Targeted Learning with an Undersmoothed Lasso Propensity Score Model for Large-Scale Covariate Adjustment in Healthcare Database Studies.

Lasso regression is widely used for large-scale propensity score (PS) estimation in healthcare database studies. In these settings, previous work has shown that undersmoothing (overfitting) Lasso PS models can improve confounding control, but it can also cause problems of non-overlap in covariate distributions. It remains unclear how to select the degree of undersmoothing when fitting large-scale Lasso PS models to improve confounding control while avoiding issues that can result from reduced covariate overlap. Here, we used simulations to evaluate the performance of using collaborative-controlled targeted learning to data-adaptively select the degree of undersmoothing when fitting large-scale PS models within both singly and doubly robust frameworks to reduce bias in causal estimators. Simulations showed that collaborative learning can data-adaptively select the degree of undersmoothing to reduce bias in estimated treatment effects. Results further showed that when fitting undersmoothed Lasso PS-models, the use of cross-fitting was important for avoiding non-overlap in covariate distributions and reducing bias in causal estimates.

Prevalence of Pretreatment Testing Recommended for Patients With Chronic Inflammatory Skin Diseases.

Importance: Laboratory testing for the presence of tuberculosis, hepatitis, and other conditions before starting most systemic immunomodulatory agents is recommended in patients with chronic inflammatory skin diseases (CISD) but current testing patterns in the US are unclear. Objective: To determine the prevalence of pretreatment testing that is recommended for patients with CISD (psoriasis, hidradenitis suppurativa, or atopic dermatitis). Design, Setting, and Participants: This descriptive analysis of US commercial insurance claims databases from December 31, 2002, to December 31, 2020, included adult patients with CISD (psoriasis, hidradenitis suppurativa, or atopic dermatitis) who started an immunomodulatory agent, including methotrexate, tumor necrosis factor α inhibitors, interleukin (IL)-17Ai, ustekinumab, IL-23i, dupilumab, or apremilast. Main Outcomes and Measures: The proportion of patients who underwent the screening tests as suggested by professional societies-including for tuberculosis, hepatitis, and liver function; complete blood cell counts; and lipid panels-were determined within 6 months before and during 2 years after treatment start. Results: A total of 122 308 patients with CISDs (median [IQR] age, 49 [38-58] years; 63 663 [52.1%] male) starting systemic immunomodulatory treatment in the US were included. Treatment for patients with CISDs comprised methotrexate (28 684), tumor necrosis factor α inhibitors (40 965), ustekinumab (12 841), IL-23i (6116), IL-17Ai (9799), dupilumab (7787), or apremilast (16 116). Complete blood cell count was the most common test, performed in 41% (3161/7787) to 69% (19 659/28 684) of individuals before initiation across treatments. Between 11% (889/7787) and 59% (3613/6116) of patients had tuberculosis screening within 6 months before treatment, and 3% (149/4577) to 26% (1559/6097) had updated tests 1 year later. Between 13% (1006/7787) and 41% (16 728/40 965) had hepatitis panels before treatment. Low pretreatment testing levels before apremilast (15% [2331/16 116] to 45% [7253/16 116]) persisted a year into treatment (9% [816/8496] to 36% [2999/8496]) and were similar to dupilumab (11% [850/7787] to 41% [3161/7787] vs 3% [149/4577] to 25% [1160/4577]). Conclusions and Relevance: In this descriptive analysis of patients with CISDs starting systemic immunomodulatory treatment in the US, less than 60% received the recommended pretreatment testing. Additional research is required to understand whether variations in testing affect patient outcomes.

Characteristics of Patients Receiving Novel Muscular Dystrophy Drugs in Trials vs Routine Care.

