Negative Binomial Mixed Models Research Articles

694 Insufficient sleep linked with higher COVID-19 infection cases and deaths in the United States

IntroductionCauses of COVID-19 burden in urban, suburban, and rural counties are unclear, as early studies provide mixed results implicating high prevalence of pre-existing health risks and chronic diseases. However, poor sleep health that has been linked to infection-based pandemics may provide additional insight for place-based burden. To address this gap, we investigated the relationship between habitual insufficient sleep (sleep <7 hrs./24 hr. period) and COVID-19 cases and deaths across urban, suburban, and rural counties in the US.MethodsCounty-level variables were obtained from the 2014–2018 American community survey five-year estimates and the Center for Disease Control and Prevention. These included percent with insufficient sleep, percent uninsured, percent obese, and social vulnerability index. County level COVID-19 infection and death data through September 12, 2020 were obtained from USA Facts. Cumulative COVID-19 infections and deaths for urban (n=68), suburban (n=740), and rural (n=2331) counties were modeled using separate negative binomial mixed effects regression models with logarithmic link and random state-level intercepts. Zero-inflated models were considered for deaths among suburban and rural counties to account for excess zeros.ResultsMultivariate regression models indicated positive associations between cumulative COVID-19 infection rates and insufficient sleep in urban, suburban and rural counties. The incidence rate ratio (IRR) for urban counties was 1.03 (95% CI: 1.01 – 1.05), 1.04 (95% CI: 1.02 – 1.05) for suburban, and 1.02 (95% CI: 1.00 – 1.03) rural counties.. Similar positive associations were observed with county-level COVID-19 death rates, IRR = 1.11 (95% CI: 1.07 – 1.16) for urban counties, IRR = 1.04 (95% CI: 1.01 – 1.06) for suburban counties, and IRR = 1.03 (95% CI: 1.01 – 1.05) for rural counties. Level of urbanicity moderated the association between insufficient sleep and COVID deaths, but not for the association between insufficient sleep and COVID infection rates.ConclusionInsufficient sleep was associated with COVID-19 infection cases and mortality rates in urban, suburban and rural counties. Level of urbanicity only moderated the relationship between insufficient sleep and COVID death rates. Future studies should investigate individual-level analysis to understand the role of sleep mitigating COVID-19 infection and death rates.Support (if any)NIH (K07AG052685, R01MD007716, R01HL142066, K01HL135452, R01HL152453

Air pollution, methane super-emitters, and oil and gas wells in Northern California: the relationship with migraine headache prevalence and exacerbation

BackgroundMigraine–an episodic disorder characterized by severe headache that can lead to disability–affects over 1 billion people worldwide. Prior studies have found that short-term exposure to fine particulate matter (PM2.5), nitrogen dioxide (NO2), and ozone increases risk of migraine-related emergency department (ED) visits. Our objective was to characterize the association between long-term exposure to sources of harmful emissions and common air pollutants with both migraine headache and, among patients with migraine, headache severity.MethodsFrom the Sutter Health electronic health record database, we identified 89,575 prevalent migraine cases between 2014 and 2018 using a migraine probability algorithm (MPA) score and 270,564 frequency-matched controls. Sutter Health delivers care to 3.5 million patients annually in Northern California. Exposures included 2015 annual average block group-level PM2.5 and NO2 concentrations, inverse-distance weighted (IDW) methane emissions from 60 super-emitters located within 10 km of participant residence between 2016 and 2018, and IDW active oil and gas wells in 2015 within 10 km of each participant. We used logistic and negative binomial mixed models to evaluate the association between environmental exposures and (1) migraine case status; and (2) migraine severity (i.e., MPA score > 100, triptan prescriptions, neurology visits, urgent care migraine visits, and ED migraine visits per person-year). Models controlled for age, sex, race/ethnicity, Medicaid use, primary care visits, and block group-level population density and poverty.ResultsIn adjusted analyses, for each 5 ppb increase in NO2, we observed 2% increased odds of migraine case status (95% CI: 1.00, 1.05) and for each 100,000 kg/hour increase in IDW methane emissions, the odds of case status also increased (OR = 1.04, 95% CI: 1.00, 1.08). We found no association between PM2.5 or oil and gas wells and migraine case status. PM2.5 was linearly associated with neurology visits, migraine-specific urgent care visits, and MPA score > 100, but not triptans or ED visits. NO2 was associated with migraine-specific urgent care and ED visits, but not other severity measures. We observed limited or null associations between continuous measures of methane emissions and proximity to oil and gas wells and migraine severity.ConclusionsOur findings illustrate the potential role of long-term exposure to multiple ambient air pollutants for prevalent migraine and migraine severity.

