How handling missing data may impact conclusions: A comparison of six different imputation methods for categorical questionnaire data

Prediction of Early Adverse Events After THA: A Comparison of Different Machine-Learning Strategies Based on 262,356 Observations From the Nordic Arthroplasty Register Association (NARA) Dataset

OBJECTIVE

Preoperative risk prediction models can support shared decision-making before total hip arthroplasties (THAs). Here, we compare different machine-learning (ML) approaches to predict the six-month risk of adverse events following primary THA to obtain accurate yet simple-to-use risk prediction models.

METHODS

We extracted data on primary THAs (N = 262,356) between 2010 and 2018 from the Nordic Arthroplasty Register Association dataset. We benchmarked a variety of ML algorithms in terms of the area under the receiver operating characteristic curve (AUROC) for predicting the risk of revision caused by periprosthetic joint infection (PJI), dislocation or periprosthetic fracture (PPF), and death. All models were internally validated against a randomly selected test cohort (one-third of the data) that was not used for training the models.

RESULTS

The incidences of revisions because of PJI, dislocation, and PPF were 0.8%, 0.4%, and 0.3%, respectively, and the incidence of death was 1.2%. Overall, Lasso regression with stable iterative variable selection (SIVS) produced models using only four to five input variables but with AUROC comparable to more complex models using all 32 variables available. The SIVS-based Lasso models based on age, sex, preoperative diagnosis, bearing couple, fixation, and surgical approach predicted the risk of revisions caused by PJI, dislocations, and PPF, as well as death, with AUROCs of 0.61, 0.67, 0.76, and 0.86, respectively.

CONCLUSION

Our study demonstrates that satisfactory predictive potential for adverse events following THA can be reached with parsimonious modeling strategies. The SIVS-based Lasso models may serve as simple-to-use tools for clinical risk assessment in the future.

Differences by ethnicity in the association between unpaid caring and health trajectories over 10 years in the UK Household Longitudinal Study

BACKGROUND

Unpaid carers deliver critical social care. We aimed to examine differences by ethnicity in (1) profiles of unpaid caring and (2) associations between caring and physical and mental health trajectories.

METHODS

We used 10 waves of data from 47 015 participants from the UK Household Longitudinal Study (2009-2020). Our outcomes were 12-item Short Form Health Survey physical and mental component scores. We performed bivariate comparison of profiles of caring by ethnicity. We used multilevel linear mixed effects models to estimate associations between caring and health trajectories and assess for heterogeneity by ethnicity.

RESULTS

We found that caring profiles differed by ethnicity. The proportion caring for someone within their household ranged from 39.7% of White carers to 70.1% of Pakistani and 74.8% of Bangladeshi carers. The proportion providing 20+ hours/week of care ranged from 26.9% of White carers to 40.6% of Pakistani and 43.3% of Black African carers. Ethnicity moderated associations between caring and physical but not mental health trajectories (test for interaction: p=0.038, p=0.75). Carers showed worse physical health compared with non-carers among Black African (-1.93; -3.52, -0.34), Bangladeshi (-2.01; -3.25, -0.78), Indian (-1.30; -2.33, -0.27) and Pakistani carers (-1.16; -2.25, -0.08); Bangladeshi carers' trajectories converged with non-carers over time (0.24; -0.02, 0.51). White carers showed better baseline physical health than non-carers (0.35; 0.10, 0.60), followed by worsening trajectories versus non-carers (-0.14; -0.18, -0.10).

CONCLUSIONS

There are differences by ethnicity in profiles of caring and associations between caring and physical health trajectories. Future research should account for ethnicity to ensure applicability across groups.

Development and validation of a nomogram for predicting advanced liver fibrosis in patients with chronic hepatitis B

Background

The progression of chronic hepatitis B (CHB) to liver fibrosis and even cirrhosis is often unknown to patients, but noninvasive markers capable of effectively identifying advanced liver fibrosis remains absent.

Objective

Based on the results of liver biopsy, we aimed to construct a new nomogram to validate the stage of liver fibrosis in CHB patients by the basic information of CHB patients and routine laboratory tests.

Methods

Patients with CHB diagnosed for the first time in the First Affiliated Hospital of Anhui Medical University from 2010 to 2018 were selected, and their basic information, laboratory tests and liver biopsy information were collected. Eventually, 974 patients were enrolled in the study, while all patients were randomized into a training cohort (n = 732) and an internal validation cohort (n = 242) according to a 3:1 ratio. In the training cohort, least absolute shrinkage and selection operator (Lasso) regression were used for predictor variable screening, and binary logistic regression analysis was used to build the diagnostic model, which was ultimately presented as a nomogram. The predictive accuracy of the nomograms was analyzed by running operating characteristic curve (ROC) to calculate area under curve (AUC), and the calibration was evaluated. Decision curve analysis (DCA) was used to determine patient benefit. In addition, we validated the built models with internal as well as external cohort (n = 771), respectively.

