Patterns and Predictors of Television Viewing and Computer Use Among Women Living in Socioeconomically Disadvantaged Neighborhoods: A Prospective Cohort Study.

BACKGROUND: Socioeconomically disadvantaged women are at an increased risk of sedentary behaviors including television (TV) viewing and computer use, so identifying determinants of these behaviors is important. METHODS: Women (n = 4349) self-reported weekly TV and computer time (in minutes per week), sociodemographic, and health data at 3 time points (2007-2013). Mixed-effect negative binomial regression was used to determine the baseline determinants of TV viewing and computer use over time, adjusting for confounders. RESULTS: Over 5 years, median TV viewing decreased while median computer time increased. Cross-sectionally TV viewing was highest among participants classified as obese, with poorer health, current smokers, with lower education, not working, with no income, without partners and with no children and computer time was greater among younger women, living in urban areas, working full time, with higher education, without partners and with no children. Average computer time per year increased among those not working (7%), with lower education (5%), and with children (5%) but decreased among those with higher education (1%). However, no factors were associated with a change in TV viewing over time. CONCLUSION: Among socioeconomically disadvantaged women, interventions aimed at preventing increases in computer time should consider women with lower education, not working, and with children in their design.

Highway safety assessment and improvement through crash prediction by injury severity and vehicle damage using Multivariate Poisson-Lognormal model and Joint Negative Binomial-Generalized Ordered Probit Fractional Split model.

INTRODUCTION: Predicting crash counts by severity plays a dominant role in identifying roadway sites that experience overrepresented crashes, or an increase in the potential for crashes with higher severity levels. Valid and reliable methodologies for predicting highway accidents by severity are necessary in assessing contributing factors to severe highway crashes, and assisting the practitioners in allocating safety improvement resources. METHODS: This paper uses urban and suburban intersection data in Connecticut, along with two sophisticated modeling approaches, i.e. a Multivariate Poisson-Lognormal (MVPLN) model and a Joint Negative Binomial-Generalized Ordered Probit Fractional Split (NB-GOPFS) model to assess the methodological rationality and accuracy by accommodating for the unobserved factors in predicting crash counts by severity level. Furthermore, crash prediction models based on vehicle damage level are estimated using the same two methodologies to supplement the injury severity in estimating crashes by severity when the sample mean of severe injury crashes (e.g., fatal crashes) is very low. RESULTS: The model estimation results highlight the presence of correlations of crash counts among severity levels, as well as the crash counts in total and crash proportions by different severity levels. A comparison of results indicates that injury severity and vehicle damage are highly consistent. CONCLUSIONS: Crash severity counts are significantly correlated and should be accommodated in crash prediction models. Practical application: The findings of this research could help select sound and reliable methodologies for predicting highway accidents by injury severity. When crash data samples have challenges associated with the low observed sampling rates for severe injury crashes, this research also confirmed that vehicle damage can be appropriate as an alternative to injury severity in crash prediction by severity.

Asynchronous dissipative filtering for nonhomogeneous Markov switching neural networks with variable packet dropouts.

This work focuses on the problem of asynchronous filtering for nonhomogeneous Markov switching neural networks with variable packet dropouts (VPDs). The discrete-time nonhomogeneous Markov process is adopted to depict the modes switching of target plant, where time-varying transition probabilities are revealed by utilizing a polytope technology. By means of the Bernoulli distributed sequence, the randomly occurring packet dropouts are presented, where VPD rates are mode-dependent and remain variable. Unlike the existing results, the hidden Markov model scheme is formulated to describe the asynchronization between nonhomogeneous neural networks and filter, and resilient filters are presented, which makes the designed filters more general. Eventually, a simulation example is established to verify the effectiveness of the developed filter scheme.

Modeling the effect of area deprivation on COVID-19 incidences: a study of Chennai megacity, India.

