Development and validation of a patient-level model to predict dementia across a network of observational databases.

BACKGROUND: A prediction model can be a useful tool to quantify the risk of a patient developing dementia in the next years and take risk-factor-targeted intervention. Numerous dementia prediction models have been developed, but few have been externally validated, likely limiting their clinical uptake. In our previous work, we had limited success in externally validating some of these existing models due to inadequate reporting. As a result, we are compelled to develop and externally validate novel models to predict dementia in the general population across a network of observational databases. We assess regularization methods to obtain parsimonious models that are of lower complexity and easier to implement. METHODS: Logistic regression models were developed across a network of five observational databases with electronic health records (EHRs) and claims data to predict 5-year dementia risk in persons aged 55-84. The regularization methods L1 and Broken Adaptive Ridge (BAR) as well as three candidate predictor sets to optimize prediction performance were assessed. The predictor sets include a baseline set using only age and sex, a full set including all available candidate predictors, and a phenotype set which includes a limited number of clinically relevant predictors. RESULTS: BAR can be used for variable selection, outperforming L1 when a parsimonious model is desired. Adding candidate predictors for disease diagnosis and drug exposure generally improves the performance of baseline models using only age and sex. While a model trained on German EHR data saw an increase in AUROC from 0.74 to 0.83 with additional predictors, a model trained on US EHR data showed only minimal improvement from 0.79 to 0.81 AUROC. Nevertheless, the latter model developed using BAR regularization on the clinically relevant predictor set was ultimately chosen as best performing model as it demonstrated more consistent external validation performance and improved calibration. CONCLUSIONS: We developed and externally validated patient-level models to predict dementia. Our results show that although dementia prediction is highly driven by demographic age, adding predictors based on condition diagnoses and drug exposures further improves prediction performance. BAR regularization outperforms L1 regularization to yield the most parsimonious yet still well-performing prediction model for dementia.

Assessing Covariate Balance with Small Sample Sizes.

Propensity score adjustment addresses confounding by balancing covariates in subject treatment groups through matching, stratification, inverse probability weighting, etc. Diagnostics ensure that the adjustment has been effective. A common technique is to check whether the standardized mean difference for each relevant covariate is less than a threshold like 0.1. For small sample sizes, the probability of falsely rejecting the validity of a study because of chance imbalance when no underlying balance exists approaches 1. We propose an alternative diagnostic that checks whether the standardized mean difference statistically significantly exceeds the threshold. Through simulation and real-world data, we find that this diagnostic achieves a better trade-off of type 1 error rate and power than standard nominal threshold tests and not testing for sample sizes from 250 to 4000 and for 20 to 100,000 covariates. In network studies, meta-analysis of effect estimates must be accompanied by meta-analysis of the diagnostics or else systematic confounding may overwhelm the estimated effect. Our procedure for statistically testing balance at both the database level and the meta-analysis level achieves the best balance of type-1 error rate and power. Our procedure supports the review of large numbers of covariates, enabling more rigorous diagnostics.

An open source knowledge graph ecosystem for the life sciences.

Translational research requires data at multiple scales of biological organization. Advancements in sequencing and multi-omics technologies have increased the availability of these data, but researchers face significant integration challenges. Knowledge graphs (KGs) are used to model complex phenomena, and methods exist to construct them automatically. However, tackling complex biomedical integration problems requires flexibility in the way knowledge is modeled. Moreover, existing KG construction methods provide robust tooling at the cost of fixed or limited choices among knowledge representation models. PheKnowLator (Phenotype Knowledge Translator) is a semantic ecosystem for automating the FAIR (Findable, Accessible, Interoperable, and Reusable) construction of ontologically grounded KGs with fully customizable knowledge representation. The ecosystem includes KG construction resources (e.g., data preparation APIs), analysis tools (e.g., SPARQL endpoint resources and abstraction algorithms), and benchmarks (e.g., prebuilt KGs). We evaluated the ecosystem by systematically comparing it to existing open-source KG construction methods and by analyzing its computational performance when used to construct 12 different large-scale KGs. With flexible knowledge representation, PheKnowLator enables fully customizable KGs without compromising performance or usability.

