A Review of Hot Deck Imputation for Survey Non-response

Cumulative incidence of SARS-CoV-2 infections among adults in Georgia, USA, August-December 2020

BACKGROUND

Reported COVID-19 cases underestimate true SARS-CoV-2 infections. Data on all infections, including asymptomatic infections, are needed to guide programs. To minimize biases in estimates from reported cases and seroprevalence surveys, we conducted a household-based probability survey in Georgia and estimated cumulative incidence of SARS-CoV-2 infections adjusted for antibody waning.

METHODS

From August to December 2020, we mailed specimen collection kits (nasal swabs and blood spots) to a random sample of Georgia addresses. One household adult completed a survey and returned specimens for virus and antibody testing. We estimated cumulative incidence of SARS-CoV-2 infections adjusted for waning antibodies, reported fraction, and infection fatality ratio (IFR). Differences in seropositivity among demographic, geographic and clinical subgroups were explored with weighted prevalence ratios (PR).

RESULTS

Among 1,370 participants, adjusted cumulative incidence of SARS-CoV-2 was 16.1% (95% credible interval (CrI): 13.5-19.2%) as of November 16, 2020. The reported fraction was 26.6% and IFR was 0.78%. Non-Hispanic Black (PR: 2.03, CI 1.0, 4.1) and Hispanic adults (PR: 1.98, CI 0.74, 5.31) were more likely than non-Hispanic White adults to be seropositive.

CONCLUSIONS

As of mid-November 2020, one in 6 adults in Georgia had been infected with SARS-CoV-2. The COVID-19 epidemic in Georgia is likely substantially underestimated by reported cases.

What are the determinants of childhood infections in India's peri-urban slums? A case study of eight cities

BACKGROUND

Respiratory Tract Infections (RTIs) and Gastro-Intestinal (GI) infections are the leading causes of child mortality and morbidity. This study investigates the associations between the individual, household and slum-level determinants of children's health and vulnerability to RTIs and GI infections in peri-urban slums in India; an area of research interest at the Childhood Infections and Pollution Consortium.

METHODS

The 2015-16 Indian National Family Health Survey was used for data analysis on children aged 0-5 years. NFHS-4 includes data on slums in eight Indian cities, including Delhi, Meerut, Kolkata, Indore, Mumbai, Nagpur, Hyderabad, Chennai. The outcome variables, having fever and cough (FeCo) and diarrhoea in the last two weeks, were used to define the phenotype of infections; for this analysis fever and cough were measures of RTIs and diarrhoea was used to measure GI infections. Exposures considered in this study include variables at the individual, household and slum level and were all informed by existing literature. Multilevel models were used to estimate the association between exposures and outcomes variables; a prior of Cauchy distribution with a scale of 2.5 was selected when building the multilevel logistic models.

RESULTS

The total sample size of the number of children included in the analysis was n = 1,424. Data was imputed to account for missingness, and the original and imputed sample showing similar distributions. Results showed that diarrhoea and FeCo were both found to be more present in younger children than older children by a few months. In fixed effects, the odds of developing FeCo were higher if the mother perceives the child was born smaller than average (AOR 4.41, 1.13-17.17, P<0.05) at individual level. On the other hand, the odds of the diarrhoea outcome were lower if the child was older (AOR 0.97, 0.96-0.98, P<0.05) at individual level, and household's water source was public tap or standpipe (AOR 0.54, 0.31-0.96, P<0.05) at household level.

CONCLUSION

The determinants of health, both social and related to health care, at all levels demonstrated linkages to child morbidity in RTIs and GI infections. The empirical evidence highlights the need for contextualised ideas at each level, including one health approach when designing interventions to improve child health.

Living in a block group with a higher eviction rate is associated with increased odds of preterm delivery

BACKGROUND

Housing instability is associated with adverse pregnancy outcomes. Recent studies indicate that eviction, which may affect a larger segment of the population than other forms of housing instability, is also associated with adverse pregnancy outcomes. However, these studies evaluate eviction across large areas, such as counties, so it remains unclear whether these patterns extend to individual-level pregnancy outcomes.

METHODS

We used data on a cohort of all singleton live births at a single Chicago hospital between March 2008 and March 2018 to investigate the associations between block-group eviction rates and individual adverse pregnancy outcomes. Eviction data were obtained from the Eviction Lab at Princeton University. Generalised estimating equations were used to estimate associations and account for correlations among individuals living in the same block groups.

RESULTS

Individuals living in block groups in the highest quartile for eviction filing rate were 1.17 times as likely to deliver preterm (95% CI: 1.08 to 1.27) and 1.13 times as likely to deliver a small for gestational age infant (95% CI: 1.03 to 1.25) as compared with individuals living in block groups in the lowest quartile. Further, tests for linear trend indicated that for each quartile increase in eviction filing rate, there was a corresponding increase in odds of adverse outcomes (p<0.05). Results were strongest in magnitude for those with low neighbourhood and individual socioeconomic status, who are most likely to be renters and affected by local eviction policies.

CONCLUSION

Our results suggest that individuals living in block groups with higher eviction rates are more likely to deliver preterm. Future research should explore associations of individual experience with eviction on adverse pregnancy outcomes and examine whether policies to improve tenant protections also impact pregnancy outcomes.

Beyond question wording: How survey design and administration shape estimates of disability

BACKGROUND

Between 2008 and 2014, annual estimates of disability prevalence among U.S. adults varied somewhat across federal surveys that use a standardized measure of disability, but trends over-time were relatively stable and consistent. In 2014, however, estimates of disability from the Survey of Income and Program Participation (SIPP) increased markedly relative to previous years and were much higher than disability estimates from other federal surveys.

OBJECTIVE

To examine why disability prevalence among adults aged 40 and older substantially increased in the first wave of the 2014 SIPP Panel.

METHODS

We consider three factors that may have contributed to the rise in disability: data processing, context effects linked to question order, and sampling. To do so, we compare estimates with and without survey weights and imputed data, analyze supplemental disability-related data collected among SIPP participants, and employ decomposition analysis to assess what proportion of the increase in disability can be attributed to changes in sample composition.

RESULTS

We find evidence that differences in sample composition contributed to the observed rise in disability prevalence in SIPP between 2011 and 2014. There is less evidence that weighting and imputation or context effects played a role.

CONCLUSIONS

Previous studies emphasize differences in operationalization and conceptualization of disability as the major factor driving discrepancies in disability estimates. This study suggests that other factors related to survey design and administration may influence disability measurement. Such aspects of surveys, like question order and sampling, may be difficult to standardize, leading to meaningful cross-survey differences in disability estimates.

Snapshot Models of Undocumented Immigration

Accurately estimating the size of the undocumented immigrant population is a critical component of assessing the health and security risks of undocumented immigration to the United States. To provide one such estimate, we use data from the Mexican Migration Project (MMP), a study that includes samples of undocumented Mexican immigrants to the United States after their return to Mexico. Of particular interest are the departure and return dates of a sampled migrant's most recent sojourn in the United States, and the total number of such journeys undertaken by that migrant household, for these data enable the construction of data-driven undocumented immigration models. However, such data are subject to an extreme physical bias, for to be included in such a sample, a migrant must have returned to Mexico by the time of the survey, excluding those undocumented immigrants still in the United States. In our analysis, we account for this bias by jointly modeling trip timing and duration to produce the likelihood of observing the data in such a "snapshot" sample. Our analysis characterizes undocumented migration flows including single-visit migrants, repeat visitors, and "retirement" from circular migration. Starting with 1987, we apply our models to 30 annual random snapshot surveys of returned undocumented Mexican migrants accounting for undocumented Mexican migration from 1980 to 2016. Scaling to population quantities and supplementing our analysis of southern border crossings with estimates of visa overstays, we produce lower bounds on the total number of undocumented immigrants that are much larger than conventional estimates based on U.S.-based census-linked surveys, and broadly consistent with the more recent estimates reported by Fazel-Zarandi, Feinstein, and Kaplan.

