Pharmacoepidemiology and Drug Safety

BackgroundClinical Practice Research Datalink (CPRD) GOLD is an invaluable resource for clinical research. However, some exposures are difficult to capture, including continuous subcutaneous insulin infusion pump systems ("insulin pumps"). We present a strategy we developed to classify insulin pump users and to estimate the duration of pump use in CPRD GOLD. This was done to study adverse skin events in new adult pump users. MethodsInsulin pump users were defined as patients who had a specific insulin pump code (prescription for an insulin pump cartridge or clinical code for continuous insulin infusion) in their record. Duration of use was defined as the continuous use of any insulin formulation commonly used in pump systems before and after the insulin pump specific code. Each patients pump start and end dates were calculated programmatically and then confirmed by manual review of the patients CPRD record. ResultsThere were 1032 patients with an insulin pump specific code recorded in CPRD GOLD through December 2018, of which 302 met the inclusion criteria for our safety study. Due to high variability in the patterns of insulin use, programmatic determination of pump start and end dates was insufficient. The start and/or end dates of >50% of patients required adjustment upon manual review. ConclusionsInsulin pump users in CPRD GOLD could be easily identified using this strategy, but we may have missed additional insulin pump users without specific pump codes. The duration of pump use, however, was difficult to capture. This strategy, though time intensive, is a useful tool for the study of insulin pumps. FundingThis study was funded by AbbVie. Conflicts of Interest StatementBoston Collaborative Drug Surveillance Program (BCDSP) received funding from AbbVie for this study. Susan Jick and Rebecca Persson are employees of BCDSP. Todd Sponholtz and Brenda Baak were interns at BCDSP. Authors retain full and scientific control over the content of this manuscript. DisclosuresThis manuscript has not been peer-reviewed. We provide this information as a reference for other users of Clinical Practice Research Datalink GOLD. This study is based in part on data from the Clinical Practice Research Datalink obtained under license from the UK Medicines and Healthcare products Regulatory Agency. The data is provided by patients and collected by the NHS as part of their care and support. The interpretation and conclusions contained in this study are those of the authors alone. This study was approved by the Independent Scientific Advisory Committee (ISAC) for Medicines and Healthcare products Regulatory Agency (protocol no: 19_216R).

BackgroundQuantitative bias analyses often rely on unrealistic assumptions and do not fully reflect the complexities of healthcare data. MethodsWe describe a plasmode simulation-based bias analysis for residual confounding from unmeasured variables by leveraging granular information from a subset of cohort members. We generated 500 simulated cohorts based on individual-level claims and linked electronic health record (EHR) data identifying new users of varenicline and bupropion from the Mass General Brigham site of the FDA Sentinel Real World Evidence Data Enterprise. Two adverse outcomes were simulated: 1) neuropsychiatric hospitalizations and 2) major adverse cardiovascular events (MACE), and measured confounding factors, identified from information available in claims including demographics, comorbid conditions, and comedications, were tailored to each outcome. Residual confounding was simulated using potential confounders measured in EHRs but unmeasured in claims including suicidal ideation for the neuropsychiatric outcomes and body mass index (BMI), blood pressure (BP), and smoking pack-years for the MACE outcome. These simulations retained the correlation between claims and EHR-based confounders observed in empirical data for realistic reflection of proxy adjustment of unmeasured confounders. Analyses were conducted in simulated data with and without adjustment for the EHR-based covariates to evaluate the extent of residual confounding in claims-only analyses. ResultsAfter 500 simulations, the median absolute standardized mean difference (ASMD) between treatment groups in the unadjusted sample was 0.16 for suicidal ideation; while <0.1 for BMI, BP, and smoking pack-years. For both outcomes, adjustment using claims-based variables provided relative bias close to 0, leading to the conclusion that EHR-measured confounders that were unmeasured in claims were unlikely to result in strong residual confounding within realistic simulations informed by empirical data. ConclusionThe proposed approach provides a method for quantifying bias in non-randomized studies threatened by unavailability of potentially important confounding variables. Key pointsO_LIResidual confounding by unmeasured factors is a central threat in pharmacoepidemiology that is almost always acknowledged in published studies but seldom quantified. C_LIO_LIWe describe a plasmode-simulation based approach to systematically design quantitative bias analyses that reflect the complexities of routinely collected healthcare data by leveraging detailed electronic health records from a subset. C_LIO_LIWe provide open-source software code to enable other researchers to adopt this method in future studies and improve the reliability of their findings. C_LI Plain language summaryThis study introduces a new way for researchers to better understand and measure bias caused by missing health information in large insurance databases. Using detailed hospital records alongside insurance claims data, we created realistic computer simulations to test how much of the observed risk in safety studies could be explained away by missing important health factors, like depression or smoking habits, that arent always recorded in insurance data. The approach is flexible, uses real patient data, and helps researchers make stronger, more reliable conclusions about risks and benefits of treatments, even when some patient information is not available in all records.