Similar Risk of Kidney Failure among Patients with Blinding Diseases Who Receive Ranibizumab, Aflibercept, and Bevacizumab: An Observational Health Data Sciences and Informatics Network Study.

PURPOSE: To characterize the incidence of kidney failure associated with intravitreal anti-VEGF exposure; and compare the risk of kidney failure in patients treated with ranibizumab, aflibercept, or bevacizumab. DESIGN: Retrospective cohort study across 12 databases in the Observational Health Data Sciences and Informatics (OHDSI) network. SUBJECTS: Subjects aged ≥ 18 years with ≥ 3 monthly intravitreal anti-VEGF medications for a blinding disease (diabetic retinopathy, diabetic macular edema, exudative age-related macular degeneration, or retinal vein occlusion). METHODS: The standardized incidence proportions and rates of kidney failure while on treatment with anti-VEGF were calculated. For each comparison (e.g., aflibercept versus ranibizumab), patients from each group were matched 1:1 using propensity scores. Cox proportional hazards models were used to estimate the risk of kidney failure while on treatment. A random effects meta-analysis was performed to combine each database's hazard ratio (HR) estimate into a single network-wide estimate. MAIN OUTCOME MEASURES: Incidence of kidney failure while on anti-VEGF treatment, and time from cohort entry to kidney failure. RESULTS: Of the 6.1 million patients with blinding diseases, 37 189 who received ranibizumab, 39 447 aflibercept, and 163 611 bevacizumab were included; the total treatment exposure time was 161 724 person-years. The average standardized incidence proportion of kidney failure was 678 per 100 000 persons (range, 0-2389), and incidence rate 742 per 100 000 person-years (range, 0-2661). The meta-analysis HR of kidney failure comparing aflibercept with ranibizumab was 1.01 (95% confidence interval [CI], 0.70-1.47; P = 0.45), ranibizumab with bevacizumab 0.95 (95% CI, 0.68-1.32; P = 0.62), and aflibercept with bevacizumab 0.95 (95% CI, 0.65-1.39; P = 0.60). CONCLUSIONS: There was no substantially different relative risk of kidney failure between those who received ranibizumab, bevacizumab, or aflibercept. Practicing ophthalmologists and nephrologists should be aware of the risk of kidney failure among patients receiving intravitreal anti-VEGF medications and that there is little empirical evidence to preferentially choose among the specific intravitreal anti-VEGF agents. FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Comparative Effectiveness of Second-line Antihyperglycemic Agents for Cardiovascular Outcomes: A Large-scale, Multinational, Federated Analysis of the LEGEND-T2DM Study.

Background: SGLT2 inhibitors (SGLT2is) and GLP-1 receptor agonists (GLP1-RAs) reduce major adverse cardiovascular events (MACE) in patients with type 2 diabetes mellitus (T2DM). However, their effectiveness relative to each other and other second-line antihyperglycemic agents is unknown, without any major ongoing head-to-head trials. Methods: Across the LEGEND-T2DM network, we included ten federated international data sources, spanning 1992-2021. We identified 1,492,855 patients with T2DM and established cardiovascular disease (CVD) on metformin monotherapy who initiated one of four second-line agents (SGLT2is, GLP1-RAs, dipeptidyl peptidase 4 inhibitor [DPP4is], sulfonylureas [SUs]). We used large-scale propensity score models to conduct an active comparator, target trial emulation for pairwise comparisons. After evaluating empirical equipoise and population generalizability, we fit on-treatment Cox proportional hazard models for 3-point MACE (myocardial infarction, stroke, death) and 4-point MACE (3-point MACE + heart failure hospitalization) risk, and combined hazard ratio (HR) estimates in a random-effects meta-analysis. Findings: Across cohorts, 16·4%, 8·3%, 27·7%, and 47·6% of individuals with T2DM initiated SGLT2is, GLP1-RAs, DPP4is, and SUs, respectively. Over 5·2 million patient-years of follow-up and 489 million patient-days of time at-risk, there were 25,982 3-point MACE and 41,447 4-point MACE events. SGLT2is and GLP1-RAs were associated with a lower risk for 3-point MACE compared with DPP4is (HR 0·89 [95% CI, 0·79-1·00] and 0·83 [0·70-0·98]), and SUs (HR 0·76 [0·65-0·89] and 0·71 [0·59-0·86]). DPP4is were associated with a lower 3-point MACE risk versus SUs (HR 0·87 [0·79-0·95]). The pattern was consistent for 4-point MACE for the comparisons above. There were no significant differences between SGLT2is and GLP1-RAs for 3-point or 4-point MACE (HR 1·06 [0·96-1·17] and 1·05 [0·97-1·13]). Interpretation: In patients with T2DM and established CVD, we found comparable cardiovascular risk reduction with SGLT2is and GLP1-RAs, with both agents more effective than DPP4is, which in turn were more effective than SUs. These findings suggest that the use of GLP1-RAs and SGLT2is should be prioritized as second-line agents in those with established CVD. Funding: National Institutes of Health, United States Department of Veterans Affairs.