OBJECTIVES: Socio-economic inequalities may affect coronavirus disease 2019 (COVID-19) incidence. The goal of the research was to explore the association between deprivation of socio-economic status (SES) and spatial patterns of COVID-19 incidence in Chennai megacity for unfolding the disease epidemiology. STUDY DESIGN: This is an ecological (or contextual) study for electoral wards (subcities) of Chennai megacity. METHODS: Using data of confirmed COVID-19 cases from May 15, 2020, to May 21, 2020, for 155 electoral wards obtained from the official website of the Chennai Municipal Corporation, we examined the incidence of COVID-19 using two count regression models, namely, Poisson regression (PR) and negative binomial regression (NBR). As explanatory factors, we considered area deprivation that represented the deprivation of SES. An index of multiple deprivations (IMD) was developed to measure the area deprivation using an advanced local statistic, geographically weighted principal component analysis. Based on the availability of appropriately scaled data, five domains (i.e., poor housing condition, low asset possession, poor availability of WaSH services, lack of household amenities and services, and gender disparity) were selected as components of the IMD in this study. RESULTS: The hot spot analysis revealed that area deprivation was significantly associated with higher incidences of COVID-19 in Chennai megacity. The high variations (adjusted R2: 72.2%) with the lower Bayesian Information Criteria (BIC) (124.34) and Akaike's Information Criteria (AIC) (112.12) for NBR compared with PR suggests that the NBR model better explains the relationship between area deprivation and COVID-19 incidences in Chennai megacity. NBR with two-sided tests and P <0.05 were considered statistically significant. The outcome of the PR and NBR models suggests that when all other variables were constant, according to NBR, the relative risk (RR) of COVID-19 incidences was 2.19 for the wards with high housing deprivation or, in other words, the wards with high housing deprivation having 119% higher probability (RR = e0.786 = 2.19, 95% confidence interval [CI] = 1.98 to 2.40), compared with areas with low deprivation. Similarly, in the wards with poor availability of WaSH services, chances of having COVID-19 incidence was 90% higher than in the wards with good WaSH services (RR = e0.642 = 1.90, 95% CI = 1.79 to 2.00). Spatial risks of COVID-19 were predominantly concentrated in the wards with higher levels of area deprivation, which were mostly located in the northeastern parts of Chennai megacity. CONCLUSIONS: We formulated an area-based IMD, which was substantially related to COVID-19 incidences in Chennai megacity. This study highlights that the risks of COVID-19 tend to be higher in areas with low SES and that the northeastern part of Chennai megacity is predominantly high-risk areas. Our results can guide measures of COVID-19 control and prevention by considering spatial risks and area deprivation.

Chance-level hit rates in closed-set, forced-choice audiometry and a novel utility for the significance test-based detection of malingering.

The primary aim of this study was to extend existing theory on the relationship between chance-level performance and the number of alternatives and trials in closed-set, forced-choice speech audiometry and sound localization methods. When calculating chance performance for closed-set, forced-choice experiments with multiple trials, the binomial distribution should be preferred over the simple 1/a probability, as the latter is appropriate only for single trial experiments. The historical use of constant hit rates for determining chance performance has been based upon the assumption that random hits are distributed evenly across multiple trials. For any closed-set, forced-choice task with 2 to 10 alternatives and 2 to 100 trials, we calculated the probability of obtaining any given hit rate due to random guessing alone according to the binomial distribution. Hit rates with probabilities p > 0.05 were interpreted as being likely to occur due to random chance alone, whereas hit rates with probabilities of p ≤ 0.05 were interpreted as being unlikely to occur due to chance alone. For sound localization experiments with speakers at fixed positions, the expected probability of a random hit was also calculated using the binomial distribution. The expected angular root mean square (rms) error in sound localization resulting from the random selection of sound sources was investigated using Monte Carlo simulations. A new aspect in the interpretation of test results was identified for situations in which the observed number of hits is much lower than would be expected due to chance alone. For test methods incorporating a relatively low number of alternatives and a sufficiently high, yet clinically feasible, number of trials, both upper and lower thresholds for chance-level performance could be identified. This lower threshold represents the lowest hit rate which can be expected through random chance alone. Extending interpretation of results to include this lower threshold affords the ability to not only identify performance significantly superior to that of chance, but also that significantly poorer than chance and thereby represents a simple method for the objective detection of malingering.

An Analysis of State-Level Policies and Opioid Overdose Deaths.

Many legislative and regulatory changes have occurred at the state level in response to the opioid crisis in an attempt to decrease overdose deaths. We conducted a negative binomial, regression-based, interrupted time series analysis to evaluate the effects of policies on opioid overdose death counts for 2008-2017 in five Midwestern states: Illinois, Indiana, Kentucky, Michigan, and Ohio. Except for the Good Samaritan laws in Illinois, no single policy change was statistically significant in decreasing overdose deaths. Governmental involvement as a whole was significantly associated with an increase in overdose deaths in Indiana. Policies created to address opioid overdose mortality have had minimal impact in these five Midwestern states. Most of the legislation and regulation that have been created lack educational components for prescribers and community members, which may explain why these policies have not had the intended effect.