BackgroundThe U.S. FDA authorized the monovalent third primary series or booster doses of COVID-19 mRNA vaccines in August 2021 for persons 18 years and older. Monitoring of outcomes following updated authorizations is critical to evaluate vaccine safety and can provide early detection of rare adverse events (AEs) not identified in pre-licensure trials. MethodsWe evaluated the risk of 17 AEs following third doses of COVID-19 mRNA vaccines from August 2021 through early 2022 among adults aged 18-64 years in three commercial databases (Optum, Carelon Research, CVS Health) and adults aged >65 years in Medicare Fee-For-Service. We compared observed AE incidence rates to historical (expected) rates prior to the pandemic, estimated incidence rate ratios (IRRs) for the Medicare database and pooled IRR across the three commercial databases. Analyses were also stratified by prior history of COVID-19 diagnosis. Estimates exceeding a pre-defined threshold were considered statistical signals. ResultsFour AEs met the threshold for statistical signals for BNT162b2 and mRNA-1273 vaccines including Bells Palsy and pulmonary embolism in Medicare, and anaphylaxis and myocarditis/pericarditis in commercial databases. Nine AEs and three AEs signaled among adults with and without prior COVID-19 diagnosis, respectively. ConclusionsThis early monitoring study identified statistical signals for AEs following third doses of COVID-19 mRNA vaccination. Since this method is intended for screening purposes and generates crude results, results do not establish a causal association between the vaccines and AEs. FDAs public health assessment remains consistent that the benefits of COVID-19 vaccination outweigh the risks of vaccination.

ImportanceGabapentin is frequently prescribed off-label for pain management following an acute ischemic stroke. However, little is known about optimal gabapentin prescribing practices, such as appropriate treatment duration for specific indications, particularly off-label uses. ObjectiveWe examined gabapentin treatment patterns among older adults who initiated gabapentin within 30 days of discharge from hospitalization for acute ischemic stroke. DesignThis is an observational, retrospective study of existing administrative claims data of Medicare beneficiaries with acute ischemic stroke (AIS) hospitalizations. SettingA national 20% random sample of US Medicare beneficiaries. ParticipantsPatients aged 65 years and older who were hospitalized for their first acute ischemic stroke between 2013 and 2021. ExposuresGabapentin initiation within 30 days of discharge. Main Outcomes and MeasuresA novel approach that combines a time-varying proportion of days covered, calculated every two months, with a latent class mixed model to identify and characterize gabapentin treatment patterns during the first 12 months after initiation, accounting for prescription overlap and hospitalizations. An 80% proportion of days covered threshold was applied within each interval to distinguish high versus low medication coverage. ResultsThe analytic cohort (N=1,628) had a mean age of 76.4 (IQR 70-82) years and was 60% female and 76.5% non-Hispanic White. The latent class mixed model identified three distinct gabapentin treatment patterns: 692 patients (42.5%) experienced rapid low medication coverage (proportion of days covered <80%) within two months after initiation; 96 patients (5.9%) experienced gradually declining medication coverage over 8 months after initiation; and 840 patients (51.6%) maintained high and stable medication coverage for at least one year. Conclusions and RelevanceIn this nationwide sample, half of older adults hospitalized for acute ischemic stroke and who initiated gabapentin within 30 days of discharge had gabapentin coverage for 12 months or longer after initiation. Key PointsO_ST_ABSQuestionC_ST_ABSAre stroke survivors aged 65 and older receiving a gabapentin prescription for longer than 12 months? Findingsafter analyzing a 20% sample of Medicare beneficiaries, 51.6% of older adults hospitalized for an acute ischemic stroke initiated gabapentin within 30 days of discharge and had gabapentin coverage for 12 months or longer after initiation. MeaningMost stroke survivors aged 65 and older continue receiving a gabapentin prescription 12 months after their stroke.

