Handling Of Missing Data Research Articles

Management of Mental Health Disorders, Substance Use Disorders, and Suicide in Adults with Spinal Cord Injury: Clinical Practice Guideline for Healthcare Providers.

Management of Mental Health Disorders, Substance Use Disorders, and Suicide in Adults with Spinal Cord Injury: Clinical Practice Guideline for Healthcare Providers.

Machine Learning Data Imputation and Prediction of Foraging Group Size in a Kleptoparasitic Spider

Cost–benefit analysis is widely used to elucidate the association between foraging group size and resource size. Despite advances in the development of theoretical frameworks, however, the empirical systems used for testing are hindered by the vagaries of field surveys and incomplete data. This study developed the three approaches to data imputation based on machine learning (ML) algorithms with the aim of rescuing valuable field data. Using 163 host spider webs (132 complete data and 31 incomplete data), our results indicated that the data imputation based on random forest algorithm outperformed classification and regression trees, the k-nearest neighbor, and other conventional approaches (Wilcoxon signed-rank test and correlation difference have p-value from < 0.001–0.030). We then used rescued data based on a natural system involving kleptoparasitic spiders from Taiwan and Vietnam (Argyrodes miniaceus, Theridiidae) to test the occurrence and group size of kleptoparasites in natural populations. Our partial least-squares path modelling (PLS-PM) results demonstrated that the size of the host web (T = 6.890, p = 0.000) is a significant feature affecting group size. The resource size (T = 2.590, p = 0.010) and the microclimate (T = 3.230, p = 0.001) are significant features affecting the presence of kleptoparasites. The test of conformation of group size distribution to the ideal free distribution (IFD) model revealed that predictions pertaining to per-capita resource size were underestimated (bootstrap resampling mean slopes <IFD predicted slopes, p < 0.001). These findings highlight the importance of applying appropriate ML methods to the handling of missing field data.

Characterization of immunization secondary analyses using demographic and health surveys (DHS) and multiple indicator cluster surveys (MICS), 2006\u20132018

BackgroundInfant immunization coverage worldwide has plateaued at about 85%. Using existing survey data to conduct analyses beyond estimating coverage may help immunization programmes better tailor strategies to reach un- and under-immunized children. The Demographic and Health Survey (DHS) and the Multiple Indicators Cluster Survey (MICS), routinely conducted in low and middle-income countries (LMICs), collect immunization data, yet vaccination coverage is often the only indicator reported and used. We conducted a review of published immunization-related analyses to characterize and quantify immunization secondary analyses done using DHS and MICS databases.MethodsWe conducted a systematic search of the literature, of immunization-related secondary analyses from DHS or MICS published between 2006 and August 2018. We searched 15 electronic databases without language restrictions. For the articles included, relevant information was extracted and analyzed to summarize the characteristics of immunization-related secondary analyses. Results are presented following the PRISMA guidelines.ResultsAmong 1411 papers identified, 115 met our eligibility criteria; additionally, one article was supplemented by the Pan American Health Organization. The majority were published since 2012 (77.6%), and most (68.9%) had a first or corresponding author affiliated with institutions in high-income countries (as opposed to LMICs where these surveys are conducted). The median delay between survey implementation and publication of the secondary analysis was 5.4 years, with papers with authors affiliated to institutions in LMIC having a longer median publication delay (p < 0.001). Over 80% of the published analyses looked at factors associated with a specific vaccine or with full immunization. Quality proxies, such as reporting percent of immunization data from cards vs recall; occurrence and handling of missing data; whether survey analyses were weighted; and listing of potential biases or limitations of the original survey or analyses, were infrequently mentioned.ConclusionOur review suggests that more needs to be done to increase the increase the utilization of existing DHS and MICS datasets and improve the quality of the analyses to inform immunization programmes. This would include increasing the proportion of analyses done in LMICs, reducing the time lag between survey implementation and publication of additional analyses, and including more qualitative information about the survey in the publications to better interpret the results.

