Search

Tracey G Simon,1,2 Sebastian Schneeweiss,2 Richard Wyss,2 Zhigang Lu,2 Lily G Bessette,2 Cassandra York,2 Kueiyu Joshua Lin1,2 1Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; 2Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USACorrespondence: Kueiyu Joshua Lin, Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St. Suite 3030, Boston, MA, 02120, USA, Tel +1 617 278-0930, Fax +1 617 232-8602, Email jklin@bwh.harvard.eduBackground: 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.Keywords: cirrhosis, phenotyping, administrative data, claims

Santosh K Verma,1,2 Joanna Huang,3 Howard G Hutchinson,4 Irisdaly Estevez,1 Kammy Kuang,1 Shannon L Reynolds,1 Sebastian Schneeweiss1,5 1Science, Aetion, Inc, New York, NY, USA; 2Department of Family Medicine and Community Health, University of Massachusetts Medical School, Worcester, MA, USA; 3Payer Evidence, AstraZeneca, Wilmington, DE, USA; 4Global Medicines Development, AstraZeneca, Wilmington, DE, USA; 5Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USACorrespondence: Santosh K Verma, Director, Science, Aetion, Inc, 50 Congress Street, Suite 1025, Boston, MA, 02109, USA, Tel +1 401-765-8817, Email santosh.verma@aetion.comIntroduction: While serious liver injury among statin users is extremely rare, baseline liver enzyme testing is still recommended prior to initiating therapy. The benefit of such screening should be reevaluated based on empirical evidence. This study compared the risk of severe acute liver injury (SALI) between statin initiators with an elevated ALT (> 35U/L) matched to statin initiators with a normal ALT level (≤ 35U/L). Statin initiators with an elevated ALT were additionally compared against matched non-users.Methods: The study created cohorts from Optum and MarketScan claims data. Exposed and comparison cohorts were propensity score (PS) matched in each dataset and findings were pooled using meta-analysis. Proportional hazards regression was used to estimate hazard ratios (HRs), and a prespecified non-inferiority margin for SALI was set at a HR of 1.8.Results: 232,889 patients with elevated ALT were PS-matched to 232,889 with normal ALT level. The overall incidence rate of SALI was about 19/100,000 person-years among statin initiators. Statin initiators with elevated ALT had no meaningfully increased risk of SALI compared to those with normal ALT (HR=1.15; 95% CI 0.75 to 1.75). Comparing statin initiators with non-initiators with elevated ALT values equally yielded no increased risk (HR=0.76; 95% CI 0.52 to 1.11).Conclusion: In this large population-based study, SALI in statin users was rare. Importantly, the results showed no evidence that baseline ALT status is a reliable indicator for an increased risk of severe liver injury among statin initiators.Keywords: statins, liver injury, liver failure, alanine aminotransferase

David Merola,1,2 Sebastian Schneeweiss,1,2 Yinzhu Jin,1 Joyce Lii,1 Kueiyu Joshua Lin1,3 1Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 2Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; 3Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USACorrespondence: Kueiyu Joshua Lin, Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St. Suite 3030, Boston, MA, 02120, USA, Tel +1 617 278-0930, Fax +1 617 232-8602, Email jklin@bwh.harvard.eduBackground: Identifying high data-continuity patients in an electronic health record (EHR) system may facilitate selecting cohorts with a lower degree of variable misclassification and promote study validity. We updated a previously developed algorithm for identifying patients with high EHR data-completeness by adding demographic and health utilization factors to improve adaptability to networks serving patients of diverse backgrounds. We also expanded the algorithm to accommodate data in the ICD-10 era.Methods: We used Medicare claims linked with EHR data to identify individuals aged ≥ 65 years. EHR-continuity was defined as the proportion of encounters captured in EHR data relative to claims. We compared the model with additional demographic factors and their interaction terms with other predictors with the original algorithm and assessed the performance by area under the ROC curve (AUC) and net reclassification index (NRI).Results: The study cohort consisted of 264,099 subjects. The updated prediction model had an AUC of 0.93 in the validation set. Compared to the previous model, the new model had an NRI of 37.4% (p< 0.001) for EHR-continuity classification. Interaction terms between demographic variables and other predictors did not improve the performance. Patients within the top 20% of predicted EHR-continuity had four times less misclassification of key variables compared to the remaining population.Conclusion: Adding demographic and healthcare utilization variables significantly improved the model performance. Patients with high predicted EHR-continuity had less misclassification of study variables compared to the remaining population in both ICD-9 and 10 eras.Keywords: electronic medical records, comparative effectiveness research, information bias, data continuity

