Relationship Between Pain and Delirium in Critically Ill Adults.

OBJECTIVES: Although opioids are frequently used to treat pain, and are an important risk for ICU delirium, the association between ICU pain itself and delirium remains unclear. We sought to evaluate the relationship between ICU pain and delirium. DESIGN: Prospective cohort study. SETTING: A 32-bed academic medical-surgical ICU. PATIENTS: Critically ill adults (n = 4,064) admitted greater than or equal to 24 hours without a condition hampering delirium assessment. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Daily mental status was classified as arousable without delirium, delirium, or unarousable. Pain was assessed six times daily in arousable patients using a 0-10 Numeric Rating Scale (NRS) or the Critical Care Pain Observation Tool (CPOT); daily peak pain score was categorized as no (NRS = 0/CPOT = 0), mild (NRS = 1-3/CPOT = 1-2), moderate (NRS = 4-6/CPOT = 3-4), or severe (NRS = 7-10/CPOT = 5-8) pain. To address missingness, a Multiple Imputation by Chained Equations approach that used available daily pain severity and 19 pain predictors was used to generate 25 complete datasets. Using a first-order Markov model with a multinomial logistic regression analysis, that controlled for 11 baseline/daily delirium risk factors and considered the competing risks of unarousability and ICU discharge/death, the association between peak daily pain and next-day delirium in each complete dataset was evaluated. RESULTS: Among 14,013 ICU days (contributed by 4,064 adults), delirium occurred on 2,749 (19.6%). After pain severity imputation on 1,818 ICU days, mild, moderate, and severe pain were detected on 2,712 (34.1%), 1,682 (21.1%), and 894 (11.2%) of the no-delirium days, respectively, and 992 (36.1%), 513 (18.6%), and 27 (10.1%) of delirium days (p = 0.01). The presence of any pain (mild, moderate, or severe) was not associated with a transition from awake without delirium to delirium (aOR 0.96; 95% CI, 0.76-1.21). This association was similar when days with only mild, moderate, or severe pain were considered. All results were stable after controlling for daily opioid dose. CONCLUSIONS: After controlling for multiple delirium risk factors, including daily opioid use, pain may not be a risk factor for delirium in the ICU. Future prospective research is required.

"Multivariate analysis of the impact of sleep and working hours on medical errors: a MICE approach".

BACKGROUND: The main objective of this study was to describe the relationship between working conditions, sleep and psycho-affective variables and medical errors. METHODS: This was an observational, analytical and cross-sectional study in which 661 medical residents answered questionnaires about working conditions, sleep and psycho-affective variables. Actigraphic sleep parameters and peripheral temperature circadian rhythm were measured in a subgroup of 38 subjects. Bivariate and multivariate predictors of medical errors were assessed. RESULTS: Medical residents reported working 66.2 ± 21.9 weekly hours. The longest continuous shift was of 28.4 ± 10.9 h. They reported sleeping 6.1 ± 1.6 h per day, with a sleep debt of 94 ± 129 min in workdays. A high percentage of them reported symptoms related to psycho-affective disorders. The longest continuous shift duration (OR = 1.03 [95% CI, 1.00-1.05], p = 0.01), working more than six monthly on-call shifts (OR = 1.87 [95% CI, 1.16-3.02], p = 0.01) and sleeping less than six hours per working day (OR = 1.66 [95% CI, 1.10-2.51], p = 0.02) were independently associated with self-reported medical errors. The report of medical errors was associated with an increase in the percentage of diurnal sleep (2.2% [95% CI, 0.1-4.3] vs 14.5% [95% CI, 5.9-23.0]; p = 0.01) in the actigraphic recording. CONCLUSIONS: Medical residents have a high working hour load that affect their sleep opportunities, circadian rhythms and psycho-affective health, which are also related to the report of medical errors. These results highlight the importance of implementing multidimensional strategies to improve medical trainees' sleep and wellbeing, increasing in turn their own and patients' safety.

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.

