Roman Jandarov

Exposure assessment models for elemental components of particulate matter in an urban environment: A comparison of regression and random forest approaches

Exposure assessment for elemental components of particulate matter (PM) using land use modeling is a complex problem due to the high spatial and temporal variations in pollutant concentrations at the local scale. Land use regression (LUR) models may fail to capture complex interactions and non-linear relationships between pollutant concentrations and land use variables. The increasing availability of big spatial data and machine learning methods present an opportunity for improvement in PM exposure assessment models. In this manuscript, our objective was to develop a novel land use random forest (LURF) model and compare its accuracy and precision to a LUR model for elemental components of PM in the urban city of Cincinnati, Ohio. PM smaller than 2.5 μm (PM2.5) and eleven elemental components were measured at 24 sampling stations from the Cincinnati Childhood Allergy and Air Pollution Study (CCAAPS). Over 50 different predictors associated with transportation, physical features, community socioeconomic characteristics, greenspace, land cover, and emission point sources were used to construct LUR and LURF models. Cross validation was used to quantify and compare model performance. LURF and LUR models were created for aluminum (Al), copper (Cu), iron (Fe), potassium (K), manganese (Mn), nickel (Ni), lead (Pb), sulfur (S), silicon (Si), vanadium (V), zinc (Zn), and total PM2.5 in the CCAAPS study area. LURF utilized a more diverse and greater number of predictors than LUR and LURF models for Al, K, Mn, Pb, Si, Zn, TRAP, and PM2.5 all showed a decrease in fractional predictive error of at least 5% compared to their LUR models. LURF models for Al, Cu, Fe, K, Mn, Pb, Si, Zn, TRAP, and PM2.5 all had a cross validated fractional predictive error less than 30%. Furthermore, LUR models showed a differential exposure assessment bias and had a higher prediction error variance. Random forest and other machine learning methods may provide more accurate exposure assessment.

Secondhand Smoke Exposure and Pediatric Healthcare Visits and Hospitalizations

INTRODUCTION This study assessed the relationship between secondhand smoke exposure (SHSe) as measured by serum cotinine and healthcare utilization among children. METHODS In 2016, the 2009-2012 National Health and Nutrition Examination Survey data were analyzed including 4,985 children aged 3-19 years. Associations between SHSe and having a routine place for healthcare, type of place, and hospital utilization were examined using logistic regression models. Poisson regression analyses assessed the relationship between SHSe and number of hospital admissions. Relationships between SHSe and acute care visits and hospital utilization were examined among asthmatic children. RESULTS SHSe level did not differ by having a routine place for healthcare, although children with high SHSe indicative of active smoking (cotinine ≥3 ng/mL) were 3.49 times (95% CI=1.77, 6.89) more likely to use an emergency department. Children with high SHSe were 2.85 times (95% CI=1.87, 4.34) more likely to have had an overnight hospital stay. Children with high SHSe had 2.05 times (95% CI=1.46, 2.87) the risk of having a higher number of hospital admissions for overnight stays versus children with no SHSe (cotinine <0.05 ng/mL). Among asthmatic children, those with high SHSe and low SHSe (cotinine 0.05-2.99 ng/mL) were more likely to have an acute care visit, overnight hospital stay, and higher number of hospital admissions than asthmatic children with no SHSe. CONCLUSIONS High SHSe is associated with increased healthcare utilization. The emergency department and inpatient settings are important venues in which to routinely offer cessation and SHSe reduction interventions.

Factors Contributing to Depression and Chronic Pain in Patients with Hidradenitis Suppurativa: Results from a Single-Center Retrospective Review

Background: Hidradenitis suppurativa (HS) is a debilitating chronic disease that leads to inflammation and abscess formation in the involved skin, along with a malodorous discharge. Pain is a considerable aspect of HS and significantly impacts quality of life. In addition, HS is significantly associated with depression. A better understanding of contributing factors to depression and pain in patients with HS can identify opportunities to improve care for patients. Objective: To identify factors that contribute to depression and chronic pain in patients with HS. Methods: This is a retrospective chart review of 283 patients seen at dermatology clinics of an academic health center for HS from July 2012 to December 2015. The association between HS and depression and chronic pain was assessed in multivariate models using logistic regression analyses. Results: Patients with a greater number of areas of involvement were more likely to have both chronic pain and depression. Limitations: This is a single-center retrospective chart review with a limited sample size. Conclusion: This study suggests that the extent of disease rather than severity plays a role in reducing the quality of life in HS patients.