Improving Outcome Predictions for Patients Receiving Mechanical Circulatory Support by Optimizing Imputation of Missing Values

BACKGROUND

Risk prediction models play an important role in clinical decision making. When developing risk prediction models, practitioners often impute missing values to the mean. We evaluated the impact of applying other strategies to impute missing values on the prognostic accuracy of downstream risk prediction models, that is, models fitted to the imputed data. A secondary objective was to compare the accuracy of imputation methods based on artificially induced missing values. To complete these objectives, we used data from the Interagency Registry for Mechanically Assisted Circulatory Support.

METHODS

We applied 12 imputation strategies in combination with 2 different modeling strategies for mortality and transplant risk prediction following surgery to receive mechanical circulatory support. Model performance was evaluated using Monte-Carlo cross-validation and measured based on outcomes 6 months following surgery using the scaled Brier score, concordance index, and calibration error. We used Bayesian hierarchical models to compare model performance.

RESULTS

Multiple imputation with random forests emerged as a robust strategy to impute missing values, increasing model concordance by 0.0030 (25th-75th percentile: 0.0008-0.0052) compared with imputation to the mean for mortality risk prediction using a downstream proportional hazards model. The posterior probability that single and multiple imputation using random forests would improve concordance versus mean imputation was 0.464 and >0.999, respectively.

CONCLUSIONS

Selecting an optimal strategy to impute missing values such as random forests and applying multiple imputation can improve the prognostic accuracy of downstream risk prediction models.

Age, sex, and inflammatory markers predict chronic conditions, cardiac disease, and mortality among captive western lowland gorillas (Gorilla gorilla gorilla)

In humans, inflammatory markers predict health risks. As great apes experience many similar conditions, measuring inflammation may provide valuable health information. We examined four serum inflammatory markers in zoo-housed gorillas (n = 48): albumin, CRP, IL-6, and TNF-α. We first analyzed age- and sex-associated patterns, then used multimodel inference to evaluate models with age, sex, and inflammatory markers as predictors of all-cause morbidity, cardiac disease, and mortality. Older gorillas had lower albumin and higher IL-6, and males had higher albumin, lower CRP, and lower TNF-α. All-cause morbidity was best predicted by age, sex, and TNF-α, but the second model containing only age and sex was equivalent. Cardiac disease was best predicted by TNF-α alongside age and sex, with lower levels associated with increased risk. When outliers were removed, the model with TNF-α was second to the model containing only age and sex. Finally, mortality risk was best predicted by the model with only age and sex. Other models containing individual inflammatory markers were within top model sets for each health outcome. Our results indicate that age and sex are robust for predicting all-cause morbidity and mortality risk in gorillas; while models which include individual inflammatory markers also predict risk, they may not improve predictions over age and sex alone. However, given the prevalence of cardiac disease in great apes, these results suggest that TNF-α warrants further investigation. With their potential to provide valuable health information, data on inflammatory markers may contribute to the care and management of gorillas in human care.

Health Disparities of Older Adults in California: The Role of Sexual Identity and Latinx Ethnicity

BACKGROUND AND OBJECTIVES

We examined the health disparities of older adults (age 50 and older) in California at the intersection of sexual identity and Latinx ethnicity, by comparing the prevalence of health outcomes of 4 groups: LGB (lesbian, gay, and bisexual) Latinx, straight Latinx, LGB non-Latinx, and straight non-Latinx older adults.

RESEARCH DESIGN AND METHODS

Data were from the 2015-2016 California Health Interview Survey. Multivariable logistic regressions tested differences among the 4 groups and the effect of covariates on prevalence of mental and physical health outcomes. We compared LGB and straight people within the same ethnic groups and Latinx and non-Latinx people within the same sexual identity groups to understand the intersectional effect of Latinx ethnicity and LGB identity.

RESULTS

Tests by sexual identity showed that among Latinx older adults, more LGB than non-LGB people experienced serious psychological distress. Among non-Latinx older adults, there were no health disparities due to sexual identity. Tests by Latinx ethnicity showed that among LGB older adults, more Latinx than non-Latinx people were obese. Among straight people, more Latinx than non-Latinx older people had poor health, diabetes, and obesity.

DISCUSSION AND IMPLICATIONS

The compounded effect of Latinx and LGB identity on psychological distress is notable. However, most health disparities were among straight older adults, between Latinx and non-Latinx people, indicating that Latinx, not sexual identity, nor their intersection, was most influential. Given the importance of sociodemographic factors on health outcomes, programs targeting LGB older adults should take a comprehensive approach to understand their experiences as ethnic minorities.

Methodological Challenges and Statistical Approaches in the COMprehensive Post-Acute Stroke Services Study

BACKGROUND

The COMprehensive Post-Acute Stroke Services study was a cluster-randomized pragmatic trial designed to evaluate a comprehensive care transitions model versus usual care. The data collected during this trial were complex and analysis methodology was required that could simultaneously account for the cluster-randomized design, missing patient-level covariates, outcome nonresponse, and substantial nonadherence to the intervention.

OBJECTIVE

The objective of this study was to discuss an array of complementary statistical methods to evaluate treatment effectiveness that appropriately addressed the challenges presented by the complex data arising from this pragmatic trial.

METHODS

We utilized multiple imputation combined with inverse probability weighting to account for missing covariate and outcome data in the estimation of intention-to-treat effects (ITT). The ITT estimand reflects the effectiveness of assignment to the COMprehensive Post-Acute Stroke Services intervention compared with usual care (ie, it does not take into account intervention adherence). Per-protocol analyses provide complementary information about the effect of treatment, and therefore are relevant for patients to inform their decision-making. We describe estimation of the complier average causal effect using an instrumental variables approach through 2-stage least squares estimation. For all preplanned analyses, we also discuss additional sensitivity analyses.

DISCUSSION

Pragmatic trials are well suited to inform clinical practice. Care should be taken to proactively identify the appropriate balance between control and pragmatism in trial design. Valid estimation of ITT and per-protocol effects in the presence of complex data requires application of appropriate statistical methods and concerted efforts to ensure high-quality data are collected.

Asymptotic theory and inference of predictive mean matching imputation using a superpopulation model framework

Predictive mean matching imputation is popular for handling item nonresponse in survey sampling. In this article, we study the asymptotic properties of the predictive mean matching estimator for finite-population inference using a superpopulation model framework. We also clarify conditions for its robustness. For variance estimation, the conventional bootstrap inference is invalid for matching estimators with a fixed number of matches due to the nonsmoothness nature of the matching estimator. We propose a new replication variance estimator, which is asymptotically valid. The key strategy is to construct replicates directly based on the linear terms of the martingale representation for the matching estimator, instead of individual records of variables. Simulation studies confirm that the proposed method provides valid inference.

