A comparison of multiple imputation methods for handling missing values in longitudinal data in the presence of a time-varying covariate with a non-linear association with time: a simulation study

Association Of Estimated Plasma Volume with New Onset Acute Kidney Injury in Hospitalized COVID-19 Patients

PURPOSE

To explore the association of estimated plasma volume (ePV) and plasma volume status (PVS) as surrogates of volume status with new-onset AKI and in-hospital mortality among hospitalized COVID-19 patients.

MATERIALS AND METHODS

We performed a retrospective multi-center study on COVID-19-related ARDS patients who were admitted to the Mayo Clinic Enterprise health system. Plasma volume was calculated using the formulae for ePV and PVS, and longitudinal analysis was performed to find the association of ePV and PVS with new-onset AKI during hospitalization as the primary outcome and in-hospital mortality as a secondary outcome.

RESULTS

Our analysis included 7616 COVID-19 patients with new-onset AKI occurring in 1365 (17.9%) and a mortality rate of 25.96% among them. A longitudinal multilevel multivariate analysis showed both ePV (OR 1.162; 95% CI 1.048-1.288, p=0.004) and PVS (OR 1.032; 95% CI 1.012-1.050, p=0.001) were independent predictors of new onset AKI. Higher PVS was independently associated with increased in-hospital mortality (OR 1.038, 95% CI 1.007-1.070, p=0.017), but not ePV (OR 0.868, 95% CI 0.740-1.018, p=0.082).

CONCLUSION

A higher PVS correlated with a higher incidence of new-onset AKI and worse outcomes in our cohort of hospitalized COVID-19 patients. Further large-scale and prospective studies are needed to understand its utility.

Acute kidney injury in patients before and after extracorporeal membrane oxygenation (ECMO) - Retrospective longitudinal analysis of the hospital outcomes

PURPOSE

Acute Kidney Injury (AKI) occurs in up to 85% of patients managed by ECMO support. Limited data are available comparing the outcomes among patients who develop AKI before and after ECMO initiation.

METHODS

A retrospective longitudinal observational study was performed on all adult patients placed on ECMO from January 2000 to December 2015 at our institution. Longitudinal multivariate logistic regressional analysis was performed to identify the variables that are associated with the outcome measures (post-ECMO AKI and in-hospital mortality).

RESULTS

A total of 329 patients were included in our analysis in which AKI occurred in 176 (53%) and 137 (42%) patients before and after ECMO, respectively. In the multivariate analysis, increasing age, pre-existing chronic kidney disease (CKD), increasing bilirubin, decreasing fibrinogen, and use of LVAD had significant association with post-ECMO AKI. In-hospital mortality was seen in 128 out of 176 (73%) patients in the pre-ECMO AKI group and 32 out of 137 (42%) in the post-ECMO AKI group. In the multivariate analysis, age, interstitial lung disease, pre-ECMO AKI, and post-ECMO RRT requirement were independently associated with mortality.

CONCLUSION

AKI before ECMO initiation and the need for RRT post-ECMO are independently associated with poor patient survival.

Association of plasma volume status with outcomes in hospitalized Covid-19 ARDS patients: A retrospective multicenter observational study

PURPOSE

To evaluate the association of estimated plasma volume (ePV) and plasma volume status (PVS) on admission with the outcomes in COVID-19-related acute respiratory distress syndrome (ARDS) patients.

MATERIALS AND METHODS

We performed a retrospective multi-center study on COVID-19-related ARDS patients who were admitted to the Mayo Clinic Enterprise health system. Plasma volume was calculated using the formulae for ePV and PVS, and these variables were analyzed for correlation with patient outcomes.

RESULTS

Our analysis included 1298 patients with sequential organ failure assessment (SOFA) respiratory score ≥ 2 (PaO2/FIO2 ≤300 mmHg) and a mortality rate of 25.96%. A Cox proportional multivariate analysis showed PVS but not ePV as an independent correlation with 90-day mortality after adjusting for the covariates (HR: 1.015, 95% CI: 1.005-1.025, p = 0.002 and HR 1.054, 95% CI 0.958-1.159, p = 0.278 respectively).

CONCLUSION

A lower PVS on admission correlated with a greater chance of survival in COVID-19-related ARDS patients. The role of PVS in guiding fluid management should be investigated in future prospective studies.

Russian-language translation and cultural adaptation of the Norwegian 'Patient Experience Questionnaire'

The availability of patient-reported experience measures (PREM) is an unmet need in Russian healthcare.

Objective

To translate, adapt culturally, and validate PREM for outpatients.

Methods

A core set of questions from the Patient Experience Questionnaire (PEQ, in Norwegian, available in English) was translated to Russian (forward-backward translation). Acceptability, construct validity, and reliability were assessed. Patients aged ≥18 y.o. were invited to complete the questionnaire via QR-code within 24 h after a medical encounter.

Results

A questionnaire with adequate conceptual and linguistic equivalence was obtained. For four questions, a rating scale was replaced by Likert-type. A total of 308 responses were received (median age 55 y.o., 52% females). The correlation matrix was factorable. Four factors were extracted using varimax rotation: 1) outcome of this specific visit; 2) communication experiences; 3) communication competency; 4) emotions after this visit. These explained 65.4% of the total variance. Three items were excluded. The model was confirmed to be adequate. The Cronbach alpha was >0.9. Item-total correlation confirmed discriminative validity.

Conclusion

These preliminary results show that the Russian version of PEQ, adapted to national features, shows good psychometric properties. External validation is needed for the broad implementation of this PREM.

Innovation

This research is first attempt to use PREM in the Russian Federation. The use of quick response codes is feasible and eases survey conduction. The more PREMs are used the higher the quality of healthcare.

Comparison of Imputation Strategies for Incomplete Longitudinal Data in Life-Course Epidemiology

Incomplete longitudinal data are common in life-course epidemiology and may induce bias leading to incorrect inference. Multiple imputation (MI) is increasingly preferred for handling missing data, but few studies explore MI-method performance and feasibility in real-data settings. We compared 3 MI methods using real data under 9 missing-data scenarios, representing combinations of 10%, 20%, and 30% missingness and missing completely at random, at random, and not at random. Using data from Health and Retirement Study (HRS) participants, we introduced record-level missingness to a sample of participants with complete data on depressive symptoms (1998-2008), mortality (2008-2018), and relevant covariates. We then imputed missing data using 3 MI methods (normal linear regression, predictive mean matching, variable-tailored specification), and fitted Cox proportional hazards models to estimate effects of 4 operationalizations of longitudinal depressive symptoms on mortality. We compared bias in hazard ratios, root mean square error, and computation time for each method. Bias was similar across MI methods, and results were consistent across operationalizations of the longitudinal exposure variable. However, our results suggest that predictive mean matching may be an appealing strategy for imputing life-course exposure data, given consistently low root mean square error, competitive computation times, and few implementation challenges.