A global association between Covid-19 cases and airborne particulate matter at regional level

Evidences of an association between air pollution and Covid-19 infections are mixed and inconclusive. We conducted an ecological analysis at regional scale of long-term exposure to air-borne particle matter and spread of Covid-19 cases during the first wave of epidemics. Global air pollution and climate data were calculated from satellite earth observation data assimilated into numerical models at 10 km resolution. Main outcome was defined as the cumulative number of cases of Covid-19 in the 14 days following the date when > 10 cumulative cases were reported. Negative binomial mixed effect models were applied to estimate the associations between the outcome and long-term exposure to air pollution at the regional level (PM10, PM2.5), after adjusting for relevant regional and country level covariates and spatial correlation. In total we collected 237,749 Covid-19 cases from 730 regions, 63 countries and 5 continents at May 30, 2020. A 10 μg/m3 increase of pollution level was associated with 8.1% (95% CI 5.4%, 10.5%) and 11.5% (95% CI 7.8%, 14.9%) increases in the number of cases in a 14 days window, for PM2.5 and PM10 respectively. We found an association between Covid-19 cases and air pollution suggestive of a possible causal link among particulate matter levels and incidence of COVID-19.

Teachers' work-related non-literature-known building-related symptoms are also connected to indoor toxicity: A cross-sectional study.

A previous study showed that classical building-related symptoms (BRS) were related to indoor dust and microbial toxicity via boar sperm motility assay, a sensitive method for measuring mitochondrial toxicity. In this cross-sectional study, we analyzed whether teachers' most common work-related non-literature-known BRS (nBRS) were also associated with dust or microbial toxicity. Teachers from 15 schools in Finland completed a questionnaire evaluating 20 nBRS including general, eye, respiratory, hearing, sleep, and mental symptoms. Boar sperm motility assay was used to measure the toxicity of extracts from wiped dust and microbial fallout samples collected from teachers' classrooms. 231 teachers answered a questionnaire and their classroom toxicity data were recorded. A negative binomial mixed model showed that teachers' work-related nBRS were 2.9-fold (95% CI: 1.2-7.3) higher in classrooms with highly toxic dust samples compared to classrooms with non-toxic dust samples (p=0.024). The RR of work-related nBRS was 1.8 (95% CI: 1.1-2.9) for toxic microbial samples (p=0.022). Teachers' BRS appeared to be broader than reported in the literature, and the work-related nBRS were associated with toxic dusts and microbes in classrooms.

Changes in Physical Activity among United Kingdom University Students Following the Implementation of Coronavirus Lockdown Measures.

Coronavirus disease (COVID-19) and resulting restrictions have significantly impacted physical activity levels. However, objectively measured changes in physical activity levels among UK university students during lockdown are understudied. Using data collected via remote measurement technology from a mobile physical activity tracker, this study aimed to describe the longitudinal trajectories of physical activity following the start of lockdown among students at a large UK university, and to investigate whether these trajectories varied according to age, gender, and ethnicity. Continuous physical activity data for steps walked per week (n = 730) and miles run per week (n = 264) were analysed over the first period of lockdown and subsequent restriction easing using negative binomial mixed models for repeated measures. Throughout the observation period, more steps were walked by males compared to females, and by White groups compared to all other ethnic groups combined. However, there was a gradual increase in the number of steps walked per week following the commencement of lockdown, irrespective of sociodemographic characteristics. For females only, there was a decrease in the number of miles run per week following lockdown. The long-term impact of the pandemic on physical health is unknown, but our results highlight changes in physical activity which could have implications for physical health.