Bayesian safety surveillance with adaptive bias correction.

Postmarket safety surveillance is an integral part of mass vaccination programs. Typically relying on sequential analysis of real-world health data as they accrue, safety surveillance is challenged by sequential multiple testing and by biases induced by residual confounding in observational data. The current standard approach based on the maximized sequential probability ratio test (MaxSPRT) fails to satisfactorily address these practical challenges and it remains a rigid framework that requires prespecification of the surveillance schedule. We develop an alternative Bayesian surveillance procedure that addresses both aforementioned challenges using a more flexible framework. To mitigate bias, we jointly analyze a large set of negative control outcomes that are adverse events with no known association with the vaccines in order to inform an empirical bias distribution, which we then incorporate into estimating the effect of vaccine exposure on the adverse event of interest through a Bayesian hierarchical model. To address multiple testing and improve on flexibility, at each analysis timepoint, we update a posterior probability in favor of the alternative hypothesis that vaccination induces higher risks of adverse events, and then use it for sequential detection of safety signals. Through an empirical evaluation using six US observational healthcare databases covering more than 360 million patients, we benchmark the proposed procedure against MaxSPRT on testing errors and estimation accuracy, under two epidemiological designs, the historical comparator and the self-controlled case series. We demonstrate that our procedure substantially reduces Type 1 error rates, maintains high statistical power and fast signal detection, and provides considerably more accurate estimation than MaxSPRT. Given the extensiveness of the empirical study which yields more than 7 million sets of results, we present all results in a public R ShinyApp. As an effort to promote open science, we provide full implementation of our method in the open-source R package EvidenceSynthesis.

Evaluating the impact of alternative phenotype definitions on incidence rates across a global data network.

Objective: Developing accurate phenotype definitions is critical in obtaining reliable and reproducible background rates in safety research. This study aims to illustrate the differences in background incidence rates by comparing definitions for a given outcome. Materials and Methods: We used 16 data sources to systematically generate and evaluate outcomes for 13 adverse events and their overall background rates. We examined the effect of different modifications (inpatient setting, standardization of code set, and code set changes) to the computable phenotype on background incidence rates. Results: Rate ratios (RRs) of the incidence rates from each computable phenotype definition varied across outcomes, with inpatient restriction showing the highest variation from 1 to 11.93. Standardization of code set RRs ranges from 1 to 1.64, and code set changes range from 1 to 2.52. Discussion: The modification that has the highest impact is requiring inpatient place of service, leading to at least a 2-fold higher incidence rate in the base definition. Standardization showed almost no change when using source code variations. The strength of the effect in the inpatient restriction is highly dependent on the outcome. Changing definitions from broad to narrow showed the most variability by age/gender/database across phenotypes and less than a 2-fold increase in rate compared to the base definition. Conclusion: Characterization of outcomes across a network of databases yields insights into sensitivity and specificity trade-offs when definitions are altered. Outcomes should be thoroughly evaluated prior to use for background rates for their plausibility for use across a global network.

Scalable and interpretable alternative to chart review for phenotype evaluation using standardized structured data from electronic health records.

