Sparse modeling of spatial environmental variables associated with asthma

Application of Spatial Analysis on Electronic Health Records to Characterize Patient Phenotypes: Systematic Review

BACKGROUND

Electronic health records (EHRs) commonly contain patient addresses that provide valuable data for geocoding and spatial analysis, enabling more comprehensive descriptions of individual patients for clinical purposes. Despite the widespread use of EHRs in clinical decision support and interventions, no systematic review has examined the extent to which spatial analysis is used to characterize patient phenotypes.

OBJECTIVE

This study reviews advanced spatial analyses that used individual-level health data from EHRs within the United States to characterize patient phenotypes.

METHODS

We systematically evaluated English-language, peer-reviewed studies from the PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar databases from inception to August 20, 2023, without imposing constraints on study design or specific health domains.

RESULTS

A substantial proportion of studies (>85%) were limited to geocoding or basic mapping without implementing advanced spatial statistical analysis, leaving only 49 studies that met the eligibility criteria. These studies used diverse spatial methods, with a predominant focus on clustering techniques, while spatiotemporal analysis (frequentist and Bayesian) and modeling were less common. A noteworthy surge (n=42, 86%) in publications was observed after 2017. The publications investigated a variety of adult and pediatric clinical areas, including infectious disease, endocrinology, and cardiology, using phenotypes defined over a range of data domains such as demographics, diagnoses, and visits. The primary health outcomes investigated were asthma, hypertension, and diabetes. Notably, patient phenotypes involving genomics, imaging, and notes were limited.

CONCLUSIONS

This review underscores the growing interest in spatial analysis of EHR-derived data and highlights knowledge gaps in clinical health, phenotype domains, and spatial methodologies. We suggest that future research should focus on addressing these gaps and harnessing spatial analysis to enhance individual patient contexts and clinical decision support.

Application of Spatial Analysis for Electronic Health Records: Characterizing Patient Phenotypes and Emerging Trends

Background

Electronic health records (EHR) commonly contain patient addresses that provide valuable data for geocoding and spatial analysis, enabling more comprehensive descriptions of individual patients for clinical purposes. Despite the widespread use of EHR in clinical decision support and interventions, no systematic review has examined the extent to which spatial analysis is used to characterize patient phenotypes.

Objective

This study reviews advanced spatial analyses that employed individual-level health data from EHR within the US to characterize patient phenotypes.

Methods

We systematically evaluated English-language peer-reviewed articles from PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar databases from inception to August 20, 2023, without imposing constraints on time, study design, or specific health domains.

Results

Only 49 articles met the eligibility criteria. These articles utilized diverse spatial methods, with a predominant focus on clustering techniques, while spatiotemporal analysis (frequentist and Bayesian) and modeling were relatively underexplored. A noteworthy surge (n = 42, 85.7%) in publications was observed post-2017. The publications investigated a variety of adult and pediatric clinical areas, including infectious disease, endocrinology, and cardiology, using phenotypes defined over a range of data domains, such as demographics, diagnoses, and visits. The primary health outcomes investigated were asthma, hypertension, and diabetes. Notably, patient phenotypes involving genomics, imaging, and notes were rarely utilized.

Conclusions

This review underscores the growing interest in spatial analysis of EHR-derived data and highlights knowledge gaps in clinical health, phenotype domains, and spatial methodologies. Additionally, this review proposes guidelines for harnessing the potential of spatial analysis to enhance the context of individual patients for future clinical decision support.

Application of Spatial Analysis for Electronic Health Records: Characterizing Patient Phenotypes and Emerging Trends

Background

Electronic health records (EHR) commonly contain patient addresses that provide valuable data for geocoding and spatial analysis, enabling more comprehensive descriptions of individual patients for clinical purposes. Despite the widespread use of EHR in clinical decision support and interventions, no systematic review has examined the extent to which spatial analysis is used to characterize patient phenotypes.

Objective

This study reviews advanced spatial analyses that employed individual-level health data from EHR within the US to characterize patient phenotypes.

Methods

We systematically evaluated English-language peer-reviewed articles from PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar databases from inception to August 20, 2023, without imposing constraints on time, study design, or specific health domains.