Results

Ultimately, the training cohort, the internal validation cohort, and the external validation cohort contained sample sizes of 188, 53, and 149, respectively, for advanced liver fibrosis. Gender, albumin (Alb), globulin (Glb), platelets (PLT), alkaline phosphatase (AKP), glutamyl transpeptidase (GGT), and prothrombin time (PT) were screened as independent predictors. Compared with the aminotransferase-to-platelet ratio index (APRI), fibrosis-4 index (FIB-4), and King's score, the model in the training cohort (AUC = 0.834, 95% CI 0.800-0.868, p < 0.05) and internal validation cohort (AUC = 0.804, 95% CI 0.742-0.866, p < 0.05) showed the best discrimination and the best predictive performance. In addition, DCA showed that the clinical benefit of the nomogram was superior to the APRI, FIB-4 and King's scores in all cohorts.

Conclusion

This study constructed a validated nomogram model with predictors screened from clinical variables which could be easily used for the diagnosis of advanced liver fibrosis in CHB patients.

Machine Learning-Based Risk Prediction of Discharge Status for Sepsis

As a severe inflammatory response syndrome, sepsis presents complex challenges in predicting patient outcomes due to its unclear pathogenesis and the unstable discharge status of affected individuals. In this study, we develop a machine learning-based method for predicting the discharge status of sepsis patients, aiming to improve treatment decisions. To enhance the robustness of our analysis against outliers, we incorporate robust statistical methods, specifically the minimum covariance determinant technique. We utilize the random forest imputation method to effectively manage and impute missing data. For feature selection, we employ Lasso penalized logistic regression, which efficiently identifies significant predictors and reduces model complexity, setting the stage for the application of more complex predictive methods. Our predictive analysis incorporates multiple machine learning methods, including random forest, support vector machine, and XGBoost. We compare the prediction performance of these methods with Lasso penalized logistic regression to identify the most effective approach. Each method's performance is rigorously evaluated through ten iterations of 10-fold cross-validation to ensure robust and reliable results. Our comparative analysis reveals that XGBoost surpasses the other models, demonstrating its exceptional capability to navigate the complexities of sepsis data effectively.

Sex Disparity in Stroke Mortality among Adults: A Time Series Analysis in the Greater Vitoria Region, Brazil (2000-2021)

The disparity between the sexes in stroke mortality has been demonstrated in people from different locations. The objective of this study was to analyze the disparity between sexes in stroke mortality in adults in the metropolitan area of Greater Vitoria between 2000 and 2021. Ecological time series design was conducted with a database of the Brazilian Health System Informatics Department. The annual percentage change and average annual percentage change were calculated through joinpoint regression. Pairwise comparisons using parallelism and coincidence tests were applied to compare temporal trends between men and women. Men had higher mortality rates in most years between 2000 and 2021. In contrast, women had higher proportional mortality values in all years evaluated from 2000 to 2021. The paired comparison revealed a disparity between the sexes in the proportional mortality time series (parallelism test: p = 0.003; coincidence test: p < 0.001). However, the time series of the mortality rates showed no disparity between the sexes (parallelism test: p = 0.114; coincidence test: p = 0.093). From 2000 to 2021, there was a disparity in proportional mortality from stroke between the sexes of the population in the metropolitan area of Greater Vitoria, Brazil. However, the time series of mortality rates between the sexes did not reveal any disparity in the study period.

IRTCI: Item Response Theory for Categorical Imputation

Most datasets suffer from partial or complete missing values, which has downstream limitations on the available models on which to test the data and on any statistical inferences that can be made from the data. Several imputation techniques have been designed to replace missing data with stand in values. The various approaches have implications for calculating clinical scores, model building and model testing. The work showcased here offers a novel means for categorical imputation based on item response theory (IRT) and compares it against several methodologies currently used in the machine learning field including k-nearest neighbors (kNN), multiple imputed chained equations (MICE) and Amazon Web Services (AWS) deep learning method, Datawig. Analyses comparing these techniques were performed on three different datasets that represented ordinal, nominal and binary categories. The data were modified so that they also varied on both the proportion of data missing and the systematization of the missing data. Two different assessments of performance were conducted: accuracy in reproducing the missing values, and predictive performance using the imputed data. Results demonstrated that the new method, Item Response Theory for Categorical Imputation (IRTCI), fared quite well compared to currently used methods, outperforming several of them in many conditions. Given the theoretical basis for the new approach, and the unique generation of probabilistic terms for determining category belonging for missing cells, IRTCI offers a viable alternative to current approaches.

The impact of co-occurring mental health problems on referral to and initiation of treatment among youth under probation supervision: Findings from a cluster randomized trial

INTRODUCTION

Many youth under community supervision have substance use and co-occurring mental health issues. Yet, access to treatment is limited, and many programs cannot address co-occurring disorders. This study examines how co-occurring symptoms among youth on probation affect referral to and initiation of treatment. We hypothesize that both referral and initiation rates will be lower for youth with any co-occurring indicators.

METHODS

This study collected administrative data from 14 sites in three states between March 2014 and November 2017 using JJ-TRIALS, a cluster randomized trial. Among 8552 youth in need of treatment (screened as having a substance use problem, drug possession arrest, positive drug test, etc.), 2069 received a referral to treatment and 1630 initiated treatment among those referred. A co-occurring indicator (n = 2828) was based on symptoms of an internalizing and/or externalizing issue. Descriptive analyses compared referral and initiation by behavioral health status. Two-level mixed effects logistic regression models estimated effects of site-level variables.