OBJECTIVES: Chart review as the current gold standard for phenotype evaluation cannot support observational research on electronic health records and claims data sources at scale. We aimed to evaluate the ability of structured data to support efficient and interpretable phenotype evaluation as an alternative to chart review. MATERIALS AND METHODS: We developed Knowledge-Enhanced Electronic Profile Review (KEEPER) as a phenotype evaluation tool that extracts patient's structured data elements relevant to a phenotype and presents them in a standardized fashion following clinical reasoning principles. We evaluated its performance (interrater agreement, intermethod agreement, accuracy, and review time) compared to manual chart review for 4 conditions using randomized 2-period, 2-sequence crossover design. RESULTS: Case ascertainment with KEEPER was twice as fast compared to manual chart review. 88.1% of the patients were classified concordantly using charts and KEEPER, but agreement varied depending on the condition. Missing data and differences in interpretation accounted for most of the discrepancies. Pairs of clinicians agreed in case ascertainment in 91.2% of the cases when using KEEPER compared to 76.3% when using charts. Patient classification aligned with the gold standard in 88.1% and 86.9% of the cases respectively. CONCLUSION: Structured data can be used for efficient and interpretable phenotype evaluation if they are limited to relevant subset and organized according to the clinical reasoning principles. A system that implements these principles can achieve noninferior performance compared to chart review at a fraction of time.

Development and evaluation of an algorithm to link mothers and infants in two US commercial healthcare claims databases for pharmacoepidemiology research.

BACKGROUND: Administrative healthcare claims databases are used in drug safety research but are limited for investigating the impacts of prenatal exposures on neonatal and pediatric outcomes without mother-infant pair identification. Further, existing algorithms are not transportable across data sources. We developed a transportable mother-infant linkage algorithm and evaluated it in two, large US commercially insured populations. METHODS: We used two US commercial health insurance claims databases during the years 2000 to 2021. Mother-infant links were constructed where persons of female sex 12-55 years of age with a pregnancy episode ending in live birth were associated with a person who was 0 years of age at database entry, who shared a common insurance plan ID, had overlapping insurance coverage time, and whose date of birth was within ± 60-days of the mother's pregnancy episode live birth date. We compared the characteristics of linked vs. non-linked mothers and infants to assess similarity. RESULTS: The algorithm linked 3,477,960 mothers to 4,160,284 infants in the two databases. Linked mothers and linked infants comprised 73.6% of all mothers and 49.1% of all infants, respectively. 94.9% of linked infants' dates of birth were within ± 30-days of the associated mother's pregnancy episode end dates. Characteristics were largely similar in linked vs. non-linked mothers and infants. Differences included that linked mothers were older, had longer pregnancy episodes, and had greater post-pregnancy observation time than mothers with live births who were not linked. Linked infants had less observation time and greater healthcare utilization than non-linked infants. CONCLUSIONS: We developed a mother-infant linkage algorithm and applied it to two US commercial healthcare claims databases that achieved a high linkage proportion and demonstrated that linked and non-linked mother and infant cohorts were similar. Transparent, reusable algorithms applied to large databases enable large-scale research on exposures during pregnancy and pediatric outcomes with relevance to drug safety. These features suggest studies using this algorithm can produce valid and generalizable evidence to inform clinical, policy, and regulatory decisions.

Multinational patterns of second line antihyperglycaemic drug initiation across cardiovascular risk groups: federated pharmacoepidemiological evaluation in LEGEND-T2DM.

