Multipotent adult progenitor cells prevent functional impairment and improve development in inflammation driven detriment of preterm ovine lungs.

Background: Perinatal inflammation increases the risk for bronchopulmonary dysplasia in preterm neonates, but the underlying pathophysiological mechanisms remain largely unknown. Given their anti-inflammatory and regenerative capacity, multipotent adult progenitor cells (MAPC) are a promising cell-based therapy to prevent and/or treat the negative pulmonary consequences of perinatal inflammation in the preterm neonate. Therefore, the pathophysiology underlying adverse preterm lung outcomes following perinatal inflammation and pulmonary benefits of MAPC treatment at the interface of prenatal inflammatory and postnatal ventilation exposures were elucidated. Methods: Instrumented ovine fetuses were exposed to intra-amniotic lipopolysaccharide (LPS 5 mg) at 125 days gestation to induce adverse systemic and peripheral organ outcomes. MAPC (10 × 106 cells) or saline were administered intravenously two days post LPS exposure. Fetuses were delivered preterm five days post MAPC treatment and either killed humanely immediately or mechanically ventilated for 72 h. Results: Antenatal LPS exposure resulted in inflammation and decreased alveolar maturation in the preterm lung. Additionally, LPS-exposed ventilated lambs showed continued pulmonary inflammation and cell junction loss accompanied by pulmonary edema, ultimately resulting in higher oxygen demand. MAPC therapy modulated lung inflammation, prevented loss of epithelial and endothelial barriers and improved lung maturation in utero. These MAPC-driven improvements remained evident postnatally, and prevented concomitant pulmonary edema and functional loss. Conclusion: In conclusion, prenatal inflammation sensitizes the underdeveloped preterm lung to subsequent postnatal inflammation, resulting in injury, disturbed development and functional impairment. MAPC therapy partially prevents these changes and is therefore a promising approach for preterm infants to prevent adverse pulmonary outcomes.

Tumour infiltrating lymphocytes and survival after adjuvant chemotherapy in patients with gastric cancer: post-hoc analysis of the CLASSIC trial.

BACKGROUND: Only a subset of gastric cancer (GC) patients with stage II-III benefits from chemotherapy after surgery. Tumour infiltrating lymphocytes per area (TIL density) has been suggested as a potential predictive biomarker of chemotherapy benefit. METHODS: We quantified TIL density in digital images of haematoxylin-eosin (HE) stained tissue using deep learning in 307 GC patients of the Yonsei Cancer Center (YCC) (193 surgery+adjuvant chemotherapy [S + C], 114 surgery alone [S]) and 629 CLASSIC trial GC patients (325 S + C and 304 S). The relationship between TIL density, disease-free survival (DFS) and clinicopathological variables was analysed. RESULTS: YCC S patients and CLASSIC S patients with high TIL density had longer DFS than S patients with low TIL density (P = 0.007 and P = 0.013, respectively). Furthermore, CLASSIC patients with low TIL density had longer DFS if treated with S + C compared to S (P = 0.003). No significant relationship of TIL density with other clinicopathological variables was found. CONCLUSION: This is the first study to suggest TIL density automatically quantified in routine HE stained tissue sections as a novel, clinically useful biomarker to identify stage II-III GC patients deriving benefit from adjuvant chemotherapy. Validation of our results in a prospective study is warranted.

Effects of an Explicit Value Clarification Method With Computer-Tailored Advice on the Effectiveness of a Web-Based Smoking Cessation Decision Aid: Findings From a Randomized Controlled Trial.

