Multiple imputation for missing continuous and dichotomous data: The effects of patterns of missingness and variable correlations on type I error with small samples

The purpose of this study is to investigate the effects of predictor variable correlations and patterns of missingness with dichotomous and/or continuous data in small samples when missing data is multiply imputed. Missing data of predictor variables is multiply imputed under three different multivariate models: the multivariate normal model for continuous data, the multinomial model for dichotomous data and the general location model for mixed dichotomous and continuous data. Subsequent to the multiple imputation process, Type I error rates of the regression coefficients obtained with logistic regression analysis are estimated under various conditions of correlation structure, sample size, type of data and patterns of missing data. The distributional properties of average mean, variance and correlations among the predictor variables are assessed after the multiple imputation process. ^ For continuous predictor data under the multivariate normal model, Type I error rates are generally within the nominal values with samples of size n = 100. Smaller samples of size n = 50 resulted in more conservative estimates (i.e., lower than the nominal value). Correlation and variance estimates of the original data are retained after multiple imputation with less than 50% missing continuous predictor data. For dichotomous predictor data under the multinomial model, Type I error rates are generally conservative, which in part is due to the sparseness of the data. The correlation structure for the predictor variables is not well retained on multiply-imputed data from small samples with more than 50% missing data with this model. For mixed continuous and dichotomous predictor data, the results are similar to those found under the multivariate normal model for continuous data and under the multinomial model for dichotomous data. With all data types, a fully-observed variable included with variables subject to missingness in the multiple imputation process and subsequent statistical analysis provided liberal (larger than nominal values) Type I error rates under a specific pattern of missing data. It is suggested that future studies focus on the effects of multiple imputation in multivariate settings with more realistic data characteristics and a variety of multivariate analyses, assessing both Type I error and power. ^

Survey of shift and work tolerance of City of Houston emergency medical service personnel

This study of ambulance workers for the emergency medical services of the City of Houston studied the factors related to shiftwork tolerance and intolerance. The EMS personnel work a 24-hour shift with rotating days of the week. Workers are assigned to A, B, C, D shift, each of which rotate 24-hours on, 24-hours off, 24-hours on and 4 days off. One-hundred and seventy-six male EMTs, paramedics and chauffeurs from stations of varying levels of activity were surveyed. The sample group ranged in age from 20 to 45. The average tenure on the job was 8.2 years. Over 68% of the workers held a second job, the majority of which worked over 20 hours a week at the second position.^ The survey instrument was a 20-page questionnaire modeled after the Folkard Standardized Shiftwork Index. In addition to demographic data, the survey tool provided measurements of general job satisfaction, sleep quality, general health complaints, morningness/eveningness, cognitive and somatic anxiety, depression, and circadian types. The survey questionnaire included an EMS-specific scaler of stress.^ A conceptual model of Shiftwork Tolerance was presented to identify the key factors examined in the study. An extensive list of 265 variables was reduced to 36 key variables that related to: (1) shift schedule and demographic/lifestyle factors, (2) individual differences related to traits and characteristics, and (3) tolerance/intolerance effects. Using the general job satisfaction scaler as the key measurement of shift tolerance/intolerance, it was shown that a significant relationship existed between this dependent variable and stress, number of years working a 24-hour shift, sleep quality, languidness/vigorousness. The usual amount of sleep received during the shift, general health complaints and flexibility/rigidity (R$\sp2$ =.5073).^ The sample consisted of a majority of morningness-types or extreme-morningness types, few evening-types and no extreme-evening types, duplicating the findings of Motohashi's previous study of ambulance workers. The level of activity by station was not significant on any of the dependent variables examined. However, the shift worked had a relationship with sleep quality, despite the fact that all shifts work the same hours and participate in the same rotation schedule. ^

Estimating the mean and standard deviation of concentration distributions using censored samples