Combining individual and aggregated data to investigate the role of socioeconomic disparities on cancer burden in Italy.

Quantifying socioeconomic disparities and understanding the roots of inequalities are growing topics in cancer research. However, socioeconomic differences are challenging to investigate mainly due to the lack of accurate data at individual-level, while aggregate indicators are only partially informative. We implemented a multiple imputation algorithm within a statistical matching framework that combines diverse sources of data to estimate individual-level associations between income and risk of breast and lung cancer, adjusting for potential confounding factors in Italy. The framework is computationally flexible and can be adapted to similar contexts.

Effect of economic crisis on suicide cases: An ARDL bounds testing approach.

BACKGROUND: Suicide events observed in various groups, community or countries, especially in the periods of economic recession. It is thought that suicide cases increase when people's income decreases dramatically and they lose their jobs. AIM/OBJECTIVE: In this study, it was aimed to investigate whether the 2008 economic crisis had any effect on suicides in the United States. METHODS: Autoregressive distributed lag method was used. For the purpose of the study, the number of suicide-related deaths was taken as the dependent variable, while unemployment rates and 2008 economic crisis were taken as independent variables. FINDINGS: The short-term and long-term relationships obtained within the scope of the study indicated that the 2008 economic crisis had a statistically significant effect on suicide cases in the United States. RESULTS AND CONCLUSION: It can be said that the results of this study are consistent with the information which emphasizes that economic crises increase suicide cases in the literature.

Energy-efficient data routing in cooperative UAV swarms for medical assistance after a disaster.

Medical assistance is crucial to disaster management. In particular, the situation of survivors as well as the environmental information after disasters should be collected and sent back to cloud/data centers immediately for further interpretation and analysis. Recently, unmanned aerial vehicle (UAV)-aided disaster management has been considered a promising approach to enhance the efficiency of searching and rescuing survivors after a disaster, in which a group of UAVs collaborates to accomplish the search and rescue task. However, the battery capacity of UAVs is a critical shortcoming that limits their usage. Worse still, the unstable network connectivity of disaster sites could lead to high latency of data transmission from UAV to remote data centers, which could make significant challenges on real-time data collecting and processing. To solve the above problems, in this paper, we investigate an energy-efficient multihop data routing algorithm with the guarantee of quality-of-service for UAV-aided medical assistance. Specifically, we first study the data routing problem to minimize the energy consumption considering transmission rate, time delay, and life cycle of the UAV swarms. Then, we formulate the issue as a mixed-integer nonlinear programming model. Because of the Non-deterministic Polynomial-hardness of this problem, we propose a polynomial time algorithm based on a genetic algorithm to solve the problem. To achieve high efficiency, we further enhance our algorithm based on DBSCAN and adaptive techniques. Extensive experiments show that our approach can outperform the state-of-the-art methods.

Ensemble confidence intervals for binomial proportions.

We propose two measures of performance for a confidence interval for a binomial proportion p: the root mean squared error and the mean absolute deviation. We also devise a confidence interval for p based on the actual coverage function that combines several existing approximate confidence intervals. This "Ensemble" confidence interval has improved statistical properties over the constituent confidence intervals. Software in an R package, which can be used in devising and assessing these confidence intervals, is available on CRAN.

A Joint Poisson State-Space Modelling Approach to Analysis of Binomial Series with Random Cluster Sizes.

Serially correlation binomial data with random cluster sizes occur frequently in environmental and health studies. Such data series have traditionally been analyzed using binomial state-space or hidden Markov models without appropriately accounting for the randomness in the cluster sizes. To characterize correlation and extra-variation arising from the random cluster sizes properly, we introduce a joint Poisson state-space modelling approach to analysis of binomial series with random cluster sizes. This approach enables us to model the marginal counts and binomial proportions simultaneously. An optimal estimation of our model has been developed using the orthodox best linear unbiased predictors. This estimation method is computationally efficient and robust since it depends only on the first- and second- moment assumptions of unobserved random effects. Our proposed approach is illustrated with analysis of birth delivery data.

Migraine day frequency in migraine prevention: longitudinal modelling approaches.