RESULTS

Among youth in need with co-occurring internal, external, or both indicators, only 16 %, 18 %, and 20 % were referred to treatment and of those referred, 63 %, 69 %, and 57 % initiated treatment, respectively. Comparatively, 27 % and 83 % of youth with a substance use only indicator were referred and initiated treatment respectively. Multi-level multivariate models found that, contrary to our hypothesis, co-occurring-both (p = 0.00, OR 1.44) and co-occurring-internal indicators (p = 0.06, OR 1.25) predicted higher referral but there were no differences in initiation rates. However, there was substantial site-level variation.

CONCLUSIONS

Youth on probation in need of substance use treatment with co-occurring issues have low referral rates. Behavioral health status may influence youth referral to treatment depending on where a youth is located. Depending on the site, there may be a lack of community programs that can adequately treat youth with co-occurring issues and reduce unmet service needs.

Comparison of the effects of imputation methods for missing data in predictive modelling of cohort study datasets

BACKGROUND

Missing data is frequently an inevitable issue in cohort studies and it can adversely affect the study's findings. We assess the effectiveness of eight frequently utilized statistical and machine learning (ML) imputation methods for dealing with missing data in predictive modelling of cohort study datasets. This evaluation is based on real data and predictive models for cardiovascular disease (CVD) risk.

METHODS

The data is from a real-world cohort study in Xinjiang, China. It includes personal information, physical examination data, questionnaires, and laboratory biochemical results from 10,164 subjects with a total of 37 variables. Simple imputation (Simple), regression imputation (Regression), expectation-maximization(EM), multiple imputation (MICE) , K nearest neighbor classification (KNN), clustering imputation (Cluster), random forest (RF), and decision tree (Cart) were the chosen imputation methods. Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) are utilised to assess the performance of different methods for missing data imputation at a missing rate of 20%. The datasets processed with different missing data imputation methods were employed to construct a CVD risk prediction model utilizing the support vector machine (SVM). The predictive performance was then compared using the area under the curve (AUC).

RESULTS

The most effective imputation results were attained by KNN (MAE: 0.2032, RMSE: 0.7438, AUC: 0.730, CI: 0.719-0.741) and RF (MAE: 0.3944, RMSE: 1.4866, AUC: 0.777, CI: 0.769-0.785). The subsequent best performances were achieved by EM, Cart, and MICE, while Simple, Regression, and Cluster attained the worst performances. The CVD risk prediction model was constructed using the complete data (AUC:0.804, CI:0.796-0.812) in comparison with all other models with p<0.05.

CONCLUSION

KNN and RF exhibit superior performance and are more adept at imputing missing data in predictive modelling of cohort study datasets.

Development and validation of multicentre study on novel Artificial Intelligence-based Cardiovascular Risk Score (AICVD)

OBJECTIVE

Cardiovascular diseases (CVD) are one of the most prevalent diseases in India amounting for nearly 30% of total deaths. A dearth of research on CVD risk scores in Indian population, limited performance of conventional risk scores and inability to reproduce the initial accuracies in randomised clinical trials has led to this study on large-scale patient data. The objective is to develop an Artificial Intelligence-based Risk Score (AICVD) to predict CVD event (eg, acute myocardial infarction/acute coronary syndrome) in the next 10 years and compare the model with the Framingham Heart Risk Score (FHRS) and QRisk3.

METHODS

Our study included 31 599 participants aged 18-91 years from 2009 to 2018 in six Apollo Hospitals in India. A multistep risk factors selection process using Spearman correlation coefficient and propensity score matching yielded 21 risk factors. A deep learning hazards model was built on risk factors to predict event occurrence (classification) and time to event (hazards model) using multilayered neural network. Further, the model was validated with independent retrospective cohorts of participants from India and the Netherlands and compared with FHRS and QRisk3.

RESULTS

The deep learning hazards model had a good performance (area under the curve (AUC) 0.853). Validation and comparative results showed AUCs between 0.84 and 0.92 with better positive likelihood ratio (AICVD -6.16 to FHRS -2.24 and QRisk3 -1.16) and accuracy (AICVD -80.15% to FHRS 59.71% and QRisk3 51.57%). In the Netherlands cohort, AICVD also outperformed the Framingham Heart Risk Model (AUC -0.737 vs 0.707).

CONCLUSIONS

This study concludes that the novel AI-based CVD Risk Score has a higher predictive performance for cardiac events than conventional risk scores in Indian population.

TRIAL REGISTRATION NUMBER

CTRI/2019/07/020471.

Studying missingness in spinal cord injury data: challenges and impact of data imputation

BACKGROUND

In the last decades, medical research fields studying rare conditions such as spinal cord injury (SCI) have made extensive efforts to collect large-scale data. However, most analysis methods rely on complete data. This is particularly troublesome when studying clinical data as they are prone to missingness. Often, researchers mitigate this problem by removing patients with missing data from the analyses. Less commonly, imputation methods to infer likely values are applied.

OBJECTIVE

Our objective was to study how handling missing data influences the results reported, taking the example of SCI registries. We aimed to raise awareness on the effects of missing data and provide guidelines to be applied for future research projects, in SCI research and beyond.

METHODS

Using the Sygen clinical trial data (n = 797), we analyzed the impact of the type of variable in which data is missing, the pattern according to which data is missing, and the imputation strategy (e.g. mean imputation, last observation carried forward, multiple imputation).