Importance: The US Food and Drug Administration approved eteplirsen for Duchenne muscular dystrophy (DMD) in 2016 based on a controversial pivotal study that demonstrated a limited effect on the surrogate measure of dystrophin production. Other DMD treatments in the same class followed. Objective: To assess how patients receiving novel DMD treatments in postapproval clinical settings compare with patients in the clinical trials. Design, Setting, and Participants: This cross-sectional study collected data on patients who initiated 1 of 4 novel DMD treatments (eteplirsen, golodirsen, viltolarsen, and casimersen) using national claims databases of commercially insured (Merative MarketScan and Optum's Clinformatics Data Mart Database [CDM]) and Medicaid patients between September 19, 2016, and March 31, 2022. Patients were followed for 1 year after the date of first use of any novel DMD treatment. In addition, patients in pivotal DMD drug trials were identified for comparison. Exposures: Age, sex, race and ethnicity, region, and DMD stage of patients receiving novel DMD treatment. Main Outcome and Measures: The main outcome was health care costs and drug discontinuation as measured using descriptive statistics. Results: A total of 223 routine care patients initiating novel DMD drugs (58 in MarketScan, 35 in CDM, and 130 in Medicaid) were identified. Among the 106 patients in the pivotal trials, the mean (SD) age was 8.5 (2.0) years (range, 4.0-13.0 years), which was younger than the mean age of patients in routine care (MarketScan: 13.7 [7.0] years [range, 1.8-33.3 years; P < .001]; CDM: 11.9 [5.7] years [range, 0.6-23.6 years; P < .001]; Medicaid: 13.4 [6.5] years [range, 1.8-46.1 years; P < .001]). The proportion of female patients identified in postapproval clinical settings was 2.9% (n = 1) in CDM (vs 34 male patients [97.1%]) and 1.5% (n = 2) in Medicaid (vs 128 male patients [98.5%]), which was not different from the pivotal trials. While nearly all patients in the pivotal trials had DMD disease stage 1 or 2 when initiating the DMD treatments (103 [97.2%]), in the postapproval clinical setting, slightly more than one-third of patients were in disease stage 3 or 4 (MarketScan, 17 [36.2%; P < .001]; CDM, 13 [41.9%; P < .001]; Medicaid, 54 [47.0%; P < .001]). The payer's cost for novel DMD treatments varied across the databases, with a mean (SD) of $634 764 ($607 101) in MarketScan, $482 749 ($582 350) in CDM, and $384 023 ($1 165 730) in Medicaid. Approximately one-third of routine care patients discontinued the treatments after approximately 7 months (mean [SD], 6.1 [4.4], 6.9 [3.9], and 7.2 [4.3] months in MarketScan, CDM, and Medicaid, respectively). Conclusions and Relevance: These findings raise questions about the translation of DMD drug trial findings to routine care settings, with patients in routine care discontinuing the treatment within 1 year and payers incurring substantial expenses for these medications. More data are needed on whether these high costs are accompanied by corresponding clinical benefits.

Prenatal Exposure to Opioids and Neurodevelopmental Disorders in Children: A Bayesian Mediation Analysis.

This study explores natural direct and joint natural indirect effects (JNIE) of prenatal opioid exposure on neurodevelopmental disorders (NDDs) in children mediated through pregnancy complications, major and minor congenital malformations, and adverse neonatal outcomes, using Medicaid claims linked to vital statistics in Rhode Island, United States, 2008-2018. A Bayesian mediation analysis with elastic net shrinkage prior was developed to estimate mean time to NDD diagnosis ratio using posterior mean and 95% credible intervals (CrIs) from Markov chain Monte Carlo algorithms. Simulation studies showed desirable model performance. Of 11,176 eligible pregnancies, 332 had ≥2 dispensations of prescription opioids anytime during pregnancy, including 200 (1.8%) having ≥1 dispensation in the first trimester (T1), 169 (1.5%) in the second (T2), and 153 (1.4%) in the third (T3). A significant JNIE of opioid exposure was observed in each trimester (T1, JNIE = 0.97, 95% CrI: 0.95, 0.99; T2, JNIE = 0.97, 95% CrI: 0.95, 0.99; T3, JNIE = 0.96, 95% CrI: 0.94, 0.99). The proportion of JNIE in each trimester was 17.9% (T1), 22.4% (T2), and 56.3% (T3). In conclusion, adverse pregnancy and birth outcomes jointly mediated the association between prenatal opioid exposure and accelerated time to NDD diagnosis. The proportion of JNIE increased as the timing of opioid exposure approached delivery.

