Multiple Imputation Techniques Research Articles

IMPUTING THE MISSING VALUES IN IOT USING ESTCP MODEL

In Internet of Things, occurrence of missing data is inevitable due to its intrinsic characteristics. This missing data phenomenon occurs due to a variety of reasons such as uneven network communication, synchronization difficulties, untrustworthy sensor devices, environmental aspects and other device malfunctions which often resulted in data incompleteness. A robust approach to missing data is an indispensible component of analysis to promote the perfect explanation of research findings.As the data generated by the IoT devices is usually correlated in space and time, in this paper it is demonstrated experimentally that substituting missing sensor values with spatially and temporally correlated sensor readings using thenovel extended spatial and temporal correlated proximate missing data imputation model (ESTCP)has considerably improved the accuracy than that of the previously proposed STCP model and the existing single imputation and multiple imputation techniques.

Implementation of a Smartphone application in medical education: a randomised trial (iSTART)

Smartphones are popular technologies that combine telephone communications and informatics in portable devices. Limited evidence exists regarding their effectiveness in improving academic performance among medical students. This study aims to assess whether a smartphone application could improve academic performance in multiple-choice tests. A double-masked randomised trial was held among interns at the School of Medicine of the Universidad de Valparaiso. Participants were randomised to receive an application designed to review key concepts in Internal Medicine and its subspecialties using clinical vignettes. Contents were selected and provided in a format akin to a mandatory national examination required for practising medicine in Chile. Analyses were undertaken under the intention to treat principle and missing data were handled using multiple imputation techniques. Eighty interns volunteered to participate in this trial, most were female (48 students, 60%) and had a mean age of 25.3 ± 2.2 years. Participants showed significant experience with smartphones, with a median use of 4 years (IQR 3–6 years) and 67 (83.7%) reporting routine use in clinical practice. Intention-to-treat analyses showed significant improvements in performance amongst students allocated to the smartphone application (mean increase of 14.5 ± 8.9 vs 9.4 ± 11.6points, p = 0.03). A reduction in total time and mean time per question was also found, which was significant in complete-case analyses (p = 0.04). Smartphones were popular among medical trainees. Academic performance was significantly improved by the use of our application, although the overall effect was smaller than expected from previous trials. This study provides evidence that smartphone-based interventions can assist in teaching internal medicine. ClinicalTrials NCT02723136 .

Prehospital sodium bicarbonate use could worsen long term survival with favorable neurological recovery among patients with out-of-hospital cardiac arrest

BackgroundSodium bicarbonate (SB) is widely used for resuscitation in out-of- hospital cardiac arrest (OHCA); however, its effect on long term outcomes is unclear. MethodsFrom 2005–2016, we prospectively conducted a province-wide population-based observational study including adult non-traumatic OHCA patients managed by paramedics. SB was administered by paramedics based on their clinical assessments. To examine the association of SB administration and survival and favorable neurological outcome to hospital discharge, defined as modified Rankin scale of 3 or less, we performed a multivariable logistic regression analysis: (1) within propensity score matched comparison groups, and; (2) within the full cohort with missing variables addressed by multiple imputation techniques. ResultsOf 15 601 OHCA patients, 13,865 were included in this study with 5165 (37.3%) managed with SB. In the SB treated group, 118 (2.3%) patients survived and 62 (1.2%) had favorable neurological outcomes to hospital discharge, compared to 1699 (19.8%) and 831 (10.6%) in the non-SB treated group, respectively. In the 1:1 propensity matched cohort including 5638 OHCA patients, SB was associated with decreased probability of outcomes (adjusted OR for survival: 0.64, 95% CI 0.45–0.91, and adjusted OR for favorable neurological outcome: 0.59, 95% CI 0.39–0.88, respectively). The association remained consistent in the multiply imputed cohort (adjusted OR 0.48, 95 CI 0.36–0.64, and adjusted OR 0.54, 95% CI 0.38–0.76, respectively). ConclusionsIn OHCA patients, prehospital SB administration was associated with worse survival rate and neurological outcomes to hospital discharge.