RESULTS

Our simulations show that mean imputation may lead to results strongly deviating from the underlying expected results. For repeated measures missing at late stages (> = 6 months after injury in this simulation study), carrying the last observation forward seems the preferable option for the imputation. This simulation study could show that a one-size-fit-all imputation strategy falls short in SCI data sets.

CONCLUSIONS

Data-tailored imputation strategies are required (e.g., characterisation of the missingness pattern, last observation carried forward for repeated measures evolving to a plateau over time). Therefore, systematically reporting the extent, kind and decisions made regarding missing data will be essential to improve the interpretation, transparency, and reproducibility of the research presented.

Statistical Methods for Pre-Post Intervention Design

The author provides an overview of options for statistical tests in pre-post analysis.

The impact of mortality underreporting on the association of ambient temperature and PM10 with mortality risk in time series study

Properly analyzing and reporting data remains a challenging task in epidemiologic research, as underreporting of data is often overlooked. The evaluation on the effect of underreporting remains understudied. In this study, we examined the effect of different scenarios of mortality underreporting on the relationship between PM₁₀, temperature, and mortality. Mortality data, PM₁₀, and temperature data in seven cities were obtained from Provincial Center for Disease Control and Prevention (CDC), China Meteorological Data Sharing Service System, and China National Environmental Monitoring Center, respectively. A time-series design with a distributed lag nonlinear model (DLNM) was used to examine the effects of five mortality underreporting scenarios: 1) Random underreporting of mortality; 2) Underreporting is monotonically increasing (MI) or monotonically decreasing (MD); 3) Underreporting due to holiday and weekends; 4) Underreporting occurs before the 20th day of each month, and these underreporting will be added after the 20th day of the month; and 5) Underreporting due to holiday, weekends, MI, and MD. We observed that underreporting at random (UAR) scenario had little effect on the association between PM₁₀, temperature, and daily mortality. However, other four underreporting not at random (UNAR) scenarios mentioned above had varying degrees of influence on the association between PM₁₀, temperature, and daily mortality. Additionally, in addition to imputation under UAR, the variation of minimum mortality temperature (MMT) and attributable fraction (AF) of mortality attributed to temperature in the same imputation scenarios is inconsistent in different cities. Finally, we observed that the pooled excess risk (ER) below MMT was negatively associated with mortality and the pooled ER above MMT was positively associated with mortality. This study showed that UNAR impacted the association between PM₁₀, temperature, and mortality, and potential underreporting should be dealt with before analyzing data to avoid drawing invalid conclusions.

Sociodemographic disparities in preterm birth and low birthweight in the State of Georgia: Results from the 2017-2018 Pregnancy Risk Assessment Monitoring System

PURPOSE

To update the overall prevalence of preterm birth (PTB) (<37 weeks gestation) and low birthweight (LBW) (<2,500 g) in the State of Georgia, including rural and urban counties.

METHODS

A sample was drawn from the 2017-2018 Georgia Pregnancy Risk Assessment Monitoring System (PRAMS). In the complete-case data of singleton births (n=1,258), we estimated the weighted percentage prevalence of PTB, LBW, early/late PTB, and moderately/very LBW subcategories in association with maternal sociodemographic characteristics, and the prevalence stratified by rural/urban county of residence. Univariate and multivariate logistic regression models were fitted to estimate the odds ratios (ORs) of PTB and LBW adjusting for selected covariates. Logistic regression results from multiple imputation by chained equations (MICE) were used for comparison.

FINDINGS

The overall rate for PTB was 9.3% and 6.8% for LBW and among them, 2.3% were early PTB, 7.0% were late PTB, 5.4% were moderately LBW (MLBW), and 1.3% were very LBW (VLBW). Non-Hispanic Black women had the highest prevalence of PTB, LBW, early PTB, MLBW, and VLBW, as well as PTB and LBW in urban counties and LBW in rural counties. The odds of PTB (aOR 1.38; 95% CI: 0.81, 2.35) and LBW (aOR 2.68; 95% CI: 1.32, 5.43) were also higher among non-Hispanic Black relative to non-Hispanic White women and among women who received adequate-plus prenatal care compared to inadequate prenatal care.

CONCLUSIONS

Socioeconomic and health disparities created by disadvantage should be a focus of state policy to improve neonatal outcomes in the State of Georgia.

Importance of missingness in baseline variables: A case study of the All of Us Research Program

OBJECTIVE

The All of Us Research Program collects data from multiple information sources, including health surveys, to build a national longitudinal research repository that researchers can use to advance precision medicine. Missing survey responses pose challenges to study conclusions. We describe missingness in All of Us baseline surveys.

STUDY DESIGN AND SETTING

We extracted survey responses between May 31, 2017, to September 30, 2020. Missing percentages for groups historically underrepresented in biomedical research were compared to represented groups. Associations of missing percentages with age, health literacy score, and survey completion date were evaluated. We used negative binomial regression to evaluate participant characteristics on the number of missed questions out of the total eligible questions for each participant.