Results

Only 49 articles met the eligibility criteria. These articles utilized diverse spatial methods, with a predominant focus on clustering techniques, while spatiotemporal analysis (frequentist and Bayesian) and modeling were relatively underexplored. A noteworthy surge (n = 42, 85.7%) in publications was observed post-2017. The publications investigated a variety of adult and pediatric clinical areas, including infectious disease, endocrinology, and cardiology, using phenotypes defined over a range of data domains, such as demographics, diagnoses, and visits. The primary health outcomes investigated were asthma, hypertension, and diabetes. Notably, patient phenotypes involving genomics, imaging, and notes were rarely utilized.

Conclusions

This review underscores the growing interest in spatial analysis of EHR-derived data and highlights knowledge gaps in clinical health, phenotype domains, and spatial methodologies. Additionally, this review proposes guidelines for harnessing the potential of spatial analysis to enhance the context of individual patients for future clinical decision support.

Neighborhood Opportunity and Vulnerability and Incident Asthma Among Children

Background

The extent to which physical and social attributes of neighborhoods play a role in childhood asthma remains understudied.

Objective

To examine associations of neighborhood-level opportunity and social vulnerability measures with childhood asthma incidence.

Design, Setting, and Participants

This cohort study used data from children in 46 cohorts participating in the Environmental Influences on Child Health Outcomes (ECHO) Program between January 1, 1995, and August 31, 2022. Participant inclusion required at least 1 geocoded residential address from birth and parent or caregiver report of a physician's diagnosis of asthma. Participants were followed up to the date of asthma diagnosis, date of last visit or loss to follow-up, or age 20 years.

Exposures

Census tract-level Child Opportunity Index (COI) and Social Vulnerability Index (SVI) at birth, infancy, or early childhood, grouped into very low (<20th percentile), low (20th to <40th percentile), moderate (40th to <60th percentile), high (60th to <80th percentile), or very high (≥80th percentile) COI or SVI.

Main Outcomes and Measures

The main outcome was parent or caregiver report of a physician's diagnosis of childhood asthma (yes or no). Poisson regression models estimated asthma incidence rate ratios (IRRs) associated with COI and SVI scores at each life stage.

Results

The study included 10 516 children (median age at follow-up, 9.1 years [IQR, 7.0-11.6 years]; 52.2% male), of whom 20.6% lived in neighborhoods with very high COI and very low SVI. The overall asthma incidence rate was 23.3 cases per 1000 child-years (median age at asthma diagnosis, 6.6 years [IQR, 4.1-9.9 years]). High and very high (vs very low) COI at birth, infancy, or early childhood were associated with lower subsequent asthma incidence independent of sociodemographic characteristics, parental asthma history, and parity. For example, compared with very low COI, the adjusted IRR for asthma was 0.87 (95% CI, 0.75-1.00) for high COI at birth and 0.83 (95% CI, 0.71-0.98) for very high COI at birth. These associations appeared to be attributable to the health and environmental and the social and economic domains of the COI. The SVI during early life was not significantly associated with asthma incidence. For example, compared with a very high SVI, the adjusted IRR for asthma was 0.88 (95% CI, 0.75-1.02) for low SVI at birth and 0.89 (95% CI, 0.76-1.03) for very low SVI at birth.

Conclusions

In this cohort study, high and very high neighborhood opportunity during early life compared with very low neighborhood opportunity were associated with lower childhood asthma incidence. These findings suggest the need for future studies examining whether investing in health and environmental or social and economic resources in early life would promote health equity in pediatric asthma.