Development and Validation of a Claims-Based Model to Predict Categories of Obesity.

We developed and validated a claims-based algorithm that classifies patients into obesity categories. Using Medicare (2007-2017) and Medicaid (2000-2014) claims data linked to 2 electronic health record (EHR) systems in Boston, Massachusetts, we identified a cohort of patients with an EHR-based body mass index (BMI) measurement (calculated as weight (kg)/height (m)2). We used regularized regression to select from 137 variables and built generalized linear models to classify patients with BMIs of ≥25, ≥30, and ≥40. We developed the prediction model using EHR system 1 (training set) and validated it in EHR system 2 (validation set). The cohort contained 123,432 patients in the Medicare population and 40,736 patients in the Medicaid population. The model comprised 97 variables in the Medicare set and 95 in the Medicaid set, including BMI-related diagnosis codes, cardiovascular and antidiabetic drugs, and obesity-related comorbidities. The areas under the receiver-operating-characteristic curve in the validation set were 0.72, 0.75, and 0.83 (Medicare) and 0.66, 0.66, and 0.70 (Medicaid) for BMIs of ≥25, ≥30, and ≥40, respectively. The positive predictive values were 81.5%, 80.6%, and 64.7% (Medicare) and 81.6%, 77.5%, and 62.5% (Medicaid), for BMIs of ≥25, ≥30, and ≥40, respectively. The proposed model can identify obesity categories in claims databases when BMI measurements are missing and can be used for confounding adjustment, defining subgroups, or probabilistic bias analysis.

Iterative Causal Forest: A Novel Algorithm for Subgroup Identification.

Precisely and efficiently identifying subgroups with heterogeneous treatment effects (HTEs) in real-world evidence studies remains a challenge. Based on the causal forest (CF) method, we developed an iterative CF (iCF) algorithm to identify HTEs in subgroups defined by important variables. Our method iteratively grows different depths of the CF with important effect modifiers, performs plurality votes to obtain decision trees (subgroup decisions) for a family of CFs with different depths, then finds the cross-validated subgroup decision that best predicts the treatment effect as a final subgroup decision. We simulated 12 different scenarios and showed that the iCF outperformed other machine learning methods for interaction/subgroup identification in the majority of scenarios assessed. Using a 20% random sample of fee-for-service Medicare beneficiaries initiating sodium-glucose cotransporter-2 inhibitors (SGLT2i) or glucagon-like peptide-1 receptor agonists (GLP1RA), we implemented the iCF to identify subgroups with HTEs for hospitalized heart failure. Consistent with previous studies suggesting patients with heart failure benefit more from SGLT2i, iCF successfully identified such a subpopulation with HTEs and additive interactions. The iCF is a promising method for identifying subgroups with HTEs in real-world data where the potential for unmeasured confounding can be limited by study design.

An application of the Causal Roadmap in two safety monitoring case studies: Causal inference and outcome prediction using electronic health record data.

Background: Real-world data, such as administrative claims and electronic health records, are increasingly used for safety monitoring and to help guide regulatory decision-making. In these settings, it is important to document analytic decisions transparently and objectively to assess and ensure that analyses meet their intended goals. Methods: The Causal Roadmap is an established framework that can guide and document analytic decisions through each step of the analytic pipeline, which will help investigators generate high-quality real-world evidence. Results: In this paper, we illustrate the utility of the Causal Roadmap using two case studies previously led by workgroups sponsored by the Sentinel Initiative - a program for actively monitoring the safety of regulated medical products. Each case example focuses on different aspects of the analytic pipeline for drug safety monitoring. The first case study shows how the Causal Roadmap encourages transparency, reproducibility, and objective decision-making for causal analyses. The second case study highlights how this framework can guide analytic decisions beyond inference on causal parameters, improving outcome ascertainment in clinical phenotyping. Conclusion: These examples provide a structured framework for implementing the Causal Roadmap in safety surveillance and guide transparent, reproducible, and objective analysis.