Environmental data sets of pollutant concentrations in air, water, and soil frequently include unquantified sample values reported only as being below the analytical method detection limit. These values, referred to as censored values, should be considered in the estimation of distribution parameters as each represents some value of pollutant concentration between zero and the detection limit. Most of the currently accepted methods for estimating the population parameters of environmental data sets containing censored values rely upon the assumption of an underlying normal (or transformed normal) distribution. This assumption can result in unacceptable levels of error in parameter estimation due to the unbounded left tail of the normal distribution. With the beta distribution, which is bounded by the same range of a distribution of concentrations, $\rm\lbrack0\le x\le1\rbrack,$ parameter estimation errors resulting from improper distribution bounds are avoided. This work developed a method that uses the beta distribution to estimate population parameters from censored environmental data sets and evaluated its performance in comparison to currently accepted methods that rely upon an underlying normal (or transformed normal) distribution. Data sets were generated assuming typical values encountered in environmental pollutant evaluation for mean, standard deviation, and number of variates. For each set of model values, data sets were generated assuming that the data was distributed either normally, lognormally, or according to a beta distribution. For varying levels of censoring, two established methods of parameter estimation, regression on normal ordered statistics, and regression on lognormal ordered statistics, were used to estimate the known mean and standard deviation of each data set. The method developed for this study, employing a beta distribution assumption, was also used to estimate parameters and the relative accuracy of all three methods were compared. For data sets of all three distribution types, and for censoring levels up to 50%, the performance of the new method equaled, if not exceeded, the performance of the two established methods. Because of its robustness in parameter estimation regardless of distribution type or censoring level, the method employing the beta distribution should be considered for full development in estimating parameters for censored environmental data sets. ^

The role of LMX1B and PITX2 in anterior segment development and adult ocular function

Several congenital syndromes associated with anterior segment (AS) anomalies can lead to impaired vision and glaucoma, such as nail-patella syndrome (NPS), caused by mutations in the LIM homeodomain transcription factor LMX1B and Axenfeld-Rieger's syndrome (ARS), caused by mutations in the bicoid-related homeodomain transcription factor PITX2. Targeted mutations in lmx1b and pitx2 and RNA in situ analysis reveal that both genes are required for AS development and are co-expressed within the periocular mesenchyme, suggesting they participate in a shared genetic pathway. Lmx1b homozygous mutants display iris and corneal stroma hypoplasia, and defects in ciliary body formation. In contrast, pitx2 homozygous mutants exhibit a more severe phenotype: the AS chamber, corneal endothelium, and extraocular muscles (EOM) fail to develop. The absence of EOM in pitx2 mutants suggests pitx2 acts upstream of lmx1b, or that other lmx1b family members, such as lmx1a, can compensate for lmx1b function. Lmxla/lmx1b double homozygous mutants have a reduced capacity to generate EOM, implying that lmx1 gene products have a redundant function in EOM development and that lmx1 family members may act downstream of pitx2. However, analysis of pitx2 expression in the AS tissues of lmx1b mutants and reciprocal studies of lmx1b expression in pitx2 mutants indicate that these genes do not function in a simple linear pathway. Instead, lmx1b and pitx2 may regulate a shared set of downstream targets or both genes may work in parallel transcribing unique targets required for a common biological process. Ultrastructural analysis of lmx1b and pitx2 mutant corneas indicates that collagen fibrillogenesis is perturbed, revealing a common role for both genes in the deposition of extracellular matrix. Furthermore, lmx1b/pitx2 double heterozygotes develop corneal opacities not observed in single heterozygotes demonstrating that lmx1b and pitx2 genetically interact. Data suggests that defects in the basement membrane of the corneal endothelium underlie the opacities observed in double heterozygotes. Additionally, double heterozygotes develop anterior synechias that occlude the trabecular meshwork, potentially blocking aqueous humor drainage. These data suggest that lmx1b and pitx2 are responsible for ECM deposition in multiple cell types and imply that such defects may contribute to the glaucomas observed in NPS and ARS patients. ^