Association of secondhand smoke exposure with asthma symptoms, medication use, and healthcare utilization among asthmatic adolescents

ABSTRACT Objective: To investigate the association between secondhand smoke exposure (SHSe) and asthma symptoms, medication use, and emergency department (ED)/urgent care (UC) utilization among adolescents. Methods: We performed a secondary cross-sectional analysis of Population Assessment of Tobacco and Health Study Wave 2 (2014–2015) including asthmatic adolescents (N = 2198). Logistic regression models and Poisson regression models were built. Results: Participants with SHSe ≥1 hour in the past 7 days were at increased risk of reporting shortness of breath and harder to exercise aOR, 1.22; 95% CI, 1.04–1.43), wheezing (aOR, 1.26; 95% CI, 1.01–1.56), wheezing disturbing sleep (aOR, 1.88; 95% CI, 1.35–2.63), wheezing during/after exercise (aOR, 1.41; 95% CI, 1.19–1.66), wheezing limiting speech (aOR, 2.11; 95% CI, 1.55–2.86), dry cough at night (aOR, 1.86; 95% CI, 1.54–2.24), and asthma symptoms disturbing sleep (aOR, 2.25; 95% CI, 1.81–2.79). Participants with SHSe ≥1 hour were more likely to take asthma medications (aOR, 1.25; 95% CI, 1.03–1.52), including steroids (aOR, 1.86; 95% CI, 1.19–2.91), oxygen therapy (aOR, 2.88; 95% CI, 1.82–4.54), and controlling medications (aOR, 1.50; 95% CI, 1.24–1.82). Symptoms and medications varied by living with a smoker and home SHSe. Participants with SHSe were at increased risk of having a higher number of asthma attacks that required steroid use. Participants who lived with a smoker and had home SHSe were at increased risk of having higher ED/UC visits for asthma. Conclusions: SHSe reduction efforts are needed for asthmatic adolescents, and EDs/UCs are promising venues.

Impact of a change in surveillance definition on performance assessment of a catheter-associated urinary tract infection prevention program at a tertiary care medical center

HighlightsA good surveillance definition is necessary to assess quality improvement initiatives.Outcome data could be affected by changes in surveillance definitions.The Link Nurse Program is effective in preventing hospital‐onset infections. Background: In January 2015, the Centers for Disease Control and Prevention (CDC)/National Health Safety Network (NHSN) changed the definition of catheter‐associated urinary tract infection (CAUTI). We evaluated the outcomes of a robust CAUTI prevention program when we performed surveillance using the old definition (before 2015) versus the new definition (after 2015). This is the first study to evaluate how the change in CDC/NHSN definitions affected the outcomes of a CAUTI reduction program. Methods: Baseline was from January 2012 to September 2014; the intervention period was from October 2014 to February 2016. Staff nurses were trained to be liaisons of infection prevention (Link Nurses) with clearly defined CAUTI prevention goals and with ongoing monthly activities. CAUTI incidence per 1000 catheter days was compared between the baseline and intervention periods, using the 2 definitions. Results: With the new definition, CAUTIs decreased by 33%, from 2.69 to 1.81 cases per 1000 catheter days (incidence rate ratio [IRR] = 0.67; 95% confidence interval [CI]: 0.48‐0.93; P < .016). With the old definition, CAUTIs increased by 12%, from 3.38 to 3.80 cases per 1000 catheter days (IRR = 1.12; 95% CI: 0.88‐1.43; P = .348). Conclusion: We aggressively targeted CAUTI prevention, but a reduction was observed only with the new definition. Our findings stress the importance of having a reasonably accurate surveillance definition to monitor infection prevention initiatives.

A novel principal component analysis for spatially misaligned multivariate air pollution data

We propose novel methods for predictive (sparse) PCA with spatially misaligned data. These methods identify principal component loading vectors that explain as much variability in the observed data as possible, while also ensuring the corresponding principal component scores can be predicted accurately by means of spatial statistics at locations where air pollution measurements are not available. This will make it possible to identify important mixtures of air pollutants and to quantify their health effects in cohort studies, where currently available methods cannot be used. We demonstrate the utility of predictive (sparse) PCA in simulated data and apply the approach to annual averages of particulate matter speciation data from national Environmental Protection Agency (EPA) regulatory monitors.

Adolescent Tobacco Smoke Exposure, Respiratory Symptoms, and Emergency Department Use