BACKGROUND: Smoking continues to be a driver of mortality. Various forms of evidence-based cessation assistance exist; however, their use is limited. The choice between them may also induce decisional conflict. Offering decision aids (DAs) may be beneficial; however, insights into their effective elements are lacking. OBJECTIVE: This study tested the added value of an effective element (ie, an "explicit value clarification method" paired with computer-tailored advice indicating the most fitting cessation assistance) of a web-based smoking cessation DA. METHODS: A web-based randomized controlled trial was conducted among smokers motivated to stop smoking within 6 months. The intervention group received a DA with the aforementioned elements, and the control group received the same DA without these elements. The primary outcome measure was 7-day point prevalence abstinence 6 months after baseline (time point 3 [t=3]). Secondary outcome measures were 7-day point prevalence of abstinence 1 month after baseline (time point 2 [t=2]), evidence-based cessation assistance use (t=2 and t=3), and decisional conflict (immediately after DA; time point 1). Logistic and linear regression analyses were performed to assess the outcomes. Analyses were conducted following 2 (decisional conflict) and 3 (smoking cessation) outcome scenarios: complete cases, worst-case scenario (assuming that dropouts still smoked), and multiple imputations. A priori sample size calculation indicated that 796 participants were needed. The participants were mainly recruited on the web (eg, social media). All the data were self-reported. RESULTS: Overall, 2375 participants were randomized (intervention n=1164, 49.01%), of whom 599 (25.22%; intervention n=275, 45.91%) completed the DAs, and 276 (11.62%; intervention n=143, 51.81%), 97 (4.08%; intervention n=54, 55.67%), and 103 (4.34%; intervention n=56, 54.37%) completed time point 1, t=2, and t=3, respectively. More participants stopped smoking in the intervention group (23/63, 37%) than in the control group (14/52, 27%) after 6 months; however, this was only statistically significant in the worst-case scenario (crude P=.02; adjusted P=.04). Effects on the secondary outcomes were only observed for smoking abstinence after 1 month (15/55, 27%, compared with 7/46, 15%, in the crude and adjusted models, respectively; P=.02) and for cessation assistance uptake after 1 month (26/56, 46% compared with 18/47, 38% only in the crude model; P=.04) and 6 months (38/61, 62% compared with 26/50, 52%; crude P=.01; adjusted P=.02) but only in the worst-case scenario. Nonuse attrition was 34.19% higher in the intervention group than in the control group (P<.001). CONCLUSIONS: Currently, we cannot confidently recommend the inclusion of explicit value clarification methods and computer-tailored advice. However, they might result in higher nonuse attrition rates, thereby limiting their potential. As a lack of statistical power may have influenced the outcomes, we recommend replicating this study with some adaptations based on the lessons learned. TRIAL REGISTRATION: Netherlands Trial Register NL8270; https://www.trialregister.nl/trial/8270. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.2196/21772.

Temporal trends of sex differences for COVID-19 infection, hospitalisation, severe disease, intensive care unit (ICU) admission and death: a meta-analysis of 229 studies covering over 10M patients.

Background: This review aims to investigate the association of sex with the risk of multiple COVID-19 health outcomes, ranging from infection to death. Methods: Pubmed and Embase were searched through September 2020. We considered studies reporting sex and coronavirus disease 2019 (COVID-19) outcomes. Qualitative and quantitative data were extracted using standardised electronic data extraction forms with the assessment of Newcastle Ottawa Scale for risk of bias. Pooled trends in infection, hospitalization, severity, intensive care unit (ICU) admission and death rate were calculated separately for men and women and subsequently random-effects meta-analyses on relative risks (RR) for sex was performed. Results: Of 10,160 titles, 229 studies comprising 10,417,452 patients were included in the analyses. Methodological quality of the included studies was high (6.9 out of 9). Men had a higher risk for infection with COVID-19 than women (RR = 1.14, 95%CI: 1.07 to 1.21). When infected, they also had a higher risk for hospitalization (RR = 1.33, 95%CI: 1.27 to 1.41), higher risk for severe COVID-19 (RR = 1.22, 95%CI: 1.17 to 1.27), higher need for Intensive Care (RR = 1.41, 95%CI: 1.28 to 1.55), and higher risk of death (RR = 1.35, 95%CI: 1.28 to 1.43). Within the period studied, the RR for infection and severity increased for men compared to women, while the RR for mortality decreased for men compared to women. Conclusions: Meta-analyses on 229 studies comprising over 10 million patients showed that men have a higher risk for COVID-19 infection, hospitalization, disease severity, ICU admission and death. The relative risks of infection, disease severity and death for men versus women showed temporal trends with lower relative risks for infection and severity of disease and higher relative risk for death at the beginning of the pandemic compared to the end of our inclusion period. PROSPERO registration: CRD42020180085 (20/04/2020).

Imputation of Missing Covariates in Randomized Controlled Trials with Continuous Outcomes: Simple, Unbiased and Efficient Methods.