Multiple imputation of missing data in multilevel ecological momentary assessments: an example using smoking cessation study data

Advances in digital technology have greatly increased the ease of collecting intensive longitudinal data (ILD) such as ecological momentary assessments (EMAs) in studies of behavior changes. Such data are typically multilevel (e.g., with repeated measures nested within individuals), and are inevitably characterized by some degrees of missingness. Previous studies have validated the utility of multiple imputation as a way to handle missing observations in ILD when the imputation model is properly specified to reflect time dependencies. In this study, we illustrate the importance of proper accommodation of multilevel ILD structures in performing multiple imputations, and compare the performance of a multilevel multiple imputation (multilevel MI) approach relative to other approaches that do not account for such structures in a Monte Carlo simulation study. Empirical EMA data from a tobacco cessation study are used to demonstrate the utility of the multilevel MI approach, and the implications of separating participant- and study-initiated EMAs in evaluating individuals' affective dynamics and urge.

A wide range of missing imputation approaches in longitudinal data: a simulation study and real data analysis

BACKGROUND

Missing data is a pervasive problem in longitudinal data analysis. Several single-imputation (SI) and multiple-imputation (MI) approaches have been proposed to address this issue. In this study, for the first time, the function of the longitudinal regression tree algorithm as a non-parametric method after imputing missing data using SI and MI was investigated using simulated and real data.

METHOD

Using different simulation scenarios derived from a real data set, we compared the performance of cross, trajectory mean, interpolation, copy-mean, and MI methods (27 approaches) to impute missing longitudinal data using parametric and non-parametric longitudinal models and the performance of the methods was assessed in real data. The real data included 3,645 participants older than 18 years within six waves obtained from the longitudinal Tehran cardiometabolic genetic study (TCGS). The data modeling was conducted using systolic and diastolic blood pressure (SBP/DBP) as the outcome variables and included predictor variables such as age, gender, and BMI. The efficiency of imputation approaches was compared using mean squared error (MSE), root-mean-squared error (RMSE), median absolute deviation (MAD), deviance, and Akaike information criteria (AIC).

RESULTS

The longitudinal regression tree algorithm outperformed based on the criteria such as MSE, RMSE, and MAD than the linear mixed-effects model (LMM) for analyzing the TCGS and simulated data using the missing at random (MAR) mechanism. Overall, based on fitting the non-parametric model, the performance of the 27 imputation approaches was nearly similar. However, the SI traj-mean method improved performance compared with other imputation approaches.

CONCLUSION

Both SI and MI approaches performed better using the longitudinal regression tree algorithm compared with the parametric longitudinal models. Based on the results from both the real and simulated data, we recommend that researchers use the traj-mean method for imputing missing values of longitudinal data. Choosing the imputation method with the best performance is widely dependent on the models of interest and the data structure.

School absenteeism in children with special health care needs. Results from the prospective cohort study ikidS

OBJECTIVE

Children with special health care needs (SHCN) due to a chronic health condition perform more poorly at school compared to their classmates. There is still little knowledge on the causal pathways and which factors could be targeted by interventions. We, therefore, investigated school absenteeism in children with SHCN compared to their peers.

METHODS

This study was based on data from the German population-based prospective cohort study ikidS (German for: I will start school). Children with SHCN were identified by the Children with Special Health Care Needs screener that captures five consequences of physical or mental chronic health conditions: (1) use or need of prescription medication, (2) above average use or need of medical, mental health, or educational services, (3) functional limitations compared with others of the same age, (4) use or need of specialized therapies, and (5) treatment or counseling for emotional, behavioral, or developmental problems. School absenteeism was defined as days absent from school due to illness during first grade and was reported by classroom teachers. Associations between SHCN consequences and school absenteeism were investigated by negative binomial regression models. Effect estimates were adjusted for confounding variables identified by a causal framework and directed acyclic graphs.

RESULTS

1,921 children (mean age at follow-up 7.3 years, standard deviation 0.3; 49% females) were included; of these, 14% had SHCN. Compared to their classmates, children with SHCN had more days absent (adjusted rate ratio: 1.37; 95% confidence interval 1.16, 1.62). The effect was strongest among children with i) functional limitations, ii) treatment or counseling for emotional, behavioral, or developmental problems, and iii) those who experienced two or more SHCN consequences.

CONCLUSIONS

Children with SHCN have higher school absenteeism, which could-at least partly-explain their poorer school performance and lower educational attainment. SHCN-specific targeted interventions may reduce the adverse effects of SHCN on educational outcomes in children.

Mean versus variability of lipid measurements over 6 years and incident cardiovascular events: More than a decade follow-up

Background

Lipid variability (LV) has emerged as a contributor to the incidence of cardiovascular diseases (CVD), even after considering the effect of mean lipid levels. However, these associations have not been examined among people in the Middle East and North Africa (MENA) region. We aimed to investigate the association of 6-year mean lipid levels versus lipid variability with the risk of CVD among an Iranian population.

Methods

A total of 3,700 Iranian adults aged ≥ 30 years, with 3 lipid profile measurements, were followed up for incident CVD until March 2018. Lipid variability was measured as standard deviation (SD), coefficient of variation (CV), average real variability (ARV), and variability independent of mean (VIM). The effects of mean lipid levels and LV on CVD risk were assessed using multivariate Cox proportional hazard models.

Results

During a median 14.5-year follow-up, 349 cases of CVD were recorded. Each 1-SD increase in the mean levels of total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), TC/high-density lipoprotein cholesterol (HDL-C), and non-HDL-C increased the risk of CVD by about 26-29%; for HDL-C, the risk was significantly lower by 12% (all p-values < 0.05); these associations resisted after adjustment for their different LV indices. Considering LV, each 1-SD increment in SD and ARV variability indices for TC and TC/HDL-C increased the risk of CVD by about 10%; however, these associations reached null after further adjustment for their mean values. The effect of TC/HDL-C variability (measured as SD) and mean lipid levels, except for LDL-C, on CVD risk was generally more pronounced in the non-elderly population.

Conclusion

Six-year mean lipid levels were associated with an increased future risk of incident CVD, whereas LV were not. Our findings highlight the importance of achieving normal lipid levels over time, but not necessarily consistent, for averting adverse clinical outcomes.

Development of a long-term time-weighted exposure metric that accounts for missing data in the Seychelles Child Development Study

In many studies of the health effects of toxicants, exposure is measured once even though exposure may be continuous. However, some studies collect repeated measurements on participants over an extended time with the goal of determining a long-term metric that captures the average or cumulative exposure. This can be challenging, especially when exposure is measured at irregular intervals and has some missing values. Here we describe a method for determining a measure of long-term exposure using data on postnatal mercury (Hg) from the Seychelles Child Development Study (SCDS) Main Cohort as a model. In this cohort (n = 779), we incorporate postnatal Hg values that were measured on most study participants at seven ages, three between 6 months and 5.5 years ("childhood"), and an additional four between 17 and 24 years ("early adulthood"). We develop time-weighted measures of average exposure during the childhood and the early adulthood periods and compare the strengths and weaknesses of our metric to two standard measures: overall average and cumulative exposure. We account for missing values through an imputation method that uses information about age- and sex-specific Hg means and the participant's Hg values at similar ages to estimate subject-specific missing Hg values. We compare our method to the implicit imputation assumed by these two standard methods, and to Fully Conditional Specification (FCS), an alternative method of imputing missing data. To determine the accuracy of our imputation method we use data from participants with no missing Hg values in the relevant time window. The imputed values from our proposed method are substantially closer to the observed values on average than the average or cumulative exposure, while also performing slightly better than FCS. In conclusion, time-weighted long-term exposure appears to offer advantages over cumulative exposure in longitudinal studies with repeated measures where the follow-up period for a toxicant is similar for all participants. Additionally, our method to impute missing values maximizes the number of participants for whom the overall exposure metric can be calculated and should provide a more accurate long-term exposure metric than standard methods when exposure has missing values. Our method is applicable to any study of long-term toxicant effects when longitudinal exposure measurements are available but have missing values.