Imputation of Multivariate Time Series Data - Performance Benchmarks for Multiple Imputation and Spectral Techniques

Missing data is a problem appearing ubiquitously across many fields and needs to be dealt with systematically. For multivariate time series data imputation can be a challenging problem. We consider the particular case of credit default swap time series, where missing data can pose a considerable problem preventing important value at risk estimates. We present a new framework to structure and classify missingness patterns, and generate suitable realistic test sets. We then benchmark the performance of a number of state-of-the-art imputation techniques, both stochastic multiple imputation (MI) approaches and deterministic spectral decomposition techniques. We demonstrate that for the missingness patterns under consideration, the MI package Amelia based on the expectation maximisation algorithm performs most robustly and reliably, however, other techniques like multiple singular spectral analysis can also perform well. Our results can serve as a valuable guideline for researchers and practicioners working with incomplete multivariate time series.

A robust imputation method for missing responses and covariates in sample selection models.

Sample selection arises when the outcome of interest is partially observed in a study. Although sophisticated statistical methods in the parametric and non-parametric framework have been proposed to solve this problem, it is yet unclear how to deal with selectively missing covariate data using simple multiple imputation techniques, especially in the absence of exclusion restrictions and deviation from normality. Motivated by the 2003-2004 NHANES data, where previous authors have studied the effect of socio-economic status on blood pressure with missing data on income variable, we proposed the use of a robust imputation technique based on the selection-t sample selection model. The imputation method, which is developed within the frequentist framework, is compared with competing alternatives in a simulation study. The results indicate that the robust alternative is not susceptible to the absence of exclusion restrictions - a property inherited from the parent selection-t model - and performs better than models based on the normal assumption even when the data is generated from the normal distribution. Applications to missing outcome and covariate data further corroborate the robustness properties of the proposed method. We implemented the proposed approach within the MICE environment in R Statistical Software.

Environmental arsenic exposure and risk of diabetes type 2 in Ron Phibun subdistrict, Nakhon Si Thammarat Province, Thailand: unmatched and matched case-control studies.

BackgroundThere are inconsistent findings on associations between low-to-moderate level of arsenic in water and diabetes risk from previous epidemiological reports. In Ron Phibun subdistrict, Nakhon Si Thammarat Province, Thailand, a low level of arsenic exposure among population was observed and increased diabetes mellitus (DM) rate was identified.ObjectivesWe aimed to investigate the association between determinants (including low-level water arsenic exposure) of DM type 2 risk among residents of three villages of Ron Phibun subdistrict, Nakhon Si Thammarat Province.Materials and methodsSecondary data from two previous community based-studies, conducted in 2000 and 2008, were utilized. Data on independent variables relating to arsenic exposure and sociodemographic characteristics were taken from questionnaires and worksheets for health-risk screening. Water samples collected during household visit were sent for analysis of arsenic level at certified laboratories. Diabetes cases (N=185) were those who had been diagnosed with DM type 2. Two groups of controls, one unmatched to cases (n=200) and one pair matched on age and gender (n=200), were selected for analysis as unmatched and matched case–control studies, respectively. A multiple imputation technique was used to impute missing values of independent variables. Multivariable logistic regression models, with independent variables for arsenic exposure and sociodemographic characteristics, were constructed. The unmatched and matched data sets were analyzed using unconditional and conditional logistic analyses, respectively.ResultsOlder age, body mass index (BMI), having a history of illness in siblings and parents, and drinking were associated with increased DM type 2 risk. We found no convincing association between DM type 2 risk and water arsenic concentration in either study.ConclusionWe did not observe meaningful association between diabetes risk and the low-to-moderate arsenic levels observed in this study. Further research is needed to confirm this finding in the study area and elsewhere in Thailand.