Objective: To assess the uptake of second line antihyperglycaemic drugs among patients with type 2 diabetes mellitus who are receiving metformin. Design: Federated pharmacoepidemiological evaluation in LEGEND-T2DM. Setting: 10 US and seven non-US electronic health record and administrative claims databases in the Observational Health Data Sciences and Informatics network in eight countries from 2011 to the end of 2021. Participants: 4.8 million patients (≥18 years) across US and non-US based databases with type 2 diabetes mellitus who had received metformin monotherapy and had initiated second line treatments. Exposure: The exposure used to evaluate each database was calendar year trends, with the years in the study that were specific to each cohort. Main outcomes measures: The outcome was the incidence of second line antihyperglycaemic drug use (ie, glucagon-like peptide-1 receptor agonists, sodium-glucose cotransporter-2 inhibitors, dipeptidyl peptidase-4 inhibitors, and sulfonylureas) among individuals who were already receiving treatment with metformin. The relative drug class level uptake across cardiovascular risk groups was also evaluated. Results: 4.6 million patients were identified in US databases, 61 382 from Spain, 32 442 from Germany, 25 173 from the UK, 13 270 from France, 5580 from Scotland, 4614 from Hong Kong, and 2322 from Australia. During 2011-21, the combined proportional initiation of the cardioprotective antihyperglycaemic drugs (glucagon-like peptide-1 receptor agonists and sodium-glucose cotransporter-2 inhibitors) increased across all data sources, with the combined initiation of these drugs as second line drugs in 2021 ranging from 35.2% to 68.2% in the US databases, 15.4% in France, 34.7% in Spain, 50.1% in Germany, and 54.8% in Scotland. From 2016 to 2021, in some US and non-US databases, uptake of glucagon-like peptide-1 receptor agonists and sodium-glucose cotransporter-2 inhibitors increased more significantly among populations with no cardiovascular disease compared with patients with established cardiovascular disease. No data source provided evidence of a greater increase in the uptake of these two drug classes in populations with cardiovascular disease compared with no cardiovascular disease. Conclusions: Despite the increase in overall uptake of cardioprotective antihyperglycaemic drugs as second line treatments for type 2 diabetes mellitus, their uptake was lower in patients with cardiovascular disease than in people with no cardiovascular disease over the past decade. A strategy is needed to ensure that medication use is concordant with guideline recommendations to improve outcomes of patients with type 2 diabetes mellitus.

Serially Combining Epidemiological Designs Does Not Improve Overall Signal Detection in Vaccine Safety Surveillance.

INTRODUCTION: Vaccine safety surveillance commonly includes a serial testing approach with a sensitive method for 'signal generation' and specific method for 'signal validation.' The extent to which serial testing in real-world studies improves or hinders overall performance in terms of sensitivity and specificity remains unknown. METHODS: We assessed the overall performance of serial testing using three administrative claims and one electronic health record database. We compared type I and II errors before and after empirical calibration for historical comparator, self-controlled case series (SCCS), and the serial combination of those designs against six vaccine exposure groups with 93 negative control and 279 imputed positive control outcomes. RESULTS: The historical comparator design mostly had fewer type II errors than SCCS. SCCS had fewer type I errors than the historical comparator. Before empirical calibration, the serial combination increased specificity and decreased sensitivity. Type II errors mostly exceeded 50%. After empirical calibration, type I errors returned to nominal; sensitivity was lowest when the methods were combined. CONCLUSION: While serial combination produced fewer false-positive signals compared with the most specific method, it generated more false-negative signals compared with the most sensitive method. Using a historical comparator design followed by an SCCS analysis yielded decreased sensitivity in evaluating safety signals relative to a one-stage SCCS approach. While the current use of serial testing in vaccine surveillance may provide a practical paradigm for signal identification and triage, single epidemiological designs should be explored as valuable approaches to detecting signals.

Ontologizing health systems data at scale: making translational discovery a reality.

Common data models solve many challenges of standardizing electronic health record (EHR) data but are unable to semantically integrate all of the resources needed for deep phenotyping. Open Biological and Biomedical Ontology (OBO) Foundry ontologies provide computable representations of biological knowledge and enable the integration of heterogeneous data. However, mapping EHR data to OBO ontologies requires significant manual curation and domain expertise. We introduce OMOP2OBO, an algorithm for mapping Observational Medical Outcomes Partnership (OMOP) vocabularies to OBO ontologies. Using OMOP2OBO, we produced mappings for 92,367 conditions, 8611 drug ingredients, and 10,673 measurement results, which covered 68-99% of concepts used in clinical practice when examined across 24 hospitals. When used to phenotype rare disease patients, the mappings helped systematically identify undiagnosed patients who might benefit from genetic testing. By aligning OMOP vocabularies to OBO ontologies our algorithm presents new opportunities to advance EHR-based deep phenotyping.