CD38 in Advanced Prostate Cancers

BackgroundCD38, a druggable ectoenzyme, is involved in the generation of adenosine, which is implicated in tumour immune evasion. Its expression and role in prostate tumour-infiltrating immune cells (TIICs) have not been elucidated. ObjectiveTo characterise CD38 expression on prostate cancer (PC) epithelial cells and TIICs, and to associate this expression with clinical outcomes. Design, setting, and participantsRNAseq from 159 patients with metastatic castration-resistant prostate cancer (mCRPC) in the International Stand Up To Cancer/Prostate Cancer Foundation (SU2C/PCF) cohort and 171 mCRPC samples taken from 63 patients in the Fred Hutchinson Cancer Research Centre cohort were analysed. CD38 expression was immunohistochemically scored by a validated assay on 51 castration-resistant PC (CRPC) and matching, same-patient castration-sensitive PC (CSPC) biopsies obtained between 2016 and 2018, and was associated with retrospectively collected clinical data. Outcome measurements and statistical analysismCRPC transcriptomes were analysed for associations between CD38 expression and gene expression signatures. Multiplex immunofluorescence determined CD38 expression in PC biopsies. Differences in CD38+ TIIC densities between CSPC and CRPC biopsies were analysed using a negative binomial mixed model. Differences in the proportions of CD38+ epithelial cells between non-matched benign prostatic epithelium and PC were compared using Fisher’s exact test. Differences in the proportions of biopsies containing CD38+ tumour epithelial cells between matched CSPC and CRPC biopsies were compared by McNemar’s test. Univariable and multivariable survival analyses were performed using Cox regression models. Results and limitationsCD38 mRNA expression in mCRPC was most significantly associated with upregulated immune signalling pathways. CD38 mRNA expression was associated with interleukin (IL)-12, IL-23, and IL-27 signalling signatures as well as immunosuppressive adenosine signalling and T cell exhaustion signatures. CD38 protein was frequently expressed on phenotypically diverse TIICs including B cells and myeloid cells, but largely absent from tumour epithelial cells. CD38+ TIIC density increased with progression to CRPC and was independently associated with worse overall survival. Future studies are required to dissect TIIC CD38 function. ConclusionsCD38+ prostate TIICs associate with worse survival and immunosuppressive mechanisms. The role of CD38 in PC progression warrants investigation as insights into its functions may provide rationale for CD38 targeting in lethal PC. Patient summaryCD38 is expressed on the surface of white blood cells surrounding PC cells. These cells may impact PC growth and treatment resistance. Patients with PC with more CD38-expressing white blood cells are more likely to die earlier.

Linking the oral microbiome and salivary cytokine abundance to circadian oscillations

Saliva has immense potential as a diagnostic fluid for identification and monitoring of several systemic diseases. Composition of the microbiome and inflammation has been associated and reflective of oral and overall health. In addition, the relative ease of collection of saliva further strengthens large-scale diagnostic purposes. However, the future clinical utility of saliva cannot be fully determined without a detailed examination of daily fluctuations that may occur within the oral microbiome and inflammation due to circadian rhythm. In this study, we explored the association between the salivary microbiome and the concentration of IL-1β, IL-6 and IL-8 in the saliva of 12 healthy adults over a period of 24 h by studying the 16S rRNA gene followed by negative binomial mixed model regression analysis. To determine the periodicity and oscillation patterns of both the oral microbiome and inflammation (represented by the cytokine levels), two of the twelve subjects were studied for three consecutive days. Our results indicate that the Operational Taxonomic Units (OTUs) belonging to Prevotella, SR1 and Ruminococcaceae are significantly associated to IL-1β while Prevotella and Granulicatella were associated with IL-8. Our findings have also revealed a periodicity of both the oral microbiome (OTUs) and inflammation (cytokine levels) with identifiable patterns between IL-1β and Prevotella, and IL-6 with Prevotella, Neisseria and Porphyromonas. We believe that this study represents the first measure and demonstration of simultaneous periodic fluctuations of cytokine levels and specific populations of the oral microbiome.

Titi monkey neophobia and visual abilities allow for fast responses to novel stimuli

The Snake Detection Theory implicates constricting snakes in the origin of primates, and venomous snakes for differences between catarrhine and platyrrhine primate visual systems. Although many studies using different methods have found very rapid snake detection in catarrhines, including humans, to date no studies have examined how quickly platyrrhine primates can detect snakes. We therefore tested in captive coppery titi monkeys (Plecturocebus cupreus) the latency to detect a small portion of visible snake skin. Because titi monkeys are neophobic, we designed a crossover experiment to compare their latency to look and their duration of looking at a snake skin and synthetic feather of two lengths (2.5 cm and uncovered). To test our predictions that the latency to look would be shorter and the duration of looking would be longer for the snake skin, we used survival/event time models for latency to look and negative binomial mixed models for duration of looking. While titi monkeys looked more quickly and for longer at both the snake skin and feather compared to a control, they also looked more quickly and for longer at larger compared to smaller stimuli. This suggests titi monkeys’ neophobia may augment their visual abilities to help them avoid dangerous stimuli.