Advanced methods for missing values imputation based on similarity learning

The real-world data analysis and processing using data mining techniques often are facing observations that contain missing values. The main challenge of mining datasets is the existence of missing values. The missing values in a dataset should be imputed using the imputation method to improve the data mining methods' accuracy and performance. There are existing techniques that use k-nearest neighbors algorithm for imputing the missing values but determining the appropriate k value can be a challenging task. There are other existing imputation techniques that are based on hard clustering algorithms. When records are not well-separated, as in the case of missing data, hard clustering provides a poor description tool in many cases. In general, the imputation depending on similar records is more accurate than the imputation depending on the entire dataset's records. Improving the similarity among records can result in improving the imputation performance. This paper proposes two numerical missing data imputation methods. A hybrid missing data imputation method is initially proposed, called KI, that incorporates k-nearest neighbors and iterative imputation algorithms. The best set of nearest neighbors for each missing record is discovered through the records similarity by using the k-nearest neighbors algorithm (kNN). To improve the similarity, a suitable k value is estimated automatically for the kNN. The iterative imputation method is then used to impute the missing values of the incomplete records by using the global correlation structure among the selected records. An enhanced hybrid missing data imputation method is then proposed, called FCKI, which is an extension of KI. It integrates fuzzy c-means, k-nearest neighbors, and iterative imputation algorithms to impute the missing data in a dataset. The fuzzy c-means algorithm is selected because the records can belong to multiple clusters at the same time. This can lead to further improvement for similarity. FCKI searches a cluster, instead of the whole dataset, to find the best k-nearest neighbors. It applies two levels of similarity to achieve a higher imputation accuracy. The performance of the proposed imputation techniques is assessed by using fifteen datasets with variant missing ratios for three types of missing data; MCAR, MAR, MNAR. These different missing data types are generated in this work. The datasets with different sizes are used in this paper to validate the model. Therefore, proposed imputation techniques are compared with other missing data imputation methods by means of three measures; the root mean square error (RMSE), the normalized root mean square error (NRMSE), and the mean absolute error (MAE). The results show that the proposed methods achieve better imputation accuracy and require significantly less time than other missing data imputation methods.

SARS-CoV-2 cumulative incidence, United States, August-December 2020

Background

Reported coronavirus disease 2019 (COVID-19) cases underestimate severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections. We conducted a national probability survey of US households to estimate cumulative incidence adjusted for antibody waning.

Methods

From August–December 2020 a random sample of US addresses were mailed a survey and self-collected nasal swabs and dried blood spot cards. One adult household member completed the survey and mail specimens for viral detection and total (immunoglobulin [Ig] A, IgM, IgG) nucleocapsid antibody by a commercial, emergency use authorization–approved antigen capture assay. We estimated cumulative incidence of SARS-CoV-2 adjusted for waning antibodies and calculated reported fraction (RF) and infection fatality ratio (IFR). Differences in seropositivity among demographic, geographic, and clinical subgroups were explored.

Results

Among 39 500 sampled households, 4654 respondents provided responses. Cumulative incidence adjusted for waning was 11.9% (95% credible interval [CrI], 10.5%–13.5%) as of 30 October 2020. We estimated 30 332 842 (CrI, 26 703 753–34 335 338) total infections in the US adult population by 30 October 2020. RF was 22.3% and IFR was 0.85% among adults. Black non-Hispanics (Prevalence ratio (PR) 2.2) and Hispanics (PR, 3.1) were more likely than White non-Hispanics to be seropositive.

Conclusions

One in 8 US adults had been infected with SARS-CoV-2 by October 2020; however, few had been accounted for in public health reporting. The COVID-19 pandemic is likely substantially underestimated by reported cases. Disparities in COVID-19 by race observed among reported cases cannot be attributed to differential diagnosis or reporting of infections in population subgroups.

The Optimal Machine Learning-Based Missing Data Imputation for the Cox Proportional Hazard Model

An adequate imputation of missing data would significantly preserve the statistical power and avoid erroneous conclusions. In the era of big data, machine learning is a great tool to infer the missing values. The root means square error (RMSE) and the proportion of falsely classified entries (PFC) are two standard statistics to evaluate imputation accuracy. However, the Cox proportional hazards model using various types requires deliberate study, and the validity under different missing mechanisms is unknown. In this research, we propose supervised and unsupervised imputations and examine four machine learning-based imputation strategies. We conducted a simulation study under various scenarios with several parameters, such as sample size, missing rate, and different missing mechanisms. The results revealed the type-I errors according to different imputation techniques in the survival data. The simulation results show that the non-parametric “missForest” based on the unsupervised imputation is the only robust method without inflated type-I errors under all missing mechanisms. In contrast, other methods are not valid to test when the missing pattern is informative. Statistical analysis, which is improperly conducted, with missing data may lead to erroneous conclusions. This research provides a clear guideline for a valid survival analysis using the Cox proportional hazard model with machine learning-based imputations.

Binge drinking among Hispanic older adults: 2015-2019

Previous research has shown that older adults (i.e., individuals 50 years or older) show a high propensity to report unhealthy alcohol habits. Much is to be gleaned regarding these relationships among Hispanic older adults. The purpose of the present study was to examine correlates to binge drinking among a national sample of Hispanic older adults. Pooled data from the 2015-2019 National Survey on Drug Use and Health were analyzed among 4,152 Hispanic individuals. Findings revealed that a sizable percentage (17.9%) of individuals reported binge drinking in the past 30 days. Of the sample, 15.1% of individuals diagnosed with diabetes reported binge drinking and high co-morbid substance use was found. Findings can address critical gaps in Hispanic health care, prevention messaging, and harm reduction.

Are spatial models advantageous for predicting county-level HIV epidemiology across the United States?

Predicting human immunodeficiency virus (HIV) epidemiology is vital for achieving public health milestones. Incorporating spatial dependence when data varies by region can often provide better prediction results, at the cost of computational efficiency. However, with the growing number of covariates available that capture the data variability, the benefit of a spatial model could be less crucial. We investigate this conjecture by considering both non-spatial and spatial models for county-level HIV prediction over the US. Due to many counties with zero HIV incidences, we utilize a two-part model, with one part estimating the probability of positive HIV rates and the other estimating HIV rates of counties not classified as zero. Based on our data, the compound of logistic regression and a generalized estimating equation outperforms the candidate models in making predictions. The results suggest that considering spatial correlation for our data is not necessarily advantageous when the purpose is making predictions.

Tobacco Use Prevalence and Transitions From 2013 to 2018 Among Adults With a History of Cardiovascular Disease

Background Although tobacco product use and transitions have been characterized in the general population, few studies have focused on individuals with established cardiovascular disease (CVD) in a population-based sample. Methods and Results We examined tobacco use prevalence and longitudinal patterns of tobacco product transitions in adults (≥18 years) of the nationally representative PATH (Population Assessment of Tobacco and Health) study, from 2013 to 2014 (Wave 1) through 2016 to 2018 (Wave 4). Prevalent CVD was classified through self-report of having had a heart attack, heart failure, stroke, or other heart condition. Factors associated with tobacco product use and transitions were investigated using survey logistic regression. We examined 2615 participants with self-reported CVD at Wave 1. Overall, 28.9% reported current tobacco use, equating to ≈6.2 million adults in the United States with prevalent CVD and current tobacco use. Among adults with CVD who are current tobacco users, the most commonly used product was cigarettes (82.8%), followed by any type of cigar (23.7%), and e-cigarette use (23.3%). E-cigarette use without concurrent cigarette use among participants with prevalent CVD was uncommon (1.1%). Factors associated with tobacco use were younger age, male sex, had lower education level, and lack of knowledge about the association between smoking and CVD. Men with prevalent CVD were less likely to use e-cigarettes compared with women (odds ratio [OR], 0.7; 95% CI, 0.5-0.9). Among cigarette users with CVD, transition rates between Waves 1 and 4 demonstrated <5% decrease in cigarette, with a 0.5% increase in e-cigarette use. Only ≈10% were in formal tobacco cessation programs. Conclusions Despite known harmful cardiovascular effects, over one fourth of adults with prevalent CVD use tobacco products and few quit smoking over the 4 waves of the PATH data set.