A standardized framework for risk-based assessment of treatment effect heterogeneity in observational healthcare databases.

Treatment effects are often anticipated to vary across groups of patients with different baseline risk. The Predictive Approaches to Treatment Effect Heterogeneity (PATH) statement focused on baseline risk as a robust predictor of treatment effect and provided guidance on risk-based assessment of treatment effect heterogeneity in a randomized controlled trial. The aim of this study is to extend this approach to the observational setting using a standardized scalable framework. The proposed framework consists of five steps: (1) definition of the research aim, i.e., the population, the treatment, the comparator and the outcome(s) of interest; (2) identification of relevant databases; (3) development of a prediction model for the outcome(s) of interest; (4) estimation of relative and absolute treatment effect within strata of predicted risk, after adjusting for observed confounding; (5) presentation of the results. We demonstrate our framework by evaluating heterogeneity of the effect of thiazide or thiazide-like diuretics versus angiotensin-converting enzyme inhibitors on three efficacy and nine safety outcomes across three observational databases. We provide a publicly available R software package for applying this framework to any database mapped to the Observational Medical Outcomes Partnership Common Data Model. In our demonstration, patients at low risk of acute myocardial infarction receive negligible absolute benefits for all three efficacy outcomes, though they are more pronounced in the highest risk group, especially for acute myocardial infarction. Our framework allows for the evaluation of differential treatment effects across risk strata, which offers the opportunity to consider the benefit-harm trade-off between alternative treatments.

Reproducible variability: assessing investigator discordance across 9 research teams attempting to reproduce the same observational study.

OBJECTIVE: Observational studies can impact patient care but must be robust and reproducible. Nonreproducibility is primarily caused by unclear reporting of design choices and analytic procedures. This study aimed to: (1) assess how the study logic described in an observational study could be interpreted by independent researchers and (2) quantify the impact of interpretations' variability on patient characteristics. MATERIALS AND METHODS: Nine teams of highly qualified researchers reproduced a cohort from a study by Albogami et al. The teams were provided the clinical codes and access to the tools to create cohort definitions such that the only variable part was their logic choices. We executed teams' cohort definitions against the database and compared the number of subjects, patient overlap, and patient characteristics. RESULTS: On average, the teams' interpretations fully aligned with the master implementation in 4 out of 10 inclusion criteria with at least 4 deviations per team. Cohorts' size varied from one-third of the master cohort size to 10 times the cohort size (2159-63 619 subjects compared to 6196 subjects). Median agreement was 9.4% (interquartile range 15.3-16.2%). The teams' cohorts significantly differed from the master implementation by at least 2 baseline characteristics, and most of the teams differed by at least 5. CONCLUSIONS: Independent research teams attempting to reproduce the study based on its free-text description alone produce different implementations that vary in the population size and composition. Sharing analytical code supported by a common data model and open-source tools allows reproducing a study unambiguously thereby preserving initial design choices.

International cohort study indicates no association between alpha-1 blockers and susceptibility to COVID-19 in benign prostatic hyperplasia patients.

Purpose: Alpha-1 blockers, often used to treat benign prostatic hyperplasia (BPH), have been hypothesized to prevent COVID-19 complications by minimising cytokine storm release. The proposed treatment based on this hypothesis currently lacks support from reliable real-world evidence, however. We leverage an international network of large-scale healthcare databases to generate comprehensive evidence in a transparent and reproducible manner. Methods: In this international cohort study, we deployed electronic health records from Spain (SIDIAP) and the United States (Department of Veterans Affairs, Columbia University Irving Medical Center, IQVIA OpenClaims, Optum DOD, Optum EHR). We assessed association between alpha-1 blocker use and risks of three COVID-19 outcomes-diagnosis, hospitalization, and hospitalization requiring intensive services-using a prevalent-user active-comparator design. We estimated hazard ratios using state-of-the-art techniques to minimize potential confounding, including large-scale propensity score matching/stratification and negative control calibration. We pooled database-specific estimates through random effects meta-analysis. Results: Our study overall included 2.6 and 0.46 million users of alpha-1 blockers and of alternative BPH medications. We observed no significant difference in their risks for any of the COVID-19 outcomes, with our meta-analytic HR estimates being 1.02 (95% CI: 0.92-1.13) for diagnosis, 1.00 (95% CI: 0.89-1.13) for hospitalization, and 1.15 (95% CI: 0.71-1.88) for hospitalization requiring intensive services. Conclusion: We found no evidence of the hypothesized reduction in risks of the COVID-19 outcomes from the prevalent-use of alpha-1 blockers-further research is needed to identify effective therapies for this novel disease.