COVID-19 incidence and mortality in Lombardy, Italy: An ecological study on the role of air pollution, meteorological factors, demographic and socioeconomic variables

Lombardy, the most populated and industrialized Italian region, was the epicentre of the first wave (March and April 2020) of COVID-19 in Italy and it is among the most air polluted areas of Europe. We carried out an ecological study to assess the association between long-term exposure to particulate matter (PM) and nitrogen dioxide (NO2) on COVID-19 incidence and all-cause mortality after accounting for demographic, socioeconomic and meteorological variables.The study was based on publicly available data. Multivariable negative binomial mixed regression models were fitted, and results were reported in terms of incidence rate ratios (IRRs) and standardized mortality ratios (SMR). The effect of winter temperature and humidity was modelled through restricted cubic spline.Data from 1439 municipalities out of 1507 (95%) were included in the analyses, leading to a total of 61,377 COVID-19 cases and 40,401 deaths from all-causes collected from February 20th to April 16th and from March 1st to April 30th, 2020, respectively. Several demographic and socioeconomic variables resulted significantly associated with COVID-19 incidence and all-cause mortality in a multivariable fashion. An increase in average winter temperature was associated with a nonlinear decrease in COVID-19 incidence and all-cause mortality, while an opposite trend emerged for the absolute humidity. An increase of 10 μg/m3 in the mean annual concentrations of PM2.5 and PM10 over the previous years was associated with a 58% and 34% increase in COVID-19 incidence rate, respectively. Similarly, a 10 μg/m3 increase of annual mean PM2.5 concentration was associated with a 23% increase in all-cause mortality. An inverse association was found between NO2 levels and COVID-19 incidence and all-cause mortality.Our ecological study showed that exposure to PM was significantly associated with the COVID-19 incidence and excess mortality during the first wave of the outbreak in Lombardy, Italy.

Near-term climate change impacts on sub-national malaria transmission

The role of climate change on global malaria is often highlighted in World Health Organisation reports. We modelled a Zambian socio-environmental dataset from 2000 to 2016, against malaria trends and investigated the relationship of near-term environmental change with malaria incidence using Bayesian spatio-temporal, and negative binomial mixed regression models. We introduced the diurnal temperature range (DTR) as an alternative environmental measure to the widely used mean temperature. We found substantial sub-national near-term variations and significant associations with malaria incidence-trends. Significant spatio-temporal shifts in DTR/environmental predictors influenced malaria incidence-rates, even in areas with declining trends. We highlight the impact of seasonally sensitive DTR, especially in the first two quarters of the year and demonstrate how substantial investment in intervention programmes is negatively impacted by near-term climate change, most notably since 2010. We argue for targeted seasonally-sensitive malaria chemoprevention programmes.

Effect of Ivermectin Treatment on the Frequency of Seizures in Persons with Epilepsy Infected with Onchocerca volvulus.

A clinical trial performed in the Democratic Republic of Congo (DRC), among persons with epilepsy (PWE) infected with Onchocerca volvulus treated with anti-seizure medication suggested that ivermectin reduces the seizure frequency. We assessed the effect of ivermectin treatment on seizure frequency in PWE with and without anti-seizure medication in three onchocerciasis endemic areas (Maridi, South Sudan; Aketi, DRC; and Mahenge, Tanzania). Pre- and 3–5 months post-ivermectin microfilariae densities in skin snips and seizure frequency were assessed. After ivermectin, the median (IQR) percentage reduction in seizure frequency in the study sites ranged from 73.4% (26.0–90.0) to 100% (50.0–100.0). A negative binomial mixed model showed that ivermectin significantly reduced the seizure frequency, with a larger decrease in PWE with a high baseline seizure frequency. Mediation analysis showed that ivermectin reduced the seizure frequencies indirectly through reduction in microfilariae densities but also that ivermectin may have a direct anti-seizure effect. However, given the short half-life of ivermectin and the fact that ivermectin does not penetrate the healthy brain, such a direct anti-seizure effect is unlikely. A randomized controlled trial assessing the ivermectin effect in people infected with O. volvulus who are also PWE on a stable anti-seizure regimen may be needed to clarify the causal relationship between ivermectin and seizure frequency.

A joint modeling approach for longitudinal microbiome data improves ability to detect microbiome associations with disease.