We examined the association between TSE and related respiratory symptoms and ED and/or UC use among nonsmoking adolescents without asthma diagnoses. OBJECTIVES: Our objective was to examine the relationship between distinct tobacco smoke exposure (TSE) measures and TSE-related symptoms and emergency department (ED) and/or urgent care (UC) use among nonsmoking adolescents without asthma diagnoses. METHODS: We performed a secondary analysis of 7389 adolescents who completed the Population Assessment of Tobacco and Health Study wave 2. Logistic regression and Poisson regression models were built. RESULTS: Adolescents with TSE were at increased risk of reporting: shortness of breath, finding it hard to exercise, wheezing during or after exercise, and dry cough at night. Adolescents who lived with a smoker and had home TSE were at increased odds of reporting wheezing or whistling in the chest, and only adolescents with home TSE were at increased risk of reporting wheezing that disturbed sleep. Adolescents with TSE were less likely to report very good or excellent overall health and physical health but were more likely to report they sometimes, often, or very often missed school because of illness. Participants who lived with a smoker and had TSE ≥1 hour were more likely to have had an ED and/or UC visit. Participants with any TSE were at increased risk of having a higher number of ED and/or UC visits. CONCLUSIONS: Different TSE measures uniquely increased the risk of TSE-related symptoms, but any TSE increased the risk of having a higher number of ED and/or UC visits. The providers at these high-volume settings should offer interventions to adolescents who are exposed to tobacco smoke and their families to decrease these symptoms and related morbidity.

Assessing the accuracy of commercially available gas sensors for the measurement of ambient ozone and nitrogen dioxide

Abstract The objective of the National Institute for Occupational Safety and Health (NIOSH) accuracy criterion is to ensure that measurements from monitoring devices are within ±25% of the true concentration of the analyte with 95% certainty. To determine whether NO2 and O3 sensors meet this criterion, three commercially available units (Cairclip O3/NO2, Aeroqual NO2, and Aeroqual O3 sensors) were co-located three times with validated instruments (NOx chemiluminescence [NO2mon] and photometric O3 analyzers [O3mon]) at an outdoor monitoring station. As cofactors of sensor performance such as temperature (T) and relative humidity (RH) potentially influence the response of NO2 and O3 sensors, corrections for cofactors were made by using T, RH, and the sensor measurements to predict measurements made by NO2mon and O3mon during the first co-location period (training dataset). The developed models were tested in the merged data obtained from the second and third co-location periods (testing dataset). In the training and testing datasets, the mean NO2 as measured by NO2mon was 4.6 ppb (range = 0.4–35 ppb) and 9.4 ppb (range = 1–37 ppb), respectively. The mean O3 in the training and testing datasets as measured by O3mon was 38.8 ppb (range = 1–65 ppb) and 35.7 ppb (range = 1–61 ppb), respectively. None of the sensor measurements in the training dataset were within the NIOSH accuracy criterion (mean error ≥25%). After correcting for cofactors of sensor performance, the accuracy of the Cairclip O3/NO2 and the Aeroqual O3 sensors considerably improved when tested with the testing dataset (mean error = -1% and 14%, respectively). However, the Aeroqual NO2 sensor had an error that was not within ±25%. Raw measurements from the tested sensors may be unsuitable for assessing workers’ exposure to NO2 and O3. Corrections for cofactors of Cairclip O3/NO2 and Aeroqual O3 sensor performance are required for more accurate occupational exposure assessment.

Predicting Daily Urban Fine Particulate Matter Concentrations Using a Random Forest Model

The short-term and acute health effects of fine particulate matter less than 2.5 μm (PM2.5) have highlighted the need for exposure assessment models with high spatiotemporal resolution. Here, we utilize satellite, meteorologic, atmospheric, and land-use data to train a random forest model capable of accurately predicting daily PM2.5 concentrations at a resolution of 1 × 1 km throughout an urban area encompassing seven counties. Unlike previous models based on aerosol optical density (AOD), we show that the missingness of AOD is an effective predictor of ground-level PM2.5 and create an ensemble model that explicitly deals with AOD missingness and is capable of predicting with complete spatial and temporal coverage of the study domain. Our model performed well with an overall cross-validated root mean squared error (RMSE) of 2.22 μg/m3 and a cross-validated R2 of 0.91. We illustrate the daily changing spatial patterns of PM2.5 concentrations across our urban study area made possible by our accurate, high-resolution model. The model will facilitate high-resolution assessment of both long-term and acute PM2.5 exposures in order to quantify their associations with related health outcomes.

A comparison of resampling and recursive partitioning methods in random forest for estimating the asymptotic variance using the infinitesimal jackknife: Effect of RF variations on prediction variance

The infinitesimal jackknife (IJ) has recently been applied to the random forest to estimate its prediction variance. These theorems were verified under a traditional random forest framework which uses classification and regression trees (CART) and bootstrap resampling. However, random forests using conditional inference (CI) trees and subsampling have been found to be not prone to variable selection bias. Here, we conduct simulation experiments using a novel approach to explore the applicability of the IJ to random forests using variations on the resampling method and base learner. Test data points were simulated and each trained using random forest on one hundred simulated training data sets using different combinations of resampling and base learners. Using CI trees instead of traditional CART trees as well as using subsampling instead of bootstrap sampling resulted in a much more accurate estimation of prediction variance when using the IJ. The random forest variations here have been incorporated into an open source software package for the R programming language.