A note on multiple imputation under complex sampling

SUMMARY Multiple imputation is popular for handling item nonresponse in survey sampling. Current multiple imputation techniques with complex survey data assume that the sampling design is ignorable. In this paper, we propose a new multiple imputation procedure for parametric inference without this assumption. Instead of using the sample-data likelihood, we use the sampling distribution of the pseudo maximum likelihood estimator to derive the posterior distribution of the parameters. The asymptotic properties of the proposed method are investigated. A simulation study confirms that the new procedure provides unbiased point estimation and valid confidence intervals with correct coverage properties whether or not the sampling design is ignorable.

Missing Data in the Field of Otorhinolaryngology and Head & Neck Surgery: Need for Improvement.

Clinical studies are often facing missing data. Data can be missing for various reasons, for example, patients moved, certain measurements are only administered in high-risk groups, and patients are unable to attend clinic because of their health status. There are various ways to handle these missing data (e.g., complete cases analyses, mean substitution). Each of these techniques potentially influences both the analyses and the results of a study. The first aim of this structured review was to analyze how often researchers in the field of otorhinolaryngology/head & neck surgery report missing data. The second aim was to systematically describe how researchers handle missing data in their analyses. The third aim was to provide a solution on how to deal with missing data by means of the multiple imputation technique. With this review, we aim to contribute to a higher quality of reporting in otorhinolaryngology research. Clinical studies among the 398 most recently published research articles in three major journals in the field of otorhinolaryngology/head & neck surgery were analyzed based on how researchers reported and handled missing data. Of the 316 clinical studies, 85 studies reported some form of missing data. Of those 85, only a small number (12 studies, 3.8%) actively handled the missingness in their data. The majority of researchers exclude incomplete cases, which results in biased outcomes and a drop in statistical power. Within otorhinolaryngology research, missing data are largely ignored and underreported, and consequently, handled inadequately. This has major impact on the results and conclusions drawn from this research. Based on the outcomes of this review, we provide solutions on how to deal with missing data. To illustrate, we clarify the use of multiple imputation techniques, which recently became widely available in standard statistical programs.

Errors in meta-analyses of selection.

Errors in meta-analyses of selection.

The Missing Dimension of the Political Resource Curse Debate

Given the methodological sophistication of the debate over the “political resource curse”—the purported negative relationship between natural resource wealth (in particular oil wealth) and democracy—it is surprising that scholars have not paid more attention to the basic statistical issue of how to deal with missing data. This article highlights the problems caused by the most common strategy for analyzing missing data in the political resource curse literature—listwise deletion—and investigates how addressing such problems through the best-practice technique of multiple imputation affects empirical results. I find that multiple imputation causes the results of a number of influential recent studies to converge on a key common finding: A political resource curse does exist, but only since the widespread nationalization of petroleum industries in the 1970s. This striking finding suggests that much of the controversy over the political resource curse has been caused by a neglect of missing-data issues.

A proposed methodology for deriving tsunami fragility functions for buildings using optimum intensity measures

