Comparison of different definitions of traumatic brain injury: implications for cohort characteristics and survival in women, Philadelphia, USA.

BACKGROUND: Traumatic brain injury (TBI) is an acute injury that is understudied in civilian cohorts, especially among women, as TBI has historically been considered to be largely a condition of athletes and military service people. Both the Centres for Disease Control and Prevention (CDC) and Department of Defense (DOD)/Veterans Affairs (VA) have developed case definitions to identify patients with TBI from medical records; however, their definitions differ. We sought to re-examine these definitions to construct an expansive and more inclusive definition among a cohort of women with TBI. METHODS: In this study, we use electronic health records (EHR) from a single healthcare system to study the impact of using different case definitions to identify patients with TBI. Specifically, we identified adult female patients with TBI using the CDC definition, DOD/VA definition and a combined and expanded definition herein called the Penn definition. RESULTS: We identified 4446 adult-female TBI patients meeting the CDC definition, 3619 meeting the DOD/VA definition, and together, 6432 meeting our expanded Penn definition that includes the CDC ad DOD/VA definitions. CONCLUSIONS: Using the expanded definition identified almost two times as many patients, enabling investigations to more fully characterise these patients and related outcomes. Our expanded TBI case definition is available to other researchers interested in employing EHRs to investigate TBI.

Towards deep phenotyping pregnancy: a systematic review on artificial intelligence and machine learning methods to improve pregnancy outcomes.

OBJECTIVE: Development of novel informatics methods focused on improving pregnancy outcomes remains an active area of research. The purpose of this study is to systematically review the ways that artificial intelligence (AI) and machine learning (ML), including deep learning (DL), methodologies can inform patient care during pregnancy and improve outcomes. MATERIALS AND METHODS: We searched English articles on EMBASE, PubMed and SCOPUS. Search terms included ML, AI, pregnancy and informatics. We included research articles and book chapters, excluding conference papers, editorials and notes. RESULTS: We identified 127 distinct studies from our queries that were relevant to our topic and included in the review. We found that supervised learning methods were more popular (n = 69) than unsupervised methods (n = 9). Popular methods included support vector machines (n = 30), artificial neural networks (n = 22), regression analysis (n = 17) and random forests (n = 16). Methods such as DL are beginning to gain traction (n = 13). Common areas within the pregnancy domain where AI and ML methods were used the most include prenatal care (e.g. fetal anomalies, placental functioning) (n = 73); perinatal care, birth and delivery (n = 20); and preterm birth (n = 13). Efforts to translate AI into clinical care include clinical decision support systems (n = 24) and mobile health applications (n = 9). CONCLUSIONS: Overall, we found that ML and AI methods are being employed to optimize pregnancy outcomes, including modern DL methods (n = 13). Future research should focus on less-studied pregnancy domain areas, including postnatal and postpartum care (n = 2). Also, more work on clinical adoption of AI methods and the ethical implications of such adoption is needed.

Postpartum complications increased in women with polycystic ovary syndrome.