Investigating the association of environmental exposures and all-cause mortality in the UK Biobank using sparse principal component analysis

Multicollinearity refers to the presence of collinearity between multiple variables and renders the results of statistical inference erroneous (Type II error). This is particularly important in environmental health research where multicollinearity can hinder inference. To address this, correlated variables are often excluded from the analysis, limiting the discovery of new associations. An alternative approach to address this problem is the use of principal component analysis. This method, combines and projects a group of correlated variables onto a new orthogonal space. While this resolves the multicollinearity problem, it poses another challenge in relation to interpretability of results. Standard hypothesis testing methods can be used to evaluate the association of projected predictors, called principal components, with the outcomes of interest, however, there is no established way to trace the significance of principal components back to individual variables. To address this problem, we investigated the use of sparse principal component analysis which enforces a parsimonious projection. We hypothesise that this parsimony could facilitate the interpretability of findings. To this end, we investigated the association of 20 environmental predictors with all-cause mortality adjusting for demographic, socioeconomic, physiological, and behavioural factors. The study was conducted in a cohort of 379,690 individuals in the UK. During an average follow-up of 8.05 years (3,055,166 total person-years), 14,996 deaths were observed. We used Cox regression models to estimate the hazard ratio (HR) and 95% confidence intervals (CI). The Cox models were fitted to the standardised environmental predictors (a) without any transformation (b) transformed with PCA, and (c) transformed with SPCA. The comparison of findings underlined the potential of SPCA for conducting inference in scenarios where multicollinearity can increase the risk of Type II error. Our analysis unravelled a significant association between average noise pollution and increased risk of all-cause mortality. Specifically, those in the upper deciles of noise exposure have between 5 and 10% increased risk of all-cause mortality compared to the lowest decile.

Systematic review of approaches to use of neighborhood-level risk factors with clinical data to predict clinical risk and recommend interventions

BACKGROUND

Despite a large body of literature investigating how the environment influences health outcomes, most published work to date includes only a limited subset of the rich clinical and environmental data that is available and does not address how these data might best be used to predict clinical risk or expected impact of clinical interventions.

OBJECTIVE

Identify existing approaches to inclusion of a broad set of neighborhood-level risk factors with clinical data to predict clinical risk and recommend interventions.

METHODS

A systematic review of scientific literature published and indexed in PubMed, Web of Science, Association of Computing Machinery (ACM) and SCOPUS from 2010 through October 2020 was performed. To be included, articles had to include search terms related to Electronic Health Record (EHR) data Neighborhood-Level Risk Factors (NLRFs), and Machine Learning (ML) Methods. Citations of relevant articles were also reviewed for additional articles for inclusion. Articles were reviewed and coded by two independent reviewers to capture key information including data sources, linkage of EHR to NRLFs, methods, and results. Articles were assessed for quality using a modified Quality Assessment Tool for Systematic Reviews of Observational Studies (QATSO).

RESULTS

A total of 334 articles were identified for abstract review. 36 articles were identified for full review with 19 articles included in the final analysis. All but two of the articles included socio-demographic data derived from the U.S. Census and we found great variability in sources of NLRFs beyond the Census. The majority or the articles (14 of 19) included broader clinical (e.g. medications, labs and co-morbidities) and demographic information about the individual from the EHR in addition to the clinical outcome variable. Half of the articles (10) had a stated goal to predict the outcome(s) of interest. While results of the studies reinforced the correlative association of NLRFs to clinical outcomes, only one article found that adding NLRFs into a model with other data added predictive power with the remainder concluding either that NLRFs were of mixed value depending on the model and outcome or that NLRFs added no predictive power over other data in the model. Only one article scored high on the quality assessment with 13 scoring moderate and 4 scoring low.

CONCLUSIONS

In spite of growing interest in combining NLRFs with EHR data for clinical prediction, we found limited evidence that NLRFs improve predictive power in clinical risk models. We found these data and methods are being used in four ways. First, early approaches to include broad NLRFs to predict clinical risk primarily focused on dimension reduction for feature selection or as a data preparation step to input into regression analysis. Second, more recent work incorporates NLRFs into more advanced predictive models, such as Neural Networks, Random Forest, and Penalized Lasso to predict clinical outcomes or predict value of interventions. Third, studies that test how inclusion of NLRFs predict clinical risk have shown mixed results regarding the value of these data over EHR or claims data alone and this review surfaced evidence of potential quality challenges and biases inherent to this approach. Finally, NLRFs were used with unsupervised learning to identify underlying patterns in patient populations to recommend targeted interventions. Further access to computable, high quality data is needed along with careful study design, including sub-group analysis, to better determine how these data and methods can be used to support decision making in a clinical setting.