Trajectories of alcohol consumption in relation to all-cause mortality in patients with cardiovascular disease: a 35-year prospective cohort study

BACKGROUND AND AIMS

Research into alcohol consumption and cardiovascular disease (CVD) patients' prognosis has largely ignored the longitudinal dynamics in drinking behaviour. This study measured the association between alcohol consumption trajectories and mortality risk in CVD patients.

DESIGN

Prospective cohort study.

SETTING

UK-based Whitehall II Study.

PARTICIPANTS

A total of 1306 participants with incident non-fatal CVD (coronary heart disease/stroke) events.

MEASUREMENTS

Up to eight repeated measures of alcohol intake were available for each patient from the most recent assessment phase pre-incident CVD and all subsequent phases post-incident CVD, spanning up to three decades. Six trajectory groups of alcohol consumption were identified using group-based trajectory modelling and related to the risk of all-cause mortality, adjusting for demographics and changes in life-style and health status.

FINDINGS

Three hundred and eighty deaths were recorded during a median follow-up of 5 years after patients' last alcohol assessment. Compared with patients who consistently drank moderately (≤ 14 units/week), former drinkers had a greater risk of mortality (hazard ratio = 1.74, 95% confidence interval = 1.19-2.54) after adjustment for covariates. There was no significantly increased risk of mortality in long-term abstainers, reduced moderate drinkers, stable or unstable heavy drinkers. Cross-sectional analyses based only on drinking information at patients' last assessment found no significant differences in mortality risk for abstainers, former or heavy drinkers versus moderate drinkers.

CONCLUSIONS

Cardiovascular disease patients who consistently drink ≤ 14 units/week appear to have a similar risk of mortality to those who are long-term abstainers, which does not support a protective effect of moderate drinking on total mortality. Cardiovascular disease patients who stop drinking appear to have increased mortality risk compared with continuous moderate drinkers, but this may be linked to poor self-rated health before cardiovascular disease onset.

Analytical challenges in estimating the effect of exposures that are bounded by follow-up time: experiences from the Blood Stream Infection-Focus on Outcomes study

Objective

To illustrate the challenges of estimating the effect of an exposure that is bounded by duration of follow-up on all-cause 28-day mortality, whilst simultaneously addressing missing data and time-varying covariates.

Study design and methods

BSI-FOO is a multicentre cohort study with the primary aim of quantifying the effect of modifiable risk factors, including time to initiation of therapy, on all-cause 28-day mortality in patients with bloodstream infection. The primary analysis involved two Cox proportional hazard models, first one for non-modifiable risk factors and second one for modifiable risk factors, with a risk score calculated from the first model included as a covariate in the second model. Modifiable risk factors considered in this study were recorded daily for a maximum of 28 days after infection. Follow-up was split at daily intervals from day 0 to 28 with values of daily collected data updated at each interval (i.e., one row per patient per day).

Analytical challenges

Estimating the effect of time to initiation of treatment on survival is analytically challenging since only those who survive to time t can wait until time t to start treatment, introducing immortal time bias. Time-varying covariates representing cumulative counts were used for variables bounded by survival time e.g. the cumulative count of days before first receipt of treatment. Multiple imputation using chained equations was used to impute missing data, using conditional imputation to avoid imputing non-applicable data e.g. ward data after discharge.

Conclusion

Using time-varying covariates represented by cumulative counts within a one row per day per patient framework can reduce the risk of bias in effect estimates. The approach followed uses established methodology and is easily implemented in standard statistical packages.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01393-9.

Derivation and validation of a machine learning risk score using biomarker and electronic patient data to predict progression of diabetic kidney disease

AIM

Predicting progression in diabetic kidney disease (DKD) is critical to improving outcomes. We sought to develop/validate a machine-learned, prognostic risk score (KidneyIntelX™) combining electronic health records (EHR) and biomarkers.

METHODS

This is an observational cohort study of patients with prevalent DKD/banked plasma from two EHR-linked biobanks. A random forest model was trained, and performance (AUC, positive and negative predictive values [PPV/NPV], and net reclassification index [NRI]) was compared with that of a clinical model and Kidney Disease: Improving Global Outcomes (KDIGO) categories for predicting a composite outcome of eGFR decline of ≥5 ml/min per year, ≥40% sustained decline, or kidney failure within 5 years.

RESULTS

In 1146 patients, the median age was 63 years, 51% were female, the baseline eGFR was 54 ml min-1 [1.73 m]-2, the urine albumin to creatinine ratio (uACR) was 6.9 mg/mmol, follow-up was 4.3 years and 21% had the composite endpoint. On cross-validation in derivation (n = 686), KidneyIntelX had an AUC of 0.77 (95% CI 0.74, 0.79). In validation (n = 460), the AUC was 0.77 (95% CI 0.76, 0.79). By comparison, the AUC for the clinical model was 0.62 (95% CI 0.61, 0.63) in derivation and 0.61 (95% CI 0.60, 0.63) in validation. Using derivation cut-offs, KidneyIntelX stratified 46%, 37% and 17% of the validation cohort into low-, intermediate- and high-risk groups for the composite kidney endpoint, respectively. The PPV for progressive decline in kidney function in the high-risk group was 61% for KidneyIntelX vs 40% for the highest risk strata by KDIGO categorisation (p < 0.001). Only 10% of those scored as low risk by KidneyIntelX experienced progression (i.e., NPV of 90%). The NRIevent for the high-risk group was 41% (p < 0.05).

CONCLUSIONS

KidneyIntelX improved prediction of kidney outcomes over KDIGO and clinical models in individuals with early stages of DKD.

A Novel Approach Integrating Hierarchical Clustering and Weighted Combination for Association Study of Multiple Phenotypes and a Genetic Variant

As a pivotal research tool, genome-wide association study has successfully identified numerous genetic variants underlying distinct diseases. However, these identified genetic variants only explain a small proportion of the phenotypic variation for certain diseases, suggesting that there are still more genetic signals to be detected. One of the reasons may be that one-phenotype one-variant association study is not so efficient in detecting variants of weak effects. Nowadays, it is increasingly worth noting that joint analysis of multiple phenotypes may boost the statistical power to detect pathogenic variants with weak genetic effects on complex diseases, providing more clues for their underlying biology mechanisms. So a Weighted Combination of multiple phenotypes following Hierarchical Clustering method (WCHC) is proposed for simultaneously analyzing multiple phenotypes in association studies. A series of simulations are conducted, and the results show that WCHC is either the most powerful method or comparable with the most powerful competitor in most of the simulation scenarios. Additionally, we evaluated the performance of WCHC in its application to the obesity-related phenotypes from Atherosclerosis Risk in Communities, and several associated variants are reported.