A causal roadmap for generating high-quality real-world evidence.

Increasing emphasis on the use of real-world evidence (RWE) to support clinical policy and regulatory decision-making has led to a proliferation of guidance, advice, and frameworks from regulatory agencies, academia, professional societies, and industry. A broad spectrum of studies use real-world data (RWD) to produce RWE, ranging from randomized trials with outcomes assessed using RWD to fully observational studies. Yet, many proposals for generating RWE lack sufficient detail, and many analyses of RWD suffer from implausible assumptions, other methodological flaws, or inappropriate interpretations. The Causal Roadmap is an explicit, itemized, iterative process that guides investigators to prespecify study design and analysis plans; it addresses a wide range of guidance within a single framework. By supporting the transparent evaluation of causal assumptions and facilitating objective comparisons of design and analysis choices based on prespecified criteria, the Roadmap can help investigators to evaluate the quality of evidence that a given study is likely to produce, specify a study to generate high-quality RWE, and communicate effectively with regulatory agencies and other stakeholders. This paper aims to disseminate and extend the Causal Roadmap framework for use by clinical and translational researchers; three companion papers demonstrate applications of the Causal Roadmap for specific use cases.

The impact of electronic health record discontinuity on prediction modeling.

BACKGROUND: To determine the impact of electronic health record (EHR)-discontinuity on the performance of prediction models. METHODS: The study population consisted of patients with a history of cardiovascular (CV) comorbidities identified using US Medicare claims data from 2007 to 2017, linked to EHR from two networks (used as model training and validation set, respectively). We built models predicting one-year risk of mortality, major CV events, and major bleeding events, stratified by high vs. low algorithm-predicted EHR-continuity. The best-performing models for each outcome were chosen among 5 commonly used machine-learning models. We compared model performance by Area under the ROC curve (AUROC) and Area under the precision-recall curve (AUPRC). RESULTS: Based on 180,950 in the training and 103,061 in the validation set, we found EHR captured only 21.0-28.1% of all the non-fatal outcomes in the low EHR-continuity cohort but 55.4-66.1% of that in the high EHR-continuity cohort. In the validation set, the best-performing model developed among high EHR-continuity patients had consistently higher AUROC than that based on low-continuity patients: AUROC was 0.849 vs. 0.743 when predicting mortality; AUROC was 0.802 vs. 0.659 predicting the CV events; AUROC was 0.635 vs. 0.567 predicting major bleeding. We observed a similar pattern when using AUPRC as the outcome metric. CONCLUSIONS: Among patients with CV comorbidities, when predicting mortality, major CV events, and bleeding outcomes, the prediction models developed in datasets with low EHR-continuity consistently had worse performance compared to models developed with high EHR-continuity.

Venous Thromboembolism in Patients With Bullous Pemphigoid.

Importance: Studies have linked bullous pemphigoid (BP) with venous thromboembolism (VTE) across several data sources finding 6-fold to 15-fold increased incidence rates. Objective: To determine the incidence of VTE in patients with BP compared with similar controls. Design, Setting, and Participants: This cohort study used insurance claims data from a nationwide US health care database from January 1, 2004, through January 1, 2020. Patients with dermatologist-recorded BP were identified (≥2 diagnoses of BP [International Classification of Diseases, Ninth Revision (ICD-9) 694.5 and ICD-10 L12.0] recorded by dermatologists within 1 year). Risk-set sampling identified comparator patients without BP and free of other chronic inflammatory skin diseases. Patients were followed-up until the first of the following events occurred: VTE, death, disenrollment, or end of data stream. Exposures: Patients with BP compared with those without BP and free of other chronic inflammatory skin diseases (CISD). Main outcome: Venous thromboembolism events were identified and incidence rates were computed before and after propensity-score (PS) matching to account for VTE risk factors. Hazard ratios (HRs) compared the incidence of VTE in BP vs non-CISD. Results: Overall, 2654 patients with BP and 26 814 comparator patients without BP or another CISD were identified. The mean (SD) age in the BP group was 73.0 (12.6) years and 55.0 (18.9) years in the non-CSID group. With a median follow-up time was 2 years, the unadjusted incidence rate (per 1000 person-years) of outpatient or inpatient VTE was 8.5 in the BP group compared with 1.8 in patients without a CISD. Adjusted rates were 6.7 in the BP group compared with 3.0 in the non-CISD group. Age-specific adjusted incidence rates (per 1000 person-years) in patients aged 50 to 74 years was 6.0 (vs 2.9 in the non-CISD group) and in those aged 75 years or older was 7.1 (vs 4.53 in the non-CISD group). After 1:1 propensity-score matching including 60 VTE risk factors and severity markers, BP was associated with a 2-fold increased risk of VTE (2.24 [1.26-3.98]) vs those in the non-CISD group. When restricting to patients aged 50 years or older, the adjusted relative risk of VTE was 1.82 (1.05-3.16) for the BP vs non-CISD groups. Conclusions: In this nationwide US cohort study, BP was associated with a 2-fold increased incidence of VTE after controlling for VTE risk factors in a dermatology patient population.