BACKGROUND: Health economic models are critical tools to inform reimbursement agencies on health care interventions. Many clinical trials report outcomes using the frequency of an event over a set period of time, for example, the primary efficacy outcome in most clinical trials of migraine prevention is mean change in the frequency of migraine days (MDs) per 28 days (monthly MDs [MMD]) relative to baseline for active treatment versus placebo. Using these cohort-level endpoints in economic models, accounting for variation among patients is challenging. In this analysis, parametric models of change in MMD for migraine preventives were assessed using data from erenumab clinical studies. METHODS: MMD observations from the double-blind phases of two studies of erenumab were used: one in episodic migraine (EM) (NCT02456740) and one in chronic migraine (CM) (NCT02066415). For each trial, two longitudinal regression models were fitted: negative binomial and beta binomial. For a thorough comparison we also present the fitting from the standard multilevel Poisson and the zero inflated negative binomial. RESULTS: Using the erenumab study data, both the negative binomial and beta-binomial models provided unbiased estimates relative to observed trial data with well-fitting distribution at various time points. CONCLUSIONS: This proposed methodology, which has not been previously applied in migraine, has shown that these models may be suitable for estimating MMD frequency. Modelling MMD using negative binomial and beta-binomial distributions can be advantageous because these models can capture intra- and inter-patient variability so that trial observations can be modelled parametrically for the purposes of economic evaluation of migraine prevention. Such models have implications for use in a wide range of disease areas when assessing repeated measured utility values.

Economic crises and suicides between 1970 and 2011: time trend study in 21 developed countries.

BACKGROUND: Existing research on the relationship between economic recessions and suicides has almost completely concentrated on the most recent global financial crisis (2008). We provide the most comprehensive explanation to date of how different types of economic/financial crises since 1970 have affected suicides in developed countries. METHODS: Negative binomial regressions were used to estimate what the suicide rates would have been during and 1 year after each crisis began in 21 Organisation for Economic Co-operation and Development countries from 1970 to 2011 if the suicide rates had followed the pre-crisis trends. RESULTS: We found that every economic/financial crisis since 1970, except the European Exchange Rate Mechanism crisis in 1992, led to excess suicides in developed countries. Among males, the excess suicide rate (per 100 000 persons) varied from 1.1 (95% CI 0.7 to 1.5) to 9.5 (7.6 to 11.2) and, among females, from 0 to 2.4 (1.9 to 2.9). For both sexes, suicides increased mostly due to stock market crashes and banking crises. In terms of actual numbers, the post-1969 economic/financial crises caused >60 000 excess suicides in the 21 developed countries. The Asian financial crisis in 1997 was the most damaging crisis when assessed based on excess suicides. CONCLUSIONS: Evidence indicates that, when considered in terms of effects on suicide mortality, the most recent global financial crisis is not particularly severe compared with previous global economic/financial crises. The distinct types of crises (ie, banking, currency and inflation crises, and stock market crashes) have different effects on suicide.

Assessment of Susceptibility Risk Factors for ADHD in Imaging Genetic Studies.

OBJECTIVE: ADHD consists of a count of symptoms that often presents heterogeneity due to overdispersion and excess of zeros. Statistical inference is usually based on a dichotomous outcome that is underpowered. The main goal of this study was to determine a suited probability distribution to analyze ADHD symptoms in Imaging Genetic studies. METHOD: We used two independent population samples of children to evaluate the consistency of the standard probability distributions based on count data for describing ADHD symptoms. RESULTS: We showed that the zero-inflated negative binomial (ZINB) distribution provided the best power for modeling ADHD symptoms. ZINB reveals a genetic variant, rs273342 (Microtubule-Associated Protein [MAPRE2]), associated with ADHD ( p value = 2.73E-05). This variant was also associated with perivascular volumes (Virchow-Robin spaces; p values < 1E-03). No associations were found when using dichotomous definition. CONCLUSION: We suggest that an appropriate modeling of ADHD symptoms increases statistical power to establish significant risk factors.

Comparative assessment of different familial aggregation methods in the context of large and unstructured pedigrees.