The literature on dealing with missing covariates in nonrandomized studies advocates the use of sophisticated methods like multiple imputation (MI) and maximum likelihood (ML)-based approaches over simple methods. However, these methods are not necessarily optimal in terms of bias and efficiency of treatment effect estimation in randomized studies, where the covariate of interest (treatment group) is independent of all baseline (pre-randomization) covariates due to randomization. This has been shown in the literature, but only for missingness on a single baseline covariate. Here, we extend the situation to multiple baseline covariates with missingness and evaluate the performance of MI and ML compared with simple alternative methods under various missingness scenarios in RCTs with a quantitative outcome. We first derive asymptotic relative efficiencies of the simple methods under the missing completely at random (MCAR) scenario and then perform a simulation study for non-MCAR scenarios. Finally, a trial on chronic low back pain is used to illustrate the implementation of the methods. The results show that all simple methods give unbiased treatment effect estimation but with increased mean squared residual. It also turns out that mean imputation and the missing-indicator method are most efficient under all covariate missingness scenarios and perform at least as well as MI and LM in each scenario.

Ethnic minority status as social determinant for COVID-19 infection, hospitalisation, severity, ICU admission and deaths in the early phase of the pandemic: a meta-analysis.

INTRODUCTION: Early literature on the COVID-19 pandemic indicated striking ethnic inequalities in SARS-CoV-2-related outcomes. This systematic review and meta-analysis aimed to describe the presence and magnitude of associations between ethnic groups and COVID-19-related outcomes. METHODS: PubMed and Embase were searched from December 2019 through September 2020. Studies reporting extractable data (ie, crude numbers, and unadjusted or adjusted risk/ORs) by ethnic group on any of the five studied outcomes: confirmed COVID-19 infection in the general population, hospitalisation among infected patients, and disease severity, intensive care unit (ICU) admission and mortality among hospitalised patients with SARS-CoV-2 infection, were included using standardised electronic data extraction forms. We pooled data from published studies using random-effects meta-analysis. RESULTS: 58 studies were included from seven countries in four continents, mostly retrospective cohort studies, covering a total of almost 10 million individuals from the first wave until the summer of 2020. The risk of diagnosed SARS-CoV-2 infection was higher in most ethnic minority groups than their White counterparts in North American and Europe with the differences remaining in the US ethnic minorities after adjustment for confounders and explanatory factors. Among people with confirmed infection, African-Americans and Hispanic-Americans were also more likely than White-Americans to be hospitalised with SARS-CoV-2 infection. No increased risk of COVID-19 outcomes (ie, severe disease, ICU admission and death) was found among ethnic minority patients once hospitalised, except for a higher risk of death among ethnic minorities in Brazil. CONCLUSION: The risk of SARS-CoV-2 diagnosis was higher in most ethnic minorities, but once hospitalised, no clear inequalities exist in COVID-19 outcomes except for the high risk of death in ethnic minorities in Brazil. The findings highlight the necessity to tackle disparities in social determinants of health, preventative opportunities and delay in healthcare use. Ethnic minorities should specifically be considered in policies mitigating negative impacts of the pandemic. PROSPERO REGISTRATION NUMBER: CRD42020180085.

BMI trajectories after primary school-based lifestyle intervention: Unravelling an uncertain future. A mixed methods study.

This mixed methods study aimed to examine plausible body mass index (BMI) trajectories after exposure to a primary school-based lifestyle intervention to aid in estimating the long-term intervention benefits. BMI trajectories for children at control schools (mean 7.6 years of age) were modelled until 20 years of age through extrapolating trial evidence (N = 1647). A reference scenario assumed that the observed 2-year effects of the 'Healthy Primary Schools of the Future' (HPSF) and 'Physical Activity Schools' (PAS) were fully maintained over time. This was modelled by applying the observed 2-year BMI effects until 20 years of age. Expert opinions on likely trends in effect maintenance after the 2-year intervention period were elicited qualitatively and quantitatively, and were used for developing alternative scenarios. Expert elicitation revealed three scenarios: (a) a constant exposure-effect and an uncontrolled environment with effect decay scenario, (b) a household multiplier and an uncontrolled environment with effect decay scenario, and (c) a household multiplier and maintainer scenario. The relative effect of HPSF at 20 years of age was -0.21 kg/m2 under the reference scenario, and varied from -0.04 kg/m2 (a) to -0.06 kg/m2 (b), and -0.50 kg/m2 (c). For PAS, the relative effect was -0.17 kg/m2 under the reference scenario, and varied from -0.04 kg/m2 (a, b), to -0.21 kg/m2 (c). The mixed methods approach proved to be useful in modelling plausible BMI trajectories and specifying uncertainty on effect maintenance. Further observations until adulthood could reduce the uncertainty around future benefits. This trial was retrospectively registered at Clinicaltrials.gov (NCT02800616).

Demographic risk factors for COVID-19 infection, severity, ICU admission and death: a meta-analysis of 59 studies.