Ethnic inequalities in health-related quality of life among older adults in England: secondary analysis of a national cross-sectional survey

The population of older adults (ie, those aged ≥55 years) in England is becoming increasingly ethnically diverse. Previous reports indicate that ethnic inequalities in health exist among older adults, but information is limited by the paucity of data from small minority ethnic groups. This study aimed to analyse inequalities in health-related quality of life (HRQoL) and five determinants of health in older adults across all ethnic groups in England. In this cross-sectional study, we analysed data from five waves (July 1, 2014, to April 7, 2017) of the nationally representative English General Practice Patient Survey (GPPS). Study participants were adults aged 55 years or older who were registered with general practices in England. We used regression models (age-adjusted and stratified by gender) to estimate the association between ethnicity and HRQoL, measured by use of the EQ-5D-5L index and its domains (mobility, self-care, usual activities, pain or discomfort, and anxiety or depression). We also estimated associations between ethnicity and five determinants of health (presence of long-term conditions or multimorbidity, experience of primary care, degree of support from local services, patient self-confidence in managing own health, and degree of area-level social deprivation). We examined robustness to differential handling of missing data, alternative EQ-5D-5L value sets, and differences in area-level social deprivation. There were 1 416 793 GPPS respondents aged 55 years and older. 1 394 361 (98·4%) respondents had complete data on ethnicity and gender and were included in our analysis. Of these, 152 710 (11·0%) self-identified as belonging to minority ethnic groups. HRQoL was worse for men or women, or both, in 15 (88·2%) of 17 minority ethnic groups than the White British ethnic group. In both men and women, inequalities were widest for Gypsy or Irish Traveller (linear regression coefficient -0·192 [95% CI -0·318 to -0·066] in men; -0·264 [-0·354 to -0·173] in women), Bangladeshi (-0·111 [-0·136 to -0·087] in men; -0·209 [-0·235 to -0·184] in women), Pakistani (-0·084 [-0·096 to -0·073] in men; -0·206 [-0·219 to -0·193] in women), and Arab (-0·061 [-0·086 to -0·035] in men; -0·145 [-0·180 to -0·110] in women) ethnic groups, with magnitudes generally greater for women than men. Differentials tended to be widest for the self-care EQ-5D-5L domain. Ethnic inequalities in HRQoL were accompanied by increased prevalence of long-term conditions or multimorbidity, poor experiences of primary care, insufficient support from local services, low patient self-confidence in managing their own health, and high area-level social deprivation, compared with the White British group. We found evidence of wide ethnic inequalities in HRQoL and five determinants of health for older adults in England. Outcomes varied between minority ethnic groups, highlighting heterogeneity in the direction and magnitude of associations. We recommend further research to understand the drivers of inequalities, together with policy changes to improve equity of socioeconomic opportunity and access to services for older adults from minority ethnic groups. University of Manchester and National Institute for Health Research.

Continuous(ly) missing outcome data in network meta-analysis: A one-stage pattern-mixture model approach.

Appropriate handling of aggregate missing outcome data is necessary to minimise bias in the conclusions of systematic reviews. The two-stage pattern-mixture model has been already proposed to address aggregate missing continuous outcome data. While this approach is more proper compared with the exclusion of missing continuous outcome data and simple imputation methods, it does not offer flexible modelling of missing continuous outcome data to investigate their implications on the conclusions thoroughly. Therefore, we propose a one-stage pattern-mixture model approach under the Bayesian framework to address missing continuous outcome data in a network of interventions and gain knowledge about the missingness process in different trials and interventions. We extend the hierarchical network meta-analysis model for one aggregate continuous outcome to incorporate a missingness parameter that measures the departure from the missing at random assumption. We consider various effect size estimates for continuous data, and two informative missingness parameters, the informative missingness difference of means and the informative missingness ratio of means. We incorporate our prior belief about the missingness parameters while allowing for several possibilities of prior structures to account for the fact that the missingness process may differ in the network. The method is exemplified in two networks from published reviews comprising a different amount of missing continuous outcome data.

Potential problems in the use of patient reported outcome measures (PROMs) and reporting of PROM data in sports science.