BACKGROUND: Women with polycystic ovary syndrome are at a higher risk of cardiometabolic and psychiatric comorbidities and preconception and antepartum complications, but the impact of polycystic ovary syndrome during the postpartum period is unknown. OBJECTIVE: This study aimed to investigate the risk of postpartum cardiovascular disease complications and perinatal and postpartum depression. STUDY DESIGN: This was a retrospective cohort study conducted using a United States insurance claims database. Women with and without polycystic ovary syndrome aged 18 to 50 years enrolled continuously in a single health plan during the preconception, antepartum, and postpartum periods between 2000 and 2016 were included. The primary outcome was postpartum cardiovascular disease and depression (perinatal and postpartum). Multivariable logistic regression was used to adjust for covariates including age, geographic location, preterm delivery, assisted reproductive technology use, multiple births, prepregnancy depression, prepregnancy diabetes, prepregnancy hypertension, gestational diabetes, gestational hypertension, obesity, history of hyperlipidemia, smoking, and race. RESULTS: We identified 42,391 unique women with polycystic ovary syndrome and 795,480 women without polycystic ovary syndrome. In multivariable models, women with polycystic ovary syndrome had significantly higher odds of cardiovascular disease complications, including postpartum preeclampsia (adjusted odds ratio, 1.30; 95% confidence interval, 1.17-1.45), eclampsia (adjusted odds ratio, 1.45; 95% confidence interval, 1.14-1.86) cardiomyopathy (adjusted odds ratio, 1.26; 95% confidence interval, 1.03-1.54), hypertensive heart disease (adjusted odds ratio, 1.32: 95% confidence interval, 1.07-1.64), thrombotic disease (adjusted odds ratio, 1.50; 95% confidence interval, 1.20-1.87), congestive heart failure (adjusted odds ratio, 1.35; 95% confidence interval, 1.13-1.61), and cerebrovascular accidents (adjusted odds ratio, 1.21; 95% confidence interval, 1.14-1.29), than those without polycystic ovary syndrome, as well as both perinatal (adjusted odds ratio, 1.27; 95% confidence interval, 1.22-1.33) and postpartum depression (adjusted odds ratio, 1.46; 95% confidence interval, 1.36-1.57). Nonobese women with polycystic ovary syndrome had higher odds of postpartum eclampsia (adjusted odds ratio 1.72; 95% confidence interval, 1.31-2.26), peripartum cardiomyopathy (adjusted odds ratio, 1.43; 95% confidence interval, 1.14-1.79), and cerebrovascular accidents (adjusted odds ratio, 1.28; 95% confidence interval, 1.19-1.38) than nonobese women without polycystic ovary syndrome. In the group of women without prepregnancy depression, the odds of perinatal depression (adjusted odds ratio, 1.32; 95% confidence interval, 1.26-1.39) and postpartum depression (adjusted odds ratio, 1.50; 95% confidence interval, 1.39-1.62) were higher in women with polycystic ovary syndrome than those without polycystic ovary syndrome. CONCLUSION: In a large United States cohort, our study found that women with polycystic ovary syndrome are at increased risk of both cardiovascular and psychiatric complications during the postpartum period. Polycystic ovary syndrome should be recognized as an at-risk condition; our findings underscore the need for routine screening and early interventions for these major comorbidities.

Enabling pregnant women and their physicians to make informed medication decisions using artificial intelligence.

The role of artificial intelligence (AI) in healthcare for pregnant women. To assess the role of AI in women's health, discover gaps, and discuss the future of AI in maternal health. A systematic review of English articles using EMBASE, PubMed, and SCOPUS. Search terms included pregnancy and AI. Research articles and book chapters were included, while conference papers, editorials and notes were excluded from the review. Included papers focused on pregnancy and AI methods, and pertained to pharmacologic interventions. We identified 376 distinct studies from our queries. A final set of 31 papers were included for the review. Included papers represented a variety of pregnancy concerns and multidisciplinary applications of AI. Few studies relate to pregnancy, AI, and pharmacologics and therefore, we review carefully those studies. External validation of models and techniques described in the studies is limited, impeding on generalizability of the studies. Our review describes how AI has been applied to address maternal health, throughout the pregnancy process: preconception, prenatal, perinatal, and postnatal health concerns. However, there is a lack of research applying AI methods to understand how pharmacologic treatments affect pregnancy. We identify three areas where AI methods could be used to improve our understanding of pharmacological effects of pregnancy, including: (a) obtaining sound and reliable data from clinical records (15 studies), (b) designing optimized animal experiments to validate specific hypotheses (1 study) to (c) implementing decision support systems that inform decision-making (11 studies). The largest literature gap that we identified is with regards to using AI methods to optimize translational studies between animals and humans for pregnancy-related drug exposures.

The digital revolution in phenotyping.

