using discrimination and calibration measures. Among ATP males, older age at diagnosis was associated with an earlier stage at diagnosis, while full- or part-time employment, prostate-specific antigen testing, and former/current smoking were associated with a later stage at diagnosis. Among ATP females, mammogram and sigmoidoscopy or colonoscopy were associated with an earlier stage at diagnosis, while older age at diagnosis, number of pregnancies, and hysterectomy were associated with a later stage at diagnosis. On external validation, discrimination results were poor for both males and females while calibration results indicated that the models did not over- or under-fit to derivation data or over- or under-predict risk. Multiple factors associated with cancer stage at diagnosis were identified among ATP participants. While the prediction model calibration was acceptable, discrimination was poor when applied to BCGP data. Updating our models with additional predictors may help improve predictive performance. [ABSTRACT FROM AUTHOR]

Type 1 diabetes mellitus (T1DM) patients are a significant threat to chronic kidney disease (CKD) development during their life. However, there is always a high chance of delay in CKD detection because CKD can be asymptomatic, and T1DM patients bypass traditional CKD tests during their routine checkups. This study aims to develop and validate a prediction model and nomogram of CKD in T1DM patients using readily available routine checkup data for early CKD detection. This research utilized 1375 T1DM patients' sixteen years of longitudinal data from multi-center Epidemiology of Diabetes Interventions and Complications (EDIC) clinical trials conducted at 28 sites in the USA and Canada and considered 17 routinely available features. Three feature ranking algorithms, extreme gradient boosting (XGB), random forest (RF), and extremely randomized trees classifier (ERT), were applied to create three feature ranking lists, and logistic regression analyses were performed to develop CKD prediction models using these ranked feature lists to identify the best performing top-ranked features combination. Finally, the most significant features were selected to develop a multivariate logistic regression-based CKD prediction model for T1DM patients. This model was evaluated using sensitivity, specificity, accuracy, precision, and F1 score on train and test data. A nomogram of the final model was further generated for easy application in clinical practices. Hypertension, duration of diabetes, drinking habit, triglycerides, ACE inhibitors, low-density lipoprotein (LDL) cholesterol, age, and smoking habit were the top-8 features ranked by the XGB model and identified as the most important features for predicting CKD in T1DM patients. These eight features were selected to develop the final prediction model using multivariate logistic regression, which showed 90.04% and 88.59% accuracy in internal and test data validation. The proposed model showed excellent performance and can be used for CKD identification in T1DM patients during routine checkups. [ABSTRACT FROM AUTHOR]

This paper investigates how to best compare algorithms for predicting chronic homelessness for the purpose of identifying good candidates for housing programs. Predictive methods can rapidly refer potentially chronic shelter users to housing but also sometimes incorrectly identify individuals who will not become chronic (false positives). We use shelter access histories to demonstrate that these false positives are often still good candidates for housing. Using this approach, we compare a simple threshold method for predicting chronic homelessness to the more complex logistic regression and neural network algorithms. While traditional binary classification performance metrics show that the machine learning algorithms perform better than the threshold technique, an examination of the shelter access histories of the cohorts identified by the three algorithms show that they select groups with very similar characteristics. This has important implications for resource constrained not-for-profit organizations since the threshold technique can be implemented using much simpler information technology infrastructure than the machine learning algorithms. [ABSTRACT FROM AUTHOR]

Background: Predictive cancer tools focus on survival; none predict severe symptoms. Aim: To develop and validate a model that predicts the risk for having low performance status and severe symptoms in cancer patients. Design: Retrospective, population-based, predictive study Setting/Participants: We linked administrative data from cancer patients from 2008 to 2015 in Ontario, Canada. Patients were randomly selected for model derivation (60%) and validation (40%). Using the derivation cohort, we developed a multivariable logistic regression model to predict the risk of an outcome at 6 months following diagnosis and recalculated after each of four annual survivor marks. Model performance was assessed using discrimination and calibration plots. Outcomes included low performance status (i.e. 10–30 on Palliative Performance Scale), severe pain, dyspnea, well-being, and depression (i.e. 7–10 on Edmonton Symptom Assessment System). Results: We identified 255,494 cancer patients (57% female; median age of 64; common cancers were breast (24%); and lung (13%)). At diagnosis, the predicted risk of having low performance status, severe pain, well-being, dyspnea, and depression in 6-months is 1%, 3%, 6%, 13%, and 4%, respectively for the reference case (i.e. male, lung cancer, stage I, no symptoms); the corresponding discrimination for each outcome model had high AUCs of 0.807, 0.713, 0.709, 0.790, and 0.723, respectively. Generally these covariates increased the outcome risk by >10% across all models: lung disease, dementia, diabetes; radiation treatment; hospital admission; pain; depression; transitional performance status; issues with appetite; or homecare. Conclusions: The model accurately predicted changing cancer risk for low performance status and severe symptoms over time. [ABSTRACT FROM AUTHOR]