Perinatal Depressive Symptoms, HIV Suppression, and the Underlying Role of ART Adherence: A Longitudinal Mediation Analysis in the IMPAACT P1025 Cohort

Background

Women with HIV have higher risk of depressive symptoms in the perinatal period. Evidence on how perinatal depressive symptoms affect viral suppression (VS) and adherence to antiretroviral therapy (ART) remains limited.

Methods

Perinatal depressive symptoms were assessed using 6 items from the AIDS Clinical Trials Group (ACTG) Quality of Life questionnaire. VS (viral load <400 copies/mL) was the outcome. Adherence was defined as no missed dose in the past 1–4 weeks using the ACTG Adherence Questionnaire. Generalized mixed-effects structural equation models estimated the association of depressive symptoms on VS and the mediating role of ART adherence among women enrolled in the IMPAACT P1025 Perinatal Core Protocol (2002–2013).

Results

Among 1869 participants, 47.6% were 21–29 years, 57.6% non-Hispanic Black. In the third trimester, the mean depressive symptoms score was 14.0 (±5.2), 68.0% had consistent adherence, and 77.3% achieved VS. At 6 months postpartum, depressive symptoms declined while adherence and VS fell to 59.8% and 53.0%, respectively. In the fully adjusted model, a 1-SD increase in depressive symptoms was associated with a 3.8-percentage-point (95% CI: −5.7, −1.9) decline in VS. This effect is the sum of the indirect effect of depressive symptoms on VS via ART adherence (−0.4; 95% CI: −.7, −.2) and the direct effect through other pathways (−3.4; −5.2, −1.5). The decline in adherence driven by depressive symptoms accounted for ≥11% of the total negative effect of depressive symptoms on VS.

Conclusions

Perinatal depressive symptoms were associated with decreased adherence and VS, highlighting the need to screen for, diagnose, and treat perinatal depression to optimize maternal outcomes.

Clinical Trials Registration

NCT00028145.

Do antimicrobial and antithrombogenic peripherally inserted central catheter (PICC) materials prevent catheter complications? An analysis of 42,562 hospitalized medical patients

OBJECTIVE

To examine the effectiveness of antimicrobial and antithrombogenic materials incorporated into peripherally inserted central catheters (PICCs) to prevent bloodstream infection, thrombosis, and catheter occlusion.

METHODS

Prospective cohort study involving 52 hospitals participating in the Michigan Hospital Medicine Safety Consortium. Sample included adult hospitalized medical patients who received a PICC between January 2013 and October 2019. Coated and impregnated catheters were identified by name, brand, and device marketing or regulatory materials. Multivariable Cox proportional hazards models with robust sandwich standard error estimates accounting for the clustered nature of data were used to identify factors associated with PICC complications in coated versus noncoated devices across general care, intensive care unit (ICU), and oncology patients. Results were expressed as hazard ratios (HRs) with corresponding 95% confidence intervals (CIs).

RESULTS

Of 42,562 patients with a PICC, 39,806 (93.5%) were plain polyurethane, 2,263 (5.3%) incorporated antimicrobial materials, and 921 (2.2%) incorporated antithrombogenic materials. Most were inserted in general ward settings (n = 28,111, 66.0%), with 12, 078 (28.4%) and 1,407 (3.3%) placed in ICU and oncological settings, respectively. Within the entire cohort, 540 (1.3%) developed thrombosis, 745 (1.8%) developed bloodstream infection, and 4,090 (9.6%) developed catheter occlusion. Adjusting for known risk factors, antimicrobial PICCs were not associated with infection reduction (HR, 1.16; 95% CI, 0.82-1.64), and antithrombogenic PICCs were not associated with reduction in thrombosis and occlusion (HR, 1.15; 95% CI, 0.92-1.44). Results were consistent across populations and care settings.

CONCLUSIONS

Antimicrobial and antithrombogenic PICCs were not associated with a reduction in major catheter complications. Guidance aimed at informing use of these devices, balancing benefits against cost, appear necessary.

Prevalence of Missing Data in the National Cancer Database and Association With Overall Survival

IMPORTANCE

Cancer registries are important real-world data sources consisting of data abstraction from the medical record; however, patients with unknown or missing data are underrepresented in studies that use such data sources.

OBJECTIVE

To assess the prevalence of missing data and its association with overall survival among patients with cancer.

DESIGN, SETTING, AND PARTICIPANTS

In this retrospective cohort study, all variables within the National Cancer Database were reviewed for missing or unknown values for patients with the 3 most common cancers in the US who received diagnoses from January 1, 2006, to December 31, 2015. The prevalence of patient records with missing data and the association with overall survival were assessed. Data analysis was performed from February to August 2020.

EXPOSURES

Any missing data field within a patient record among 63 variables of interest from more than 130 total variables in the National Cancer Database.

MAIN OUTCOMES AND MEASURES

Prevalence of missing data in the medical records of patients with cancer and associated 2-year overall survival.

RESULTS

A total of 1 198 749 patients with non-small cell lung cancer (mean [SD] age, 68.5 [10.9] years; 628 811 men [52.5%]), 2 120 775 patients with breast cancer (mean [SD] age, 61.0 [13.3] years; 2 101 758 women [99.1%]), and 1 158 635 patients with prostate cancer (mean [SD] age, 65.2 [9.0] years; 100% men) were included in the analysis. Among those with non-small cell lung cancer, 851 295 patients (71.0%) were missing data for variables of interest; 2-year overall survival was 33.2% for patients with missing data and 51.6% for patients with complete data (P < .001). Among those with breast cancer, 1 161 096 patients (54.7%) were missing data for variables of interest; 2-year overall survival was 93.2% for patients with missing data and 93.9% for patients with complete data (P < .001). Among those with prostate cancer, 460 167 patients (39.7%) were missing data for variables of interest; 2-year overall survival was 91.0% for patients with missing data and 95.6% for patients with complete data (P < .001).

CONCLUSIONS AND RELEVANCE

This study found that within a large cancer registry-based real-world data source, there was a high prevalence of missing data that were unable to be ascertained from the medical record. The prevalence of missing data among patients with cancer was associated with heterogeneous differences in overall survival. Improvements in documentation and data quality are necessary to make optimal use of real-world data for clinical advancements.

Missing data in longitudinal studies: Comparison of multiple imputation methods in a real clinical setting

RATIONALE, AIMS, AND OBJECTIVES

Missing data represent a challenge in longitudinal studies. The aim of the study is to compare the performance of the multivariate normal imputation and the fully conditional specification methods, using real data set with missing data partially completed 2 years later.