Phenotypes have gained increased notoriety in the clinical and biological domain owing to their application in numerous areas such as the discovery of disease genes and drug targets, phylogenetics and pharmacogenomics. Phenotypes, defined as observable characteristics of organisms, can be seen as one of the bridges that lead to a translation of experimental findings into clinical applications and thereby support 'bench to bedside' efforts. However, to build this translational bridge, a common and universal understanding of phenotypes is required that goes beyond domain-specific definitions. To achieve this ambitious goal, a digital revolution is ongoing that enables the encoding of data in computer-readable formats and the data storage in specialized repositories, ready for integration, enabling translational research. While phenome research is an ongoing endeavor, the true potential hidden in the currently available data still needs to be unlocked, offering exciting opportunities for the forthcoming years. Here, we provide insights into the state-of-the-art in digital phenotyping, by means of representing, acquiring and analyzing phenotype data. In addition, we provide visions of this field for future research work that could enable better applications of phenotype data.

Improving condition severity classification with an efficient active learning based framework.

Classification of condition severity can be useful for discriminating among sets of conditions or phenotypes, for example when prioritizing patient care or for other healthcare purposes. Electronic Health Records (EHRs) represent a rich source of labeled information that can be harnessed for severity classification. The labeling of EHRs is expensive and in many cases requires employing professionals with high level of expertise. In this study, we demonstrate the use of Active Learning (AL) techniques to decrease expert labeling efforts. We employ three AL methods and demonstrate their ability to reduce labeling efforts while effectively discriminating condition severity. We incorporate three AL methods into a new framework based on the original CAESAR (Classification Approach for Extracting Severity Automatically from Electronic Health Records) framework to create the Active Learning Enhancement framework (CAESAR-ALE). We applied CAESAR-ALE to a dataset containing 516 conditions of varying severity levels that were manually labeled by seven experts. Our dataset, called the "CAESAR dataset," was created from the medical records of 1.9 million patients treated at Columbia University Medical Center (CUMC). All three AL methods decreased labelers' efforts compared to the learning methods applied by the original CAESER framework in which the classifier was trained on the entire set of conditions; depending on the AL strategy used in the current study, the reduction ranged from 48% to 64% that can result in significant savings, both in time and money. As for the PPV (precision) measure, CAESAR-ALE achieved more than 13% absolute improvement in the predictive capabilities of the framework when classifying conditions as severe. These results demonstrate the potential of AL methods to decrease the labeling efforts of medical experts, while increasing accuracy given the same (or even a smaller) number of acquired conditions. We also demonstrated that the methods included in the CAESAR-ALE framework (Exploitation and Combination_XA) are more robust to the use of human labelers with different levels of professional expertise.

Clustering clinical trials with similar eligibility criteria features.

OBJECTIVES: To automatically identify and cluster clinical trials with similar eligibility features. METHODS: Using the public repository ClinicalTrials.gov as the data source, we extracted semantic features from the eligibility criteria text of all clinical trials and constructed a trial-feature matrix. We calculated the pairwise similarities for all clinical trials based on their eligibility features. For all trials, by selecting one trial as the center each time, we identified trials whose similarities to the central trial were greater than or equal to a predefined threshold and constructed center-based clusters. Then we identified unique trial sets with distinctive trial membership compositions from center-based clusters by disregarding their structural information. RESULTS: From the 145,745 clinical trials on ClinicalTrials.gov, we extracted 5,508,491 semantic features. Of these, 459,936 were unique and 160,951 were shared by at least one pair of trials. Crowdsourcing the cluster evaluation using Amazon Mechanical Turk (MTurk), we identified the optimal similarity threshold, 0.9. Using this threshold, we generated 8806 center-based clusters. Evaluation of a sample of the clusters by MTurk resulted in a mean score 4.331±0.796 on a scale of 1-5 (5 indicating "strongly agree that the trials in the cluster are similar"). CONCLUSIONS: We contribute an automated approach to clustering clinical trials with similar eligibility features. This approach can be potentially useful for investigating knowledge reuse patterns in clinical trial eligibility criteria designs and for improving clinical trial recruitment. We also contribute an effective crowdsourcing method for evaluating informatics interventions.