Mouse retinal development: Role of cell adhesion and mechanism of gene regulation

Cell differentiation and pattern formation are fundamental processes in animal development that are under intense investigation. The mouse retina is a good model to study these processes because it has seven distinct cell types, and three well-laminated nuclear layers that form during embryonic and postnatal life. β-catenin functions as both the nuclear effector for the canonical Wnt pathway and a cell adhesion molecule, and is required for the development of various organs. To study the function of β-catenin in retinal development, I used a Cre-loxP system to conditionally ablate β-catenin in the developing retina. Deletion of β-catenin led to disrupted laminar structure but did not affect the differentiation of any of the seven cell types. Eliminating β-catenin did not reduce progenitor cell proliferation, although enhanced apoptosis was observed. Further analysis showed that disruption of cell adhesion was the major cause of the observed patterning defects. Overexpression of β-catenin during retinal development also disrupted the normal retinal lamination and caused a transdifferentiation of neurons into pigmented cells. The results indicate that β-catenin functions as a cell adhesion molecule but not as a Wnt pathway component during retinal neurogenesis, and is essential for lamination but not cell differentiation. The results further imply that retinal lamination and cell differentiation are genetically separable processes. ^ Sonic hedgehog (shh) is expressed in retinal ganglion cells under the control of transcription factor Pou4f2 during retinal development. Previous studies identified a phylogenetically conserved region in the first intron of shh containing a Pou4f2 binding site. Transgenic reporter mice in which reporter gene expression was driven by this region showed that this element can direct gene expression specifically in the retina, but expression was not limited to the ganglion cells. From these data I hypothesized that this element is required for shh expression in the retina but is not sufficient for specific ganglion cell expression. To further test this hypothesis, I created a conditional allele by flanking this region with two loxP sites. Lines carrying this allele will be crossed with retinal-specific Cre lines to remove this element in the retina. My hypothesis predicts that alteration in shh expression and subsequent retinal defects will occur in the retinas of these mice. ^

Gender disparities in utilization, unmet need and expenditure for mental health care

Objectives: To compare mental health care utilization regarding the source, types, and intensity of mental health services received, unmet need for services, and out of pocket cost among non-institutionalized psychologically distressed women and men. ^ Method: Cross-sectional data for 19,325 non-institutionalized mentally distressed adult respondents to the “The National Survey on Drug Use and Health” (NSDUH), for the years 2006 -2008, representing over twenty-nine millions U.S. adults was analyzed. To assess the relative odds for women compared to men, logistic regression analysis was used for source of service, for types of barriers, for unmet need and cost; zero inflated negative binomial regression for intensity of utilization; and ordinal logistic regression analysis for quantifying out-of-pocket expenditure. ^ Results: Overall, 43% of mentally distressed adults utilized a form of mental health treatment; representing 12.6 million U.S psychologically distressed adults. Females utilized more mental health care compared to males in the previous 12 months (OR: 1. 70; 95% CI: 1.54, 1.83). Similarly, females were 54% more likely to get help for psychological distress in an outpatient setting and females were associated with an increased probability of using medication for mental distress (OR: 1.72; 95% CI: 1.63, 1.98). Women were 1.25 times likelier to visit a mental health center (specialty care) than men. ^ Females were positively associated with unmet needs (OR: 1.50; 95% CI: 1.29, 1.75) after taking into account predisposing, enabling, and need (PEN) characteristics. Women with perceived unmet needs were 23% (OR: 0.77; 95% CI: 0.59, 0.99) less likely than men to report societal accommodation (stigma) as a barrier to mental health care. At any given cutoff point, women were 1.74 times likelier to be in the higher payment categories for inpatient out of pocket cost when other variables in the model are held constant. Conclusions: Women utilize more specialty mental healthcare, report more unmet need, and pay more inpatient out of pocket costs than men. These gender disparities exist even after controlling for predisposing, enabling, and need variables. Creating policies that not only provide mental health care access but also de-stigmatize mental illness will bring us one step closer to eliminating gender disparities in mental health care.^

Data Management Policies and Issues with Life Science Data

Data management and sharing are relatively new concepts in the health and life sciences fields. This presentation will cover some basic policies as well as the impediments to data sharing unique to health and life sciences data.