IntroductionThe decision between metformin and a DPP-4 inhibitor or sulfonylurea for first-line type 2 diabetes treatment relies on many factors, including estimated glomerular filtration rate (eGFR), history of heart failure, age, sex, and even provider preferences. This study evaluates variation in this treatment decision across two factors: eGFR and provider preferences. Research Design and MethodsUsing health insurance claims data, we defined a cohort based on observation prior to first-line treatment, availability of eGFR results, and no type 1 or gestational diabetes (n=10,643). We performed a chi-squared test to verify the association between eGFR and treatment. The cohort was then restricted to providers with at least 10 patients (n=2,271 patients). We conducted a novel statistical analysis to assess variation across providers. We fitted two models to predict treatment--one using only patient characteristics (age, eGFR, sex, history of heart failure, and treatment date) and another using both patient characteristics and provider-specific random effects. With these models, we performed a generalized likelihood ratio test (GLRT) to assess whether including provider-specific random effects improved fit. ResultsThe chi-squared test confirmed significant association between treatment and eGFR (p < 0.0001). The GLRT in our novel statistical analysis found significant variation existed across providers even after accounting for patient characteristics (p < 0.0001). Visualizations of the observed treatment decisions and treatment policy models show that most of this variation across providers occurred at low eGFR levels, where the level of kidney damage at which metformin should be contraindicated is unclear. ConclusionsWhile some variation in first-line type 2 diabetes treatment was associated with eGFR, some variation may be due to provider preferences that cannot be explained by treatment guidelines. Further studies can elucidate whether such variation across providers is appropriate. Our approach can be applied to other treatment decisions to improve diabetes management. Key MessagesO_ST_ABSWhat is already known on this topicC_ST_ABSGuidelines for first-line type 2 diabetes treatments recommend metformin unless there are contraindications, such as kidney damage indicated by low estimated glomerular filtration rate (eGFR). What this study addsThis study uses a health insurance claims dataset to verify that first-line treatment is significantly associated with eGFR levels. Then, we propose a novel statistical analysis to assess whether significant variation exists across providers even after accounting for patient age, eGFR, sex, history of heart failure, and treatment date. By fitting two random effects models--one with only patient characteristics and one that also utilizes provider-specific random effects--and comparing the likelihoods of the observed treatment decisions under the two models, we find that the treatment decisions can be explained significantly better when accounting for differences among providers in treatment preferences and eGFR considerations. How this study might affect research, practice, or policyOur results suggest future studies about whether the significant variation across providers found in our analysis is appropriate may help improve first-line type 2 diabetes treatment decisions, and our novel statistical approach can be applied to evaluate variation across providers throughout the diabetes management process.

IntroductionSpontaneous abortion is a common pregnancy outcome, but incomplete recording and missing gestational age in health databases pose challenges for research. Accurate timing of the start of pregnancy is critical information in drug safety studies. ObjectivesTo review the literature on database algorithms to estimate gestational length for spontaneous abortions and clinical studies than can inform such algorithms. To estimate the average gestational age for algorithm-identified spontaneous abortions in Norway using interrupted time series analysis. MethodsWe used an algorithm to identify pregnancies registered in Norway from 2010-2020 and restricted to spontaneous abortions identified from registers of primary and specialist care, and births from the Medical Birth Registry of Norway. For births, we calculated the LMP by subtracting the recorded gestational age from the birth date. We assigned spontaneous abortions gestational ages ranging from 7 to 11 weeks and a corresponding LMP. We identified prescriptions from 70 days before to 97 days after LMP and calculated the number of antidepressant prescriptions per 10,000 pregnancies per day. We applied two-sample interrupted time series analysis with intervention points set at 28 and 55 days after LMP and compared antidepressant prescription trends after 28 gestational days for spontaneous abortions versus births. ResultsDatabase algorithms have used estimates for the gestational age at spontaneous abortion ranging from 8-10 weeks, and clinical studies suggest the mean or median gestational age at spontaneous abortion of around 9-10 weeks. In our interrupted time series analysis including 122,495 spontaneous abortions and 631,929 births, the 7-week assumption showed no post-intervention trend, suggesting underestimation. The 9-week assumption closely matched the trend for births (-0.051 prescriptions/day, 95% CI -0.090 to -0.013 vs. -0.056, 95% CI: -0.067 to - 0.046). The 8, 10, and 11-week assumptions showed less precise alignment. The best alignment occurred with the 64-day assumption (9.1 weeks). ConclusionOur study provides an empirically derived estimate for the average gestational age for algorithm-identified spontaneous abortions which can be applied in future research using the same pregnancy algorithm in Norway. While the 64-day estimate seems most accurate for our dataset, further validation studies are necessary to confirm its applicability in other contexts.