Changes in the composition of the microbiome over time are associated with myriad human illnesses. Unfortunately, the lack of analytic techniques has hindered researchers' ability to quantify the association between longitudinal microbial composition and time-to-event outcomes. Prior methodological work developed the joint model for longitudinal and time-to-event data to incorporate time-dependent biomarker covariates into the hazard regression approach to disease outcomes. The original implementation of this joint modeling approach employed a linear mixed effects model to represent the time-dependent covariates. However, when the distribution of the time-dependent covariate is non-Gaussian, as is the case with microbial abundances, researchers require different statistical methodology. We present a joint modeling framework that uses a negative binomial mixed effects model to determine longitudinal taxon abundances. We incorporate these modeled microbial abundances into a hazard function with a parameterization that not only accounts for the proportional nature of microbiome data, but also generates biologically interpretable results. Herein we demonstrate the performance improvements of our approach over existing alternatives via simulation as well as a previously published longitudinal dataset studying the microbiome during pregnancy. The results demonstrate that our joint modeling framework for longitudinal microbiome count data provides a powerful methodology to uncover associations between changes in microbial abundances over time and the onset of disease. This method offers the potential to equip researchers with a deeper understanding of the associations between longitudinal microbial composition changes and disease outcomes. This new approach could potentially lead to new diagnostic biomarkers or inform clinical interventions to help prevent or treat disease.

PSIV-17 Muscular Tension as a Marker of Stress in Horses

Abstract Effects of social isolation, sham clipping, and novel object exposure on muscular and behavioral responses were investigated in horses. Humans increase muscular tension under psychologically or physically demanding situations and thus, similar conditions were expected to alter stress responses in horses. Eight mature Standardbreds (4 mares, 4 geldings) were exposed to 3 min of social isolation (ISO), sham clipping (CLIP), novel object (NOV), and control (CON) conditions in a replicated 4×4 Latin Square experimental design with 10 min washout periods. Surface electromyography (EMG) and stress-related behaviors were recorded continuously. Median frequency (MF) and average rectified value (ARV) of the EMG signal were calculated for the first, middle, and final 10 sec of each period. EMG data were log transformed prior to analysis with a mixed model, repeated measures ANOVA. Behavior data were analyzed using a negative binomial distribution mixed model ANOVA. Significantly different means were separated using Tukey’s method. More stress-related behaviors (P &amp;lt; 0.05) were observed during ISO (3.25 ± 0.26, LSM ± SE) than CON (1.46 ± 0.29) or CLIP (1.50 ± 0.36). ISO tended (P = 0.054) to produce more stress-related behaviors than NOV (2.31 ± 0.28). CLIP and ISO produced higher (P &amp;lt; 0.05) ARV than CON or NOV in the left and right masseter. CLIP elicited the highest (P &amp;lt; 0.05) MF in the left and right masseter with ISO resulting in lower (P &amp;lt; 0.05) MF than CON. ARV was higher (P &amp;lt; 0.05) in the left cervical trapezius during ISO as compared to all other conditions and in the right as compared to CON and CLIP. In the right cervical trapezius, ISO elicited higher (P &amp;lt; 0.05) MF than CON. Increased stress-related behaviors indicate a stress response was induced. Elevated muscular activity suggests muscular tension can be used in assessing stress.

20 Vaginal microbiome composition is associated with sow longevity

Abstract The vaginal microbiome of gilts vaccinated for porcine reproductive and respiratory syndrome (PRRS) has been previously associated with first-parity reproductive performance in absence of PRRS outbreak. However, associations using multiparous sows have not been investigated. The objective of this study was to associate the vaginal microbiota of PRRS-vaccinated gilts with longevity. Vaginal swabs from 251 commercial F1 gilts (Landrace/Large-White) were collected on days 4 and 52 after PRRS vaccination (dpv) for 16S rRNA sequencing. Sequences were clustered into operational taxonomic unit (OTU). Sows were assigned to one of four groups according to the maximum parity reached (number of animals): without parity (13), first-parity (45), second-parity (47), and 3 or more parities (146). A negative binomial mixed model including fixed effects of group, dpv, group*dpv, and collection age (covariate), and random effect of animal, was used to identify OTUs with differential abundance. The false discovery rate method was used for multiple test correction. Canonical discriminant analysis (CDA) was performed to classify animals into the four parity groups by including significant OTUs in stepwise selection (P &amp;lt; 0.05) using the whole data. A leave-one-out-cross-validation was used to assess the predictive ability of OTUs to correctly classify animals into the parity groups. Abundance of five and four OTUs was associated (q &amp;lt; 0.05) with group and with group*dpv (q &amp;lt; 0.05), respectively. In the CDA analysis, 220 OTUs identified in the vaginal microbiome were included (P &amp;lt; 0.05). The first and second canonical variables explained 96.6 and 3.4% of the variation, respectively. All animals were correctly classified into their respective parity groups. This study indicates that the vaginal microbiome composition of gilts collected after PRRS-vaccination may be used to predict longevity in commercial sow herds. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brazil (Capes) – Finance Code 001.