From expert-derived user needs to user-perceived ease of use and usefulness: a two-phase mixed-methods evaluation framework.

Underspecified user needs and frequent lack of a gold standard reference are typical barriers to technology evaluation. To address this problem, this paper presents a two-phase evaluation framework involving usability experts (phase 1) and end-users (phase 2). In phase 1, a cross-system functionality alignment between expert-derived user needs and system functions was performed to inform the choice of "the best available" comparison system to enable a cognitive walkthrough in phase 1 and a comparative effectiveness evaluation in phase 2. During phase 2, five quantitative and qualitative evaluation methods are mixed to assess usability: time-motion analysis, software log, questionnaires - System Usability Scale and the Unified Theory of Acceptance of Use of Technology, think-aloud protocols, and unstructured interviews. Each method contributes data for a unique measure (e.g., time motion analysis contributes task-completion-time; software log contributes action transition frequency). The measures are triangulated to yield complementary insights regarding user-perceived ease-of-use, functionality integration, anxiety during use, and workflow impact. To illustrate its use, we applied this framework in a formative evaluation of a software called Integrated Model for Patient Care and Clinical Trials (IMPACT). We conclude that this mixed-methods evaluation framework enables an integrated assessment of user needs satisfaction and user-perceived usefulness and usability of a novel design. This evaluation framework effectively bridges the gap between co-evolving user needs and technology designs during iterative prototyping and is particularly useful when it is difficult for users to articulate their needs for technology support due to the lack of a baseline.

Development and Optimization of Machine Learning Algorithms for Predicting In-hospital Patient Charges for Congestive Heart Failure Exacerbations, Chronic Obstructive Pulmonary Disease Exacerbations and Diabetic Ketoacidosis.

Background: Hospitalizations for exacerbations of congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD) and diabetic ketoacidosis (DKA) are costly in the United States. The purpose of this study was to predict in-hospital charges for each condition using machine learning (ML) models. Results: We conducted a retrospective cohort study on national discharge records of hospitalized adult patients from January 1st, 2016, to December 31st, 2019. We used numerous ML techniques to predict in-hospital total cost. We found that linear regression (LM), gradient boosting (GBM) and extreme gradient boosting (XGB) models had good predictive performance and were statistically equivalent, with training R-square values ranging from 0.49-0.95 for CHF, 0.56-0.95 for COPD, and 0.32-0.99 for DKA. We identified important key features driving costs, including patient age, length of stay, number of procedures. and elective/nonelective admission. Conclusions: ML methods may be used to accurately predict costs and identify drivers of high cost for COPD exacerbations, CHF exacerbations and DKA. Overall, our findings may inform future studies that seek to decrease the underlying high patient costs for these conditions.

Correction: Publicly available data reveals association between asthma hospitalizations and unconventional natural gas development in Pennsylvania.

[This corrects the article DOI: 10.1371/journal.pone.0265513.].

Identifying clusters of patient comorbidities associated with obstructive sleep apnea using electronic health records.