PurposeThis study evaluated the association between coronavirus disease 2019 (COVID-19) diagnosis and adverse events (AEs), including cardiovascular AEs and appendicitis, in US adults before the introduction of COVID-19 vaccines. Real-world studies of AEs after COVID-19 suggest that diagnoses of AEs and COVID-19 frequently occur on the same day and may be a source of bias. MethodsCohort and self-controlled risk interval (SCRI) designs were used in 2 US administrative claims data sources--Merative MarketScan(R) (ages 18-64 years) and Medicare (ages [&ge;] 65 years). AEs included stroke (nonhemorrhagic and hemorrhagic), acute myocardial infarction, myocarditis/pericarditis, deep vein thrombosis, pulmonary embolism (PE), disseminated intravascular coagulation (DIC), unusual-site and common-site thrombosis with thrombocytopenia syndrome, and appendicitis. In cohort analyses, weighted hazard ratios (HRs) and 95% confidence intervals (CIs) compared adults with a COVID-19 diagnosis and matched comparators. In SCRI analysis, relative incidences (RIs) and 95% CIs compared risk and reference windows within individuals. Analyses were performed starting follow-up on Time 0 and Day 1. ResultsFor cardiovascular AEs, all estimates starting follow-up on Day 1 were above 1.0 in both data sources. For cohort analyses, the strongest associations were for inpatient PE in both databases: MarketScan, HR=8.65 (95% CI, 6.06-12.35), Medicare HR=3.06 (95% CI, 2.88-3.26). For SCRI analyses, the strongest association in MarketScan was for DIC: RI=32.28 (95% CI, 17.06-61.09) and in Medicare was for myocarditis/pericarditis: RI=4.53 (95% CI, 3.89-5.27). AEs diagnosed concurrently with COVID-19 (ie, on Time 0) were common; including Time 0 in follow-up/risk windows resulted in higher RIs, as well as higher HRs for some AEs. However, some AEs (eg, stroke) were more common on Time 0 in the comparator group resulting in lower HRs. ConclusionCOVID-19 diagnoses had moderate to strong associations with cardiovascular AEs and weak or inconsistent associations with appendicitis, although estimates varied by design and methodology.

PurposeIncidence rate estimates are sensitive to a range of factors, including age, sex, and geographical setting (data source). The magnitude of the impact of drug indication on incidence rates remains underexplored. MethodsWe conducted an observational cohort study using 13 healthcare databases to estimate the incidence rates of 73 health outcomes across 8 drug classes with multiple indications. We calculated incidence rates for each drug-outcome pair and performed random-effects meta-analyses to pool results across databases. Then, we conducted variance components analysis to find the proportions of variability attributed to database, age, sex, and indication. We reported the median of the variance components across all 73 health outcomes as a measure of the magnitude of differences across indications, age, sex, and database, per drug class. ResultsAdjusting for database, age, and sex differences, the drug classes with the highest median VC were trimethoprim (0.49), SGLT-2 inhibitors (0.26), and beta blockers (0.10), while the drug class with the lowest VC was GLP-1 agonists (<0.01). Within each drug class, and adjusting for all other factors, age was frequently the strongest contributor to incidence variation (for 5/8 drug classes, the highest class-wide median VC was the age median VC), followed by database, indication, then biological sex. ConclusionThis study showed that for some drug classes, there exists substantial variation in incidence rates estimates across indications even after accounting for heterogeneity due to age, biological sex, and data source. As many drugs have multiple indications in clinical practice, it may be important to consider drug indication when estimating incidence rates in observational studies for the purpose of patient safety evaluations.

AimsTo identify an efficient file format for data delivery from large administrative registers for research, testing the full workflow from data extraction to research data management. MethodsThrough collaboration between a data delivery and a research department within the same institute, we evaluated each step from data extraction and delivery to management and usage, comparing CSV and Parquet formats. ResultsSwitching from the unstructured, text-based CSV format to the highly structured Parquet format significantly optimized all processes by reducing file sizes and saving processing time. The Parquet format also provided access to advanced data management techniques, simplifying further work. Despite these advantages, the basic programming required for Parquet format is not very different from that for CSV. We provide a tutorial and examples as online supplement. ConclusionsWe strongly recommend replacing CSV files with contemporary data formats. The Parquet file format proved to be an excellent option throughout the entire process from data extraction to research implementation