Evaluation of elevator shafts as a pathway for fungal spores and particles to enter a hospital housing immuno-compromised patients

Background. It is estimated that hospitals spend between 28 and 33 billion dollars per year as a result of hospital-acquired infections. (Scott, 2009) The costs continue to rise despite the guidance and controls provided by hospital infection control staff to reduce patient exposures to fungal spores and other infectious agents. With all processes and controls in place, the vented elevator shaft represents an unprotected opening from the top of the building to the lower floors. The hypothesis for this prospective study is that there is a positive correlation between the number of Penicillium/Aspergillus-like spores, Cladosporium, ascospores, basidiospores in spores/m3 as individual spore categories found in the hoistway vent of an elevator shaft and the levels of the same spores, sampled near-simultaneously in the outdoor intake of the elevator shaft. Specific aims of this study include determining if external Penicillium/Aspergillus-like spores are entering the healthcare facility via the elevator shaft and hoistway vents. Additional aims include determining levels of Penicillium/Aspergillus-like spores outdoors, in the elevator shafts, and indoors in areas possibly affected by elevator shaft air; and, finally, to evaluate whether any effect is observed due to the installation of a hoistway vent damper, installed serendipitously during this study. ^ Methods. Between April 2010 and September 2010, a total of 3,521 air samples were collected, including 363 spore trap samples analyzed microscopically for seven spore types, and polymerase chain reaction analyses on 254 air samples. 2178 particle count measurements, 363 temperature readings and 363 relative humidity readings were also obtained from 7 different locations potentially related to the path of air travel inside and near a centrally-located and representative elevator shaft. ^ Results. Mean Penicillium/Aspergillus-like spore values were higher outside the building (530 spores/m3 of air) than inside the hoistway (22.8 spores/m3) during the six month study. Mean values inside the hospital were lower than outside throughout the study, ranging from 15 to 73 spores/m3 of air. Mean Penicillium/Aspergillus-like spore counts inside the hoistway decreased from 40.1 spores/m3 of air to 9 spores/m3 of air following the installation of a back draft damper between the outside air and the elevator shaft. Comparison of samples collected outside the building and inside the hoistway vent prior to installing the damper indicated a strong positive correlation (Spearman's Rho=0.8008, p=0.0001). The similar comparison following the damper installation indicated a moderate non-significant inverse correlation (Spearman's rho = −0.2795, p=0.1347). ^ Conclusion. Elevator shafts are one pathway for mold spores to enter a healthcare facility. A significant correlation was detected between spores and particle counts inside the hoistway and outside prior to changes in the ventilation system. The insertion of the back draft damper appeared to lower the spore counts inside the hoistway and inside the building. The mold spore counts in air outside the study building were higher in the period following the damper installation while the levels inside the hoistway and hospital decreased. Cladosporium and Penicillium/Aspergillus -like spores provided a method for evaluating indoor air quality as a natural tracer from outside the building to inside the building. Ascospores and basidiospores were not a valuable tracer due to low levels of detection during this study. ^ Installation of a back draft damper provides additional protection for the indoor environment of a hospital or healthcare facility, including in particular patients who may be immunocompromised. Current design standards and references do not require the installation of a back draft damper, but evaluation of adding language to relevant building codes should be considered. The data indicate a reduction in levels of Penicillium/Aspergillus -like spores, particle counts and a reduction in relative humidity inside of the elevator shaft after damper installation.^

Application of the general linear model to assess the effect of missing data on bone marrow mononuclear cell therapy with infusion timing and follow-up MRI