OBJECTIVE: We aimed to describe the associations of age and sex with the risk of COVID-19 in different severity stages ranging from infection to death. DESIGN: Systematic review and meta-analysis. DATA SOURCES: PubMed and Embase through 4 May 2020. STUDY SELECTION: We considered cohort and case-control studies that evaluated differences in age and sex on the risk of COVID-19 infection, disease severity, intensive care unit (ICU) admission and death. DATA EXTRACTION AND SYNTHESIS: We screened and included studies using standardised electronic data extraction forms and we pooled data from published studies and data acquired by contacting authors using random effects meta-analysis. We assessed the risk of bias using the Newcastle-Ottawa Scale. RESULTS: We screened 11.550 titles and included 59 studies comprising 36.470 patients in the analyses. The methodological quality of the included papers was high (8.2 out of 9). Men had a higher risk for infection with COVID-19 than women (relative risk (RR) 1.08, 95% CI 1.03 to 1.12). When infected, they also had a higher risk for severe COVID-19 disease (RR 1.18, 95% CI 1.10 to 1.27), a higher need for intensive care (RR 1.38, 95% CI 1.09 to 1.74) and a higher risk of death (RR 1.50, 95% CI 1.18 to 1.91). The analyses also showed that patients aged 70 years and above have a higher infection risk (RR 1.65, 95% CI 1.50 to 1.81), a higher risk for severe COVID-19 disease (RR 2.05, 95% CI 1.27 to 3.32), a higher need for intensive care (RR 2.70, 95% CI 1.59 to 4.60) and a higher risk of death once infected (RR 3.61, 95% CI 2.70 to 4.84) compared with patients younger than 70 years. CONCLUSIONS: Meta-analyses on 59 studies comprising 36.470 patients showed that men and patients aged 70 and above have a higher risk for COVID-19 infection, severe disease, ICU admission and death. PROSPERO REGISTRATION NUMBER: CRD42020180085.

Imputation of missing covariate in randomized controlled trials with a continuous outcome: Scoping review and new results.

In this article, we first review the literature on dealing with missing values on a covariate in randomized studies and summarize what has been done and what is lacking to date. We then investigate the situation with a continuous outcome and a missing binary covariate in more details through simulations, comparing the performance of multiple imputation (MI) with various simple alternative methods. This is finally extended to the case of time-to-event outcome. The simulations consider five different missingness scenarios: missing completely at random (MCAR), at random (MAR) with missingness depending only on the treatment, and missing not at random (MNAR) with missingness depending on the covariate itself (MNAR1), missingness depending on both the treatment and covariate (MNAR2), and missingness depending on the treatment, covariate and their interaction (MNAR3). Here, we distinguish two different cases: (1) when the covariate is measured before randomization (best practice), where only MCAR and MNAR1 are plausible, and (2) when it is measured after randomization but before treatment (which sometimes occurs in nonpharmaceutical research), where the other three missingness mechanisms can also occur. The proposed methods are compared based on the treatment effect estimate and its standard error. The simulation results suggest that the patterns of results are very similar for all missingness scenarios in case (1) and also in case (2) except for MNAR3. Furthermore, in each scenario for continuous outcome, there is at least one simple method that performs at least as well as MI, while for time-to-event outcome MI is best.

Histological intratumoral heterogeneity in pretreatment esophageal cancer biopsies predicts survival benefit from neoadjuvant chemotherapy: results from the UK MRC OE02 trial.

Despite the use of multimodal treatment, survival of esophageal cancer (EC) patients remains poor. One proposed explanation for the relatively poor response to cytotoxic chemotherapy is intratumor heterogeneity. The aim was to establish a statistical model to objectively measure intratumor heterogeneity of the proportion of tumor (IHPoT) and to use this newly developed method to measure IHPoT in the pretreatment biopsies from from EC patients recruited to the OE02 trial. A statistical mixed effect model (MEM) was established for estimating IHPoT based on variation in hematoxylin/eosin (HE) stained pretreatment biopsy pieces from the same individual in 218 OE02 trial patients (103 treated by chemotherapy and surgery (chemo+surgery); 115 patients treated by surgery alone). The relationship between IHPoT, prognosis, chemotherapy survival benefit, and clinicopathological variables was assessed. About 97 (44.5%) and 121 (55.5%) ECs showed high and low IHPoT, respectively. There was no significant difference in IHPoT between surgery (median [range], 0.1637 [0-3.17]) and chemo+surgery (median [range], 0.1692 [0-2.69]) patients (P = 0.43). Chemo+surgery patients with low IHPoT had a significantly longer survival than surgery patients (HR = 1.81, 95% CI: 1.20-2.75, P = 0.005). There was no survival difference between chemo+surgery and surgery patients with high IHPoT (HR = 1.15, 95% CI: 0.72-1.81, P = 0.566). This is the first study suggesting that IHPoT measured in the pretreatment biopsy can predict chemotherapy survival benefit in EC patients. IHPoT may represent a clinically useful biomarker for patient treatment stratification. Future studies should determine if pathologists can reliably estimate IHPoT.