A Novel Chronic Kidney Disease Phenotyping Algorithm Using Combined Electronic Health Record and Claims Data.

Purpose: Because chronic kidney disease (CKD) is often under-coded as a diagnosis in claims data, we aimed to develop claims-based prediction models for CKD phenotypes determined by laboratory results in electronic health records (EHRs). Patients and Methods: We linked EHR from two networks (used as training and validation cohorts, respectively) with Medicare claims data. The study cohort included individuals ≥65 years with a valid serum creatinine result in the EHR from 2007 to 2017, excluding those with end-stage kidney disease or on dialysis. We used LASSO regression to select among 134 predictors for predicting continuous estimated glomerular filtration rate (eGFR). We assessed the model performance when predicting eGFR categories of <60, <45, <30 mL/min/1.73m2 in terms of area under the receiver operating curves (AUC). Results: The model training cohort included 117,476 patients (mean age 74.8 years, female 58.2%) and the validation cohort included 56,744 patients (mean age 73.8 years, female 59.6%). In the validation cohort, the AUC of the primary model (with 113 predictors and an adjusted R2 of 0.35) for predicting eGFR <60, eGFR<45, and eGFR <30 mL/min/1.73m2 categories was 0.81, 0.88, and 0.92, respectively, and the corresponding positive predictive values for these 3 phenotypes were 0.80 (95% confidence interval: 0.79, 0.81), 0.79 (0.75, 0.84), and 0.38 (0.30, 0.45), respectively. Conclusion: We developed a claims-based model to determine clinical phenotypes of CKD stages defined by eGFR values. Researchers without access to laboratory results can use the model-predicted phenotypes as a proxy clinical endpoint or confounder and to enhance subgroup effect assessment.

Development and Validation of a Novel Tool to Predict Model for End-Stage Liver Disease (MELD) Scores in Cirrhosis, Using Administrative Datasets.

Background: The Model for End-Stage Liver Disease (MELD) score predicts disease severity and mortality in cirrhosis. To improve cirrhosis phenotyping in administrative databases lacking laboratory data, we aimed to develop and externally validate claims-based MELD prediction models, using claims data linked to electronic health records (EHR). Methods: We included adults with established cirrhosis in two Medicare-linked EHR networks (training and internal validation; 2007-2017), and a Medicaid-linked EHR network (external validation; 2000-2014). Using least absolute shrinkage and selection operator (LASSO) with 5-fold cross-validation, we selected among 146 investigator-specified variables to develop models for predicting continuous MELD and relevant MELD categories (MELD<10, MELD≥15 and MELD≥20), with observed MELD calculated from laboratory data. Regression coefficients for each model were applied to the validation sets to predict patient-level MELD and assess model performance. Results: We identified 4501 patients in the Medicare training set (mean age 75.1 years, 18.5% female, mean MELD=13.0), and 2435 patients in the Medicare validation set (mean age: 74.3 years, 31.7% female, mean MELD=12.3). Our final model for predicting continuous MELD included 112 variables, explaining 58% of observed MELD variability; in the Medicare validation set, the area-under-the-receiver operating characteristic curves (AUC) for MELD<10 and MELD≥15 were 0.84 and 0.90, respectively; the AUC for the model predicting MELD≥20 (using 27 variables) was 0.93. Overall, these models correctly classified 77% of patients with MELD<10 (95% CI=0.75-0.78), 85% of patients with MELD≥15 (95% CI=0.84-0.87), and 87% of patients with MELD≥20 (95% CI=0.86-0.88). Results were consistent in the external validation set (n=2240). Conclusion: Our MELD prediction tools can be used to improve cirrhosis phenotyping in administrative datasets lacking laboratory data.