Evaluation of Eight-Item Vancomycin Prescribing Confidence Questionnaire Among Junior Doctors

Background: Assessing the preparedness of junior doctors to use vancomycin is important in medical education. Preparedness is typically evaluated by self-reported confidence surveys.

Materials and Methods: An eight-item vancomycin prescribing confidence questionnaire was developed, piloted, and evaluated. The questionnaire responses were collected from 195 junior doctors and a series of statistical techniques, such as principal component analysis and confirmatory factor analysis, and were implemented to examine the validity and reliability.

Results: The principal component analysis supported a one-factor structure, which was fed into a confirmatory factor analysis model resulting in a good fit [comparative fit index (CFI) = 0.99, Tucker–Lewis index (TLI) = 0.99, root mean square error of approximation (RMSEA) = 0.08, standardized root mean square residual (SRMR) = 0.04]. Ordinal-based α was 0.95, and various ωs were all above 0.93, indicating a high reliability level. The questionnaire responses were further proved to be robust to extreme response patterns via item response tree modeling. Jonckheere–Terpstra test results (z = 6.5237, p = 3.429e−11) showed that vancomycin prescribing confidence differed based on the experience in order (i.e., four ordinal independent groups: “≤10 times,” “11–20 times,” “21–30 times,” and “≥31 times”) and therefore provided external validity evidences for the questionnaire.

Conclusions: The questionnaire is valid and reliable such that teaching hospitals can consider using it to assess junior doctors' vancomycin prescribing confidence. Further investigation of the questionnaire can point to the relationship between the prescribing confidence and the actual performance.

Associations between elementary and middle school teachers' physical activity promoting practices and teacher- and school-level factors

BACKGROUND

Few studies have evaluated teacher- and school-level characteristics associated with implementation of recommended physical activity (PA) promoting practices. The purpose of this study is to examine associations between teachers' PA practices and: [1] teacher-level factors, including their own PA, and [2] school-level factors.

METHODS

This cross-sectional study examined time spent daily in light PA (LPA) and moderate-vigorous PA (MVPA) in association with 7 teacher PA practices among 288 classroom/special area teachers and teaching assistants in 20 urban, suburban and rural schools (recruited through a school wellness trial) in 4 districts. LPA and MVPA was assessed using 24-h ankle accelerometry (up to seven consecutive days). A sum score for teacher PA practices was assessed via survey (7 items; sum score range: 7-35; Cronbach's alpha = 0.73; higher scores indicate more PA promoting practices). Teacher-level factors included gender, race, self-reported height/weight, years teaching, and education. School-level factors included school type, free-and-reduced-price meal eligibility, student racial/ethnic composition, and urbanicity. Analyses included multilevel regression models, accounting for clustering within schools and adjusting for demographic covariates and school district.

RESULTS

Teachers were 91% female, 63% elementary, 60% white, mean age 43.2 years (SD = 11.3), and 41% obese). Teachers wore accelerometers an average of 5.8 days, spent 399.6 min in LPA (SD = 85.0) per day, 24.1 min in MVPA (SD = 14.4) per day, and the mean teacher PA practices sum score was 22.4 (SD = 5.0). Every 15-min increase in MVPA was related to an increase in teacher PA practices sum score (coeff =1.07; SE = 0.28; p < 0.001). Female gender (versus males; coeff = - 1.95; SE = 0.92, p = 0.034), an obese weight status (versus non-obese; coeff = - 1.38; SE = 0.54, p = 0.010), and teaching in a middle school (versus elementary; coeff = - 3.86; SE = 0.54, p < 0.001) were associated with lower teacher PA practices scores. LPA was not associated with teacher PA promoting practices.

CONCLUSIONS

Teachers with higher MVPA, but not higher LPA, and those without obesity were more likely to implement PA promoting practices that could positively impact their students' PA. Similar to prior studies, these practices were more commonly implemented in elementary schools and by male teachers. Future studies in schools should explore whether improvement of teacher health behaviors subsequently impacts student health behaviors.

TRIAL REGISTRATION

Clinical Trials, NCT03432715 ; Registered on 02/2/2018.

Perceived Injustice Helps Explain the Association Between Chronic Pain Stigma and Movement-Evoked Pain in Adults with Nonspecific Chronic Low Back Pain

OBJECTIVE

For most patients with chronic low back pain (cLBP), the cause is "nonspecific," meaning there is no clear association between pain and identifiable pathology of the spine or associated tissues. Laypersons and providers alike are less inclined to help, feel less sympathy, dislike patients more, suspect deception, and attribute lower pain severity to patients whose pain does not have an objective basis in tissue pathology. Because of these stigmatizing responses from others, patients with cLBP may feel that their pain is particularly unjust and unfair. These pain-related injustice perceptions may subsequently contribute to greater cLBP severity. The purpose of this study was to examine whether perceived injustice helps explain the relationship between chronic pain stigma and movement-evoked pain severity among individuals with cLBP.

METHODS

Participants included 105 patients with cLBP who completed questionnaires assessing chronic pain stigma and pain-related injustice perception, as well as a short physical performance battery for the assessment of movement-evoked pain and physical function.

RESULTS

Findings revealed that perceived injustice significantly mediated the association between chronic pain stigma and cLBP severity (indirect effect = 6.64, 95% confidence interval [CI] = 2.041 to 14.913) and physical function (indirect effect = -0.401, 95% CI = -1.029 to -0.052). Greater chronic pain stigma was associated with greater perceived injustice (P = 0.001), which in turn was associated with greater movement-evoked pain severity (P = 0.003).

CONCLUSIONS

These results suggest that perceived injustice may be a means through which chronic pain stigma impacts nonspecific cLBP severity and physical function.

The role of cognitive and non-cognitive factors in dental anxiety: A mediation model

Dental anxiety is a crucial problem for dentistry because it may represent a significant risk to oral health. Different factors, whether non‐cognitive (e.g., traumatic dental events) or cognitive (e.g., the patient's subjective perceptions), may cause dental anxiety. However, previous studies have assessed these factors as independent predictors of dental anxiety, without providing any exploration of potential mediational pathways. The current study assessed the role of certain cognitive dimensions (i.e., the dentist's perceived professionalism and communicational attitudes, and the patient's perceived lack of control) as mediators between traumatic dental events and dental anxiety. The sample comprised 253 patients who had accessed a public university hospital dental surgery. The mediation analysis used a structural equation modeling. Traumatic dental events were positively associated with dental anxiety but, among the cognitive factors, only lack of control was. Furthermore, lack of control mediated the relationship between traumatic dental events and dental anxiety, although this mediation was only partial. This study sheds light on the mechanisms through which non‐cognitive and cognitive factors may affect dental anxiety. The clinical implications for dental practice, in terms of improving the psychological well‐being of patients, are discussed.

Balancing Inferential Integrity and Disclosure Risk via Model Targeted Masking and Multiple Imputation

There is a growing expectation that data collected by government-funded studies should be openly available to ensure research reproducibility, and so is the concern on data-privacy. A strategy to protect individuals' identity is to release multiply imputed (MI) synthetic datasets with masked sensitivity values (Rubin, 1993). However, information loss or incorrectly specified imputation models can weaken or invalidate the inferences obtained from the MI-datasets. Studying a restricted-use Canadian Scleroderma Research Group (CSRG) dataset, the authors investigate the use of a new masking framework with a data-augmentation (DA) component and a tuning mechanism that balances between protecting identity-disclosure and preserving data-utility. They found, respectively in a work-disability and an interstitial lung disease study, using this DA-MI strategy reached 0% identity disclosure-risk, preserved all inferential conclusions, and on average produced 98.5% and 95.5% confidence intervals (CI) overlaps when compared to the 95% CIs constructed using the generic CSGR dataset; the lowest CI-overlap value is 91%. In contrast, the same is not true for the currently used methods; with the CI-overlap values ranging from 73.9% to 91.8% and the lowest value being 28.1%. These findings indicate that the DA-MI masking framework facilitates sharing of useful research data while protecting participants' identities.