David Merola,1,2 Jessica Young,2,3 Deborah Schrag,4 Kueiyu Joshua Lin,1,2,5 Nicholas Robert,6 Sebastian Schneeweiss1,2 1Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 2Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; 3Department of Population Medicine, Harvard Medical School & Harvard Pilgrim Healthcare Institute, Boston, MA, USA; 4Department of Medicine, Memorial Sloan Kettering Cancer Center, Weill Cornell Medical School New York, New York, NY, USA; 5Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; 6Ontada, Irving, TX, USACorrespondence: David Merola, Email davemerola@gmail.comBackground: The use of electronic health records (EHR) data to assess drug effectiveness in clinical oncology practice is of great interest to regulators, clinicians, and payers. However, the utility of EHR data in clinical effectiveness studies may be limited by missing data, unmeasured confounding, and imperfect outcome surveillance. This study sought to emulate and compare the results of a randomized controlled trial investigating the efficacy of palbociclib with fulvestrant vs letrozole in advanced breast cancer.Methods: This was a cohort study using longitudinal EHR data derived from outpatient oncology practices in the United States. Eligibility criteria from the PARSIFAL trial were emulated as closely as possible. Patients were included if they had hormone-positive, human epidermal growth factor receptor – 2 (HER-2) negative metastatic breast cancer and had no record of prior treatment for metastatic disease. Patients initiating first-line treatment with palbociclib and fulvestrant following their first record of metastasis were compared to those initiating palbociclib and letrozole on the same day. Treatments were ascertained by oncology medication ordering records in the data source. The primary outcome was death as recorded in the oncologists’ EHR systems.Results: There were 1886 eligible women in the study cohort. Although the 3-year survival was meaningfully lower in clinical practice (59%) compared to the randomized trial (78%), the relative effect size was a hazard ratio (HR) of 1.07 (95% CI: 0.86– 1.35), similar to the randomized trial (HR = 1.00; 95% CI: 0.68– 1.48).Conclusion: Despite common challenges encountered in EHR-based studies, it is possible to achieve similar conclusions to emulated randomized trials with the application of analytic approaches that address missing data, confounding, and selection bias. This is a promising finding in light of other emulations and ongoing efforts to improve data from clinical practice and causal analytics.Keywords: real-world evidence, comparative effectiveness research, oncology, pharmacoepidemiology, healthcare databases

Shirley V Wang, Sebastian Schneeweiss Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USACorrespondence: Shirley V Wang, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St, Suite 303, Boston, MA, 02120, USA, Tel +1 617-525-8376, Email Swang1@bwh.harvard.eduBackground: There is growing interest in using evidence generated from clinical practice data to support regulatory, coverage and other healthcare decision-making. A graphical framework for depicting longitudinal study designs to mitigate this barrier was introduced and has found wide acceptance. We sought to enhance the framework to contain information that helps readers assess the appropriateness of the source data in which the study design was applied.Methods: For the enhanced graphical framework, we added a simple visualization of data type and observability to capture differences between electronic health record (EHR) and other registry data that may have limited data continuity and insurance claims data that have enrollment files.Results: We illustrate the revised graphical framework with 2 example studies conducted using different data sources, including administrative claims only, EHR only, linked claims and EHR, as well as specialty community based EHRs with and without external linkages.Conclusion: The enhanced visualization framework is important because evaluation of study validity needs to consider the triad of study question, design, and data together. Any given data source or study design may be appropriate for some questions but not others.Keywords: study design, real world evidence, real world data, visualization, methods, bias