Risk of Infection in Children With Psoriasis Receiving Treatment With Ustekinumab, Etanercept, or Methotrexate Before and After Labeling Expansion.

Importance: Psoriasis in children is increasingly treated with systemic medications, yet their risk of serious infection is not well characterized in clinical practice. Pediatric clinical trials for these medications were often small and placebo controlled. Objective: To estimate the 6-month rate of infections among children with psoriasis who started treatment with ustekinumab, etanercept, or methotrexate. Design, Setting, and Participants: This cohort study used insurance claims data from clinical practices across the US on children aged 17 years or younger with psoriasis who were receiving treatment with a topical medication for psoriasis and started new treatment with ustekinumab, etanercept, or methotrexate. The analysis was stratified by the time before pediatric labeling (2009-2015) and after pediatric approval (2016-2021). Patient follow-up started 1 day after initiating treatment and ended at 6 months. Exposures: New treatment with ustekinumab, etanercept, and methotrexate. Main Outcomes and Measures: During follow-up, the frequency of inpatient serious infections and outpatient infections requiring treatment was compared. Event rates and rate ratios were estimated after propensity score decile stratification. Results: After exclusions, we identified 2338 patients (1368 girls [57.8%]) who initiated new treatment with a targeted immunomodulating agent. In all, 379 patients started treatment with ustekinumab, 779 patients started treatment with etanercept, and 1180 patients started treatment with methotrexate from 2009 through 2021. The propensity score-adjusted incidence rate of serious infection was 18.4 per 1000 person-years (3 events) for ustekinumab users, 25.6 per 1000 person-years (9 events) for etanercept users, and 14.9 per 1000 person-years (8 events) for methotrexate users. The adjusted rate of outpatient infections was 254.9 per 1000 person-years (39 events) for ustekinumab users, 435.7 per 1000 person-years (139 events) for etanercept users, and 433.6 per 1000 person-years (209 events) for methotrexate users. The adjusted rate ratio of outpatient infections was 0.58 (95% CI, 0.41-0.83) for ustekinumab vs etanercept, 0.66 (95% CI, 0.48-0.91) for ustekinumab vs methotrexate, and 0.95 (95% CI, 0.75-1.21) for etanercept vs methotrexate. Rate ratios were similar during the off-label use era and after pediatric labeling. Conclusions and Relevance: Among children with psoriasis who started treatment with immunomodulating agents, serious infections were infrequent. This cohort study suggests that there was no increase in the risk of outpatient infections for children who started treatment with ustekinumab compared with etanercept or methotrexate.

Scalable Feature Engineering from Electronic Free Text Notes to Supplement Confounding Adjustment of Claims-Based Pharmacoepidemiologic Studies.

Natural language processing (NLP) tools turn free-text notes (FTNs) from electronic health records (EHRs) into data features that can supplement confounding adjustment in pharmacoepidemiologic studies. However, current applications are difficult to scale. We used unsupervised NLP to generate high-dimensional feature spaces from FTNs to improve prediction of drug exposure and outcomes compared with claims-based analyses. We linked Medicare claims with EHR data to generate three cohort studies comparing different classes of medications on the risk of various clinical outcomes. We used "bag-of-words" to generate features for the top 20,000 most prevalent terms from FTNs. We compared machine learning (ML) prediction algorithms using different sets of candidate predictors: Set1 (39 researcher-specified variables), Set2 (Set1 + ML-selected claims codes), and Set3 (Set1 + ML-selected NLP-generated features), vs. Set4 (Set1 + 2 + 3). When modeling treatment choice, we observed a consistent pattern across the examples: ML models utilizing Set4 performed best followed by Set2, Set3, then Set1. When modeling the outcome risk, there was little to no improvement beyond models based on Set1. Supplementing claims data with NLP-generated features from free text notes improved prediction of prescribing choices but had little or no improvement on clinical risk prediction. These findings have implications for strategies to improve confounding using EHR data in pharmacoepidemiologic studies.