Asthma-prone areas modeling using a machine learning model

Nowadays, owing to population growth, increasing environmental pollution, and lifestyle changes, the number of asthmatics has significantly increased. Therefore, the purpose of our study was to determine the asthma-prone areas in Tehran, Iran considering environmental, spatial factors. Initially, we built a spatial database using 872 locations of children with asthma and 13 environmental factors affecting the disease—distance to parks and streets, rainfall, temperature, humidity, pressure, wind speed, particulate matter (PM 10 and PM 2.5), ozone (O₃), sulfur dioxide (SO₂), carbon monoxide (CO), and nitrogen dioxide (NO₂). Subsequently, utilizing this spatial database, a random forest (RF) machine learning model, and a geographic information system, we prepared a map of asthma-prone areas. For modeling and validation, we deployed 70% and 30%, respectively, of the locations of children with asthma. The results of spatial autocorrelation and RF model showed that the criteria of distance to parks and streets as well as PM 2.5 and PM 10 had the greatest impact on asthma occurrence in the study area. Spatial autocorrelation analyses indicated that the distribution of asthma cases was not random. According to receiver operating characteristic results, the RF model had good accuracy (the area under the curve was 0.987 and 0.921, respectively, for training and testing data).

Spatiotemporal variations of asthma admission rates and their relationship with environmental factors in Guangxi, China

OBJECTIVE

The study aimed to determine if and how environmental factors correlated with asthma admission rates in geographically different parts of Guangxi province in China.

SETTING

Guangxi, China.

PARTICIPANTS

This study was done among 7804 asthma patients.

PRIMARY AND SECONDARY OUTCOME MEASURES

Spearman correlation coefficient was used to estimate correlation between environmental factors and asthma hospitalisation rates in multiple regions. Generalised additive model (GAM) with Poisson regression was used to estimate effects of environmental factors on asthma hospitalisation rates in 14 regions of Guangxi.

RESULTS

The strongest effect of carbon monoxide (CO) was found on lag1 in Hechi, and every 10 µg/m3 increase of CO caused an increase of 25.6% in asthma hospitalisation rate (RR 1.26, 95% CI 1.02 to 1.55). According to the correlation analysis, asthma hospitalisations were related to the daily temperature, daily range of temperature, CO, nitrogen dioxide (NO2) and particulate matter (PM2.5) in multiple regions. According to the result of GAM, the adjusted R2 was high in Beihai and Nanning, with values of 0.29 and 0.21, which means that environmental factors are powerful in explaining changes of asthma hospitalisation rates in Beihai and Nanning.

CONCLUSION

Asthma hospitalisation rate was significantly and more strongly associated with CO than with NO2, SO2 or PM2.5 in Guangxi. The risk factors of asthma exacerbations were not consistent in different regions, indicating that targeted measures should differ between regions.

Neighborhood Risk and Hospital Use for Pediatric Asthma, Rhode Island, 2005-2014

INTRODUCTION

Studies consistently show that children living in poor neighborhoods have worse asthma outcomes. The objective of our study was to assess the association between negative neighborhood factors (ie, neighborhood risk) and pediatric asthma hospital use.

METHODS

This retrospective study used data from children aged 2 to 17 years in a statewide (Rhode Island) hospital network administrative database linked to US Census Bureau data. We defined an asthma visit as an International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code of 493 in any diagnosis field. We used 8 highly correlated measures for each census-block group to construct an index of neighborhood risk. We used maps and linear regression to assess the association of neighborhood risk with average annual census-block-group rates of asthma emergency department visits and hospitalizations. We used multivariable analyses to identify child characteristics and neighborhood risk associated with an asthma revisit, accounting for the child's sociodemographic information, season, and multiple measurements per child.

RESULTS

From 2005 through 2014, we counted 359,195 visits for 146,889 children. Of these, 12,699 children (8.6%) had one or more asthma visits. Linear regression results showed 1.18 (95% confidence interval, 1.06-1.30) more average annual emergency departments visits per 100 children and 0.41 (95% confidence interval, 0.34-0.47) more average annual hospitalizations per 100 children in neighborhoods in the highest-risk index quintile than in neighborhoods in the lowest-risk index quintile.