Moa P Lee,1,2 Robert J Glynn,1 Sebastian Schneeweiss,1 Kueiyu Joshua Lin,1,3 Elisabetta Patorno,1 Julie Barberio,1 Raisa Levin,1 Thomas Evers,4 Shirley V Wang,1 Rishi J Desai1 1Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital & Harvard Medical School, Boston, MA, USA; 2Department of Epidemiology, University of North Carolina, Chapel Hill, NC, USA; 3Department of Medicine, Massachusetts General Hospital & Harvard Medical School, Boston, MA, USA; 4Bayer AG, Wuppertal, GermanyCorrespondence: Rishi J DesaiDivision of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, 1620 Tremont Street, Boston, MA 02120, USATel +1 617-278-0932Fax +1 617-232-8602Email rdesai@bwh.harvard.eduBackground: The differential impact of various demographic characteristics and comorbid conditions on development of heart failure (HF) with preserved (pEF) and reduced ejection fraction (rEF) is not well studied among the elderly.Methods: Using Medicare claims data linked to electronic health records, we conducted an observational cohort study of individuals ≥ 65 years of age without HF. A Cox proportional hazards model accounting for competing risk of HFrEF and HFpEF incidence was constructed. A gradient-boosted model (GBM) assessed the relative influence (RI) of each predictor in the development of HFrEF and HFpEF.Results: Among 138,388 included individuals, 9701 developed HF (incidence rate = 20.9 per 1000 person-years). Males were more likely to develop HFrEF than HFpEF (HR = 2.07, 95% CI: 1.81– 2.37 vs. 1.11, 95% CI: 1.02– 1.20, P for heterogeneity < 0.01). Atrial fibrillation and pulmonary hypertension had stronger associations with the risk of HFpEF (HR = 2.02, 95% CI: 1.80– 2.26 and 1.66, 95% CI: 1.23– 2.22) while cardiomyopathy and myocardial infarction were more strongly associated with HFrEF (HR = 4.37, 95% CI: 3.21– 5.97 and 1.94, 95% CI: 1.23– 3.07). Age was the strongest predictor across all HF subtypes with RI from GBM > 35%. Atrial fibrillation was the most influential comorbidity for the development of HFpEF (RI = 8.4%) while cardiomyopathy was the most influential comorbidity for the development of HFrEF (RI = 20.7%).Conclusion: These findings of heterogeneous relationships between several important risk factors and heart failure types underline the potential differences in the etiology of HFpEF and HFrEF.Keywords: heart failure, epidemiology, risk factors, LVEF, HEpEF, HFrEF, competing risks, GBM

Kueiyu Joshua Lin,1,2 Gary E Rosenthal,3 Shawn N Murphy,4,5 Kenneth D Mandl,6 Yinzhu Jin,1 Robert J Glynn,1 Sebastian Schneeweiss1 1Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; 2Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; 3Department of Internal Medicine, Wake Forest School of Medicine, Winston-Salem, NC, USA; 4Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; 5Research Information Science and Computing, Partners Healthcare, Somerville, MA, USA; 6Computational Health Informatics Program, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USACorrespondence: Kueiyu Joshua LinDivision of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St. Suite 3030, Boston, MA 02120, USATel +1 617 278-0930Fax +1 617 232-8602Email jklin@mgh.harvard.eduObjective: Electronic health records (EHR) data-discontinuity, i.e. receiving care outside of a particular EHR system, may cause misclassification of study variables. We aimed to validate an algorithm to identify patients with high EHR data-continuity to reduce such bias.Materials and Methods: We analyzed data from two EHR systems linked with Medicare claims data from 2007 through 2014, one in Massachusetts (MA, n=80,588) and the other in North Carolina (NC, n=33,207). We quantified EHR data-continuity by Mean Proportion of Encounters Captured (MPEC) by the EHR system when compared to complete recording in claims data. The prediction model for MPEC was developed in MA and validated in NC. Stratified by predicted EHR data-continuity, we quantified misclassification of 40 key variables by Mean Standardized Differences (MSD) between the proportions of these variables based on EHR alone vs the linked claims-EHR data.Results: The mean MPEC was 27% in the MA and 26% in the NC system. The predicted and observed EHR data-continuity was highly correlated (Spearman correlation=0.78 and 0.73, respectively). The misclassification (MSD) of 40 variables in patients of the predicted EHR data-continuity cohort was significantly smaller (44%, 95% CI: 40– 48%) than that in the remaining population.Discussion: The comorbidity profiles were similar in patients with high vs low EHR data-continuity. Therefore, restricting an analysis to patients with high EHR data-continuity may reduce information bias while preserving the representativeness of the study cohort.Conclusion: We have successfully validated an algorithm that can identify a high EHR data-continuity cohort representative of the source population.Keywords: electronic medical records, data linkage, comparative effectiveness research, information bias, continuity, external validation