Development and Validation of Algorithms to Estimate Live Birth Gestational Age in Medicaid Analytic eXtract Data.

BACKGROUND: While healthcare utilization data are useful for postmarketing surveillance of drug safety in pregnancy, the start of pregnancy and gestational age at birth are often incompletely recorded or missing. Our objective was to develop and validate a claims-based live birth gestational age algorithm. METHODS: Using the Medicaid Analytic eXtract (MAX) linked to birth certificates in three states, we developed four candidate algorithms based on: preterm codes; preterm or postterm codes; timing of prenatal care; and prediction models - using conventional regression and machine-learning approaches with a broad range of prespecified and empirically selected predictors. We assessed algorithm performance based on mean squared error (MSE) and proportion of pregnancies with estimated gestational age within 1 and 2 weeks of the gold standard, defined as the clinical or obstetric estimate of gestation on the birth certificate. We validated the best-performing algorithms against medical records in a nationwide sample. We quantified misclassification of select drug exposure scenarios due to estimated gestational age as positive predictive value (PPV), sensitivity, and specificity. RESULTS: Among 114,117 eligible pregnancies, the random forest model with all predictors emerged as the best performing algorithm: MSE 1.5; 84.8% within 1 week and 96.3% within 2 weeks, with similar performance in the nationwide validation cohort. For all exposure scenarios, PPVs were >93.8%, sensitivities >94.3%, and specificities >99.4%. CONCLUSIONS: We developed a highly accurate algorithm for estimating gestational age among live births in the nationwide MAX data, further supporting the value of these data for drug safety surveillance in pregnancy. See video abstract at, http://links.lww.com/EDE/B989 .

Occurrence of inflammatory bowel disease in patients with chronic inflammatory skin diseases: a cohort study: Classification: Epidemiology.

BACKGROUND: Several studies have linked various chronic inflammatory skin diseases (CISDs) with inflammatory bowel disease (IBD) in a range of data sources with mixed conclusions. OBJECTIVES: We compared the incidence of IBD - ulcerative colitis (UC) and Crohn disease (CD) - in patients with a CISD vs. similar persons without a CISD. METHODS: In this cohort study using nationwide, longitudinal, commercial insurance claims data from the USA, we identified adults and children who were seen by a dermatologist between 2004 and 2020, and diagnosed with either psoriasis, atopic dermatitis, alopecia areata, vitiligo or hidradenitis suppurativa. Comparator patients were identified through risk-set sampling; they were eligible if they were seen by a dermatologist at least twice and not diagnosed with a CISD. Patient follow-up lasted until either IBD diagnosis, death, disenrolment or end of data stream, whichever came first. IBD events, UC or CD, were identified via validated algorithms: hospitalization or diagnosis with endoscopic confirmation. Incidence rates were computed before and after adjustment via propensity-score decile stratification to account for IBD risk factors. Hazard ratios (HR) and 95% confidence intervals (CIs) were estimated to compare the incidence of IBD in CISD vs. non-CISD. RESULTS: We identified patients with atopic dermatitis (n = 123 614), psoriasis (n = 83 049), alopecia areata (n = 18 135), vitiligo (n = 9003) or hidradenitis suppurativa (n = 6806), and comparator patients without a CISD (n = 2 376 120). During a median follow-up time of 718 days, and after applying propensity-score adjustment for IBD risk factors, we observed increased risk of both UC (HRUC 2·30, 95% CI 1·61-3·28) and CD (HRCD 2·70, 1·69-4·32) in patients with hidradenitis suppurativa, an increased risk of CD (HRCD 1·23, 1·03-1·46) but not UC (HRUC 1·01, 0·89-1·14) in psoriasis, and no increased risk of IBD in atopic dermatitis (HRUC 1·02, 0·92-1·12; HRCD 1·08, 0·94-1·23), alopecia areata (HRUC 1·18, 0·89-1·56; HRCD 1·26, 0·86-1·86) or vitiligo (HRUC 1·14, 0·77-1·68; HRCD 1·45, 0·87-2·41). CONCLUSIONS: IBD was increased in patients with hidradenitis suppurativa. CD alone was increased in patients with psoriasis. Neither UC nor CD was increased in patients with atopic dermatitis, alopecia areata or vitiligo. What is already known about this topic? Several studies have linked various chronic inflammatory skin diseases (CISDs) with inflammatory bowel disease (IBD) utilizing a range of data sources, with mixed conclusions. What does this study add? This large-scale, claims-based cohort study expands current knowledge by providing background rates for IBD across multiple CISDs using consistent methods and within a single, nationally representative patient population. We observed a relative increased risk of IBD in patients with hidradenitis suppurativa, but the overall incidence rate difference of IBD was generally low. Crohn disease alone was significantly increased in patients with psoriasis, and neither ulcerative colitis nor Crohn disease was increased in patients with atopic dermatitis, vitiligo or alopecia areata.