CONCLUSION

Interventions to improve asthma outcomes among children should move beyond primary care or clinic settings and involve a careful evaluation of social context and environmental triggers.

Using electronic health record data for environmental and place based population health research: a systematic review

PURPOSE

We conducted a systematic review of literature published on January 2000-May 2017 that spatially linked electronic health record (EHR) data with environmental information for population health research.

METHODS

We abstracted information on the environmental and health outcome variables and the methods and data sources used.

RESULTS

The automated search yielded 669 articles; 128 articles are included in the full review. The number of articles increased by publication year; the majority (80%) were from the United States, and the mean sample size was approximately 160,000. Most articles used cross-sectional (44%) or longitudinal (40%) designs. Common outcomes were health care utilization (32%), cardiometabolic conditions/obesity (23%), and asthma/respiratory conditions (10%). Common environmental variables were sociodemographic measures (42%), proximity to medical facilities (15%), and built environment and land use (13%). The most common spatial identifiers were administrative units (59%), such as census tracts. Residential addresses were also commonly used to assign point locations, or to calculate distances or buffer areas.

CONCLUSIONS

Future research should include more detailed descriptions of methods used to geocode addresses, focus on a broader array of health outcomes, and describe linkage methods. Studies should also explore using longitudinal residential address histories to evaluate associations between time-varying environmental variables and health outcomes.

Analysis of correlation between pediatric asthma exacerbation and exposure to pollutant mixtures with association rule mining

OBJECTIVES

Traditional studies on effects of outdoor pollution on asthma have been criticized for questionable statistical validity and inefficacy in exploring the effects of multiple air pollutants, alone and in combination. Association rule mining (ARM), a method easily interpretable and suitable for the analysis of the effects of multiple exposures, could be of use, but the traditional interest metrics of support and confidence need to be substituted with metrics that focus on risk variations caused by different exposures.

METHODS

We present an ARM-based methodology that produces rules associated with relevant odds ratios and limits the number of final rules even at very low support levels (0.5%), thanks to post-pruning criteria that limit rule redundancy and control for statistical significance. The methodology has been applied to a case-crossover study to explore the effects of multiple air pollutants on risk of asthma in pediatric subjects.

RESULTS

We identified 27 rules with interesting odds ratio among more than 10,000 having the required support. The only rule including only one chemical is exposure to ozone on the previous day of the reported asthma attack (OR=1.14). 26 combinatory rules highlight the limitations of air quality policies based on single pollutant thresholds and suggest that exposure to mixtures of chemicals is more harmful, with odds ratio as high as 1.54 (associated with the combination day0 SO₂, day0 NO, day0 NO₂, day1 PM).

CONCLUSIONS

The proposed method can be used to analyze risk variations caused by single and multiple exposures. The method is reliable and requires fewer assumptions on the data than parametric approaches. Rules including more than one pollutant highlight interactions that deserve further investigation, while helping to limit the search field.

Using Electronic Health Records for Population Health Research: A Review of Methods and Applications

The use and functionality of electronic health records (EHRs) have increased rapidly in the past decade. Although the primary purpose of EHRs is clinical, researchers have used them to conduct epidemiologic investigations, ranging from cross-sectional studies within a given hospital to longitudinal studies on geographically distributed patients. Herein, we describe EHRs, examine their use in population health research, and compare them with traditional epidemiologic methods. We describe diverse research applications that benefit from the large sample sizes and generalizable patient populations afforded by EHRs. These have included reevaluation of prior findings, a range of diseases and subgroups, environmental and social epidemiology, stigmatized conditions, predictive modeling, and evaluation of natural experiments. Although studies using primary data collection methods may have more reliable data and better population retention, EHR-based studies are less expensive and require less time to complete. Future EHR epidemiology with enhanced collection of social/behavior measures, linkage with vital records, and integration of emerging technologies such as personal sensing could improve clinical care and population health.