To use an inadequate patient reported outcome measure (PROM) or use a PROM in an inappropriate way potentially influences the quality of measurement. The objectives of this study were to define potential inadequate uses of PROMs in sports research studies and estimate how often they occur. A consensus group consisting of medical researchers, statisticians, and psychometricians identified and defined potentially irregular applications of PROMs. Occurrence of these in 349 consecutive articles in sports medicine in which PROMs were used as primary outcomes was reviewed. In all, 14 different potential problems were defined, and one or several occurred in 172 of the articles (49%). These were as follows: using a PROM that was developed for a different patient group (100 cases), using two or more PROMs with identical questions (94), aggregation of domain sum scores (82), combinations of subjective and objective measures (27), using a PROM to diagnose or evaluate the individual patient (7), using a PROM for a single limb (3), recall bias (3), exclusion of domains or items (3), construction of a PROM for a specific occasion (2), categorization of the scale (2), and mixing different versions of a PROM (1). Adaption of scale scores (e. g., to percentage) when results are reported (144) carries a risk of miscalculation and distorted impression of results. Data related to uncertainty about completing the PROM and the handling of missing data were not provided in the manuscripts. In conclusion, potential problems in the use and reporting of PROMs are common in sports research, and this can influence the validity of reported results.

Extended follow-up data: impact of selective loss to follow-up in a trial of secondary prevention of coronary artery disease

Abstract Background We performed an extended follow-up of the RESPONSE-2 trial, which evaluated a new secondary prevention strategy for patients with coronary artery disease (CAD). Selective loss to follow-up (LTFU) however, may induce biased results. Objectives To assess LTFU and its implications on the internal validity of the trial at extended follow-up. Methods RESPONSE-2 (N=824) was a randomised trial of nurse-coordinated referral of CAD patients to ≤3 commercial lifestyle interventions, on top of usual-care, or to usual-care alone. The primary outcome was assessed at 1-year follow-up, after which the interventions were discontinued. An extended follow-up was performed two years later. We evaluated overall selective LTFU, between-group selective LTFU and within-group selective LTFU, by comparing baseline characteristics, compliance with the lifestyle interventions (intervention group only) and achievement of lifestyle-related risk factor (LRF) targets at 1-year follow-up for attendees and non-attendees of the extended follow-up. In all variables a standardized mean difference (SMD) of ≥0.2 with a p-value &amp;lt;0.05, based on an independent t-test for continuous data and fisher's test for discrete data, was considered evidence of selective LTFU. Results In total, 520 patients attended the extended follow-up: 266/411 in the intervention and 254/413 in the control group. Patients LTFU were generally younger (56.5 vs. 59.3 years; SMD=0.299), more often baseline smokers (37% vs. 22%; SMD=0.334) and had less often quit smoking directly after hospitalization (17% vs. 25%; SMD=0.203). We found no evidence of between-group selective LTFU and several indications of within-group selective LTFU (table). Intervention group attendees were more compliant with the lifestyle interventions compared with non-attendees (weight loss: SMD=0.450; p≤0.05 and physical activity: SMD=0.516; p≤0.01) Conclusion Baseline LRF control, compliance with the lifestyle interventions and achievement of LRF targets were less favourable for non-attendees of the extended follow-up. Effective handling of missing data is recommended to minimize loss of contrast due to within-group selective LTFU and to ensure generalizability of results. Funding Acknowledgement Type of funding source: Private company. Main funding source(s): The RESPONSE-2 trial was sponsored by Weight Watchers International, Inc. (New York, New York) and Philips Consumer Lifestyle (the Netherlands). The sponsors had no role in the design, data collection, data analysis, data interpretation, and writing of the manuscript.

Targeted hypothermia versus targeted normothermia after out-of-hospital cardiac arrest: a statistical analysis plan