Machine learning for improving high-dimensional proxy confounder adjustment in healthcare database studies: An overview of the current literature.

PURPOSE: Supplementing investigator-specified variables with large numbers of empirically identified features that collectively serve as 'proxies' for unspecified or unmeasured factors can often improve confounding control in studies utilizing administrative healthcare databases. Consequently, there has been a recent focus on the development of data-driven methods for high-dimensional proxy confounder adjustment in pharmacoepidemiologic research. In this paper, we survey current approaches and recent advancements for high-dimensional proxy confounder adjustment in healthcare database studies. METHODS: We discuss considerations underpinning three areas for high-dimensional proxy confounder adjustment: (1) feature generation-transforming raw data into covariates (or features) to be used for proxy adjustment; (2) covariate prioritization, selection, and adjustment; and (3) diagnostic assessment. We discuss challenges and avenues of future development within each area. RESULTS: There is a large literature on methods for high-dimensional confounder prioritization/selection, but relatively little has been written on best practices for feature generation and diagnostic assessment. Consequently, these areas have particular limitations and challenges. CONCLUSIONS: There is a growing body of evidence showing that machine-learning algorithms for high-dimensional proxy-confounder adjustment can supplement investigator-specified variables to improve confounding control compared to adjustment based on investigator-specified variables alone. However, more research is needed on best practices for feature generation and diagnostic assessment when applying methods for high-dimensional proxy confounder adjustment in pharmacoepidemiologic studies.

Synthetic Negative Controls: Using Simulation to Screen Large-scale Propensity Score Analyses.

The propensity score has become a standard tool to control for large numbers of variables in healthcare database studies. However, little has been written on the challenge of comparing large-scale propensity score analyses that use different methods for confounder selection and adjustment. In these settings, balance diagnostics are useful but do not inform researchers on which variables balance should be assessed or quantify the impact of residual covariate imbalance on bias. Here, we propose a framework to supplement balance diagnostics when comparing large-scale propensity score analyses. Instead of focusing on results from any single analysis, we suggest conducting and reporting results for many analytic choices and using both balance diagnostics and synthetically generated control studies to screen analyses that show signals of bias caused by measured confounding. To generate synthetic datasets, the framework does not require simulating the outcome-generating process. In healthcare database studies, outcome events are often rare, making it difficult to identify and model all predictors of the outcome to simulate a confounding structure closely resembling the given study. Therefore, the framework uses a model for treatment assignment to divide the comparator population into pseudo-treatment groups where covariate differences resemble those in the study cohort. The partially simulated datasets have a confounding structure approximating the study population under the null (synthetic negative control studies). The framework is used to screen analyses that likely violate partial exchangeability due to lack of control for measured confounding. We illustrate the framework using simulations and an empirical example.