Motivation: Familial aggregation analysis is an important early step for characterizing the genetic determinants of phenotypes in epidemiological studies. To facilitate this analysis, a collection of methods to detect familial aggregation in large pedigrees has been made available recently. However, efficacy of these methods in real world scenarios remains largely unknown. Here, we assess the performance of five aggregation methods to identify individuals or groups of related individuals affected by a Mendelian trait within a large set of decoys. We investigate method performance under a representative set of combinations of causal variant penetrance, trait prevalence and number of affected generations in the pedigree. These methods are then applied to assess familial aggregation of familial hypercholesterolemia and stroke, in the context of the Cooperative Health Research in South Tyrol (CHRIS) study. Results: We find that in some situations statistical hypothesis testing with a binomial null distribution achieves performance similar to methods that are based on kinship information, while kinship based methods perform better when information is available on fewer generations. Potential case families from the CHRIS study are reported and the results are discussed taking into account insights from the performance assessment. Availability and implementation: The familial aggregation analysis package is freely available at the Bioconductor repository, http://www.bioconductor.org/packages/FamAgg. Supplementary information: Supplementary data are available at Bioinformatics online.

Incorporation of frailties into a cure rate regression model and its diagnostics and application to melanoma data.

Cure rate models have been widely studied to analyze time-to-event data with a cured fraction of patients. Our proposal consists of incorporating frailty into a cure rate model, as an alternative to the existing models to describe this type of data, based on the Birnbaum-Saunders distribution. Such a distribution has theoretical arguments to model medical data and has shown empirically to be a good option for their analysis. An advantage of the proposed model is the possibility to jointly consider the heterogeneity among patients by their frailties and the presence of a cured fraction of them. In addition, the number of competing causes is described by the negative binomial distribution, which absorbs several particular cases. We consider likelihood-based methods to estimate the model parameters and to derive influence diagnostics for this model. We assess local influence on the parameter estimates under different perturbation schemes. Deriving diagnostic tools is needed in all statistical modeling, which is another novel aspect of our proposal. Numerical evaluation of the considered model is performed by Monte Carlo simulations and by an illustration with melanoma data, both of which show its good performance and its potential applications. Particularly, the illustration confirms the importance of statistical diagnostics in the modeling.

Some properties of the conditioned reconstructed process with Bernoulli sampling.

In many areas of genetics it is of relevance to consider a population of individuals that is founded by a single individual in the past. One model for such a scenario is the conditioned reconstructed process with Bernoulli sampling that describes the evolution of a population of individuals that originates from a single individual. Several aspects of this reconstructed process are studied, in particular the Markov structure of the process. It is shown that at any given time in the past, the conditioned reconstructed process behaves as the original conditioned reconstructed process after a suitable time-dependent change of the sampling probability. Additionally, it is discussed how mutations accumulate in a sample of particles. It is shown that random sampling of particles at the present time has the effect of making the mutation rate look time-dependent. Conditions are given under which this sampling effect is negligible. A possible extension of the reconstructed process that allows for multiple founding particles is discussed.

Estimating negative likelihood ratio confidence when test sensitivity is 100%: A bootstrapping approach.

Objectives Assessing high-sensitivity tests for mortal illness is crucial in emergency and critical care medicine. Estimating the 95% confidence interval (CI) of the likelihood ratio (LR) can be challenging when sample sensitivity is 100%. We aimed to develop, compare, and automate a bootstrapping method to estimate the negative LR CI when sample sensitivity is 100%. Methods The lowest population sensitivity that is most likely to yield sample sensitivity 100% is located using the binomial distribution. Random binomial samples generated using this population sensitivity are then used in the LR bootstrap. A free R program, "bootLR," automates the process. Extensive simulations were performed to determine how often the LR bootstrap and comparator method 95% CIs cover the true population negative LR value. Finally, the 95% CI was compared for theoretical sample sizes and sensitivities approaching and including 100% using: (1) a technique of individual extremes, (2) SAS software based on the technique of Gart and Nam, (3) the Score CI (as implemented in the StatXact, SAS, and R PropCI package), and (4) the bootstrapping technique. Results The bootstrapping approach demonstrates appropriate coverage of the nominal 95% CI over a spectrum of populations and sample sizes. Considering a study of sample size 200 with 100 patients with disease, and specificity 60%, the lowest population sensitivity with median sample sensitivity 100% is 99.31%. When all 100 patients with disease test positive, the negative LR 95% CIs are: individual extremes technique (0,0.073), StatXact (0,0.064), SAS Score method (0,0.057), R PropCI (0,0.062), and bootstrap (0,0.048). Similar trends were observed for other sample sizes. Conclusions When study samples demonstrate 100% sensitivity, available methods may yield inappropriately wide negative LR CIs. An alternative bootstrapping approach and accompanying free open-source R package were developed to yield realistic estimates easily. This methodology and implementation are applicable to other binomial proportions with homogeneous responses.