With most clinical trials, missing data presents a statistical problem in evaluating a treatment's efficacy. There are many methods commonly used to assess missing data; however, these methods leave room for bias to enter the study. This thesis was a secondary analysis on data taken from TIME, a phase 2 randomized clinical trial conducted to evaluate the safety and effect of the administration timing of bone marrow mononuclear cells (BMMNC) for subjects with acute myocardial infarction (AMI).^ We evaluated the effect of missing data by comparing the variance inflation factor (VIF) of the effect of therapy between all subjects and only subjects with complete data. Through the general linear model, an unbiased solution was made for the VIF of the treatment's efficacy using the weighted least squares method to incorporate missing data. Two groups were identified from the TIME data: 1) all subjects and 2) subjects with complete data (baseline and follow-up measurements). After the general solution was found for the VIF, it was migrated Excel 2010 to evaluate data from TIME. The resulting numerical value from the two groups was compared to assess the effect of missing data.^ The VIF values from the TIME study were considerably less in the group with missing data. By design, we varied the correlation factor in order to evaluate the VIFs of both groups. As the correlation factor increased, the VIF values increased at a faster rate in the group with only complete data. Furthermore, while varying the correlation factor, the number of subjects with missing data was also varied to see how missing data affects the VIF. When subjects with only baseline data was increased, we saw a significant rate increase in VIF values in the group with only complete data while the group with missing data saw a steady and consistent increase in the VIF. The same was seen when we varied the group with follow-up only data. This essentially showed that the VIFs steadily increased when missing data is not ignored. When missing data is ignored as with our comparison group, the VIF values sharply increase as correlation increases.^

Data quality in trauma transfusion studies and the impact of missing data on predicting massive transfusion

Maximizing data quality may be especially difficult in trauma-related clinical research. Strategies are needed to improve data quality and assess the impact of data quality on clinical predictive models. This study had two objectives. The first was to compare missing data between two multi-center trauma transfusion studies: a retrospective study (RS) using medical chart data with minimal data quality review and the PRospective Observational Multi-center Major Trauma Transfusion (PROMMTT) study with standardized quality assurance. The second objective was to assess the impact of missing data on clinical prediction algorithms by evaluating blood transfusion prediction models using PROMMTT data. RS (2005-06) and PROMMTT (2009-10) investigated trauma patients receiving ≥ 1 unit of red blood cells (RBC) from ten Level I trauma centers. Missing data were compared for 33 variables collected in both studies using mixed effects logistic regression (including random intercepts for study site). Massive transfusion (MT) patients received ≥ 10 RBC units within 24h of admission. Correct classification percentages for three MT prediction models were evaluated using complete case analysis and multiple imputation based on the multivariate normal distribution. A sensitivity analysis for missing data was conducted to estimate the upper and lower bounds of correct classification using assumptions about missing data under best and worst case scenarios. Most variables (17/33=52%) had <1% missing data in RS and PROMMTT. Of the remaining variables, 50% demonstrated less missingness in PROMMTT, 25% had less missingness in RS, and 25% were similar between studies. Missing percentages for MT prediction variables in PROMMTT ranged from 2.2% (heart rate) to 45% (respiratory rate). For variables missing >1%, study site was associated with missingness (all p≤0.021). Survival time predicted missingness for 50% of RS and 60% of PROMMTT variables. MT models complete case proportions ranged from 41% to 88%. Complete case analysis and multiple imputation demonstrated similar correct classification results. Sensitivity analysis upper-lower bound ranges for the three MT models were 59-63%, 36-46%, and 46-58%. Prospective collection of ten-fold more variables with data quality assurance reduced overall missing data. Study site and patient survival were associated with missingness, suggesting that data were not missing completely at random, and complete case analysis may lead to biased results. Evaluating clinical prediction model accuracy may be misleading in the presence of missing data, especially with many predictor variables. The proposed sensitivity analysis estimating correct classification under upper (best case scenario)/lower (worst case scenario) bounds may be more informative than multiple imputation, which provided results similar to complete case analysis.^