Impact of Food Insecurity Nutrition Incentives on Household Fruit and Vegetable Expenditures

OBJECTIVE

Evaluate the impact of the Food Insecurity Nutrition Incentive (FINI) grant program on self-reported fruit and vegetable (FV) expenditures.

DESIGN

Pre-post quasi-experimental study design.

SETTING

Farmers markets and grocery stores in states with FINI projects.

PARTICIPANTS

A total of 2,471 Supplemental Nutrition Assistance Program (SNAP) households in 4 intervention groups who lived near a FINI retailer (farmers market or grocery store) and 4 matched comparison groups who did not live near a FINI retailer.

MAIN OUTCOME MEASURES

Awareness and use of point-of-sale incentives and changes in self-reported monthly household FV expenditures.

ANALYSIS

Ordinary least squares intent-to-treat regression model using lagged dependent variable model framework.

RESULTS

Awareness of FINI was higher among households who were near a FINI retailer and had shopped there before FINI than those who lived near a FINI retailer but had not shopped there before FINI; the number of information sources from which SNAP participants heard about FINI was positively associated with incentive receipt (P < 0.05). Among those who received incentives, the average amount of incentives received at the last shopping trip ranged from $15 to $23. The FINI program had a positive impact on the average monthly FV expenditures for those in the farmers market shopper, grocery store shopper, and grocery store general intervention groups-increases ranged from $9 to $15 (P < 0.05).

CONCLUSIONS AND IMPLICATIONS

Point-of-sale incentives were associated with an increase in FV expenditures among SNAP households. Further research is needed to examine (1) effective messaging strategies to increase incentive awareness and (2) the long-term impact of incentives on FV expenditures.

Direct Medical Costs of Advanced Breast Cancer Treatment: A Real-World Study in the Southeast of The Netherlands

OBJECTIVES

Policy makers increasingly seek to complement data from clinical trials with information from routine care. This study aims to provide a detailed account of the hospital resource use and associated costs of patients with advanced breast cancer in The Netherlands.

METHODS

Data from 597 patients with advanced breast cancer, diagnosed between 2010 and 2014, were retrieved from the Southeast Netherlands Advanced Breast Cancer Registry. Database lock for this study was in October 2017. We report the observed hospital costs for different resource categories and the lifetime costs per patient, adjusted for censoring using Lin's method. The relationship between patients' characteristics and costs was studied using multivariable regression.

RESULTS

The average (SE) lifetime hospital costs of patients with advanced breast cancer were €52 709 (405). Costs differed considerably between patient subgroups, ranging from €29 803 for patients with a triple-negative subtype to €92 272 for patients with hormone receptor positive and human epidermal growth factor receptor 2 positive cancer. Apart from the cancer subtype, several other factors, including age and survival time, were independently associated with patient lifetime costs. Overall, a large share of costs was attributed to systemic therapies (56%), predominantly to a few expensive agents, such as trastuzumab (15%), everolimus (10%), and bevacizumab (9%), as well as to inpatient hospital days (20%).

CONCLUSIONS

This real-world study shows the high degree of variability in hospital resource use and associated costs in advanced breast cancer care. The presented resource use and costs data provide researchers and policy makers with key figures for economic evaluations and budget impact analyses.

The Impact of Driving Time to Family Planning Facilities on Preventive Service Use in Ohio

INTRODUCTION

Publicly funded family planning clinics provide preventive health services to low-income populations in the U.S. In recent years, several states, including Ohio, have restricted public funds for organizations that provide or refer patients to abortion care, often resulting in clinic closures. This research evaluates the effects of such closures on preventive service use and access to care among female adults in Ohio.

METHODS

With data from the 2010 to 2015 Ohio Behavioral Risk Factor Surveillance System, trends in health service use were assessed for female respondents aged 18-45 years with household incomes <$50,000. Clinic locations were combined with restricted-access survey ZIP codes to compute respondents' driving times to the nearest family planning clinic. The association between changes in driving time and the use of routine preventive and unmet care owing to cost were assessed with linear probability models. Analyses took place from March 2019 to February 2020.

RESULTS

Each additional 10 minutes of driving time was associated with an 8.9 percentage point increase in the likelihood of avoided care owing to cost (95% CI=1.7, 16.2), a 10.4 percentage point decrease in the likelihood of mammogram receipt during the past 12 months (95% CI= -22.3, 1.5), and a 12.5 percentage point decrease in the likelihood of ever receiving a clinical breast examination (95% CI= -18.7, -6.3). Driving time had insignificant associations with other utilization outcomes. Similar results were obtained when using driving distance.

CONCLUSIONS

Reduced access to family planning clinics was associated with unmet care due to cost and a reduction in preventive service use among low-income, reproductive-aged females.

Leveraging Auxiliary Information on Marginal Distributions in Nonignorable Models for Item and Unit Nonresponse

Often, government agencies and survey organizations know the population counts or percentages for some of the variables in a survey. These may be available from auxiliary sources, for example, administrative databases or other high quality surveys. We present and illustrate a model-based framework for leveraging such auxiliary marginal information when handling unit and item nonresponse. We show how one can use the margins to specify different missingness mechanisms for each type of nonresponse. We use the framework to impute missing values in voter turnout in a subset of data from the U.S. Current Population Survey (CPS). In doing so, we examine the sensitivity of results to different assumptions about the unit and item nonresponse.

Data Processing Improvements for Estimates of Health Insurance Coverage in the Current Population Survey Annual Social and Economic Supplement

Estimates of health insurance coverage in the United States rely on household-based surveys, and these surveys seek to improve data quality amid a changing health insurance landscape. We examine postcollection processing improvements to health insurance data in the Current Population Survey Annual Social and Economic Supplement (CPS ASEC), one of the leading sources of coverage estimates. The implementation of updated data extraction and imputation procedures in the CPS ASEC marks the second stage of a two-stage improvement and the beginning of a new time series for health insurance estimates. To evaluate these changes, we compared estimates from two files that introduce the updated processing system with two files that use the legacy system. We find that updates resulted in higher rates of health insurance coverage and lower rates of dual coverage, among other differences. These results indicate that the updated data processing improves coverage estimates and addresses previously noted limitations of the CPS ASEC.

Activity Engagement and Cognitive Performance Amongst Older Adults

Research supporting cognitive reserve theory suggests that engaging in a variety of cognitive, social, and physical activities may serve as protective factors against age-related changes in mental functioning, especially if the activities are cognitively engaging. Individuals who participate in a variety of cognitive activities have been found to be more likely to maintain a higher level of cognitive functioning and be less likely to develop dementia. In this study, we explore the relationship between engaging in a variety of activities and cognitive performance amongst 206 healthy older adults between the ages of 65–85. Age and years of education were found to be the most significant predictors of a global composite representing cognitive performance, consistent with previous work linking these variables to age-related changes in cognition and the cognitive reserve. We interpret these results to suggest that age and education are better predictors of global cognitive performance in older adults than self-reported activity engagement.