Tsunami fragility curves are statistical models which form a key component of tsunami risk models, as they provide a probabilistic link between a tsunami intensity measure (TIM) and building damage. Existing studies apply different TIMs (e.g. depth, velocity, force etc.) with conflicting recommendations of which to use. This paper presents a rigorous methodology using advanced statistical methods for the selection of the optimal TIM for fragility function derivation for any given dataset. This methodology is demonstrated using a unique, detailed, disaggregated damage dataset from the 2011 Great East Japan earthquake and tsunami (total 67,125 buildings), identifying the optimum TIM for describing observed damage for the case study locations. This paper first presents the proposed methodology, which is broken into three steps: (1) exploratory analysis, (2) statistical model selection and trend analysis and (3) comparison and selection of TIMs. The case study dataset is then presented, and the methodology is then applied to this dataset. In Step 1, exploratory analysis on the case study dataset suggests that fragility curves should be constructed for the sub-categories of engineered (RC and steel) and non-engineered (wood and masonry) construction materials. It is shown that the exclusion of buildings of unknown construction material (common practice in existing studies) may introduce bias in the results; hence, these buildings are estimated as engineered or non-engineered through use of multiple imputation (MI) techniques. In Step 2, a sensitivity analysis of several statistical methods for fragility curve derivation is conducted in order to select multiple statistical models with which to conduct further exploratory analysis and the TIM comparison (to draw conclusions which are non-model-specific). Methods of data aggregation and ordinary least squares parameter estimation (both used in existing studies) are rejected as they are quantitatively shown to reduce fragility curve accuracy and increase uncertainty. Partially ordered probit models and generalised additive models (GAMs) are selected for the TIM comparison of Step 3. In Step 3, fragility curves are then constructed for a number of TIMs, obtained from numerical simulation of the tsunami inundation of the 2011 GEJE. These fragility curves are compared using K-fold cross-validation (KFCV), and it is found that for the case study dataset a force-based measure that considers different flow regimes (indicated by Froude number) proves the most efficient TIM. It is recommended that the methodology proposed in this paper be applied for defining future fragility functions based on optimum TIMs. With the introduction of several concepts novel to the field of fragility assessment (MI, GAMs, KFCV for model optimisation and comparison), this study has significant implications for the future generation of empirical and analytical fragility functions.

Multiple imputation of missing covariates for the Cox proportional hazards cure model.

We explore several approaches for imputing partially observed covariates when the outcome of interest is a censored event time and when there is an underlying subset of the population that will never experience the event of interest. We call these subjects 'cured', and we consider the case where the data are modeled using a Cox proportional hazards (CPH) mixture cure model. We study covariate imputation approaches using fully conditional specification. We derive the exact conditional distribution and suggest a sampling scheme for imputing partially observed covariates in the CPH cure model setting. We also propose several approximations to the exact distribution that are simpler and more convenient to use for imputation. A simulation study demonstrates that the proposed imputation approaches outperform existing imputation approaches for survival data without a cure fraction in terms of bias in estimating CPH cure model parameters. We apply our multiple imputation techniques to a study of patients with head and neck cancer. Copyright © 2016 John Wiley & Sons, Ltd.

Multiple imputation as one tool to provide longitudinal databases for modelling human height and weight development.

Besides large efforts regarding field work, provision of valid databases requires statistical and informational infrastructure to enable long-term access to longitudinal data sets on height, weight and related issues. To foster use of longitudinal data sets within the scientific community, provision of valid databases has to address data-protection regulations. It is, therefore, of major importance to hinder identifiability of individuals from publicly available databases. To reach this goal, one possible strategy is to provide a synthetic database to the public allowing for pretesting strategies for data analysis. The synthetic databases can be established using multiple imputation tools. Given the approval of the strategy, verification is based on the original data. Multiple imputation by chained equations is illustrated to facilitate provision of synthetic databases as it allows for capturing a wide range of statistical interdependencies. Also missing values, typically occurring within longitudinal databases for reasons of item non-response, can be addressed via multiple imputation when providing databases. The provision of synthetic databases using multiple imputation techniques is one possible strategy to ensure data protection, increase visibility of longitudinal databases and enhance the analytical potential.

Assessing the Everyday Discrimination Scale among American Indians and Alaska Natives.