RESULTS

The dataset analyzed contained data for 334,183 participants who submitted at least one baseline survey. Almost all (97.0%) of the participants completed all baseline surveys, and only 541 (0.2%) participants skipped all questions in at least one of the baseline surveys. The median skip rate was 5.0% of the questions, with an interquartile range (IQR) of 2.5% to 7.9%. Historically underrepresented groups were associated with higher missingness (incidence rate ratio (IRR) [95% CI]: 1.26 [1.25, 1.27] for Black/African American compared to White). Missing percentages were similar by survey completion date, participant age, and health literacy score. Skipping specific questions were associated with higher missingness (IRRs [95% CI]: 1.39 [1.38, 1.40] for skipping income, 1.92 [1.89, 1.95] for skipping education, 2.19 [2.09-2.30] for skipping sexual and gender questions).

CONCLUSION

Surveys in the All of Us Research Program will form an essential component of the data researchers can use to perform their analyses. Missingness was low in All of Us baseline surveys, but group differences exist. Additional statistical methods and careful analysis of surveys could help mitigate challenges to the validity of conclusions.

Machine Learning in Cardiac Surgery: Predicting Mortality and Readmission

Predicting outcomes in open-heart surgery can be challenging. Unexpected readmissions, long hospital stays, and mortality have economic implications. In this study, we investigated machine learning (ML) performance in data visualization and predicting patient outcomes associated with open-heart surgery. We evaluated 8,947 patients who underwent cardiac surgery from April 2006 to January 2018. Data visualization and classification were performed at cohort-level and patient-level using clustering, correlation matrix, and seven different predictive models for predicting three outcomes ("Discharged," "Died," and "Readmitted") at binary level. Cross-validation was used to train and test each dataset with the application of hyperparameter optimization and data imputation techniques. Machine learning showed promising performance for predicting mortality (AUC 0.83 ± 0.03) and readmission (AUC 0.75 ± 0.035). The cohort-level analysis revealed that ML performance is comparable to the Society of Thoracic Surgeons (STS) risk model even with limited number of samples ( e.g. , less than 3,000 samples for ML versus more than 100,000 samples for the STS risk models). With all cases (8,947 samples, referred as patient-level analysis), ML showed comparable performance to what has been reported for the STS models. However, we acknowledge that it remains unknown at this stage as to how the model might perform outside the institution and does not in any way constitute a comparison of the performance of the internal model with the STS model. Our study demonstrates a systematic application of ML in analyzing and predicting outcomes after open-heart surgery. The predictive utility of ML in cardiac surgery and clinical implications of the results are highlighted.

Determinants of continuance intention of using e-government services in Tanzania: the role of system interactivity as moderating factor

Purpose

The purpose of this study is to investigate the continuance intention of using e-government services in Tanzania as well as moderating effects of system interactivity.

Design/methodology/approach

A research model based on expectancy confirmation model was developed and empirically tested using 213 data collected from e-government services users who were selected using the judgemental sampling technique. The variance-based structural equation modelling technique was used for data analysis using SmartPLS 3.0.

Findings

The results of this study suggest that system interactivity, computer self-efficacy, management support, confirmation, satisfaction and perceived usefulness have a positive and significant influence on continuance intention to use e-government services. Moreover, the findings of this study indicate that system interactivity moderates the influence of perceived usefulness and satisfaction on continuance intention.

Originality/value

This study extends the expectancy confirmation model with system interactivity, management support and computer self-efficacy which are considered as important factors in continuance usage of technology. Furthermore, this study examines the moderating effect of system interactivity on the effects of perceived usefulness and satisfaction on continuance intention.

Using structured implementation interventions to improve referral to substance use treatment among justice-involved youth: Findings from a multisite cluster randomized trial

INTRODUCTION

Youth involved in the justice system have high rates of alcohol and other drug use, but limited treatment engagement. JJ-TRIALS tested implementation activities with community supervision (CS) and behavioral health (BH) agencies to improve screening, identification of substance use service need, referral, and treatment initiation and engagement, guided by the BH Services Cascade and EPIS frameworks. This paper summarizes intervention impacts on referrals to treatment among youth on CS.

METHODS

This multisite cluster-randomized trial involved 18 matched pairs of sites in 36 counties in seven states randomly assigned to core or enhanced conditions after implementing the core intervention at all sites for six months. Enhanced sites received external facilitation for local change team activities to reduce unmet treatment needs; Core sites were encouraged to form interagency workgroups. The dependent variable was percentage referred to treatment among youth in need (N = 14,012). Two-level Bayesian regression assessed factors predicting referral across all sites and time periods. Generalized linear mixed models using logit transformation tested two hypotheses: (H1) referrals will increase from baseline to the experimental period, (H2) referral increases will be larger in enhanced sites than in core sites.

RESULTS

Although the intervention significantly increased referral, condition did not significantly predict referral across all time periods. Youth who tested drug positive, had an alcohol/other drug-related or felony charge, were placed in secure detention or assigned more intensive supervision, or who were White were more likely to be referred. H1 (p < .05) and H2 (p < .0001) were both significant in the hypothesized direction. Interaction analyses comparing site pair differences showed that findings were not consistent across sites.

CONCLUSIONS

The percentage of youth referred to treatment increased compared with baseline overall, and enhanced sites showed larger increases in referrals over time. However, variations in effects suggest that site-level differences were important. Researchers should carry out mixed methods studies to further understand reasons for the inconsistent findings within randomized site pairs, and how to further improve treatment referrals across CS and BH systems. Findings also highlight that even when CS agencies work collaboratively with BH providers to improve referrals, most justice-involved youth who need SU services are not referred.