Background: Glucagon-like peptide-1 receptor agonist-based therapies (GLP) have recently emerged as promising treatments across a wide range of health conditions. These medications may have protective effects against severe long-term consequences of COVID-19 by promoting weight loss, exerting antihyperglycemic and anti-inflammatory effects, and providing cardiovascular and endothelial protection. Methods: We evaluated electronic health record data from a retrospective cohort of individuals in the National Clinical Cohort Collaborative. We included individuals with type 2 diabetes mellitus and comorbid COVID-19 who were prescribed either GLP (treatment) or a sodium-glucose co-transporter 2 inhibitor (SGLT2i) and subsequently developed acute COVID-19 between October 1, 2021, and April 1, 2023. We compared the 12-month cumulative incidence of mortality and Long COVID (Long COVID diagnosis and probable Long COVID via computational phenotype) between groups. We applied targeted maximum likelihood estimation to compare outcome risks by exposure status, controlling for covariates of interest. Results: We analyzed data from 14,215 individuals with COVID-19 and comorbid type 2 diabetes (mean age, 60 years; mean BMI, 37). Compared to SGLT2i, a prescription for GLP medication was associated with a lower risk of mortality (adjusted risk ratio [aRR] 0.71; 95% CI 0.53, 0.95), but not Long COVID diagnosis (aRR 1.01; 95% CI 0.80, 1.27) or probable Long COVID (aRR 0.94; 95% CI 0.88, 1.01). Conclusions: We found that among individuals with type 2 diabetes and comorbid COVID-19, a prescription for GLP vs. SGLT2i medications was associated with a lower risk of mortality, but not Long COVID.

AimsLimited recent data exist on prescribing patterns and patient characteristics for glucagon-like peptide 1 receptor agonists (GLP-1 RAs), an important drug class for patients with type 2 diabetes mellitus (T2D) and/or overweight or obesity. We describe trends in first-time prescribing of GLP-1 RAs. Materials and MethodsUsing aggregated US electronic health record data, we identified first-time prescriptions of GLP-1 RAs between January 2018 and June 2023 for adults receiving regular care. We describe prescribing volumes stratified by patient characteristics, specific drug, and FDA-labelled indication. The 5.5-year analysis period was divided into six-month periods. ResultsIn total, 274,562 patients were newly prescribed a GLP-1 RA, with a significant increase over time (January to June, 2018: 9,642 versus 2023: 66,569; p < 0.001). Overall, 181,860 (66.2%) patients had T2D and 229,715 (83.7%) had obesity or overweight. The proportion with T2D decreased over time (January through June, 2018: 84.1% versus 2023: 49%; p < 0.001), while the proportion with overweight or obesity increased (January through June, 2018: 75.9% versus 2023: 90.7%; p <.001). Of prescriptions with a known FDA-labelled indication (74.2%), 87% were labelled for T2D and 13% were labelled for overweight or obesity. Patients first prescribed a GLP-1 RA labelled for T2D were 59.1% female with a mean (SD) age of 58.6 (13) years, while those prescribed a GLP-1 RA labelled for overweight or obesity were 82.9% female, with a mean (SD) age of 48.2 (12.2) years. ConclusionsWe observed an increase in first-time prescribing of GLP-1 RAs overall, and a shift away from a predominately T2D population.