STUDY OBJECTIVES: The objectives of this study were to understand the relative comorbidity burden of obstructive sleep apnea (OSA), determine whether these relationships were modified by sex or age, and identify patient subtypes defined by common comorbidities. METHODS: Cases with OSA and noncases (controls) were defined using a validated electronic health record (EHR)-based phenotype and matched for age, sex, and time period of follow-up in the EHR. We compared prevalence of the 20 most common comorbidities between matched cases and controls using conditional logistic regression with and without controlling for body mass index. Latent class analysis was used to identify subtypes of OSA cases defined by combinations of these comorbidities. RESULTS: In total, 60,586 OSA cases were matched to 60,586 controls (from 1,226,755 total controls). Patients with OSA were more likely to have each of the 20 most common comorbidities compared with controls, with odds ratios ranging from 3.1 to 30.8 in the full matched set and 1.3 to 10.2 after body mass index adjustment. Associations between OSA and these comorbidities were generally stronger in females and patients with younger age at diagnosis. We identified 5 distinct subgroups based on EHR-defined comorbidities: High Comorbidity Burden, Low Comorbidity Burden, Cardiovascular Comorbidities, Inflammatory Conditions and Less Obesity, and Inflammatory Conditions and Obesity. CONCLUSIONS: Our study demonstrates the power of leveraging the EHR to understand the relative health burden of OSA, as well as heterogeneity in these relationships based on age and sex. In addition to enrichment for comorbidities, we identified 5 novel OSA subtypes defined by combinations of comorbidities in the EHR, which may be informative for understanding disease outcomes and improving prevention and clinical care. Overall, this study adds more evidence that OSA is heterogeneous and requires personalized management. CITATION: Te TT, Keenan BT, Veatch OJ, Boland MR, Hubbard RA, Pack AI. Identifying clusters of patient comorbidities associated with obstructive sleep apnea using electronic health records. J Clin Sleep Med. 2024;20(4):521-533.

An Integrated Model for Patient Care and Clinical Trials (IMPACT) to support clinical research visit scheduling workflow for future learning health systems.

We describe a clinical research visit scheduling system that can potentially coordinate clinical research visits with patient care visits and increase efficiency at clinical sites where clinical and research activities occur simultaneously. Participatory Design methods were applied to support requirements engineering and to create this software called Integrated Model for Patient Care and Clinical Trials (IMPACT). Using a multi-user constraint satisfaction and resource optimization algorithm, IMPACT automatically synthesizes temporal availability of various research resources and recommends the optimal dates and times for pending research visits. We conducted scenario-based evaluations with 10 clinical research coordinators (CRCs) from diverse clinical research settings to assess the usefulness, feasibility, and user acceptance of IMPACT. We obtained qualitative feedback using semi-structured interviews with the CRCs. Most CRCs acknowledged the usefulness of IMPACT features. Support for collaboration within research teams and interoperability with electronic health records and clinical trial management systems were highly requested features. Overall, IMPACT received satisfactory user acceptance and proves to be potentially useful for a variety of clinical research settings. Our future work includes comparing the effectiveness of IMPACT with that of existing scheduling solutions on the market and conducting field tests to formally assess user adoption.

Discovering medical conditions associated with periodontitis using linked electronic health records.

AIM: To use linked electronic medical and dental records to discover associations between periodontitis and medical conditions independent of a priori hypotheses. MATERIALS AND METHODS: This case-control study included 2475 patients who underwent dental treatment at the College of Dental Medicine at Columbia University and medical treatment at NewYork-Presbyterian Hospital. Our cases are patients who received periodontal treatment and our controls are patients who received dental maintenance but no periodontal treatment. Chi-square analysis was performed for medical treatment codes and logistic regression was used to adjust for confounders. RESULTS: Our method replicated several important periodontitis associations in a largely Hispanic population, including diabetes mellitus type I (OR = 1.6, 95% CI 1.30-1.99, p < 0.001) and type II (OR = 1.4, 95% CI 1.22-1.67, p < 0.001), hypertension (OR = 1.2, 95% CI 1.10-1.37, p < 0.001), hypercholesterolaemia (OR = 1.2, 95% CI 1.07-1.38, p = 0.004), hyperlipidaemia (OR = 1.2, 95% CI 1.06-1.43, p = 0.008) and conditions pertaining to pregnancy and childbirth (OR = 2.9, 95% CI: 1.32-7.21, p = 0.014). We also found a previously unreported association with benign prostatic hyperplasia (OR = 1.5, 95% CI 1.05-2.10, p = 0.026) after adjusting for age, gender, ethnicity, hypertension, diabetes, obesity, lipid and circulatory system conditions, alcohol and tobacco abuse. CONCLUSIONS: This study contributes a high-throughput method for associating periodontitis with systemic diseases using linked electronic records.