SuperMICE: An Ensemble Machine Learning Approach to Multiple Imputation by Chained Equations.

Researchers often face the problem of how to address missing data. Multiple imputation is a popular approach, with multiple imputation by chained equations (MICE) being among the most common and flexible methods for execution. MICE iteratively fits a predictive model for each variable with missing values, conditional on other variables in the data. In theory, any imputation model can be used to predict the missing values. However, if the predictive models are incorrectly specified, they may produce biased estimates of the imputed data, yielding inconsistent parameter estimates and invalid inference. Given the set of modeling choices that must be made in conducting multiple imputation, in this paper we propose a data-adaptive approach to model selection. Specifically, we adapt MICE to incorporate an ensemble algorithm, Super Learner, to predict the conditional mean for each missing value, and we also incorporate a local kernel-based estimate of variance. We present a set of simulations indicating that this approach produces final parameter estimates with lower bias and better coverage than other commonly used imputation methods. These results suggest that using a flexible machine learning imputation approach can be useful in settings where data are missing at random, especially when the relationships among the variables are complex.

Bayesian evidence synthesis in case of multi-cohort datasets: An illustration by multi-informant differences in self-control.

The trend toward large-scale collaborative studies gives rise to the challenge of combining data from different sources efficiently. Here, we demonstrate how Bayesian evidence synthesis can be used to quantify and compare support for competing hypotheses and to aggregate this support over studies. We applied this method to study the ordering of multi-informant scores on the ASEBA Self Control Scale (ASCS), employing a multi-cohort design with data from four Dutch cohorts. Self-control reports were collected from mothers, fathers, teachers and children themselves. The available set of reporters differed between cohorts, so in each cohort varying components of the overarching hypotheses were evaluated. We found consistent support for the partial hypothesis that parents reported more self-control problems than teachers. Furthermore, the aggregated results indicate most support for the combined hypothesis that children report most problem behaviors, followed by their mothers and fathers, and that teachers report the fewest problems. However, there was considerable inconsistency across cohorts regarding the rank order of children's reports. This article illustrates Bayesian evidence synthesis as a method when some of the cohorts only have data to evaluate a partial hypothesis. With Bayesian evidence synthesis, these cohorts can still contribute to the aggregated results.

Approximate confidence intervals for the likelihood ratios of a binary diagnostic test in the presence of partial disease verification.

The classic parameters used to assess the accuracy of a binary diagnostic test (BDT) are sensitivity and specificity. Other parameters used to describe the performance of a BDT are likelihood ratios (LRs). The LRs depend on the sensitivity and the specificity of the diagnostic test, and they reflect how much greater the probability of a positive or negative diagnostic test result for individuals with the disease than that for the individuals without the disease. In this study, several confidence intervals are studied for the LRs of a BDT in the presence of missing data. Two confidence intervals were studied through the method of maximum likelihood and seven confidence intervals were studied by applying the multiple imputation by chained equations method. A program in R software has been written that allows us to solve the estimation problem posed. The results obtained have been applied to the two real examples.

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.

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.

Similarity indices of meteo-climatic gauging stations: definition and comparison.

Space-time dependencies among monitoring network stations have been investigated to detect and quantify similarity relationships among gauging stations. In this work, besides the well-known rank correlation index, two new similarity indices have been defined and applied to compute the similarity matrix related to the Apulian meteo-climatic monitoring network. The similarity matrices can be applied to address reliably the issue of missing data in space-time series. In order to establish the effectiveness of the similarity indices, a simulation test was then designed and performed with the aim of estimating missing monthly rainfall rates in a suitably selected gauging station. The results of the simulation allowed us to evaluate the effectiveness of the proposed similarity indices. Finally, the multiple imputation by chained equations method was used as a benchmark to have an absolute yardstick for comparing the outcomes of the test. In conclusion, the new proposed multiplicative similarity index resulted at least as reliable as the selected benchmark.