BackgroundTo date, targeted temperature management (TTM) is the only neuroprotective intervention after resuscitation from cardiac arrest that is recommended by guidelines. The evidence on the effects of TTM is unclear.Methods/designThe Targeted Hypothermia Versus Targeted Normothermia After Out-of-hospital Cardiac Arrest (TTM2) trial is an international, multicentre, parallel group, investigator-initiated, randomised, superiority trial in which TTM with a target temperature of 33 °C after cardiac arrest will be compared with a strategy to maintain normothermia and active treatment of fever (≥ 37.8 °C). Prognosticators, outcome assessors, the steering group, the trial coordinating team, and trial statisticians will be blinded to treatment allocation. The primary outcome will be all-cause mortality at 180 days after randomisation. We estimate a 55% mortality in the targeted normothermia group. To detect an absolute risk reduction of 7.5% with an alpha of 0.05 and 90% power, 1900 participants will be enrolled. The secondary neurological outcome will be poor functional outcome (modified Rankin scale 4–6) at 180 days after cardiac arrest. In this paper, a detailed statistical analysis plan is presented, including a comprehensive description of the statistical analyses, handling of missing data, and assessments of underlying statistical assumptions. Final analyses will be conducted independently by two qualified statisticians following the present plan.DiscussionThis SAP, which was prepared before completion of enrolment, should increase the validity of the TTM trial by mitigation of analysis-bias.

A stacked approach for chained equations multiple imputation incorporating the substantive model.

Multiple imputation by chained equations (MICE) has emerged as a popular approach for handling missing data. A central challenge for applying MICE is determining how to incorporate outcome information into covariate imputation models, particularly for complicated outcomes. Often, we have a particular analysis model in mind, and we would like to ensure congeniality between the imputation and analysis models. We propose a novel strategy for directly incorporating the analysis model into the handling of missing data. In our proposed approach, multiple imputations of missing covariates are obtained without using outcome information. We then utilize the strategy of imputation stacking, where multiple imputations are stacked on top of each other to create a large data set. The analysis model is then incorporated through weights. Instead of applying Rubin's combining rules, we obtain parameter estimates by fitting a weighted version of the analysis model on the stacked data set. We propose a novel estimator for obtaining standard errors for this stacked and weighted analysis. Our estimator is based on the observed data information principle in Louis' work and can be applied for analyzing stacked multiple imputations more generally. Our approach for analyzing stacked multiple imputations is the first method that can be easily applied (using R package StackImpute) for a wide variety of standard analysis models and missing data settings.

Objective Image Quality Measures for Disparity Maps Evaluation

The evaluation of disparity (range) maps includes the selection of an objective image quality (or error) measure. Among existing measures, the percentage of bad matched pixels is commonly used. However, it requires a disparity error tolerance and ignores the relationship between range and disparity. In this research, twelve error measures are characterized in order to provide the bases to select accurate stereo algorithms during the evaluation process. Adaptations of objective quality measures for disparity maps’ accuracy evaluation are proposed. The adapted objective measures operate in a manner similar to the original objective measures, but allow special handling of missing data. Additionally, the adapted objective measures are sensitive to errors in range and surface structure, which cannot be measured using the bad matched pixels. Their utility was demonstrated by evaluating a set of 50 stereo disparity algorithms known in the literature. Consistency evaluation of the proposed measures was performed using the two conceptually different stereo algorithm evaluation methodologies—ordinary ranking and partition and grouping of the algorithms with comparable accuracy. The evaluation results showed that partition and grouping make a fair judgment about disparity algorithms’ accuracy.

Internet and Computer-Based Cognitive Behavioral Therapy for Anxiety and Depression in Adolescents and Young Adults: Systematic Review and Meta-Analysis.