Investigating Three Classification Methods for Per/Poly-Fluoroalkyl Substance (PFAS) Exposure from Electronic Health Records And Potential for Bias.

Per-/poly-fluoroalkyl substances (PFAS) are a group of manmade compounds with known human toxicity and evidence of contamination in drinking water throughout the US. We augmented our electronic health record data with geospatial information to classify PFAS exposure for our patients living in New Jersey. We explored the utility of three different methods for classifying PFAS exposure that are popularly used in the literature, resulting in different boundary types: public water supplier service area boundary, municipality, and ZIP code. We also explored the intersection of the three boundaries. To study the potential for bias, we investigated known PFAS exposure-disease associations, specifically hypertension, thyroid disease and parathyroid disease. We found that both the significance of the associations and the effect size varied by the method for classifying PFAS exposure. This has important implications in knowledge discovery and also environmental justice as across cohorts, we found a larger proportion of Black/African-American patients PFAS-exposed.

Risk of severe acute respiratory syndrome coronavirus 2 infection among women with polycystic ovary syndrome.

OBJECTIVE: To determine whether women with polycystic ovary syndrome (PCOS) had a higher incidence of testing positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) than those without PCOS and evaluate whether PCOS diagnosis independently increased the risk of moderate or severe disease in those with positive SARS-CoV-2 test results. DESIGN: Retrospective cohort study using the National COVID Cohort Collaborative (N3C). SETTING: National COVID Cohort Collaborative. PATIENT(S): Adult nonpregnant women (age, 18-65 years) enrolled in the N3C with confirmed SARS-CoV-2 testing for any indication. Sensitivity analyses were conducted in women aged 18-49 years and who were obese (body mass index, ≥30 kg/m2). INTERVENTION(S): The exposure was PCOS as identified by the N3C clinical diagnosis codes and concept sets, which are a compilation of terms, laboratory values, and International Classification of Diseases codes for the diagnosis of PCOS. To further capture patients with the symptoms of PCOS, we also included those who had concept sets for both hirsutism and irregular menses. MAIN OUTCOME MEASURE(S): Odds of testing positive for SARS-CoV-2 and odds of moderate or severe coronavirus disease 2019 (COVID-19) in the PCOS cohort compared with those in the non-PCOS cohort. RESULT(S): Of the 2,089,913 women included in our study, 39,459 had PCOS. In the overall cohort, the adjusted odds ratio (aOR) of SARS-CoV-2 positivity was 0.98 (95% confidence interval [CI], 0.97-0.98) in women with PCOS compared to women without PCOS. The aORs of disease severity were as follows: mild disease, 1.02 (95% CI, 1.01-1.03); moderate disease, 0.99 (95% CI, 0.98-1.00); and severe disease, 0.99 (95% CI, 0.99-1.00). There was no difference in COVID-19-related mortality (aOR, 1.00; 95% CI, 0.99-1.00). These findings were similar in the reproductive-age and obese reproductive-age cohorts. CONCLUSION(S): Women with PCOS had a similar likelihood of testing positive for SARS-CoV-2. Among those who tested positive, they were no more likely to have moderate or severe COVID-19 than the non-PCOS cohort. Polycystic ovary syndrome is a chronic condition associated with several comorbidities, including cardiovascular disease and mental health issues. Although these comorbidities are also associated with COVID-19 morbidity, our findings suggest that the comorbidities themselves, rather than PCOS, drive the risk of disease severity.

A deep learning method to detect opioid prescription and opioid use disorder from electronic health records.