Jeremy A Rassen,1 Dorothee B Bartels,2 Sebastian Schneeweiss,1,3,4 Amanda R Patrick,1 William Murk1,5 1Aetion, Inc, New York, NY, USA; 2BI X GmbH, Ingelheim, Germany; 3Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital, Boston, MA, USA; 4Harvard Medical School, Boston, MA, USA; 5Jacobs School of Medicine, University at Buffalo, Buffalo, NY, USA Background: Health care databases are natural sources for estimating prevalence and incidence of chronic conditions, but substantial variation in estimates limits their interpretability and utility. We evaluated the effects of design choices when estimating prevalence and incidence in claims and electronic health record databases. Methods: Prevalence and incidence for five chronic diseases at increasing levels of expected frequencies, from cystic fibrosis to COPD, were estimated in the Clinical Practice Research Datalink (CPRD) and MarketScan databases from 2011 to 2014. Estimates were compared using different definitions of lookback time and contributed person-time. Results: Variation in lookback time substantially affected estimates. In 2014, for CPRD, use of an all-time vs a 1-year lookback window resulted in 4.3–8.3 times higher prevalence (depending on disease), reducing incidence by 1.9–3.3 times. All-time lookback resulted in strong temporal trends. COPD prevalence between 2011 and 2014 in MarketScan increased by 25% with an all-time lookback but stayed relatively constant with a 1-year lookback. Varying observability did not substantially affect estimates. Conclusion: This framework draws attention to the underrecognized potential for widely varying incidence and prevalence estimates, with implications for care planning and drug development. Though prevalence and incidence are seemingly straightforward concepts, careful consideration of methodology is required to obtain meaningful estimates from health care databases. Keywords: epidemiology, epidemiologic methods, epidemiological monitoring, sentinel surveillance, pharmacoepidemiology, cross-sectional studies, secondary databases, prevalence, prevalence studies, incidence

Sebastian Schneeweiss1,2 1Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, 2Harvard Medical School, Boston, MA, USA Background: Decision makers in health care increasingly rely on nonrandomized database analyses to assess the effectiveness, safety, and value of medical products. Health care data scientists use data-adaptive approaches that automatically optimize confounding control to study causal treatment effects. This article summarizes relevant experiences and extensions. Methods: The literature was reviewed on the uses of high-dimensional propensity score (HDPS) and related approaches for health care database analyses, including methodological articles on their performance and improvement. Articles were grouped into applications, comparative performance studies, and statistical simulation experiments. Results: The HDPS algorithm has been referenced frequently with a variety of clinical applications and data sources from around the world. The appeal of HDPS for database research rests in 1) its superior performance in situations of unobserved confounding through proxy adjustment, 2) its predictable efficiency in extracting confounding information from a given data source, 3) its ability to automate estimation of causal treatment effects to the extent achievable in a given data source, and 4) its independence of data source and coding system. Extensions of the HDPS approach have focused on improving variable selection when exposure is sparse, using free text information and time-varying confounding adjustment. Conclusion: Semiautomated and optimized confounding adjustment in health care database analyses has proven successful across a wide range of settings. Machine-learning extensions further automate its use in estimating causal treatment effects across a range of data scenarios. Keywords: high-dimensional data, confounding (epidemiology), health care databases, real-world data, confounding adjustment, propensity scores, automation, causal conclusions, artificial intelligence, machine learning