Residential Neighborhood Amenities and Physical Activity Among U.S. Children with Special Health Care Needs.

Objectives Children with special health care needs (SHCN) have or are at elevated risk for chronic physical, developmental, behavioral or emotional conditions and therefore require specialized health care services. This study examines the relationship between residential neighborhood amenities and physical activity among U.S. children with SHCN. Methods A nationally representative sample of 113,767 children aged 6-17 years was taken from National Survey of Children's Health (NSCH) 2007-2008 and 2011-2012. Residential neighborhood amenities were defined by parent-reported presence or absence of sidewalks, parks/playgrounds, and recreation center. Physical activity was measured by parent-reported number of physically active days (0-7), defined as 20 min or longer during the past week. Negative binomial regressions were performed to estimate the associations between residential neighborhood amenities and physical activity among U.S. children with and without SHCN, adjusting for various neighborhood (detracting condition and safety) and individual characteristics and NSCH sampling design. Results Approximately 23% of young children aged 6-11 years and adolescents aged 12-17 years had SHCN. Number of weekly physically active days was both 4.8 among young children with and without SHCN, and 3.7 and 4.1 among adolescents with and without SHCN, respectively. Among young children with SHCN, neighborhood availability of a recreation center was associated with increased weekly physically active days by 0.23; whereas among young children without SHCN, number of weekly physically active days was not associated with the availability of any amenity. Among adolescents with SHCN, neighborhood availability of parks/playgrounds was associated with increased weekly physically active days by 0.33, whereas neighborhood availability of sidewalks was associated with reduced weekly physically active days by 0.21. Conversely, among adolescents without SHCN, neighborhood availability of a recreation center was associated with increased weekly physically active days by 0.22. Conclusions for Practice Vulnerable health status and high dependence on health care may prevent children with SHCN from being physically active. Provision of adequate amenities in residential neighborhoods could be essential in promoting physical activity and preventing obesity among children/adolescents with SHCN.

Variation in type and frequency of diagnostic imaging during trauma care across multiple time points by patient insurance type.

BACKGROUND: Research has shown that uninsured patients receive fewer radiographic studies during trauma care, but less is known as to whether differences in care are present among other insurance groups or across different time points during hospitalization. Our objective was to examine the number of radiographic studies administered to a cohort of trauma patients over the entire hospital stay as well as during the first 24-hours of care. METHODS: Patient data were obtained from an American College of Surgeons (ACS) verified Level I Trauma Center between January 1, 2011 and December 31, 2012. We used negative binomial regression to construct relative risk (RR) ratios for type and frequency of radiographic imaging received among persons with Medicare, Medicaid, no insurance, or government insurance plans in reference to those with commercial indemnity plans. The analysis was adjusted for patient age, sex, race/ethnicity, injury severity score, injury mechanism, comorbidities, complications, hospital length of stay, and Intensive Care Unit (ICU) admission. RESULTS: A total of 3621 records from surviving patients age > =18 years were assessed. After adjustment for potential confounders, the expected number of radiographic studies decreased by 15 % among Medicare recipients (RR 0.85, 95 % CI 0.78-0.93), 11 % among Medicaid recipients (0.89, 0.81-0.99), 10 % among the uninsured (0.90, 0.85-0.96) and 19 % among government insurance groups (0.81, 0.72-0.90), compared with the reference group. This disparity was observed during the first 24-hours of care among patients with Medicare (0.78, 0.71-0.86) and government insurance plans (0.83, 0.74-0.94). Overall, there were no differences in the number of radiographic studies among the uninsured or among Medicaid patients during the first 24-hours of care compared with the reference group, but differences were observed among the uninsured in a sub-analysis of severely injured patients (ISS > 15). CONCLUSIONS: Both uninsured and insured patients treated at a not-for-profit verified Level I Trauma Center receive fewer radiographic studies than patients with commercial indemnity plans, even after adjusting for clinical and demographic confounders. There is less disparity in care during the first 24-hours, which suggests that patient pathology is the determining factor for radiographic evaluation during the acute care phase. Results from this study offer initial evidence of disparity in diagnostic imaging across multiple insurance groups over different periods of trauma care.