PheValuator 2.0: Methodological improvements for the PheValuator approach to semi-automated phenotype algorithm evaluation.

PURPOSE: Phenotype algorithms are central to performing analyses using observational data. These algorithms translate the clinical idea of a health condition into an executable set of rules allowing for queries of data elements from a database. PheValuator, a software package in the Observational Health Data Sciences and Informatics (OHDSI) tool stack, provides a method to assess the performance characteristics of these algorithms, namely, sensitivity, specificity, and positive and negative predictive value. It uses machine learning to develop predictive models for determining a probabilistic gold standard of subjects for assessment of cases and non-cases of health conditions. PheValuator was developed to complement or even replace the traditional approach of algorithm validation, i.e., by expert assessment of subject records through chart review. Results in our first PheValuator paper suggest a systematic underestimation of the PPV compared to previous results using chart review. In this paper we evaluate modifications made to the method designed to improve its performance. METHODS: The major changes to PheValuator included allowing all diagnostic conditions, clinical observations, drug prescriptions, and laboratory measurements to be included as predictors within the modeling process whereas in the prior version there were significant restrictions on the included predictors. We also have allowed for the inclusion of the temporal relationships of the predictors in the model. To evaluate the performance of the new method, we compared the results from the new and original methods against results found from the literature using traditional validation of algorithms for 19 phenotypes. We performed these tests using data from five commercial databases. RESULTS: In the assessment aggregating all phenotype algorithms, the median difference between the PheValuator estimate and the gold standard estimate for PPV was reduced from -21 (IQR -34, -3) in Version 1.0 to 4 (IQR -3, 15) using Version 2.0. We found a median difference in specificity of 3 (IQR 1, 4.25) for Version 1.0 and 3 (IQR 1, 4) for Version 2.0. The median difference between the two versions of PheValuator and the gold standard for estimates of sensitivity was reduced from -39 (-51, -20) to -16 (-34, -6). CONCLUSION: PheValuator 2.0 produces estimates for the performance characteristics for phenotype algorithms that are significantly closer to estimates from traditional validation through chart review compared to version 1.0. With this tool in researcher's toolkits, methods, such as quantitative bias analysis, may now be used to improve the reliability and reproducibility of research studies using observational data.

Phenotype Algorithms for the Identification and Characterization of Vaccine-Induced Thrombotic Thrombocytopenia in Real World Data: A Multinational Network Cohort Study.

INTRODUCTION: Vaccine-induced thrombotic thrombocytopenia (VITT) has been identified as a rare but serious adverse event associated with coronavirus disease 2019 (COVID-19) vaccines. OBJECTIVES: In this study, we explored the pre-pandemic co-occurrence of thrombosis with thrombocytopenia (TWT) using 17 observational health data sources across the world. We applied multiple TWT definitions, estimated the background rate of TWT, characterized TWT patients, and explored the makeup of thrombosis types among TWT patients. METHODS: We conducted an international network retrospective cohort study using electronic health records and insurance claims data, estimating background rates of TWT amongst persons observed from 2017 to 2019. Following the principles of existing VITT clinical definitions, TWT was defined as patients with a diagnosis of embolic or thrombotic arterial or venous events and a diagnosis or measurement of thrombocytopenia within 7 days. Six TWT phenotypes were considered, which varied in the approach taken in defining thrombosis and thrombocytopenia in real world data. RESULTS: Overall TWT incidence rates ranged from 1.62 to 150.65 per 100,000 person-years. Substantial heterogeneity exists across data sources and by age, sex, and alternative TWT phenotypes. TWT patients were likely to be men of older age with various comorbidities. Among the thrombosis types, arterial thrombotic events were the most common. CONCLUSION: Our findings suggest that identifying VITT in observational data presents a substantial challenge, as implementing VITT case definitions based on the co-occurrence of TWT results in large and heterogeneous incidence rate and in a cohort of patints with baseline characteristics that are inconsistent with the VITT cases reported to date.