Longitudinal Patterns of Material Hardship among US Families

Material hardship has emerged as a direct measure of deprivation in the United States and an important complement to income poverty, providing different evidence about the ways in which deprivation may affect wellbeing. This study addresses gaps in our knowledge about deprivation as the first to examine patterns of material hardship over time. Using data from the Fragile Families and Child Well-Being Study, this study examined five material hardship types (food, housing, medical, utility, and bill-paying) experienced at five timepoints over 15 years. Employing latent class analysis and latent transition analysis, this study identified six longitudinal patterns of material hardship experience, characterized by trajectories of stability or movement and relative severity of material hardship experience over time. These findings improve our conceptual understanding of deprivation and move us towards understanding the impacts of material hardship on wellbeing and identifying policy approaches to prevent deprivation or mitigate negative consequences.

Preventive planning against mercury over-exposure among Moroccan dentists using multidimensional statistical methods

BACKGROUND: Mercury used in dental amalgams constitutes a significant source of chronic exposure to this heavy metal among dentists. Thus, the safety of dental amalgam remains a controversial issue despite its long history of use. In Morocco, most studies about dental mercury were mainly focused on the environmental risk related to the management of mercury-contaminated waste. OBJECTIVE: In order to evaluate the occupational exposure to mercury among liberal dentists practicing in two Moroccan regions, a multidimensional statistical approach was used to analyze the collected data. The main objective was to help establishing a targeted prevention plan aiming to reduce the mercury exposure among Moroccan dentists. METHODS: Fifteen variables from 146 dentists were elected for a three-step classification procedure: a multiple correspondence analysis followed by a hierarchical ascendant clustering consolidated by the k-Means algorithm. RESULTS: Three homogenous clusters were identified. The most important one includes 57.5% of the population as well as the majority of the risky factors. The characterization of these clusters allows proposing concise guidelines for a targeted preventive plan. CONCLUSIONS: A real mercurial risk has been observed in the studied population. However, its impact on health as well as the efficiency of simple preventive recommendations remains to be unveiled.

Evaluating a Sport-Based Mental Health Literacy Intervention in Australian Amateur Sporting Adolescents

Youth amateur sporting environments present an untapped, under-researched, and potentially vital avenue for targeted mental health intervention programs. This study evaluates such an intervention in 12 sporting clubs, comprising of 330 Australian youth aged 12-15 years (M = 13.73, SD = 0.79). Mental health literacy, help-seeking intentions, and help-seeking behaviors were measured throughout the season using a repeated-measures experimental-control design. Multilevel modelling revealed the intervention successfully improved mental health literacy and help-seeking intentions in particular cohorts, such as youth scoring low in these constructs pre-intervention and youth who had not previously received the intervention. This study demonstrates the efficacy of interventions to effect positive change in amateur sporting youth, highlighting a convenient method to improve mental health in young people.

Handling missing data in a rheumatoid arthritis registry using random forest approach

Missing data in clinical epidemiological research violate the intention-to-treat principle, reduce the power of statistical analysis, and can introduce bias if the cause of missing data is related to a patient's response to treatment. Multiple imputation provides a solution to predict the values of missing data. The main objective of this study is to estimate and impute missing values in patient records. The data from the Kuwait Registry for Rheumatic Diseases was used to deal with missing values among patient records. A number of methods were implemented to deal with missing data; however, choosing the best imputation method was judged by the lowest root mean square error (RMSE). Among 1735 rheumatoid arthritis patients, we found missing values vary from 5% to 65.5% of the total observations. The results show that sequential random forest method can estimate these missing values with a high level of accuracy. The RMSE varied between 2.5 and 5.0. missForest had the lowest imputation error for both continuous and categorical variables under each missing data rate (10%, 20%, and 30%) and had the smallest prediction error difference when the models used the imputed laboratory values.

Machine Learning Models for Classification of Cushing's Syndrome Using Retrospective Data

Accurate classification of Cushing's Syndrome (CS) plays a critical role in providing the early and correct diagnosis of CS that may facilitate treatment and improve patient outcomes. Diagnosis of CS is a complex process, which requires careful and concurrent interpretation of signs and symptoms, multiple biochemical test results, and findings of medical imaging by physicians with a high degree of specialty and knowledge to make correct judgments. In this article, we explore the state of the art machine learning algorithms to demonstrate their potential as a clinical decision support system to analyze and classify CS to facilitate the diagnosis, prognosis, and treatment of CS. Prominent algorithms are compared using nested cross-validation and various class comparison strategies including multiclass, one vs. all, and one vs. one binary classification. Our findings show that Random Forest (RF) algorithm is most suitable for the classification of CS. We demonstrate that the proposed approach can classify CS with an average accuracy of 92% and an average F1 score of 91.5%, depending on the class comparison strategy and selected features. RF-based one vs. all binary classification model achieves sensitivity of 97.6%, precision of 91.1%, and specificity of 87.1% to discriminate CS from non-CS on the test dataset. RF-based multiclass classification model achieves average per class sensitivity of 91.8%, average per class specificity of 97.1%, and average per class precision of 92.1% to classify different subtypes of CS on the test dataset. Clinical performance evaluation suggests that the developed models can help improve physicians' judgment in diagnosing CS.