Assessment of Family, Peers, and Externalising Behaviour Dimensions in Adolescence: The Proposal of a Comprehensive Instrument (FPEB)

In the context of externalising behaviour problems, risk factor research (RFR) focuses on risk and protective factors of juvenile delinquency, which can pertain to individual, system, and societal levels. Several instruments aiming at measuring these factors have been developed, but a comprehensive research tool is missing. The aim of the present study was to develop and validate a questionnaire, the "Family, Peers, and Externalising Behaviour in adolescence" (FPEB) as a tool for assessing adolescents' tendency of externalising behaviour, the quality of relation with their parents, and peer-relations. FPEB was administered to 835 Italian students (36.8% males, age M = 13.81, SD = 1.54) together with the Moral Disengagement questionnaire to test concurrent validity. Data about socio-demographics and school performance were also collected. An EFA (Promax rotation, subsample A, n = 444) resulted in a four-factor structure that was corroborated by a CFA (subsample B, n = 388). The factors were "externalising behaviour" (var 13.16%), "peer relations difficulties" (var 11.10%), "Family conflict" (var 8.32%), and "lack of family negotiation" (var 7.11%) and showed good internal consistency (all α ≥ 0.65). There were differences between males and females in the correlational patterns of the four factors. The FPEB factors also showed good concurrent validity: two of the four factors ("lack of family negotiation" and "externalising behaviour") and the total score of the scale correlated with the "Moral disengagement scale", whereas peer relation difficulties did not. Further analyses also showed gender differences (except for "peer relations difficulties") and an association between students' school performance and "externalising behaviour", "family conflict", and the total FPEB scores. We concluded that the FPEB is a tool that is potentially useful to assess risk and protective factors and to plan targeted interventions (focusing on the specific area). Limitations and suggestions for further improvements are also discussed.

Associations between supermarket availability and body size in Australia: a cross-sectional observational study comparing state and territory capital cities

Background

Residential environment features such as availability of supermarkets may shape dietary behaviour and thus overweight and obesity. This relationship may not be consistent between cities. This Australian national-level study examined: 1) the relationship between supermarket availability and body size; and 2) whether this relationship varied by capital city.

Methods

This study used 2017–18 Australian National Health Survey data including individual-level socio-demographic information (age, sex, country of birth, education, occupation, household income), and measured body size (height and weight to derive body mass index [BMI], and waist circumference [WC]). Objectively-expressed measures of residential environments included: counts of supermarkets (major chain outlets), counts of amenities (representing walkable destinations including essential services, recreation, and entertainment), and area of public open space - each expressed within road-network buffers at 1000 m and 1500 m; population density (1km² grid cells); and neighbourhood disadvantage (Index of Relative Socioeconomic Disadvantage) expressed within Statistical Area Level 1 units. Data for adult respondents ≥18 years residing in each of Australia’s state and territory capital cities (n = 9649) were used in multilevel models to estimate associations between supermarket availability and body size sequentially accounting for individual and other environment measures. An interaction term estimated city-specific differences in associations between supermarket availability and body size. Models were consequently repeated stratified by city.

Results

Body size (BMI and WC) and supermarket availability varied between cities. Initial inverse associations between supermarket availability and body size (BMI and WC) were attenuated to null with inclusion of all covariates, except for BMI in the 1000 m buffer model (beta = − 0.148, 95%CI -0.27, − 0.01, p = 0.025). In stratified analyses, the strengths of associations varied between cities, remaining statistically significant only for some cities (BMI: Melbourne, Brisbane Hobart; WC: Brisbane, Hobart) in fully adjusted models. Different patterns of attenuation of associations with inclusion of covariates were evident for different cities.

Conclusions

For Australian capital cities, greater availability of supermarkets is associated with healthful body size. Marked between-city variations in body size, supermarket availability, and relationships between supermarket availability and body size do not, however, support universal, “one-size-fits-all” solutions to change built environments to support healthful body size.

Associations of ultrasound estimated early mid pregnancy visceral and subcutaneous fat depths and early pregnancy BMI with adverse neonatal outcomes

This study investigated whether maternal central adiposity and body mass index (BMI) were associated with neonatal hypoglycemia and adverse neonatal outcomes. A cohort study was performed at Uppsala University Hospital, Sweden, between 2015 and 2018. Visceral and subcutaneous fat depths were measured by ultrasound at the early second-trimester anomaly scan in 2771 women giving birth to singleton infants. Body mass index was assessed in early pregnancy. Logistic regression models were performed. Adjustments were made for age, BMI (not in model with BMI as exposure), smoking, maternal country of birth, and parity. Outcomes were neonatal hypoglycemia (blood glucose concentration < 2.6 mmol/l), a composite of adverse neonatal outcomes (Apgar < 7 at 5 min of age, or umbilical artery pH ≤ 7.0, or admission to neonatal intensive care unit), and the components of the composite outcome. Visceral and subcutaneous fat depths measured by ultrasound in early mid pregnancy were not associated with any of the outcomes in adjusted analyses. For every unit increase in BMI, the likelihood of neonatal hypoglycemia increased by 5% (aOR 1.05, 95% CI 1.01–1.10), the composite outcome by 5% (aOR 1.05, 95% CI 1.01–1.08), and admission to neonatal intensive care unit by 6% (aOR 1.06, 95% CI 1.02–1.10).

Prediction of evening fatigue severity in outpatients receiving chemotherapy: less may be more

Background

Fatigue is the most common and debilitating symptom experienced by oncology patients undergoing chemotherapy. Little is known about patient characteristics that predict changes in fatigue severity over time.

Purpose

To predict the severity of evening fatigue in the week following the administration of chemotherapy using machine learning approaches.

Methods

Outpatients with breast, gastrointestinal, gynecological, or lung cancer (N=1217) completed questionnaires one week prior to and one week following administration of chemotherapy. Evening fatigue was measured with the Lee Fatigue Scale (LFS). Separate prediction models for evening fatigue severity were created using clinical, symptom, and psychosocial adjustment characteristics and either evening fatigue scores or individual fatigue item scores. Prediction models were created using two regression and three machine learning approaches.

Results

Random forest (RF) models provided the best fit across all models. For the RF model using individual LFS item scores, two of the 13 individual LFS items (i.e., "worn out", "exhausted") were the strongest predictors.

Conclusion

This study is the first to use machine learning techniques to predict evening fatigue severity in the week following chemotherapy from fatigue scores obtained in the week prior to chemotherapy. Our findings suggest that the language used to assess clinical fatigue in oncology patients is important and that two simple questions may be used to predict evening fatigue severity.

Structural validity of the Child Behavior Checklist (CBCL) for preschoolers with neurogenetic syndromes

BACKGROUND

Psychologists routinely use the Child Behavior Checklist for Ages 1½-5 (CBCL) to assess challenging behaviors of preschoolers with developmental disabilities. However, the CBCL has not been thoroughly validated in neurogenetic syndromes (NGS).

AIM

We investigated the structural validity of the CBCL in NGS.

METHODS

Based on 152 preschoolers with Angelman, fragile X, Prader-Willi, and Williams syndromes, we employed confirmatory factor analysis (CFA) to evaluate the goodness-of-fit of CBCL narrowband, broadband, and DSM-oriented scales.

RESULTS

CFA models largely supported the unidimensionality of most narrowband scales and the two-factor structure of internalizing and externalizing broadband scales. However, there was limited evidence for the unidimensionality of most DSM-oriented scales.

CONCLUSIONS

Psychologists may consider using the CBCL as a psychometrically sound narrowband and broadband measure of challenging behaviors but should exercise caution when interpreting DSM-oriented scales for preschoolers with NGS. Our findings underscore a continued need to enhance assessment measures for identifying early precursors of child psychopathology in pediatric populations with atypical developmental trajectories.

Depression and antecedent medication adherence in a cohort of new metformin users

AIMS

The association between depression and poor medication adherence is based on cross-sectional studies and cohort studies that measure adherence rates after depression status is determined. However, depressive symptoms occur well before diagnosis. This study examined adherence patterns in the year before a depressive episode.