Zero-Inflated gaussian mixed models for analyzing longitudinal microbiome data

MotivationThe human microbiome is variable and dynamic in nature. Longitudinal studies could explain the mechanisms in maintaining the microbiome in health or causing dysbiosis in disease. However, it remains challenging to properly analyze the longitudinal microbiome data from either 16S rRNA or metagenome shotgun sequencing studies, output as proportions or counts. Most microbiome data are sparse, requiring statistical models to handle zero-inflation. Moreover, longitudinal design induces correlation among the samples and thus further complicates the analysis and interpretation of the microbiome data.ResultsIn this article, we propose zero-inflated Gaussian mixed models (ZIGMMs) to analyze longitudinal microbiome data. ZIGMMs is a robust and flexible method which can be applicable for longitudinal microbiome proportion data or count data generated with either 16S rRNA or shotgun sequencing technologies. It can include various types of fixed effects and random effects and account for various within-subject correlation structures, and can effectively handle zero-inflation. We developed an efficient Expectation-Maximization (EM) algorithm to fit the ZIGMMs by taking advantage of the standard procedure for fitting linear mixed models. We demonstrate the computational efficiency of our EM algorithm by comparing with two other zero-inflated methods. We show that ZIGMMs outperform the previously used linear mixed models (LMMs), negative binomial mixed models (NBMMs) and zero-inflated Beta regression mixed model (ZIBR) in detecting associated effects in longitudinal microbiome data through extensive simulations. We also apply our method to two public longitudinal microbiome datasets and compare with LMMs and NBMMs in detecting dynamic effects of associated taxa.

Negative Binomial mixed models estimated with the maximum likelihood method can be used for longitudinal RNAseq data.

Time-course RNAseq experiments, where tissues are repeatedly collected from the same subjects, e.g. humans or animals over time or under several different experimental conditions, are becoming more popular due to the reducing sequencing costs. Such designs offer the great potential to identify genes that change over time or progress differently in time across experimental groups. Modelling of the longitudinal gene expression in such time-course RNAseq data is complicated by the serial correlations, missing values due to subject dropout or sequencing errors, long follow up with potentially non-linear progression in time and low number of subjects. Negative Binomial mixed models can address all these issues. However, such models under the maximum likelihood (ML) approach are less popular for RNAseq data due to convergence issues (see, e.g. [1]). We argue in this paper that it is the use of an inaccurate numerical integration method in combination with the typically small sample sizes which causes such mixed models to fail for a great portion of tested genes. We show that when we use the accurate adaptive Gaussian quadrature approach to approximate the integrals over the random-effects terms, we can successfully estimate the model parameters with the maximum likelihood method. Moreover, we show that the boostrap method can be used to preserve the type I error rate in small sample settings. We evaluate empirically the small sample properties of the test statistics and compare with state-of-the-art approaches. The method is applied on a longitudinal mice experiment to study the dynamics in Duchenne Muscular Dystrophy. Contact:s.tsonaka@lumc.nlRoula Tsonaka is an assistant professor at the Medical Statistics, Department of Biomedical Data Sciences, Leiden University Medical Center. Her research focuses on statistical methods for longitudinal omics data. Pietro Spitali is an assistant professor at the Department of Human Genetics, Leiden University Medical Center. His research focuses on the identification of biomarkers for neuromuscular disorders.

Outcomes in Vaso-Occlusive Crisis Treatment in the Emergency Department Vs. Acute Care Observation Center