METHOD

The data used came from an ongoing randomized controlled trial with 5-year follow-up. At a certain time, we observed a number of patients with missing data and a number of patients whose data were unobserved because they were not yet eligible for a given follow-up. Both unobserved and missing data were imputed. The imputed unobserved data were compared with the corresponding real information obtained 2 years later.

RESULTS

Both imputation methods showed similar performance on the accuracy measures and produced minimally biased estimates.

CONCLUSION

Despite the large number of repeated measures with intermittent missing data and the non-normal multivariate distribution of data, both methods performed well and was not possible to determine which was better.

The dynamics of metabolic syndrome development from its isolated components among Iranian adults: findings from 17 years of the Tehran lipid and glucose study (TLGS)

Background

Evaluating the process of changes in the Metabolic Syndrome (MetS) components over time is one of the ways to study of the MetS natural history. This study aimed to determine the trend of changes in the progression of MetS from its isolated components.

Methods

This longitudinal study was performed on four follow-up periods of the Tehran Lipid and Glucose Study (TLGS) between 1999 and 2015. The research population consisted of 3905 adults over the age of 18 years. MetS was diagnosed based on the Joint Interim Statement (JIS). The considered components were abdominal obesity, hypertension, hyperglycemia, and dyslipidemia.

Results

The highest incidence of MetS from its components was related to hypertension in the short term (3.6-year intervals). In the long run, however, the highest increase in the MetS incidence occurred due to abdominal obesity. Overall, the incidence of MetS increased due to obesity and dyslipidemia, but decreased due to the other factors. Nonetheless, the trend of MetS incidence from all components increased in total. The most common components were dyslipidemia with a decreasing trend and obesity with an increasing trend during the study.

Conclusion

The results indicated that obesity and hypertension components played a more important role in the further development of MetS compared to other components in the Iranian adult population. This necessitates careful and serious attention in preventive and control planning.

The relationship between body mass index, disease activity, and exercise in ankylosing spondylitis

BACKGROUND

Ankylosing spondylitis (AS) is associated with elevated cardiovascular (CV) risk and obesity is a common, modifiable risk factor. Our aims were 1) to assess the relationship of BMI with disease activity in AS patients, and 2) to assess the extent to which the effect is mediated through exercise.

METHODS

We used data from a prospective AS cohort with a median follow-up of 7 years. To determine the association of BMI (kg/m² ) with disease activity as measured by the Ankylosing Spondylitis Disease Activity Score (ASDAS), we used generalized estimating equations with inverse probability weighting to account for repeated measures per subject and time-varying confounding. To estimate the direct effect of overweight/obese BMI on disease activity, and the indirect effect through exercise, we performed a mediation analysis.

RESULTS

There were 183 subjects with available BMI and disease activity data (77% male, 70% white, mean age 40.8 ± 13.3 years). Higher BMI was significantly associated with higher disease activity over time; on average, for a 1 kg/m² higher BMI, the ASDAS was 0.06 units higher (95% CI 0.04 - 0.08) after adjustment for important confounders. The direct effect of an overweight/obese BMI accounted for most of the total effect on disease activity, with a smaller indirect effect mediated by exercise (7%).

CONCLUSION

Higher BMI was associated with higher disease activity in a prospective AS cohort. We found that being overweight/obese largely influenced disease activity directly, rather than indirectly through exercise. Other mechanisms such as increased inflammation may better explain the obesity-disease activity association.

The dynamics of metabolic syndrome development from its isolated components among iranian children and adolescents: Findings from 17 Years of the Tehran Lipid and Glucose Study (TLGS)

BACKGROUND

Careful evaluation of the progression trend of the metabolic syndrome (MetS) in children and adolescents (C&A) is one of the important methods of studying the natural history of MetS in them. This study was performed to determine the trend of changes in the progression of MetS from its components.

METHODS

This was a longitudinal study which was performed on data from 4 follow-up periods of Tehran Lipid and Glucose Study (TLGS) between 1999 and 2015. The research population consisted of 6-18-year-old children and adolescents creating 3895-person population. The criteria for the diagnosis of MetS was joint interim statement (JIS). The considered components were central adiposity, high blood pressure, insulin resistance, and dyslipidemia.

RESULTS

In this study, in the long term, the highest increase in the MetS' incidence in boys occurred in obesity and in girls in dyslipidemia and in total mode, in obesity. But in the short term (3.6 year follow-up periods) in the first to fourth periods, in total mode, the highest incidence occurred in dyslipidemia, hyperglycemia, dyslipidemia, and obesity. In terms of trend, in total mode, the highest increase in MetS incidence was related to the obesity component. Also, the incidence of MetS from all components was declining in overall mode. Also, the most common components at the beginning and end of the study in all groups were dyslipidemia with a decreasing and obesity with an increasing trend, respectively.

CONCLUSION

It seems that in Iranian C&As, obesity and dyslipidemia components play a more important role in the further development of the MetS than other components. This matter requires careful and serious attention in preventive and control planning.

Multiple imputation methods for handling missing values in longitudinal studies with sampling weights: Comparison of methods implemented in Stata

Many analyses of longitudinal cohorts require incorporating sampling weights to account for unequal sampling probabilities of participants, as well as the use of multiple imputation (MI) for dealing with missing data. However, there is no guidance on how MI and sampling weights should be implemented together. We simulated a target population based on the Australian Bureau of Statistics Estimated Resident Population and drew 1000 random samples dependent on three design variables to mimic the Longitudinal Study of Australian Children. The target analysis was the weighted prevalence of overweight/obesity over childhood. We evaluated the performance of several MI approaches available in Stata, based on multivariate normal imputation (MVNI), fully conditional specification (FCS) and twofold FCS: a weighted imputation model, imputing missing data separately for each quintile sampling weight grouping, including the design stratum indicator in the imputation model, and using sampling weights as a covariate in the imputation model. Approaches based on available cases and inverse probability weighting (IPW), with time-varying weights, were also compared. We observed severe issues of convergence with FCS and twofold FCS. All MVNI-based approaches performed similarly, producing minimal bias and nominal coverage, except for when imputation was conducted separately for each quintile sampling weight group. IPW performed equally as well as MVNI-based approaches in terms of bias, however, was less precise. In similar longitudinal studies, we recommend using MVNI with the design stratum as a covariate in the imputation model. If this is unknown, including the sampling weight as a covariate is an appropriate alternative.

Addressing challenges in routine health data reporting in Burkina Faso through Bayesian spatiotemporal prediction of weekly clinical malaria incidence

Sub-Saharan African (SSA) countries’ health systems are often vulnerable to unplanned situations that can hinder their effectiveness in terms of data completeness and disease control. For instance, in Burkina Faso following a workers' strike, comprehensive data on several diseases were unavailable for a long period in 2019. Weather, seasonal-malaria-chemoprevention (SMC), free healthcare, and other contextual data, which are purported to influence malarial disease, provide opportunities to fit models to describe the clinical malaria data and predict the disease spread. Bayesian spatiotemporal modeling was applied to weekly malaria surveillance data from Burkina Faso (2011–2018) while considering the effects of weather, health programs and contextual factors. Then, a prediction was used to deal with weekly missing data for the entire year of 2019, and SMC and free healthcare effects were quantified. Our proposed model accurately predicted weekly clinical malaria incidence (correlation coefficient, r = 0.90). The distribution of clinical malaria incidence was heterogeneous across the country. Overall, national predicted clinical malaria incidence in 2019 (605 per 1000 [95% CrI: 360–990]) increased by 24.7% compared with the year 2015. SMC and the interaction between free healthcare and health facility attendance were associated with a reduction in clinical malaria incidence. Our modeling approach could be a useful tool for strengthening health systems’ resilience by addressing data completeness and could support SSA countries in developing appropriate targets and indicators to facilitate the subnational control effort.