Large-scale evidence generation and evaluation across a network of databases for type 2 diabetes mellitus (LEGEND-T2DM): a protocol for a series of multinational, real-world comparative cardiovascular effectiveness and safety studies.

INTRODUCTION: Therapeutic options for type 2 diabetes mellitus (T2DM) have expanded over the last decade with the emergence of cardioprotective novel agents, but without such data for older drugs, leaving a critical gap in our understanding of the relative effects of T2DM agents on cardiovascular risk. METHODS AND ANALYSIS: The large-scale evidence generations across a network of databases for T2DM (LEGEND-T2DM) initiative is a series of systematic, large-scale, multinational, real-world comparative cardiovascular effectiveness and safety studies of all four major second-line anti-hyperglycaemic agents, including sodium-glucose co-transporter-2 inhibitor, glucagon-like peptide-1 receptor agonist, dipeptidyl peptidase-4 inhibitor and sulfonylureas. LEGEND-T2DM will leverage the Observational Health Data Sciences and Informatics (OHDSI) community that provides access to a global network of administrative claims and electronic health record data sources, representing 190 million patients in the USA and about 50 million internationally. LEGEND-T2DM will identify all adult, patients with T2DM who newly initiate a traditionally second-line T2DM agent. Using an active comparator, new-user cohort design, LEGEND-T2DM will execute all pairwise class-versus-class and drug-versus-drug comparisons in each data source, producing extensive study diagnostics that assess reliability and generalisability through cohort balance and equipoise to examine the relative risk of cardiovascular and safety outcomes. The primary cardiovascular outcomes include a composite of major adverse cardiovascular events and a series of safety outcomes. The study will pursue data-driven, large-scale propensity adjustment for measured confounding, a large set of negative control outcome experiments to address unmeasured and systematic bias. ETHICS AND DISSEMINATION: The study ensures data safety through a federated analytic approach and follows research best practices, including prespecification and full disclosure of results. LEGEND-T2DM is dedicated to open science and transparency and will publicly share all analytic code from reproducible cohort definitions through turn-key software, enabling other research groups to leverage our methods, data and results to verify and extend our findings.

Learning patient-level prediction models across multiple healthcare databases: evaluation of ensembles for increasing model transportability.

BACKGROUND: Prognostic models that are accurate could help aid medical decision making. Large observational databases often contain temporal medical data for large and diverse populations of patients. It may be possible to learn prognostic models using the large observational data. Often the performance of a prognostic model undesirably worsens when transported to a different database (or into a clinical setting). In this study we investigate different ensemble approaches that combine prognostic models independently developed using different databases (a simple federated learning approach) to determine whether ensembles that combine models developed across databases can improve model transportability (perform better in new data than single database models)? METHODS: For a given prediction question we independently trained five single database models each using a different observational healthcare database. We then developed and investigated numerous ensemble models (fusion, stacking and mixture of experts) that combined the different database models. Performance of each model was investigated via discrimination and calibration using a leave one dataset out technique, i.e., hold out one database to use for validation and use the remaining four datasets for model development. The internal validation of a model developed using the hold out database was calculated and presented as the 'internal benchmark' for comparison. RESULTS: In this study the fusion ensembles generally outperformed the single database models when transported to a previously unseen database and the performances were more consistent across unseen databases. Stacking ensembles performed poorly in terms of discrimination when the labels in the unseen database were limited. Calibration was consistently poor when both ensembles and single database models were applied to previously unseen databases. CONCLUSION: A simple federated learning approach that implements ensemble techniques to combine models independently developed across different databases for the same prediction question may improve the discriminative performance in new data (new database or clinical setting) but will need to be recalibrated using the new data. This could help medical decision making by improving prognostic model performance.