METHODS

This retrospective cohort study followed new metformin users identified in Alberta Health's administrative data between 2008 and 2018. Depressive episodes starting ≥1 year after metformin initiation were identified using a validated case definition. Controls were randomly assigned a pseudo depression date. Adherence to oral antihyperglycemic medications was estimated using proportion of days covered (PDC) and group-based trajectory models to explore the association between depression and poor adherence (PDC<0.8).

RESULTS

A depressive episode occurred in 17,418 (10.6%) of 165,056 new metformin users. Individuals with depression were more likely to have poor adherence compared to controls (adjusted odds ratio 1.21; 95% CI 1.17, 1.26). Five trajectories were identified: nearly perfect adherence (PDC >0.95 [34.8% of cohort]), discontinued (PDC=0 [18.3% of cohort], poor initial adherence (PDC 0.75) that declined either rapidly (9.2% of cohort) or gradually (30.1% of cohort), and poor initial adherence (PDC 0.26) that increased gradually (7.6% of cohort). Individuals with depression were more likely to be in one of the four trajectories of poor adherence compared to controls (adjusted odds ratio 1.24; 95% CI 1.19-1.29).

CONCLUSIONS

Poor medication adherence occurs in the year before a depressive episode; therefore, poor medication use patterns could be used as an early warning sign for depression.

County-level data on U.S. opioid distributions, demographics, healthcare supply, and healthcare access

The dataset summarized in this article is a combination of several of U.S. federal data resources for the years 2006-2013, containing county-level variables for opioid pill volumes, demographics (e.g. age, race, ethnicity, income), insurance coverage, healthcare demand (e.g. inpatient and outpatient service utilization), healthcare infrastructure (e.g. number of hospital beds or hospices), and the supply of various types of healthcare providers (e.g. medical doctors, specialists, dentists, or nurse practitioners). We also include indicators for states which permitted opioid prescribing by nurse practitioners. This dataset was originally created to assist researchers in identifying which factors predict per capita opioid pill volume (PCPV) in a county, whether early state Medicaid expansions increased PCPV, and PCPV's association with opioid-related mortality. Missing data were imputed using regression analysis and hot deck imputation. Non-imputed values are also reported.

Taken together, our data provide a new level of precision that may be leveraged by scholars, policymakers, or data journalists who are interested in studying the opioid epidemic. Researchers may use this dataset to identify patterns in opioid distribution over time and characteristics of counties or states which were disproportionately impacted by the epidemic. These data may also be joined with other sources to facilitate studies on the relationships between opioid pill volume and a wide variety of health, economic, and social outcomes.

Identifying sociodemographic profiles of veterans at risk for high-dose opioid prescribing using classification and regression trees

OBJECTIVE

To identify sociodemographic profiles of patients prescribed high-dose opioids.

DESIGN

Cross-sectional cohort study.

SETTING/PATIENTS

Veterans dually-enrolled in Veterans Health Administration and Medicare Part D, with ≥1 opioid pre-scription in 2012.

MAIN OUTCOME MEASURES

We identified five patient-level demographic characteristics and 12 community variables re-flective of region, socioeconomic deprivation, safety, and internet connectivity. Our outcome was the proportion of vet-erans receiving >120 morphine milligram equivalents (MME) for ≥90 consecutive days, a Pharmacy Quality Alliance measure of chronic high-dose opioid prescribing. We used classification and regression tree (CART) methods to identify risk of chronic high-dose opioid prescribing for sociodemographic subgroups.

RESULTS

Overall, 17,271 (3.3 percent) of 525,716 dually enrolled veterans were prescribed chronic high-dose opioids. CART analyses identified 35 subgroups using four sociodemographic and five community-level measures, with high-dose opioid prescribing ranging from 0.28 percent to 12.1 percent. The subgroup (n = 16,302) with highest frequency of the outcome included veterans who were with disability, age 18-64 years, white or other race, and lived in the Western Census region. The subgroup (n = 14,835) with the lowest frequency of the outcome included veterans who were with-out disability, did not receive Medicare Part D Low Income Subsidy, were >85 years old, and lived in communities within the second and sixth to tenth deciles of community public assistance.

CONCLUSIONS

Using CART analyses with sociodemographic and community-level variables only, we identified sub-groups of veterans with a 43-fold difference in chronic high-dose opioid prescriptions. Interactions among disability, age, race/ethnicity, and region should be considered when identifying high-risk subgroups in large populations.

A survey on missing data in machine learning

Machine learning has been the corner stone in analysing and extracting information from data and often a problem of missing values is encountered. Missing values occur because of various factors like missing completely at random, missing at random or missing not at random. All these may result from system malfunction during data collection or human error during data pre-processing. Nevertheless, it is important to deal with missing values before analysing data since ignoring or omitting missing values may result in biased or misinformed analysis. In literature there have been several proposals for handling missing values. In this paper, we aggregate some of the literature on missing data particularly focusing on machine learning techniques. We also give insight on how the machine learning approaches work by highlighting the key features of missing values imputation techniques, how they perform, their limitations and the kind of data they are most suitable for. We propose and evaluate two methods, the k nearest neighbor and an iterative imputation method (missForest) based on the random forest algorithm. Evaluation is performed on the Iris and novel power plant fan data with induced missing values at missingness rate of 5% to 20%. We show that both missForest and the k nearest neighbor can successfully handle missing values and offer some possible future research direction.

Identifying profiles of need among psychiatric inpatients approaching discharge in New York City: a latent class analysis

PURPOSE

Understanding the needs of individuals transitioning to the community following a psychiatric hospitalization can inform community service planning. This study is among the first to examine the needs of a sample of psychiatric inpatients approaching discharge in a large urban area in the USA.

METHODS

Representative data were drawn from 1129 acutely hospitalized psychiatric inpatients from eight New York City hospitals. Descriptive statistics were used to estimate patient needs at discharge across nine domains: housing, employment, income, transportation, education, time use, social support, and help accessing medical and mental health care. Latent class analysis (LCA) was applied to identify subgroups of patients based on needs profiles. Multinomial logistic regression was used to investigate socio-demographic associations with class membership.

RESULTS

Respondents were most likely to have needs related to income (50.7%), housing (49.2%), and employment (48.7%). Results from the LCA suggested a five class solution of patient needs: three domain-specific classes whose members endorsed needs for 'housing and employment' (22.5%), 'social support and time use' (15.0%) and 'access to care' (6.4%) and two classes where overall member needs were high ('high needs,'18.4%) or low ('low needs,' 37.7%) across all needs. Compared to the 'low needs' class, members of the 'high needs' class had significantly greater odds of being black or Latino, male, uninsured, and parents of a child under 18 years.

CONCLUSION

Patients have unique profiles of need that are significantly associated with the socio-demographic characteristics. These findings may help practitioners and policymakers improve mental health services.

Quality of life in patients with resectable rectal cancer during the first 24 months following diagnosis

AIM

An increasing number of patients survive rectal cancer, resulting in more patients living with the side-effects of the treatment. Exploring quality of life before and after treatment enables follow-up and additional treatment to be adjusted to the patient's needs. The aim of the study was to describe the quality of life during the 24 months following diagnosis and to identify risk factors for poor quality of life.

METHOD

This is a prospective cohort study of patients with rectal cancer followed up by extensive questionnaires. Patients from 16 surgical departments in Denmark and Sweden from 2012 to 2015 were included. The self-assessed quality of life was measured with a seven-point Likert scale.