Alexander M Walker,1 Sebastian Schneeweiss,2 Maral DerSarkissian,3 Mei Sheng Duh4 1World Health Information Science Consultants, Newton, MA, USA; 2Division of Pharmacoepidemiology, Department of Medicine, Brigham & Women’s Hospital and Harvard Medical School, Boston, MA, USA; 3Analysis Group, Inc., Los Angeles, CA, USA; 4Analysis Group, Inc., Boston, MA, USA Background: When an exclusionary criterion is an imperfect screen, some ineligible patients will remain in a study. Medical record review for outcome adjudication can reveal such individuals. Objective: To ascertain the circumstances under which it is advisable to remove outcome cases first found to be ineligible on chart review. Methods: The impact on the relative risk caused by removal of ineligible outcome cases was examined under different circumstances of confounding, prevalence, and efficacy of the screening criterion for exclusions. The result is illustrated by a hospital-based cohort study in which electronic medical record diagnosis served to exclude ineligible cases, and review of text notes for putative outcome cases revealed that the codes were only 95% sensitive. Other hypothetical scenarios provide further evidence. Results: If a condition to be excluded is a confounder of the exposure–outcome relation, residual confounding will continue to bias a study after application of an imperfect screening criterion. Removal of ineligible outcome cases after chart review creates a new bias, distinct from residual confounding. The new bias does not depend on the magnitude of the confounder–outcome association, and will be small if the exclusion criterion has resulted in a low prevalence of the exclusionary condition. The new bias caused by removal of ineligible outcome cases is almost certain to be smaller than the confounding bias that can result if they are retained. Conclusions: Outcome cases first discovered at chart review to be study-ineligible should be removed from the study, even when similar scrutiny is infeasible for non-cases. Keywords: exclusions, eligibility, confounding, study design, adjudication, analysis

Background: Hidradenitis suppurativa (HS) has been associated with auto‐inflammatory conditions, yet the risk of developing connective tissue disease (CTD), morphoea and systemic vasculitis has not been well‐characterized. Objectives: We sought to evaluate the risk of developing CTD, morphoea and systemic vasculitis in patients with HS. Methods: Using claims data, we identified patients with HS and used 2 : 1 risk‐set sampling to identify patients without HS. Patients with existing CTD were excluded. Patient follow‐up lasted until first occurrence of the following events: the occurrence of outcome (i.e. systemic lupus erythematosus, morphoea, systemic sclerosis, Sjogren's Syndrome and systemic vasculitis), death, disenrolment or end of data stream. Hazard ratios (HR) of developing CTD, morphoea and systemic vasculitis were computed after 1 : 1 propensity score (PS) matching. Results: After 2 : 1 risk‐set sampling, we identified 78 122 HS patients and 156 247 non‐HS comparators. The mean follow‐up was 540 days. After PS matching, HS patients had an increased risk of systemic lupus erythematosus HR = 1.63 (1.31–2.03) and morphoea HR = 2.02 (1.32–3.11), compared to non‐HS patients. We did not observe an increased risk for systemic sclerosis HR = 0.90 (0.59–1.44), Sjogren's Syndrome HR = 0.91 (0.73–1.14) or systemic vasculitis HR = 0.87 (0.64–1.20). Conclusion: In this population‐based study, we observed an increased risk of developing systemic lupus erythematous and morphoea subsequent to a first‐recorded diagnosis of hidradenitis suppurativa. [ABSTRACT FROM AUTHOR]