I NTRODUCTION: Acute painful vaso-occlusive crises (VOC) are the leading cause of emergency department (ED) encounters and hospital admissions for those with sickle cell disease (SCD). For SCD patients, the goal of the sickle cell acute care observation unit (ACOU) at University of Illinois Health (UIH) is to improve patient outcomes by providing immediate care for an uncomplicated VOC. At our urban hospital which cares for more than 500 adult SCD patients, a considerable portion of SCD patients, despite having access to the ACOU, continue to present to the ED for treatment of an uncomplicated VOC. In order to help improve our current system, this study investigated outcomes in SCD patients who receive care for an uncomplicated VOC in the ACOU versus the ED at UIH. METHODS: By querying the electronic medical record, a retrospective study was conducted to analyze outcomes of encounters from the ACOU and ED at UIH between October 2019 and December 2019, specifically including SCD patients ≥18 years old who received morphine for treatment of an uncomplicated VOC. Encounters for complicated VOCs such as acute chest syndrome and stroke were excluded. Endpoints collected include time to first dose of morphine, total milligrams (mg) of IV morphine equivalents given, number of total morphine doses, admission rates, subsequent hospital length of stay, and 30-day inpatient admission rates. Time to the first dose (log transformed) and total dose in mg were analyzed by linear mixed effects models. The number of doses and hospitalization days were analyzed by negative binomial mixed effects model. Admissions and 30-day admissions were analyzed by logistic mixed effects models. These models adjusted for age, gender, and severe Hb genotype (HbSS or HbS beta0-thalassemia) and treated patient identity as random effect. P values were obtained from Wald- test. RESULTS: The ACOU data set contains 394 patient encounters for 79 patients with a median age of 33 years (interquartile range [IQR], 28-40), 71% female, and 73% with severe sickle genotypes. The ED data set contains 391 patient encounters for 128 patients with a median age of 30 years (IQR, 26-41), 53% female, and 74% with severe sickle genotypes. In the ACOU, the median time to first dose of morphine was 49 minutes (IQR, 39-60) compared to 107 minutes (IQR, 71-194) in the ED. The time to first dose was significantly longer in the ED compared to the ACOU (eβ=2.5, p &amp;lt;2×10-16). There was no significant difference in the total number of morphine doses received nor the total mg of morphine received between the two locations. Admission rate from ACOU was 6.6% compared to 53% from ED (OR=0.019, p=2x10-16). Of those admitted, the median number of hospitalization days from the ACOU was 4 days (IQR, 2.3-5.8) and 4 days (IQR, 2.0-6.5) from the ED. There was no significant difference in hospitalization days (p=0.6). The 30-day admission rate was 55% from the ACOU compared to 58% from the ED. 30-day admission rate however had strong intra-patient correlation (i.e., a patient was likely re-admitted multiple times): 44% of patients from the ACOU had admissions within 30 days of their ACOU visit compared to 32% from the ED. Controlling for the intra-patient correlation, ACOU visits had a higher 30-day admission rate than ED visit (OR=2.8, p=0.0015). DISCUSSION: SCD patients treated for an uncomplicated VOC at the sickle ACOU at UIH had a significantly shorter time to initial dose of IV pain medication. The wait time in the ED before first dose of IV pain medication received was more than double than those treated in the ACOU. Patients treated for an uncomplicated VOC in ACOU and ED had similar hospitalization days without a statistically significant difference. The 30-day admission rate to the inpatient setting was comparable for those treated in the ED versus the ACOU. However, given that only 6.6% of patients from the ACOU were admitted during the study period, this suggests that most patients who use both the ED and ACOU tend to be subsequently admitted from the ED. SCD patients may be presenting to the ED for treatment of VOC if capacity in the ACOU is exceeded or are presenting outside of hours of operation (currently 2 shifts Monday through Saturday). Therefore, improving access to our ACOU by increasing capacity and hours of operation may subsequently also lead to a decrease in time to first dose of medication and decrease in the overall 30-day admission rate. Disclosures Gordeuk: Imara: Research Funding; CSL Behring: Consultancy, Research Funding; Global Blood Therapeutics: Consultancy, Research Funding; Novartis: Consultancy; Ironwood: Research Funding.

NBZIMM: negative binomial and zero-inflated mixed models, with application to microbiome/metagenomics data analysis

BackgroundMicrobiome/metagenomic data have specific characteristics, including varying total sequence reads, over-dispersion, and zero-inflation, which require tailored analytic tools. Many microbiome/metagenomic studies follow a longitudinal design to collect samples, which further complicates the analysis methods needed. A flexible and efficient R package is needed for analyzing processed multilevel or longitudinal microbiome/metagenomic data.ResultsNBZIMM is a freely available R package that provides functions for setting up and fitting negative binomial mixed models, zero-inflated negative binomial mixed models, and zero-inflated Gaussian mixed models. It also provides functions to summarize the results from fitted models, both numerically and graphically. The main functions are built on top of the commonly used R packages nlme and MASS, allowing us to incorporate the well-developed analytic procedures into the framework for analyzing over-dispersed and zero-inflated count or proportion data with multilevel structures (e.g., longitudinal studies). The statistical methods and their implementations in NBZIMM particularly address the data characteristics and the complex designs in microbiome/metagenomic studies. The package is freely available from the public GitHub repository https://github.com/nyiuab/NBZIMM.ConclusionThe NBZIMM package provides useful tools for complex microbiome/metagenomics data analysis.