RESULTS

A total of 1110 patients treated with curative intent were included, and the response rate at the 24-month follow-up was 71%. Patients with rectal cancer assessed their quality of life before start of treatment as poorer than that of a reference population. At the 12- and 24-month follow-up, the quality of life on group level had recovered to the same level as for the reference population. Risk factors for poor quality of life included bother with urinary, bowel and stoma function. A reference population was used for comparison.

CONCLUSION

The quality of life of patients with resectable rectal cancer recovered to levels comparable to a reference population 12 and 24 months after diagnosis. Our results indicate that the urinary, bowel and stoma function has an impact on quality of life.

A data driven methodology for social science research with left-behind children as a case study

For decades, traditional correlation analysis and regression models have been used in social science research. However, the development of machine learning algorithms makes it possible to apply machine learning techniques for social science research and social issues, which may outperform standard regression methods in some cases. Under the circumstances, this article proposes a methodological workflow for data analysis by machine learning techniques that have the possibility to be widely applied in social issues. Specifically, the workflow tries to uncover the natural mechanisms behind the social issues through a data-driven perspective from feature selection to model building. The advantage of data-driven techniques in feature selection is that the workflow can be built without so much restriction of related knowledge and theory in social science. The advantage of using machine learning techniques in modelling is to uncover non-linear and complex relationships behind social issues. The main purpose of our methodological workflow is to find important fields relevant to the target and provide appropriate predictions. However, to explain the result still needs theory and knowledge from social science. In this paper, we trained a methodological workflow with left-behind children as the social issue case, and all steps and full results are included.

A Hybrid Approach for the Stratified Mark-Specific Proportional Hazards Model with Missing Covariates and Missing Marks, with Application to Vaccine Efficacy Trials

Deployment of the recently licensed CYD-TDV dengue vaccine requires understanding of how the risk of dengue disease in vaccine recipients depends jointly on a host biomarker measured after vaccination (neutralization titer - NAb) and on a "mark" feature of the dengue disease failure event (the amino acid sequence distance of the dengue virus to the dengue sequence represented in the vaccine). The CYD14 phase 3 trial of CYD-TDV measured NAb via case-cohort sampling and the mark in dengue disease failure events, with about a third missing marks. We addressed the question of interest by developing inferential procedures for the stratified mark-specific proportional hazards model with missing covariates and missing marks. Two hybrid approaches are investigated that leverage both augmented inverse probability weighting and nearest neighborhood hot deck multiple imputation. The two approaches differ in how the imputed marks are pooled in estimation. Our investigation shows that NNHD imputation can lead to biased estimation without properly selected neighborhood. Simulations show that the developed hybrid methods perform well with unbiased NNHD imputations from proper neighborhood selection. The new methods applied to CYD14 show that NAb is strongly inversely associated with risk of dengue disease in vaccine recipients, more strongly against dengue viruses with shorter distances.

Association of maternal central adiposity measured by ultrasound in early mid pregnancy with infant birth size

We sought to investigate whether early mid pregnancy visceral and subcutaneous fat depths measured by ultrasound were associated with infant birth size, independent of early pregnancy BMI. A cohort study was performed at Uppsala University Hospital, Sweden, between 2015–2018. Visceral and subcutaneous fat depths were measured at the early second-trimester anomaly scan in 2498 women, giving birth to singleton, term infants. Primary outcomes were birthweight and LGA (birthweight standard deviation score > 90th percentile in the cohort). Linear and logistic regression models were used, adjusted for BMI, age, smoking, parity, maternal country of birth, gestational age and infant sex. A 5-mm increase in visceral fat depth was associated with an increase in birthweight of 8.3 g [95% confidence interval (CI) 2.5 − 14.1 g], after adjustments, and a 6% increase in the adjusted odds of having an infant born LGA (OR 1.06, CI 1.02–1.11). There was no association between subcutaneous fat depth and birthweight or LGA after covariate adjustments. Hence, visceral fat depth measured by ultrasound in early mid pregnancy was associated with excessive fetal growth, independent of early pregnancy BMI, and may be useful in models for predicting LGA infants.

Mental Health Screening for COVID-19: a Proposed Cutoff Score for the Greek Version of the Fear of COVID-19 Scale (FCV-19S)

The COVID-19 pandemic elicited fear. The Fear of COVID-19 Scale (FCV-19S) is a newly developed self-reported measure, originally developed in Persian to assess COVID-19-related fear. To date, the scale has been translated and validated in 19 other languages, among which Greek. This study, conducted through an online survey, aimed to further explore the validity of the Greek FCV-19S version, as well as to identify appropriate cutoff scores. A total of 538 respondents completed the sociodemographic data sheet, the Generalized Anxiety Disorder 7-item scale, the Short Health Anxiety Inventory, and the Posttraumatic Stress Disorder-8 inventory. According to the results, a cutoff point score of 16.5 or higher revealed a significant predictive power for anxiety, health anxiety, and posttraumatic stress symptoms. Altogether, although the cutoff scores of the Greek FCV-19S version were explored to further evaluate the scale's validity, they may facilitate discrimination of adults with extreme COVID-19-related fear from those with normal fear reactions.

Measuring Statistics Attitudes at the Student and Instructor Levels: A Multilevel Construct Validity Study of the Survey of Attitudes Toward Statistics

Numerous studies have been conducted using the Survey of Attitudes Toward Statistics-36 (SATS-36). Recently, large-scale assessment studies have begun to examine the extent to which students vary in their statistics attitudes across instructors. Yet, empirical evidence linking student responses to the SATS items to instructor-level constructs is still lacking. Using multilevel confirmatory factor analysis, we investigated the factor structure underlying the measure of students’ statistics attitudes at both the student and instructor levels. Results from 13,507 college students taught by 160 introductory statistics instructors support a correlated six-factor model at each level. Additionally, there is evidence for the structural validity of a shared teacher–student attitude impacts construct that may capture meaningful patterns of teaching characteristics and competencies tied to student development of statistics attitudes. These findings provide empirical support for the use of the SATS-36 in studying contextual variables in relation to statistics instructors. Implications for educational practice are discussed.

Psychiatric healthcare utilisation among refugee adolescents and their peers in Denmark

PURPOSE

To investigate the psychiatric healthcare utilisation of refugees vis-à-vis their peers in Denmark during the ages 15-22.

METHODS

This paper utilises comprehensive full-population registry data from 1995 to 2016 to explore the psychiatric healthcare utilisation during the transition from childhood to adulthood for refugees (N = 13,027), a comparison group of children of labour migrants from Morocco, Pakistan, and Turkey (N = 13,413), and the majority population (N = 693,043) in Denmark. To test for population differences in types of admission for particular types of disorders, odds ratios for a first contact during ages 15-22 were calculated using logistic regression. For those with at least one diagnosis-specific hospital contact, differences in the amount and type of treatment were tested using negative binomial regression to estimate means ratios of days hospitalised, days in outpatient care, number of outpatient contacts, consultations with psychiatrists in private practice, and prescribed medicine purchases.

RESULTS

Refugees and the comparison group were generally less likely than the majority population to have a first contact for most disorders (adjusted ORs 0.03-0.88), but not for schizophrenia for boys (adjusted ORs 0.92-2.13). Among those who did have a first contact, youths from the ethnic minority groups tended to have more or similar inpatient and emergency room contacts (MRs 0.89-2.10), hospitalisations of refugee girls being an exception (MR 0.46; CI [0.23-0.94]), but fewer outpatient contacts, consultations with psychiatrists in private practice, and prescribed medicine purchases (MRs 0.23-0.94).

CONCLUSIONS

The results suggest that refugee and other ethnic minority groups may face barriers both to initial contact and to completing adequate treatment beyond the first contact.