Liver venous deprivation versus portal vein embolization before major hepatectomy: future liver remnant volumetric and functional changes

Background

We previously showed that embolization of portal inflow and hepatic vein (HV) outflow (liver venous deprivation, LVD) promotes future liver remnant (FLR) volume (FLR-V) and function (FLR-F) gain. Here, we compared FLR-V and FLR-F changes after portal vein embolization (PVE) and LVD.

Methods

This study included all patients referred for liver preparation before major hepatectomy over 26 months. Exclusion criteria were: unavailable baseline/follow-up imaging, cirrhosis, Klatskin tumor, two-stage hepatectomy. 99mTc-mebrofenin SPECT-CT was performed at baseline and at day 7, 14 and 21 after PVE or LVD. FLR-V and FLR-F variations were compared using multivariate generalized linear mixed models (joint modelling) with/without missing data imputation.

Results

Baseline FLR-F was lower in the LVD (n=29) than PVE group (n=22) (P<0.001). Technical success was 100% in both groups without any major complication. Changes in FLR-V at day 14 and 21 (+14.2% vs. +50%, P=0.002; and +18.6% vs. +52.6%, P=0.001), and in FLR-F at day 7, 14 and 21 (+23.1% vs. +54.3%, P=0.02; +17.6% vs. +56.1%, P=0.006; and +29.8% vs. +63.9%, P<0.001) differed between PVE and LVD group. LVD (P=0.009), age (P=0.027) and baseline FLR-V (P=0.001) independently predicted FLR-V variations, whereas only LVD (P=0.01) predicted FLR-F changes. After missing data handling, LVD remained an independent predictor of FLR-V and FLR-F variations.

Conclusions

LVD is safe and provides greater FLR-V and FLR-F increase than PVE. These results are now evaluated in the HYPERLIV-01 multicenter randomized trial.

Initial Validation of a Machine Learning-Derived Prognostic Test (KidneyIntelX) Integrating Biomarkers and Electronic Health Record Data To Predict Longitudinal Kidney Outcomes

Background

Individuals with type 2 diabetes (T2D) or the apolipoprotein L1 high-risk (APOL1-HR) genotypes are at increased risk of rapid kidney function decline (RKFD) and kidney failure. We hypothesized that a prognostic test using machine learning integrating blood biomarkers and longitudinal electronic health record (EHR) data would improve risk stratification.

Methods

We selected two cohorts from the Mount Sinai BioMe Biobank: T2D (n=871) and African ancestry with APOL1-HR (n=498). We measured plasma tumor necrosis factor receptors (TNFR) 1 and 2 and kidney injury molecule-1 (KIM-1) and used random forest algorithms to integrate biomarker and EHR data to generate a risk score for a composite outcome: RKFD (eGFR decline of ≥5 ml/min per year), or 40% sustained eGFR decline, or kidney failure. We compared performance to a validated clinical model and applied thresholds to assess the utility of the prognostic test (KidneyIntelX) to accurately stratify patients into risk categories.

Results

Overall, 23% of those with T2D and 18% of those with APOL1-HR experienced the composite kidney end point over a median follow-up of 4.6 and 5.9 years, respectively. The area under the receiver operator characteristic curve (AUC) of KidneyIntelX was 0.77 (95% CI, 0.75 to 0.79) in T2D, and 0.80 (95% CI, 0.77 to 0.83) in APOL1-HR, outperforming the clinical models (AUC, 0.66 [95% CI, 0.65 to 0.67] and 0.72 [95% CI, 0.71 to 0.73], respectively; P<0.001). The positive predictive values for KidneyIntelX were 62% and 62% versus 46% and 39% for the clinical models (P<0.01) in high-risk (top 15%) stratum for T2D and APOL1-HR, respectively. The negative predictive values for KidneyIntelX were 92% in T2D and 96% for APOL1-HR versus 85% and 93% for the clinical model, respectively (P=0.76 and 0.93, respectively), in low-risk stratum (bottom 50%).

Conclusions

In patients with T2D or APOL1-HR, a prognostic test (KidneyIntelX) integrating biomarker levels with longitudinal EHR data significantly improved prediction of a composite kidney end point of RKFD, 40% decline in eGFR, or kidney failure over validated clinical models.

How are missing data in covariates handled in observational time-to-event studies in oncology? A systematic review

Background

Missing data in covariates can result in biased estimates and loss of power to detect associations. It can also lead to other challenges in time-to-event analyses including the handling of time-varying effects of covariates, selection of covariates and their flexible modelling. This review aims to describe how researchers approach time-to-event analyses with missing data.

Methods

Medline and Embase were searched for observational time-to-event studies in oncology published from January 2012 to January 2018. The review focused on proportional hazards models or extended Cox models. We investigated the extent and reporting of missing data and how it was addressed in the analysis. Covariate modelling and selection, and assessment of the proportional hazards assumption were also investigated, alongside the treatment of missing data in these procedures.

Results

148 studies were included. The mean proportion of individuals with missingness in any covariate was 32%. 53% of studies used complete-case analysis, and 22% used multiple imputation. In total, 14% of studies stated an assumption concerning missing data and only 34% stated missingness as a limitation. The proportional hazards assumption was checked in 28% of studies, of which, 17% did not state the assessment method. 58% of 144 multivariable models stated their covariate selection procedure with use of a pre-selected set of covariates being the most popular followed by stepwise methods and univariable analyses. Of 69 studies that included continuous covariates, 81% did not assess the appropriateness of the functional form.

Conclusion

While guidelines for handling missing data in epidemiological studies are in place, this review indicates that few report implementing recommendations in practice. Although missing data are present in many studies, we found that few state clearly how they handled it or the assumptions they have made. Easy-to-implement but potentially biased approaches such as complete-case analysis are most commonly used despite these relying on strong assumptions and where often more appropriate methods should be employed. Authors should be encouraged to follow existing guidelines to address missing data, and increased levels of expectation from journals and editors could be used to improve practice.