Guidelines for multiple imputations in repeated measurements with time-dependent covariates: a case study.

OBJECTIVES: We provide guidelines for handling the most common missing data problems in repeated measurements in observational studies and deal with practicalities in producing imputations when there are many partly missing time-varying variables and repeated measurements. STUDY DESIGN AND SETTING: The Maastricht Study on long-term dementia care environments was used as a case study. The data contain 84 momentary assessments for each of 115 participants. A continuous outcome and several time-varying covariates were involved containing missing observations varying from 4% to 25% per time point. A multiple imputation procedure is advocated with restrictions imposed on the relation within and between partially missing variables over time. RESULTS: Multiple imputation is a better approach to deal with missing observations in both outcome and independent variables. Furthermore, using the statistical package R-MICE, it is possible to deal with the limitations of current statistical software in imputation of missing observations in more complex data. CONCLUSION: In observational studies, the direct likelihood approach (i.e., the standard longitudinal data methods) is sufficient to obtain valid inferences in the presence of missing data only in the outcome. In contrast, multiple imputation is required when dealing with partly missing time-varying covariates and repeated measurements.

The relation between mood, activity, and interaction in long-term dementia care.

OBJECTIVE: The aim of the study is to identify the degree of association between mood, activity engagement, activity location, and social interaction during everyday life of people with dementia (PwD) living in long-term care facilities. METHOD: An observational study using momentary assessments was conducted. For all 115 participants, 84 momentary assessments of mood, engagement in activity, location during activity, and social interaction were carried out by a researcher using the tablet-based Maastricht Electronic Daily Life Observation-tool. RESULTS: A total of 9660 momentary assessments were completed. The mean age of the 115 participants was 84 and most (75%) were women. A negative, neutral, or positive mood was recorded during 2%, 25%, and 73% of the observations, respectively. Positive mood was associated with engagement in activities, doing activities outside, and social interaction. The type of activity was less important for mood than the fact that PwD were engaged in an activity. Low mood was evident when PwD attempted to have social interaction but received no response. CONCLUSION: Fulfilling PwD's need for occupation and social interaction is consistent with a person-centred dementia care focus and should have priority in dementia care.

Hierarchical imputation of systematically and sporadically missing data: An approximate Bayesian approach using chained equations.

In health and medical sciences, multiple imputation (MI) is now becoming popular to obtain valid inferences in the presence of missing data. However, MI of clustered data such as multicenter studies and individual participant data meta-analysis requires advanced imputation routines that preserve the hierarchical structure of data. In clustered data, a specific challenge is the presence of systematically missing data, when a variable is completely missing in some clusters, and sporadically missing data, when it is partly missing in some clusters. Unfortunately, little is known about how to perform MI when both types of missing data occur simultaneously. We develop a new class of hierarchical imputation approach based on chained equations methodology that simultaneously imputes systematically and sporadically missing data while allowing for arbitrary patterns of missingness among them. Here, we use a random effect imputation model and adopt a simplification over fully Bayesian techniques such as Gibbs sampler to directly obtain draws of parameters within each step of the chained equations. We justify through theoretical arguments and extensive simulation studies that the proposed imputation methodology has good statistical properties in terms of bias and coverage rates of parameter estimates. An illustration is given in a case study with eight individual participant datasets.

The Monkey game: A computerized verbal working memory task for self-reliant administration in primary school children.