BackgroundAnxiety and depressive disorders are prevalent in adolescents and young adults. However, most young people with mental health problems do not receive treatment. Computerized cognitive behavior therapy (cCBT) may provide an accessible alternative to face-to-face treatment, but the evidence base in young people is limited.ObjectiveThe objective was to perform an up-to-date comprehensive systematic review and meta-analysis of the effectiveness of cCBT in treating anxiety and depression in adolescents and young adults compared with active treatment and passive controls. We aimed to examine posttreatment and follow-up effects and explore the moderators of treatment effects.MethodsWe conducted systematic searches in the following six electronic databases: PubMed, EMBASE, PsycINFO, CINAHL, Web of Science, and Cochrane Central Register of Controlled Trials. We included randomized controlled trials comparing cCBT with any control group in adolescents or young adults (age 12-25 years) with anxiety or depressive symptoms. The quality of included studies was assessed using the Cochrane risk-of-bias tool for randomized trials, version 2.0. Overall quality of evidence for each outcome was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Posttreatment means and SDs were compared between intervention and control groups, and pooled effect sizes (Hedges g) were calculated. Random-effects meta-analyses were conducted using Comprehensive Meta-Analysis software. Subgroup analyses and meta-regression analyses were conducted to explore whether age, guidance level, and adherence rate were associated with treatment outcome.ResultsThe search identified 7670 papers, of which 24 studies met the inclusion criteria. Most included studies (22/24) had a high risk of bias owing to self-report measures and/or inappropriate handling of missing data. Compared with passive controls, cCBT yielded small to medium posttreatment pooled effect sizes regarding depressive symptoms (g=0.51, 95% CI 0.30-0.72, number needed to treat [NNT]=3.55) and anxiety symptoms (g=0.44, 95% CI 0.23-0.65, NNT=4.10). cCBT yielded effects similar to those of active treatment controls regarding anxiety symptoms (g=0.04, 95% CI −0.23 to 0.31). For depressive symptoms, the nonsignificant pooled effect size favored active treatment controls (g=−0.70, 95% CI −1.51 to 0.11, P=.09), but heterogeneity was very high (I2=90.63%). No moderators of treatment effects were identified. At long-term follow-up, cCBT yielded a small pooled effect size regarding depressive symptoms compared with passive controls (g=0.27, 95% CI 0.09-0.45, NNT=6.58). No other follow-up effects were found; however, power was limited owing to the small number of studies.ConclusionscCBT is beneficial for reducing posttreatment anxiety and depressive symptoms in adolescents and young adults compared with passive controls. Compared with active treatment controls, cCBT yielded similar effects regarding anxiety symptoms. Regarding depressive symptoms, however, the results remain unclear. More high-quality research involving active controls and long-term follow-up assessments is needed in this population.Trial RegistrationPROSPERO CRD42019119725; https://tinyurl.com/y5acfgd9.

Progressive exercise compared with best practice advice, with or without corticosteroid injection, for the treatment of rotator cuff disorders: statistical analysis plan for the Getting it Right: Addressing Shoulder Pain (GRASP) 2 \xd7 2 factorial multicentre randomised controlled trial

BackgroundThe Getting it Right: Addressing Shoulder Pain (GRASP) trial assesses the clinical and cost-effectiveness of individually tailored, progressive exercise compared with best practice advice, with or without corticosteroid injection, in adults with a rotator cuff disorder. This article describes the statistical analysis plan for the GRASP randomised controlled trial.Methods/designGRASP is a multicentre randomised controlled trial using a 2 × 2 factorial design. Adults aged ≥ 18 years with a new episode of shoulder pain related to a rotator cuff disorder, not currently receiving physiotherapy or being considered for surgery, are randomised (centralised computer-generated 1:1:1:1 allocation ratio) to one of four interventions: (1) progressive exercise (up to 6 physiotherapy sessions), (2) best practice advice (one physiotherapy session), (3) subacromial corticosteroid injection then progressive exercise and (4) subacromial corticosteroid injection then best practice advice. The primary outcome is the mean difference in Shoulder Pain and Disability Index (SPADI) total score over 12 months. Secondary outcomes are as follows: pain and function SPADI subdomains, health-related quality of life (EuroQol EQ-5D-5L), sleep disturbance, return to activity, global impression of change, health resource use, out-of-pocket expenses and work disability. Here, we describe in detail the following: sample size calculation, descriptive statistics of the primary and secondary outcomes, statistical models used for the analysis of the main outcomes, handling of missing data, planned sensitivity and subgroup analyses. This pre-specified statistical analysis plan was written and submitted without prior knowledge of the trial results.DiscussionPublication of the statistical analysis plan for the GRASP trial aims to reduce the risk of outcome reporting bias and increase transparency of the data analysis. Any deviations or changes to the current SAP will be described and justified in the final study report and any results publications.Trial registrationInternational Standard Randomised Controlled Trial Number ISRCTN16539266. Registered on 14 June 2016. EudraCT number 2016-002991-28. Registered on 12 June 2016.

Comparison of regression imputation methods of baseline covariates that predict survival outcomes.