OBJECTIVE: As the opioid epidemic continues across the United States, methods are needed to accurately and quickly identify patients at risk for opioid use disorder (OUD). The purpose of this study is to develop two predictive algorithms: one to predict opioid prescription and one to predict OUD. MATERIALS AND METHODS: We developed an informatics algorithm that trains two deep learning models over patient Electronic Health Records (EHRs) using the MIMIC-III database. We utilize both the structured and unstructured parts of the EHR and show that it is possible to predict both challenging outcomes. RESULTS: Our deep learning models incorporate elements from EHRs to predict opioid prescription with an F1-score of 0.88 ± 0.003 and an AUC-ROC of 0.93 ± 0.002. We also constructed a model to predict OUD diagnosis achieving an F1-score of 0.82 ± 0.05 and AUC-ROC of 0.94 ± 0.008. DISCUSSION: Our model for OUD prediction outperformed prior algorithms for specificity, F1 score and AUC-ROC while achieving equivalent sensitivity. This demonstrates the importance of a) deep learning approaches in predicting OUD and b) incorporating both structured and unstructured data for this prediction task. No prediction models for opioid prescription as an outcome were found in the literature and therefore our model is the first to predict opioid prescribing behavior. CONCLUSION: Algorithms such as those described in this paper will become increasingly important to understand the drivers underlying this national epidemic.

Assessing racial residential segregation as a risk factor for severe maternal morbidity.

PURPOSE: To measure associations of area-level racial and economic residential segregation with severe maternal morbidity (SMM). METHODS: We conducted a retrospective cohort study of births at two Philadelphia hospitals between 2018 and 2020 to analyze associations of segregation, quantified using the Index of Concentration at the Extremes (ICE), with SMM. We used stratified multivariable, multilevel, logistic regression models to determine whether associations of ICE with SMM varied by self-identified race or hospital catchment. RESULTS: Of the 25,979 patients (44.1% Black, 35.8% White), 1381 (5.3%) had SMM (Black [6.1%], White [4.4%]). SMM was higher among patients residing outside (6.3%), than inside (5.0%) Philadelphia (P < .001). Overall, ICE was not associated with SMM. However, ICErace (higher proportion of White vs. Black households) was associated with lower odds of SMM among patients residing inside Philadelphia (aOR 0.87, 95% CI: 0.80-0.94) and higher odds outside Philadelphia (aOR 1.12, 95% CI: 0.95-1.31). Moran's I indicated spatial autocorrelation of SMM overall (P < .001); when stratified, autocorrelation was only evident outside Philadelphia. CONCLUSIONS: Overall, ICE was not associated with SMM. However, higher ICErace was associated with lower odds of SMM among Philadelphia residents. Findings highlight the importance of hospital catchment area and referral patterns in spatial analyses of hospital datasets.

An algorithm to identify residential mobility from electronic health-record data.

BACKGROUND: Environmental, social and economic exposures can be inferred from address information recorded in an electronic health record. However, these data often contain administrative errors and misspellings. These issues make it challenging to determine whether a patient has moved, which is integral for accurate exposure assessment. We aim to develop an algorithm to identify residential mobility events and avoid exposure misclassification. METHODS: At Penn Medicine, we obtained a cohort of 12 147 pregnant patients who delivered between 2013 and 2017. From this cohort, we identified 9959 pregnant patients with address information at both time of delivery and one year prior. We developed an algorithm entitled REMAP (Relocation Event Moving Algorithm for Patients) to identify residential mobility during pregnancy and compared it to using ZIP code differences alone. We assigned an area-deprivation exposure score to each address and assessed how residential mobility changed the deprivation scores. RESULTS: To assess the accuracy of our REMAP algorithm, we manually reviewed 3362 addresses and found that REMAP was 95.7% accurate. In this large urban cohort, 41% of patients moved during pregnancy. REMAP outperformed the comparison of ZIP codes alone (82.9%). If residential mobility had not been taken into account, absolute area deprivation would have misclassified 39% of the patients. When setting a threshold of one quartile for misclassification, 24.4% of patients would have been misclassified. CONCLUSIONS: Our study tackles an important characterization problem for exposures that are assigned based upon residential addresses. We demonstrate that methods using ZIP code alone are not adequate. REMAP allows address information from electronic health records to be used for accurate exposure assessment and the determination of residential mobility, giving researchers and policy makers more reliable information.