The Everyday Discrimination Scale (EDS) has been used widely as a measure of subjective experiences of discrimination. The usefulness of this measure for assessments of perceived experiences of discrimination by American Indian and Alaska Native (AI/AN) peoples has not been explored. Data derived from the Special Diabetes Program for Indians--Healthy Heart Demonstration Project (SDPI-HH), a large-scale initiative to reduce cardiovascular risk among AI/ANs with Type 2 diabetes. Participants (N = 3,039) completed a self-report survey that included the EDS and measures of convergent and divergent validity. Missing data were estimated by multiple imputation techniques. Reliability estimates for the EDS were calculated, yielding a single factor with high internal consistency (α = .92). Younger, more educated respondents reported greater perceived discrimination; retired or widowed respondents reported less. Convergent validity was evidenced by levels of distress, anger, and hostility, which increased as the level of perceived discrimination increased (all p < .001). Divergent validity was evidenced by the absence of an association between EDS and resilient coping. Resilient coping and insulin-specific diabetes knowledge were not significantly associated with perceived discrimination (p = .61 and 0.16, respectively). However, general diabetes-related health knowledge was significantly associated with perceived discrimination (p = .02). The EDS is a promising measure for assessing perceived experiences of discrimination among those AI/ANs who participated in the SDPI-HH.

The effectiveness of Corticosteroids on mortality in patients with acute respiratory distress syndrome or acute lung injury: a secondary analysis.

The development of acute respiratory distress syndrome (ARDS) is associated with dys-regulated inflammation. Since corticosteroids are potent anti-inflammatory drugs, they are thought to be beneficial for ARDS patients. The study aimed to investigate the effectiveness of corticosteroids on mortality outcome in ARDS patients. The study was a secondary analysis of a prospective randomized controlled trial (NCT00979121). ARDS patients with invasive mechanical ventilation were enrolled. Corticosteroids use was defined as IV or PO administration of corticosteroids totaling more than 20 mg methylprednisolone equivalents during one calendar day. Missing data were handled using multiple imputation technique. Multivariable model was built to adjust for confounding covariates. A total of 745 patients were enrolled, including 540 survivors and 205 non-survivors. Patients in the non-survivor group were more likely to use corticosteroids (38% vs. 29.8%; p = 0.032). After adjustment for other potential confounders, corticosteroids showed no statistically significant effect on mortality outcome (OR: 1.18; 95% CI: 0.81–1.71). Furthermore, we investigated the interaction between corticosteroid use and variables of vasopressor and PaO2. The result showed that there was no significant interaction. In conclusion, the study failed to identify any beneficial effects of corticosteroids on mortality outcome in patients with ARDS.

Using Classification and Regression Trees (CART) and random forests to analyze attrition: Results from two simulations.

In this article, we describe a recent development in the analysis of attrition: using classification and regression trees (CART) and random forest methods to generate inverse sampling weights. These flexible machine learning techniques have the potential to capture complex nonlinear, interactive selection models, yet to our knowledge, their performance in the missing data analysis context has never been evaluated. To assess the potential benefits of these methods, we compare their performance with commonly employed multiple imputation and complete case techniques in 2 simulations. These initial results suggest that weights computed from pruned CART analyses performed well in terms of both bias and efficiency when compared with other methods. We discuss the implications of these findings for applied researchers.

What We Miss By Missing Data: Aid Effectiveness Revisited

Missing data is an issue in many empirical applications, as it may entail efficiency losses as well as biased results. We address these problems within the literature that investigates the effect of foreign aid on welfare. Using multiple imputation techniques to account for the missing data, we find lower aid effectiveness. In addition, imputation allows for comparison of different welfare indicators within the same framework. We find that the respective indicator choice can matter for the ensuing results.

A novel approach for obtaining a near complete dataset for time-to-event outcomes in the bronchiolitis of infancy discharge study (BIDS) using an informed multiple imputation technique