Missing data are inevitable in medical research and appropriate handling of missing data is critical for statistical estimation and making inferences. Imputation is often employed in order to maximize the amount of data available for statistical analysis and is preferred over the typically biased output of complete case analysis. This article examines several types of regression imputation of missing covariates in the prediction of time-to-event outcomes subject to right censoring. We evaluated the performance of five regression methods in the imputation of missing covariates for the proportional hazards model via summary statistics, including proportional bias and proportional mean squared error. The primary objective was to determine which among the parametric generalized linear models (GLMs) and least absolute shrinkage and selection operator (LASSO), and nonparametric multivariate adaptive regression splines (MARS), support vector machine (SVM), and random forest (RF), provides the "best" imputation model for baseline missing covariates in predicting a survival outcome. LASSO on an average observed the smallest bias, mean square error, mean square prediction error, and median absolute deviation (MAD) of the final analysis model's parameters among all five methods considered. SVM performed the second best while GLM and MARS exhibited the lowest relative performances. LASSO and SVM outperform GLM, MARS, and RF in the context of regression imputation for prediction of a time-to-event outcome.

Time to get personal? The impact of researchers choices on the selection of treatment targets using the experience sampling methodology

ObjectiveOne of the promises of the experience sampling methodology (ESM) is that a statistical analysis of an individual's emotions, cognitions and behaviors in everyday-life could be used to identify relevant treatment targets. A requisite for clinical implementation is that outcomes of such person-specific time-series analyses are not wholly contingent on the researcher performing them. MethodsTo evaluate this, we crowdsourced the analysis of one individual patient's ESM data to 12 prominent research teams, asking them what symptom(s) they would advise the treating clinician to target in subsequent treatment. ResultsVariation was evident at different stages of the analysis, from preprocessing steps (e.g., variable selection, clustering, handling of missing data) to the type of statistics and rationale for selecting targets. Most teams did include a type of vector autoregressive model, examining relations between symptoms over time. Although most teams were confident their selected targets would provide useful information to the clinician, not one recommendation was similar: both the number (0–16) and nature of selected targets varied widely. ConclusionThis study makes transparent that the selection of treatment targets based on personalized models using ESM data is currently highly conditional on subjective analytical choices and highlights key conceptual and methodological issues that need to be addressed in moving towards clinical implementation.

Prognostic models for newly-diagnosed chronic lymphocytic leukaemia in adults: a systematic review and meta-analysis

Prognostic models for newly-diagnosed chronic lymphocytic leukaemia in adults: a systematic review and meta-analysis

Handling missing predictor values when validating and applying a prediction model to new patients.

Missing data present challenges for development and real‐world application of clinical prediction models. While these challenges have received considerable attention in the development setting, there is only sparse research on the handling of missing data in applied settings. The main unique feature of handling missing data in these settings is that missing data methods have to be performed for a single new individual, precluding direct application of mainstay methods used during model development. Correspondingly, we propose that it is desirable to perform model validation using missing data methods that transfer to practice in single new patients. This article compares existing and new methods to account for missing data for a new individual in the context of prediction. These methods are based on (i) submodels based on observed data only, (ii) marginalization over the missing variables, or (iii) imputation based on fully conditional specification (also known as chained equations). They were compared in an internal validation setting to highlight the use of missing data methods that transfer to practice while validating a model. As a reference, they were compared to the use of multiple imputation by chained equations in a set of test patients, because this has been used in validation studies in the past. The methods were evaluated in a simulation study where performance was measured by means of optimism corrected C‐statistic and mean squared prediction error. Furthermore, they were applied in data from a large Dutch cohort of prophylactic implantable cardioverter defibrillator patients.

When Is a Complete-Case Approach to Missing Data Valid? The Importance of Effect-Measure Modification

When estimating causal effects, careful handling of missing data is needed to avoid bias. Complete-case analysis is commonly used in epidemiologic analyses. Previous work has shown that covariate-stratified effect estimates from complete-case analysis are unbiased when missingness is independent of the outcome conditional on the exposure and covariates. Here, we assess the bias of complete-case analysis for adjusted marginal effects when confounding is present under various causal structures of missing data. We show that estimation of the marginal risk difference requires an unbiased estimate of the unconditional joint distribution of confounders and any other covariates required for conditional independence of missingness and outcome. The dependence of missing data on these covariates must be considered to obtain a valid estimate of the covariate distribution. If none of these covariates are effect-measure modifiers on the absolute scale, however, the marginal risk difference will equal the stratified risk differences and the complete-case analysis will be unbiased when the stratified effect estimates are unbiased. Estimation of unbiased marginal effects in complete-case analysis therefore requires close consideration of causal structure and effect-measure modification.