Mortality risk of loneliness in the oldest old over a 10-year follow-up

Objective: To investigate the impact of loneliness on all-cause mortality in the oldest old population over a 10-year follow-up.Method: Participants were from the third wave of the Cambridge City over-75s Cohort (CC75C) study, a population-based longitudinal study of older people aged 75 or over. Loneliness was measured two further times. At each wave, participants were asked how often they felt lonely and the answers were divided into three levels: not lonely, slightly lonely and lonely. The relationship between loneliness and all-cause mortality was examined using Cox regression with loneliness as a time-varying predictor. The association was adjusted for socio-demographic factors, number of chronic diseases, functional ability and depression.Results: Seven hundred thirteen participants were seen at wave 3 (out of 2166 at baseline), of whom 665 had data on loneliness. The prevalence of feeling slightly lonely and lonely was 16% and 25%, respectively. Vital status was followed for a further 10 years. A total of 562 participants died during the follow-up. After adjusting for age, sex and other socio-demographic factors, loneliness was associated with a 20% increased risk of mortality (HR: 1.2, 95% CI: 1.0-1.6). The association was disappeared after further adjusting for health conditions and depression (HR: 1.0, 95% CI: 0.8-1.4). Individuals who reported being slightly lonely were not at risk of mortality.Conclusions: The association between loneliness and mortality was fully explained by health conditions, suggesting that in the very old age, health problem is the proximal risk factor for mortality.

A Bayesian Vector Autoregressive Model with Nonignorable Missingness in Dependent Variables and Covariates: Development, Evaluation, and Application to Family Processes

Intensive longitudinal designs involving repeated assessments of constructs often face the problems of nonignorable attrition and selected omission of responses on particular occasions. However, time series models, such as vector autoregressive (VAR) models, are often fit to these data without consideration of nonignorable missingness. We introduce a Bayesian model that simultaneously represents the over-time dependencies in multivariate, multiple-subject time series data via a VAR model, and possible ignorable and nonignorable missingness in the data. We provide software code for implementing this model with application to an empirical data set. Moreover, simulation results comparing the joint approach with two-step multiple imputation procedures are included to shed light on the relative strengths and weaknesses of these approaches in practical data analytic scenarios.

Health and voting over the course of adulthood: Evidence from two British birth cohorts

Systematic differences in voter turnout limit the capacity of public institutions to address the needs of under-represented groups. One critical question relates to the role of health as a mechanism driving these inequalities. This study explores the associations of self-rated health (SRH) and limitations in everyday activities with voting over the course of adulthood in the 1958 National Child Development Study and the 1970 British Cohort Study. We used data from participants who reported voting in the last general election at least once between the ages of 23 and 55 in the 1958 cohort and between the ages of 30 and 42 in the 1970 cohort. We examined associations controlling for a range of early-life and adult circumstances using random-effects models. Compared with those in good or better health: those in fair health had 15% and 18% lower odds of voting in the 1958 and 1970 cohorts; those in poor or worse health had 17% and 32% lower odds of voting in the 1958 and 1970 cohorts. These effects varied with age and were most marked among those in poor health at the ages of 23/30 in the 1958 and 1970 cohorts. Controlling for SRH, having limitations in everyday activities was not associated with voting in main models. Examining age-based differences, however, we found that reporting limitations was associated with a higher probability of voting at the age of 55 in the 1958 cohort and at the age of 30 in the 1970 cohort. Building on the qualities of the British birth cohorts, we offer nuanced evidence about the role of health on voting, which involves considerable life-course processes. Future studies need to examine how these findings progress after the age of 55, extend to mental wellbeing and health practices, and contribute to explain social inequalities in voter turnout.

Imputing missing data of function and disease activity in rheumatoid arthritis registers: what is the best technique?

Objective

To compare several methods of missing data imputation for function (Health Assessment Questionnaire) and for disease activity (Disease Activity Score-28 and Clinical Disease Activity Index) in rheumatoid arthritis (RA) patients.

Methods

One thousand RA patients from observational cohort studies with complete data for function and disease activity at baseline, 6, 12 and 24 months were selected to conduct a simulation study. Values were deleted at random or following a predicted attrition bias. Three types of imputation were performed: (1) methods imputing forward in time (last observation carried forward; linear forward extrapolation); (2) methods considering data both forward and backward in time (nearest available observation-NAO; linear extrapolation; polynomial extrapolation); and (3) methods using multi-individual models (linear mixed effects cubic regression-LME3; multiple imputation by chained equation-MICE). The performance of each estimation method was assessed using the difference between the mean outcome value, the remission and low disease activity rates after imputation of the missing values and the true value.

Results

When imputing missing baseline values, all methods underestimated equally the true value, but LME3 and MICE correctly estimated remission and low disease activity rates. When imputing missing follow-up values at 6, 12, or 24 months, NAO provided the least biassed estimate of the mean disease activity and corresponding remission rate. These results were not affected by the presence of attrition bias.

Conclusion

When imputing function and disease activity in large registers of active RA patients, researchers can consider the use of a simple method such as NAO for missing follow-up data, and the use of mixed-effects regression or multiple imputation for baseline data.

Mental health problems and school performance in first graders: results of the prospective cohort study ikidS

We aimed to estimate unbiased effects of mental health problems (MHPs) on school performance in first graders, with an emphasis on rigorous adjustment for potential confounders. A population-based prospective cohort study was performed in the area of Mainz-Bingen (Germany). In 2015, all preschoolers were approached, and the presence and type of MHP (externalising/internalising) and other physical chronic health conditions were identified by the preschool health examination and study-specific questionnaires. At the end of the first grade, school performance (reading, writing, numeracy, and science) was assessed by the class teacher and rated on a four-item scale ranging from - 8 to + 8. Of 3683 children approached, 2003 (54%) were enrolled. School performance was available for 1462 children (51% boys, mean age 7.3 years). Of these, 41% had signs of at least one MHP. Compared to children without indications of mental and physical chronic health conditions, children with MHPs had lower school performance scores [adjusted mean difference - 0.98, 95% CI (- 1.35; - 0.61); P < 0.001]. Regarding the type of MHP, externalising MHPs were associated with poor school performance [adjusted mean difference - 1.44, 95% CI (- 1.83; - 1.05); P < 0.001], while internalising MHPs were not. Children with hyperactivity inattention problems were most affected [adjusted mean difference - 1.96, 95% CI (- 2.36; - 1.56); P < 0.001]. Externalising MHPs and in particular hyperactivity inattention problems may already affect school performance in early primary school. Identification of children with externalising MHPs prior to school entry may help to prevent impaired academic achievement in affected children.

Population Pharmacokinetic Modeling in the Presence of Missing Time-Dependent Covariates: Impact of Body Weight on Pharmacokinetics of Paracetamol in Neonates

Body weight is the primary covariate in pharmacokinetics of many drugs and dramatically changes during the first weeks of life of neonates. The objective of this study is to determine if missing body weights in preterm and term neonates affect estimates of model parameters and which methods can be used to improve performance of a population pharmacokinetic model of paracetamol. Data for our analysis were obtained from previously published studies on the pharmacokinetics of intravenous paracetamol in neonates. We adopted a population model of body weight change in neonates to implement three previously introduced methods of handling missing covariates based on data imputation, likelihood function modification, and full random effects modeling. All models were implemented in NONMEM 7.4, and population parameters were estimated using the FOCE method. Our major finding was that missing body weights minimally affect population estimates of pharmacokinetic parameters but do affect the covariate relationship parameters, particularly the one describing dependence of clearance on body weight. None of the tested methods changed estimates of between-subject variability nor impacted the predictive performance of the model. Our analysis shows that a modeling approach towards handling missing covariates allows borrowing information gathered in various studies as long as they target the same population. This approach is particularly useful for handling time-dependent missing covariates.