Time to cough resolution was the primary outcome for the BIDS trial (http://www.nets.nihr.ac.uk/projects/hta/099116). Patients were followed up at 7, 14, 28 days and 6 months after randomisation. Strenuous efforts were made by the research team in order to achieve a near complete data set. However there were still some parents/guardians who could not recall the date on which their infant stopped coughing, but who could recall whether their infant had stopped coughing since last followed up. We assumed that the lost event dates were missing at random. So we developed a multiple imputation technique: If it was known that the event occurred, but the precise date was unknown, a value was chosen between the date that the event was last known to not be present and the date of the follow up when it was present. Then a value was chosen at random from patients in the same treatment group whose cough stopped in a similar time frame. This process was repeated 100 times, and the analysis done on each dataset. The mean values for the estimate of the median and its CI limits were used for reporting. In this way we managed to increase the number of available event dates from 507 (82% of 615) to 589 (96% of 615). By introducing simple “yes/no event occurred” questions at each follow-up visit and the usage of programmatic tools we recovered around 13.4% of the data with the informed imputation technique. The imputed and complete case analyses reached the same conclusion.

Community-based mental health treatments for survivors of torture and militant attacks in Southern Iraq: a randomized control trial.

BackgroundSystematic violence is a long-standing problem in Iraq. Research indicates that survivors often experience multiple mental health problems, and that there is a need for more rigorous research that targets symptoms beyond post-traumatic stress (PTS). Our objective was to test the effectiveness of two counseling therapies in Southern Iraq in addressing multiple mental health problems among survivors of systematic violence: (1) a transdiagnostic intervention (Common Elements Treatment Approach or CETA); and (2) cognitive processing therapy (CPT). The therapies were provided by non-specialized health workers since few MH professionals are available to provide therapy in Iraq.MethodsThis was a randomized, parallel, two site, two-arm (1:1 allocation), single-blinded, wait-list controlled (WLC) trial of CETA in one site (99 CETA, 50 WLC), and CPT in a second site (129 CPT, 64 WLC). Eligibility criteria were elevated trauma symptoms and experience of systematic violence. The primary and secondary outcomes were trauma symptoms and dysfunction, respectively, with additional assessment of depression and anxiety symptoms. Non-specialized health workers (community mental health worker, CMHW) provided the interventions in government-run primary health centers. Treatment effects were determined using longitudinal, multilevel models with CMHW and client as random effects, and a time by group interaction with robust variance estimation, to test for the net difference in mean score for each outcome between the baseline and follow up interview. Multiple imputation techniques were used to account for missingness at the item level and the participant level. All analyses were conducted using Stata 12.ResultsThe CETA intervention showed large effect sizes for all outcomes. The CPT intervention showed moderate effects sizes for trauma and depression, with small to no effect for anxiety or dysfunction, respectively.ConclusionsBoth CETA and CPT appear to benefit survivors of systematic violence in Southern Iraq by reducing multiple mental health symptoms, with CETA providing a very large benefit across a range of symptoms. Non-specialized health workers were able to treat comorbid symptoms of trauma, depression and anxiety, and dysfunction among survivors of systematic violence who have limited access to mental health professionals. The trial further supports the use of evidence-based therapies in lower-resource settings.Trial registration and protocolThis trial was registered at ClinicalTrials.gov on 16 July 2010 with an identifier of NCT01177072 as the Study of Effectiveness of Mental Health Interventions among Torture Survivors in Southern Iraq.The study protocol can be downloaded from the following website: http://tinyurl.com/CETA-Iraq-Protocol. In the protocol, the CETA intervention is given a different name: components-based intervention or CBI.Electronic supplementary materialThe online version of this article (doi:10.1186/s12888-015-0622-7) contains supplementary material, which is available to authorized users.

Utility valuations of health states after traumatic brain injury: preliminary results

METHODS: We did a 2 year (2011-2012) analysis of National Trauma Data Bank.We included all patients with isolated severe TBI (head AIS 3 and other body AIS<3). Patients were divided into two groups: Hypothermic (admission temperature <35 C) and Normothermic (admission temperature 35 C).Our outcome measures were mortality, hospital, and ICU length of stay.We performed missing values analysis and multiple imputation technique to account for missing data. Propensity score matching was performed adjusting for age, gender, race, insurance status, h-AIS, ISS, systolic blood pressure, heart rate, and oxygen saturation.