Registering Study Analysis Plans (SAPs) Before Dissecting Your Data-Updating and Standardizing Outcome Modeling.

Public preregistration of study analysis plans (SAPs) is widely recognized for clinical trials, but adopted to a much lesser extent in observational studies. Registration of SAPs prior to analysis is encouraged to not only increase transparency and exactness but also to avoid positive finding bias and better standardize outcome modeling. Efforts to generally standardize outcome modeling, which can be based on clinical trial and/or observational data, have recently spurred. We suggest a three-step SAP concept in which investigators are encouraged to (1) Design the SAP and circulate it among the co-investigators, (2) Log the SAP with a public repository, which recognizes the SAP with a digital object identifier (DOI), and (3) Cite (using the DOI), briefly summarize and motivate any deviations from the SAP in the associated manuscript. More specifically, the SAP should include the scope (brief data and study description, co-investigators, hypotheses, primary outcome measure, study title), in addition to step-by-step details of the analysis (handling of missing data, resampling, defined significance level, statistical function, validation, and variables and parameterization).

Performance of Pediatric Mortality Prediction Models in Low- and Middle-Income Countries: A Systematic Review and Meta-Analysis

To describe the performance of prognostic models for mortality or clinical deterioration events among hospitalized children developed or validated in low- and middle-income countries. A medical librarian systematically searched EMBASE, Ovid Medline, Scopus, Cochrane Library, EBSCO Global Health, LILACS, African Index Medicus, African Journals Online, African Healthline, Med-Carib, and Global Index Medicus (from 2000 to October 2019). We included citations that described the development or validation of a pediatric prognostic model for hospital mortality or clinical deterioration events in low- and middle-income countries. In duplicate and independently, we extracted data on included populations and model prognostic performance and evaluated risk of bias using the Prediction model Risk Of Bias Assessment Tool. Of 41 279 unique citations, we included 15 studies describing 15 prognostic models for mortality and 3 models for clinical deterioration events. Six models were validated in >1 external cohort. The Lambarene Organ Dysfunction Score (0.85 [0.77-0.92]) and Signs of Inflammation in Children that Kill (0.85 [0.82-0.88]) had the highest summary C-statistics (95% CI) for discrimination. Calibration and classification measures were poorly reported. All models were at high risk of bias owing to inappropriate selection of predictor variables and handling of missing data and incomplete performance measure reporting. Several prognostic models for mortality and clinical deterioration events have been validated in single cohorts, with good discrimination. Rigorous validation that conforms to current standards for prediction model studies and updating of existing models are needed before clinical implementation.

Data Analysis: Strengthening Inferences in Quantitative Education Studies Conducted by Novice Researchers

Data analysis is a significant methodological component when conducting quantitative education studies. Guidelines for conducting data analyses in quantitative education studies are common but often underemphasize four important methodological components impacting the validity of inferences: quality of constructed measures, proper handling of missing data, proper level of measurement of a dependent variable, and model checking. This paper highlights these components for novice researchers to help ensure statistical inferences are valid. We used empirical examples involving contingency tables, group comparisons, regression analysis, and multilevel modelling to illustrate these components using the Program for International Student Assessment (PISA) data. For every example, we stated a research question and provided evidence related to the quality of constructed measures since measures with weak reliability and validity evidence can bias estimates and distort inferences. The adequate strategies for handling missing data were also illustrated. The level of measurement for the dependent variable was assessed and the proper statistical technique was utilized accordingly. Model residuals were checked for normality and homogeneity of variance. Recommendations for obtaining stronger inferences and reporting related evidence were also illustrated. This work provides an important methodological resource for novice researchers conducting data analyses by promoting improved practice and stronger inferences.