Multiple imputation methods for handling missing values in a longitudinal categorical variable with restrictions on transitions over time: a simulation study

Background

Longitudinal categorical variables are sometimes restricted in terms of how individuals transition between categories over time. For example, with a time-dependent measure of smoking categorised as never-smoker, ex-smoker, and current-smoker, current-smokers or ex-smokers cannot transition to a never-smoker at a subsequent wave. These longitudinal variables often contain missing values, however, there is little guidance on whether these restrictions need to be accommodated when using multiple imputation methods. Multiply imputing such missing values, ignoring the restrictions, could lead to implausible transitions.

Methods

We designed a simulation study based on the Longitudinal Study of Australian Children, where the target analysis was the association between (incomplete) maternal smoking and childhood obesity. We set varying proportions of data on maternal smoking to missing completely at random or missing at random. We compared the performance of fully conditional specification with multinomial and ordinal logistic imputation, and predictive mean matching, two-fold fully conditional specification, indicator based imputation under multivariate normal imputation with projected distance-based rounding, and continuous imputation under multivariate normal imputation with calibration, where each of these multiple imputation methods were applied, accounting for the restrictions using a semi-deterministic imputation procedure.

Results

Overall, we observed reduced bias when applying multiple imputation methods with restrictions, and fully conditional specification with predictive mean matching performed the best. Applying fully conditional specification and two-fold fully conditional specification for imputing nominal variables based on multinomial logistic regression had severe convergence issues. Both imputation methods under multivariate normal imputation produced biased estimates when restrictions were not accommodated, however, we observed substantial reductions in bias when restrictions were applied with continuous imputation under multivariate normal imputation with calibration.

Conclusion

In a similar longitudinal setting we recommend the use of fully conditional specification with predictive mean matching, with restrictions applied during the imputation stage.

Electronic supplementary material

The online version of this article (10.1186/s12874-018-0653-0) contains supplementary material, which is available to authorized users.

A neural network analysis of Lifeways cross-generation imputed data

OBJECTIVES

Neural networks are a powerful statistical tool that use nonlinear regression type models to obtain predictions. Their use in the Lifeways cross-generation study that examined body mass index (BMI) of children, among other measures, is explored here. Our aim is to predict the BMI of children from that of their parents and maternal and paternal grandparents. For comparison purposes, linear models will also be used for prediction. A complicating factor is the large amount of missing data. The missing data may be imputed and we examine the effects of different imputation methods on prediction. An analysis using neural networks (and also linear models) that uses all available data without imputation is also carried out, and is the gold standard by which the analyses with imputed data sets are compared.

RESULTS

Neural network models performed better than linear models and can be used as a data analytic tool to detect nonlinear and interaction effects. Using neural networks the BMI of a child can be predicted from family members to within roughly 2.84 units. Results between the imputation methods were similar in terms of mean squared error, as were results based on imputed data compared to un-imputed data.

A comparison of multiple imputation methods for missing data in longitudinal studies

Background

Multiple imputation (MI) is now widely used to handle missing data in longitudinal studies. Several MI techniques have been proposed to impute incomplete longitudinal covariates, including standard fully conditional specification (FCS-Standard) and joint multivariate normal imputation (JM-MVN), which treat repeated measurements as distinct variables, and various extensions based on generalized linear mixed models. Although these MI approaches have been implemented in various software packages, there has not been a comprehensive evaluation of the relative performance of these methods in the context of longitudinal data.

Method

Using both empirical data and a simulation study based on data from the six waves of the Longitudinal Study of Australian Children (N = 4661), we investigated the performance of a wide range of MI methods available in standard software packages for investigating the association between child body mass index (BMI) and quality of life using both a linear regression and a linear mixed-effects model.

Results

In this paper, we have identified and compared 12 different MI methods for imputing missing data in longitudinal studies. Analysis of simulated data under missing at random (MAR) mechanisms showed that the generally available MI methods provided less biased estimates with better coverage for the linear regression model and around half of these methods performed well for the estimation of regression parameters for a linear mixed model with random intercept. With the observed data, we observed an inverse association between child BMI and quality of life, with available data as well as multiple imputation.

Conclusion

Both FCS-Standard and JM-MVN performed well for the estimation of regression parameters in both analysis models. More complex methods that explicitly reflect the longitudinal structure for these analysis models may only be needed in specific circumstances such as irregularly spaced data.

Electronic supplementary material

The online version of this article (10.1186/s12874-018-0615-6) contains supplementary material, which is available to authorized users.

Chronic health conditions and school performance in first graders: A prospective cohort study

Objective

Children with chronic health conditions may perform poorer at school. Associations may be confounded by numerous social factors. We aimed to estimate the effects of a chronic health condition on overall school performance in first graders with an emphasis on rigorous adjustment for potential confounders.

Methods

A population-based cohort study was performed in the area of Mainz-Bingen (Germany). In 2015 all preschoolers were approached and the presence of a chronic health condition was assessed by parental questionnaires and preschool health examination data. The identification of a chronic health condition was based on special health care needs and presence of a doctor’s diagnosis out of 24 school-relevant diseases. At the end of the first school year, overall school performance was assessed by teachers and rated on a 5-item scale ranging from -10 to +10.

Results

Of 3683 children approached, 2003 were enrolled. Overall school performance was available for 1462 children (51% boys). Of these, 52% suffered from a chronic health condition. Compared to children without a chronic health condition, children with special health care needs (15%) performed worse at school (adjusted mean difference: -0.95, 95% CI: [-1.55; -0.35], P = 0.002). Children with a doctor’s diagnosis but without special health care needs (37%) did not perform worse at school. The effect was further analysed considering the extent of special health care needed.

Conclusions

Chronic health conditions affect overall school performance early in primary school. To identify academically at-risk children, a chronic health condition identification based on special health care needs may be used.

A novel method to test associations between a weighted combination of phenotypes and genetic variants

Many complex diseases like diabetes, hypertension, metabolic syndrome, et cetera, are measured by multiple correlated phenotypes. However, most genome-wide association studies (GWAS) focus on one phenotype of interest or study multiple phenotypes separately for identifying genetic variants associated with complex diseases. Analyzing one phenotype or the related phenotypes separately may lose power due to ignoring the information obtained by combining phenotypes, such as the correlation between phenotypes. In order to increase statistical power to detect genetic variants associated with complex diseases, we develop a novel method to test a weighted combination of multiple phenotypes (WCmulP). We perform extensive simulation studies as well as real data (COPDGene) analysis to evaluate the performance of the proposed method. Our simulation results show that WCmulP has correct type I error rates and is either the most powerful test or comparable to the most powerful test among the methods we compared. WCmulP also has an outstanding performance for identifying single-nucleotide polymorphisms (SNPs) associated with COPD-related phenotypes.