In two studies, the psychometric properties of an online self-reliant verbal working memory task (the Monkey game) for primary school children (6-12 years of age) were examined. In Study 1, children (n = 5,203) from 31 primary schools participated. The participants completed computerized verbal and visual-spatial working memory tasks (i.e., the Monkey game and the Lion game) and a paper-and-pencil version of Raven's Standard Progressive Matrices. Reading comprehension and math achievement test scores were obtained from the schools. First, the internal consistency of the Monkey game was examined. Second, multilevel modeling was used to examine the effects of classroom membership. Multilevel multivariate regression analysis was used to examine the Monkey game's concurrent relationship with the Lion game and its predictive relationships with reading comprehension and math achievement. Also, age-related differences in performance were examined. In Study 2, the concurrent relationships between the Monkey game and two tester-led computerized working memory tasks were further examined (n = 140). Also, the 1- and 2-year stability of the Monkey game was investigated. The Monkey game showed excellent internal consistency, good concurrent relationships with the other working memory measures, and significant age differences in performance. Performance on the Monkey game was also predictive of subsequent reading comprehension and mathematics performance, even after controlling for individual differences in intelligence. Performance on the Monkey game was influenced by classroom membership. The Monkey game is a reliable and suitable instrument for the online computerized and self-reliant assessment of verbal working memory in primary school children.

Better experiences with quality of care predict well-being of patients with chronic obstructive pulmonary disease in the Netherlands.

OBJECTIVE: This study was conducted to (1) identify improvements in care quality and well-being of patients with chronic obstructive pulmonary disease in the Netherlands and (2) investigate the longitudinal relationship between these factors. METHODS: This longitudinal study was conducted among patients diagnosed with chronic obstructive pulmonary disease enrolled in the Kennemer Lucht care programme in the Netherlands. Biomarker data (lung capacity) were collected at patients' health care practices in 2012. Complete case analysis was conducted, and the multiple imputation technique allowed us to report pooled results from imputed datasets. RESULTS: Surveys were filled out by 548/1303 (42%) patients at T0 (2012) and 569/996 (57%) remaining participants at T1. Quality of care improved significantly (p < 0.05). Analyses adjusted for well-being at T0, age, educational level, marital status, gender, lung function and health behaviours showed that patients' assessments of the quality of chronic care delivery at T0 (p < 0.01) and changes therein (p < 0.001) predicted patients' well-being at T1. CONCLUSION: These results clearly show that the quality of care and changes therein are important for the well-being of patients with chronic obstructive pulmonary disease in the primary care setting. PRACTICE IMPLICATIONS: To improve quality of care for chronically ill patients, multicomponent interventions may be needed.

Imputation of systematically missing predictors in an individual participant data meta-analysis: a generalized approach using MICE.

Individual participant data meta-analyses (IPD-MA) are increasingly used for developing and validating multivariable (diagnostic or prognostic) risk prediction models. Unfortunately, some predictors or even outcomes may not have been measured in each study and are thus systematically missing in some individual studies of the IPD-MA. As a consequence, it is no longer possible to evaluate between-study heterogeneity and to estimate study-specific predictor effects, or to include all individual studies, which severely hampers the development and validation of prediction models. Here, we describe a novel approach for imputing systematically missing data and adopt a generalized linear mixed model to allow for between-study heterogeneity. This approach can be viewed as an extension of Resche-Rigon's method (Stat Med 2013), relaxing their assumptions regarding variance components and allowing imputation of linear and nonlinear predictors. We illustrate our approach using a case study with IPD-MA of 13 studies to develop and validate a diagnostic prediction model for the presence of deep venous thrombosis. We compare the results after applying four methods for dealing with systematically missing predictors in one or more individual studies: complete case analysis where studies with systematically missing predictors are removed, traditional multiple imputation ignoring heterogeneity across studies, stratified multiple imputation accounting for heterogeneity in predictor prevalence, and multilevel multiple imputation (MLMI) fully accounting for between-study heterogeneity. We conclude that MLMI may substantially improve the estimation of between-study heterogeneity parameters and allow for imputation of systematically missing predictors in IPD-MA aimed at the development and validation of prediction models.

Dual imputation model for incomplete longitudinal data.

Missing values are a practical issue in the analysis of longitudinal data. Multiple imputation (MI) is a well-known likelihood-based method that has optimal properties in terms of efficiency and consistency if the imputation model is correctly specified. Doubly robust (DR) weighing-based methods protect against misspecification bias if one of the models, but not necessarily both, for the data or the mechanism leading to missing data is correct. We propose a new imputation method that captures the simplicity of MI and protection from the DR method. This method integrates MI and DR to protect against misspecification of the imputation model under a missing at random assumption. Our method avoids analytical complications of missing data particularly in multivariate settings, and is easy to implement in standard statistical packages. Moreover, the proposed method works very well with an intermittent pattern of missingness when other DR methods can not be used. Simulation experiments show that the proposed approach achieves improved performance when one of the models is correct. The method is applied to data from the fireworks disaster study, a randomized clinical trial comparing therapies in disaster-exposed children. We conclude that the new method increases the robustness of imputations.