The natural history of AVM hemorrhage in the posterior fossa: comparison of hematoma volumes and neurological outcomes in patients with ruptured infra- and supratentorial AVMs.

OBJECT: Patients with posterior fossa arteriovenous malformations (AVMs) are more likely to present with hemorrhage than those with supratentorial AVMs. Observed patients subject to the AVM natural history should be informed of the individualized effects of AVM characteristics on the clinical course following a new, first-time hemorrhage. The authors hypothesize that the debilitating effects of first-time bleeding from an AVM in a previously intact patient with an unruptured AVM are more pronounced when AVMs are located in the posterior fossa. METHODS: The University of California, San Francisco prospective registry of brain AVMs was searched for patients with a ruptured AVM who had a pre-hemorrhage modified Rankin Scale (mRS) score of 0 and a post-hemorrhage mRS score obtained within 2 days of the hemorrhagic event. A total of 154 patients met the inclusion criteria for this study. Immediate post-hemorrhage presentation mRS scores were dichotomized into nonsevere outcome (mRS ≤ 3) and severe outcome (mRS > 3). There were 77 patients in each group. Univariate and multivariate logistic regression analyses using severe outcome as the binary response were run. The authors also performed a logistic regression analysis to measure the effects of hematoma volume and AVM location on severe outcome. RESULTS: Posterior fossa location was a significant predictor of severe outcome (OR 2.60, 95% CI 1.20-5.67, p = 0.016) and the results were strengthened in a multivariate model (OR 4.96, 95% CI 1.73-14.17, p = 0.003). Eloquent location (OR 3.47, 95% CI 1.37-8.80, p = 0.009) and associated arterial aneurysms (OR 2.58, 95% CI 1.09, 6.10; p = 0.031) were also significant predictors of poor outcome. Hematoma volume for patients with a posterior fossa AVM was 10.1 ± 10.1 cm(3) compared with 25.6 ±28.0 cm(3) in supratentorial locations (p = 0.003). A logistic analysis (based on imputed hemorrhage volume values) found that posterior fossa location was a significant predictor of severe outcome (OR 8.03, 95% CI 1.20-53.77, p = 0.033) and logarithmic hematoma volume showed a positive, but not statistically significant, association in the model (p = 0.079). CONCLUSIONS: Patients with posterior fossa AVMs are more likely to have severe outcomes than those with supratentorial AVMs based on this natural history study. Age, sex, and ethnicity were not associated with an increased risk of severe outcome after AVM rupture, but posterior fossa location, associated aneurysms, and eloquent location were associated with poor post-hemorrhage mRS scores. Posterior fossa hematomas are poorly tolerated, with severe outcomes observed even with smaller hematoma volumes. These findings support an aggressive surgical posture with respect to posterior fossa AVMs, both before and after rupture.

Missing data in Wave 2 of NSHAP: prevalence, predictors, and recommended treatment.

OBJECTIVES: This report seeks to inform National Social Life, Health, and Aging Project (NSHAP) data users of the prevalence and predictors of missing data in the in-person interview (CAPI) and leave-behind questionnaire (LBQ) in Wave 2 of NSHAP, and methods to handle missingness. METHOD: Missingness is quantified at the unit and item levels separately for CAPI and LBQ data, and at the item level is assessed within domains of conceptually related variables. Logistic and negative binomial regression analyses are used to model predictors of unit- and item-level nonresponse, respectively. RESULTS: Unit-level nonresponse on the CAPI was 10.6% of those who responded at Wave 1, and LBQ nonresponse was 11.37% of those who completed the Wave 2 CAPI component. CAPI item-level missingness was less than 1% of items for most domains but 7.1% in the Employment and Finances domain. LBQ item-level missingness was 5% across domains but 8.3% in the Attitudes domain. Missingness was predicted by characteristics of the sample and features of the study design. DISCUSSION: Multiple imputation is recommended to handle unit- and item-level missingness and can be readily and flexibly conducted with multiple imputation by chained equations, inverse probability weighting, and in some instances, full-information maximum-likelihood methods.