Validation of a Tool to Evaluate Drug Prevention Programs Among Students

Background: School-based prevention programs have been implemented worldwide with the intention of reducing or delaying the onset of alcohol and drug use among adolescents. However, their effects need to be evaluated, being essential to use validated and reliable questionnaires for this purpose. This study aimed to verify the semantic validity and reliability of an instrument developed to evaluate the results of a government drug prevention program for schoolchildren called #Tamojunto2.0. Methods: This is a mixed methods study with quantitative (test-retest, confirmatory factor analysis and non-response evaluation) and qualitative analyses (focus group and field cards). The self-administered questionnaires were used for a sample of 262 eighth-grade students (elementary school II) in 11 classes of four public schools in the city of São Paulo. Results: The level of agreement was substantial (Kappa 0.60-0.79) or almost perfect (Kappa > 0.8) for almost all questions about the use of marijuana, alcohol, cigarettes, cocaine, crack, and binge drinking. The model fit indices, for almost all secondary outcomes, indicated that the modls underlying each scale, constituted by observed and latent variables, had a good fit adjustument. The focus groups and field cards provided high-quality information that helped the researchers identify the main difficulties in applying and understanding the questions. Conclusion: The questionnaire showed high factorial validity, reliability and understanding by adolescents. After the necessary changes, identified in this study, the questionnaire will be suitable to evaluate the results of the #Tamojunto2.0 program in a randomized controlled trial.

Exploring the midline soft tissue surface changes from 12 to 15 years of age in three distinct country population cohorts

INTRODUCTION

Several studies have highlighted differences in the facial features in a White European population. Genetics appear to have a major influence on normal facial variation, and environmental factors are likely to have minor influences on face shape directly or through epigenetic mechanisms.

AIM

The aim of this longitudinal cohort study is to determine the rate of change in midline facial landmarks in three distinct homogenous population groups (Finnish, Latvian, and Welsh) from 12.8 to 15.3 years of age. This age range covers the pubertal growth period for the majority of boys and girls.

METHODS

A cohort of children aged 12 were monitored for facial growth in three countries [Finland (n = 60), Latvia (n = 107), and Wales (n = 96)]. Three-dimensional facial surface images were acquired (using either laser or photogrammetric methods) at regular intervals (6-12 months) for 4 years. Ethical approval was granted in each country. Nine midline landmarks were identified and the relative spatial positions of these surface landmarks were measured relative to the mid-endocanthion (men) over a 4-year period.

RESULTS

This study reports the children who attended 95 per cent of all scanning sessions (Finland 48 out of 60; Latvia 104 out of 107; Wales 50 out of 96). Considerable facial variation is seen for all countries and sexes. There are clear patterns of growth that show different magnitudes at different age groups for the different country groups, sexes, and facial parameters. The greatest single yearly growth rate (5.4 mm) was seen for Welsh males for men-pogonion distance at 13.6 years of age. Males exhibit greater rates of growth compared to females. These variations in magnitude and timings are likely to be influenced by genetic ancestry as a result of population migration.

CONCLUSION

The midline points are a simple and valid method to assess the relative spatial positions of facial surface landmarks. This study confirms previous reports on the subtle differences in facial shapes and sizes of male and female children in different populations and also highlights the magnitudes and timings of growth for various midline landmark distances to the men point.

Initiation of Transdermal Fentanyl Among US Commercially Insured Patients Between 2007 and 2015

INTRODUCTION

This study examined patterns of initial transdermal fentanyl (TDF) claims among US commercially insured patients and explored the risk of 30-day hospitalization among patients with and without prior opioid exposure necessary to produce tolerance.

DESIGN

A retrospective cohort study of initial outpatient TDF prescriptions.

SETTING

A 10% random sample of commercially insured enrollees within the IQVIA Health Plan Claims Database (formerly known as PharMetrics Plus).

SUBJECTS

Individuals with a claim for TDF between 2007 and 2015.

METHODS

The primary exposure was a new transdermal fentanyl claim, and the primary outcome was guideline concordance based on time and dose exposure.

RESULTS

Among the 24,770 patients in the cohort, 4,848 (20%) patients had sufficient time exposure to opioids before TDF. Among those with sufficient time exposure, 3,971 (82%) had adequate opioid exposure based on the US Food and Drug Administration (FDA) package insert dosing guidance. Overall, 3,971 of the 24,770 (16%) patients received guideline-consistent TDF. An exploratory analysis of 30-day hospitalization after a TDF claim did not detect a difference in odds between guideline-consistent or -inconsistent groups when adjusted for variables known to influence the risk of opioid-induced respiratory depression.

CONCLUSIONS

A majority of patients met FDA opioid dose thresholds for TDF but had insufficient time exposure based on package insert recommendations for tolerance. Exploratory analysis did not detect a difference in odds for all-cause hospitalization or respiratory-related 30-day hospitalization between guideline-consistent or -inconsistent TDF claims. Prescribers should continue to adhere to FDA TDF labeling, although certain aspects of the labeling should be reevaluated or clarified.