Exposure to air pollution and COVID-19 mortality in the United States

United States government scientists estimate that COVID-19 may kill tens of thousands of Americans. Many of the pre-existing conditions that increase the risk of death in those with COVID-19 are the same diseases that are affected by long-term exposure to air pollution. We investigated whether long-term average exposure to fine particulate matter (PM2.5) is associated with an increased risk of COVID-19 death in the United States. We collected data for more than 3,000 counties in the United States (representing 98% of the population) up to April 22, 2020. We fit negative binomial mixed models using county-level COVID-19 deaths as the outcome and county-level long-term average of PM2.5 as the exposure. In the main analysis, we adjusted by 20 potential confounding factors including population size, age distribution, population density, time since the beginning of the outbreak, time since state’s issuance of stay-at-home order, hospital beds, number of individuals tested, weather, and socioeconomic and behavioral variables such as obesity and smoking. We included a random intercept by state to account for potential correlation in counties within the same state. We conducted more than 68 additional sensitivity analyses. We found that an increase of only 1 μg/m3 in PM2.5 is associated with an 8% increase in the COVID-19 death rate (95% confidence interval [CI]: 2%, 15%). The results were statistically significant and robust to secondary and sensitivity analyses. A small increase in long-term exposure to PM2.5 leads to a large increase in the COVID-19 death rate. Despite the inherent limitations of the ecological study design, our results underscore the importance of continuing to enforce existing air pollution regulations to protect human health both during and after the COVID-19 crisis. The data and code are publicly available so our analyses can be updated routinely.

SPARC-ing Change—The Maryland Statewide Prevention and Reduction of Clostridioides difficile (SPARC) Collaborative

Background: In 2018, the Maryland Department of Health, in collaboration with the University of Maryland and Johns Hopkins University, created the Statewide Prevention and Reduction of Clostridioides difficile (SPARC) collaborative to reduce C. difficile as specified in Healthy People 2020. Methods: The SPARC collaborative recruited hospitals contributing most cases to statewide C. difficile standardized infection ratio (SIR), according to data reported to the National Healthcare Safety Network (NHSN). SPARC developed intervention bundles around 4 domains: infection prevention, environmental cleaning, and diagnostic and antimicrobial stewardship. Each facility completed a self-assessment followed by an on-site, day-long, peer-to-peer (P2P) evaluation with 8–12 SPARC subject matter experts (SMEs) representing each domain. The SMEs met with hospital executive leadership and then led 4 domain-based group discussions with relevant hospital team leaders. To identify policy and practice gaps, SMEs visited hospital inpatient units for informal interviews with frontline staff. In a closing session, SPARC SMEs, hospital executives, and team leaders reconvened to discuss preliminary findings. This included review of covert observation data (hand hygiene, personal protective equipment compliance, environmental cleaning) obtained by SPARC team 1–2 weeks prior. Final SPARC P2P written recommendations guided development of customized interventions at each hospital. SPARC provided continuous support (follow up phone calls, educational webinars, technical support, didactic training for antimicrobial stewardship pharmacists) to enhance facility-specific implementation. For every quarter, we categorized C. difficile NHSN data for each Maryland hospital into “SPARC” or “non-SPARC” based on participation status. Using negative binomial mixed models, we analyzed difference-in-difference of pre- and postincidence rate ratios (IRRs) for SPARC and non-SPARC hospitals, which allowed estimation of change attributable to SPARC participation independent of other time-varying factors. Results: Overall, 13 of 48 (27%) hospitals in Maryland participated in the intervention. The baseline SIR for all Maryland hospitals was 0.92, and the post-SPARC SIR was 0.67. The SPARC hospitals had a greater reduction in hospital-onset C. difficile incidence; 8.6 and 4.3 events per 10,000 patient days for baseline and most recent quarter, respectively. For non-SPARC hospitals, these hospital-onset C. difficile incidences were 5.1 preintervention and 4.3 postintervention. We found a statistically significant difference-in-difference between SPARC and non-SPARC hospital C. difficile reduction rates (ratio of IRR, 0.63; 95% CI, 0.44−0.89; P = .01). Conclusions: The Maryland SPARC collaborative, a public health-academic partnership, was associated with a 25% reduction in the Maryland C. difficile SIR. Hospitals participating in SPARC demonstrated significantly reduced C. difficile incidences to match that of high-performing hospitals in Maryland.Funding: NoneDisclosure: Aaron Milstone, BD – consulting.