Early Release Estimates for SARS-CoV-2 Prevalence and Antibody Response Interim Weighting for Probability-Based Sample Surveys.. [PMID: 32995810 PMCID: PMC7523149 DOI: 10.1101/2020.09.15.20195099] [Citation(s) in RCA: 0] [Impact Index Per Article: 0]

Many months into the SARS-CoV-2 pandemic, basic epidemiologic parameters describing burden of disease are lacking. To reduce selection bias in current burden of disease estimates derived from diagnostic testing data or serologic testing in convenience samples, we are conducting a national probability-based sample SARS-CoV-2 serosurvey. Sampling from a national address-based frame and using mailed recruitment materials and test kits will allow us to estimate national prevalence of SARS-CoV-2 infection and antibodies, overall and by demographic, behavioral, and clinical characteristics. Data will be weighted for unequal selection probabilities and non-response and will be adjusted to population benchmarks. Due to the urgent need for these estimates, expedited interim weighting of serosurvey responses will be undertaken to produce early release estimates, which will be published on the study website, COVIDVu.org. Here, we describe a process for computing interim survey weights and guidelines for release of interim estimates.

Dissemination and implementation of the e-MCHHandbook, UNRWA's newly released maternal and child health mobile application: a cross-sectional study

OBJECTIVES

In April 2017, the United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNRWA) released the electronic Maternal andChildHealth Handbook, the e-MCH Handbook application. One of the first mobile health (m-Health) interventions in a refugee setting, the application gives pregnant women and mothers access to educational information and health records on smartphones. This study investigated factors associated with the dissemination and implementation of m-Health in the refugee setting.

SETTING AND PARTICIPANTS

A cross-sectional study was conducted in 9 of 25 UNRWA health centres for Palestine refugees in Jordan. Self-administered questionnaires were distributed for 1 week to pregnant women and mothers with children aged 0-5 years.

OUTCOME MEASURES

The outcomes were whether participants knew about, downloaded or used the application. Multiple regression analyses were conducted to determine factors associated with application download and usage.

RESULTS

1042 participants were included in the analysis. 979 (95.5%) had a mobile phone and 862 (86.9%) had a smartphone. 499 (51.3%) knew about, 235 (23.8%) downloaded and 172 (17.4%) used the application. Having other mobile applications (OR 6.17, p<0.01), staff knowledge of the application (OR 11.82, p<0.01), using the internet as a source of medical information (OR 1.63, p=0.01) and having internet access at home (OR 1.46, p=0.05) were associated with application download. The age of the husband was associated with application usage (OR 1.04, p=0.11).

CONCLUSIONS

Though m-Health may be a promising means of promoting health in refugees, multiple barriers may exist to its dissemination and implementation. Those who regularly use mobile applications and get medical information from the internet are potential targets of m-Health dissemination. For successful implementation of a m-Health intervention, health staff should have thorough knowledge of the application and users should have access to the internet. Husband-related factors may also play a role.

Life expectancy with and without cognitive impairment among Chilean older adults: results of the National Survey of Health (2003, 2009 and 2016)

Background

Chile has one of the highest life expectancies within Latin American. This is the first study to determine health expectancies in older populations in Chile, considering cognitive status as a health indicator.

Methods

We estimated prevalence of cognitive decline among people aged 60 years and over based on the Mini-mental State Examination and the Pfeffer Functional Activities Questionnaire, with data from the National Survey of Health (2003, 2009, 2016). Life expectancy free of cognitive impairment was calculated using the Sullivan method.

Results

At age 60, life expectancy free of cognitive impairment was more than 3 years longer for women, compared to men of the same age. Life expectancy free from cognitive impairment was higher for both men and women aged 60 in 2016 when compared to 2003 (2.1 and 2 years higher, respectively).

Conclusions

Longer life expectancy in women was accompanied by more years free of cognitive impairment. Men expected to live a similar proportion of years free of cognitive impairment, compared to women. Common and standardised assessments of health status of older people should be adopted in Latin American studies, to allow for time-trend analyses and international comparisons.

The clinical consequences of variable selection in multiple regression models: a case study of the Norwegian Opioid Maintenance Treatment program

Background: Selecting which variables to include in multiple regression models is a pervasive problem in medical research.Objectives: Based on questionnaire data (n = 18538, 69.9% men) from the Norwegian Opioid Maintenance Treatment Program, this study aims to compare the performance of different variable selection methods and the potential clinical consequences of choice of method. The effect of missing data is also explored.Methods: The dependent variable was engagement in criminal behavior while in treatment. Twenty-nine potential covariates on demographics, psychosocial factors and drug use were tested for inclusion in a multiple logistic regression model. Both complete case and multiply imputed data were considered. We compared the results from variable selection methods ranging from expert-based and purposeful variable selection, through stepwise methods, to more recently developed penalized regression using the Least Absolute Shrinkage and Selection Operator (LASSO).Results: The various variable selection methods resulted in regression models including from 9 to 22 covariates. The stepwise selection procedures generated the models with the most covariates included. The choice of variable selection method directly affected the estimated regression coefficients, both in effect size and statistical significance. For several variables the expert-based approach disagreed with all data-driven methods.Conclusions: The choice of variable selection method may strongly affect the resulting regression model, along with accompanying effect sizes and confidence intervals. This may affect clinical conclusions. The process should consequently be given sufficient consideration in model building. We recommend combining expert knowledge with a data-driven variable selection method to explore the models' robustness.