PurposeInformation from electronic health records (EHRs) may be incorporated into computable phenotype algorithms in efforts to overcome inaccuracies of algorithms based on administrative claims data alone. However, such efforts can be resource-intensive and unsuccessful. Assessing the feasibility of computable phenotyping for a health outcome of interest (HOI) before proceeding is therefore recommended. MethodsWe developed a systematic fitness-for-purpose (FFP) assessment process to implement concepts outlined in a previously described general framework for computable phenotyping incorporating EHR data. Our process includes verifying the HOI is well-defined, reviewing clinical information about the HOI, identifying existing algorithms and their performance, evaluating HOI clinical and data complexity, and determining an overall FFP conclusion and recommendation. We applied this process to ten HOIs lacking high-performing claims-based algorithms, selecting HOIs of public health importance that varied in clinical and data complexity, including neutropenia, pericardial effusion and drug-induced liver injury. ResultsHOIs assessed as having moderate (vs. easy) overall difficulty had characteristics such as the need for natural language processing, integration of multiple laboratory test results, or longitudinal EHR data. HOIs assessed as having high difficulty required using data from multiple EHR sources, ruling out many other potential causes, or relying on low-sensitivity diagnostic tests. Input from experts in EHR data and clinical care was crucial. ConclusionEHR data have potential to enhance accuracy of defining certain HOIs for research and surveillance compared to administrative claims data. The process and tools we created will support others in assessing FFP of HOIs for computable phenotyping. Five key pointsO_LIIncorporating electronic health record (EHR) data into computable phenotypes could improve accurate identification of health outcomes of interest (HOIs), but such work can be resource intensive. C_LIO_LIWe developed a systematic fitness-for-purpose (FFP) process and tools to assess the feasibility of computable phenotyping for HOIs. C_LIO_LISteps include identifying existing algorithms and their performance, ensuring the HOI is well-defined, evaluating clinical and data complexity, and determining a feasibility recommendation. C_LIO_LIDifficulty increased with a need for natural language processing, multiple laboratory tests, longitudinal EHR data, multiple EHR sources or ruling out other potential causes. C_LIO_LIInput from EHR data and clinical care experts was crucial to the FFP assessment process. C_LI Plain Language Summary (PLS)Attempts to identify diseases and health conditions by applying computer programs to information easily gleaned from insurance claims of tens of thousands of patients (such as FDAs ongoing safety monitoring of approved drugs or medical products) are often unsuccessful because the data lack nuance. Incorporating information from electronic health records (EHR) and patient chart notes may improve accurate identification of health outcomes. Because this can be resource-intensive, we designed a process and tools to assess the feasibility of including EHR data in computer algorithms to identify health outcomes. Steps included identifying existing algorithms and their performance, building familiarity with the outcome and making sure it is well-defined, evaluating clinical and data complexity, and determining a conclusion about feasibility. We applied our process to ten health outcomes of public health importance. Health outcomes were considered moderately difficult for computerized algorithms if they required natural language processing, integration of multiple laboratory tests, or EHR data from multiple timepoints. Health outcomes having high difficulty required using multiple EHR data types, ruling out many alternative causes of the HOI (other than medications), or relying on diagnostic tests of low accuracy. Input from EHR data and clinical care experts was crucial for the assessment process.

Background: Epic Cosmos is a relatively new centralized electronic health record dataset with high potential utility in perinatal epidemiologic research. Objectives: The study objectives were to develop replicable steps to create longitudinal, linked maternal-infant cohorts in Cosmos, assess completeness of key variables, evaluate potential selection bias with restrictions for longitudinal healthcare encounters, and provide an example epidemiologic analysis. Methods: We created maternal-infant cohorts by starting with live births during 2023-2024 recorded in the BirthFact data table and joining with additional data tables as needed. We selected and created variables for perinatal characteristics, common comorbidities, and routinely measured vital signs and laboratory values, and assessed variable completeness. We sequentially restricted the birth cohort for maternal-infant linkage and longitudinal healthcare from first-trimester prenatal care encounter through infant follow-up care within 12 weeks post-discharge from birth hospitalization. Finally, we conducted an example analysis of the association between high systolic blood pressure in the first trimester ([&ge;]140 mm Hg) and later onset of preeclampsia among those with chronic hypertension. Results: The total linked birth cohort included 2,624,186 pregnancies. Completeness was >90% for most variables assessed but was 77% for racial and ethnic group and 76% for body mass index at delivery. Characteristics of the cohort were similar to those reported for the entire United States birth population based on birth certificate data, including similar regional and racial-ethnic composition. Longitudinal cohort restriction requiring linked records from first trimester prenatal care through infant follow-up care reduced the cohort size to 509,148 pregnancies. However, restriction had minimal effects on cohort characteristics. In the example analysis, high systolic blood pressure was associated with increased risk of preeclampsia among those with chronic hypertension (aRR: 1.26; 95% CI: 1.22, 1.30). Conclusions: This study provides a rigorous and reproducible approach to creating longitudinal, linked maternal-infant cohorts in Epic Cosmos and the analytical findings suggest high data quality and representativeness.

BackgroundObesity is common among older adults in the US, but Medicare policies prohibit coverage of anti-obesity medications (AOMs) and may limit access to effective treatment. ObjectiveTo describe first-time prescribing and dispensing of AOM glucagon-like peptide-1 receptor agonists (GLP-1 RA) among eligible older adults, stratified by Medicare age eligibility. MethodsAdults aged 60-69 with overweight or obesity and without Type II diabetes (T2D) were identified from Truveta Data. Data included EHR (encounters, prescriptions, conditions, BMI) and medication dispensing for a collective of US healthcare systems. Eligibility required an outpatient office encounter between June 2021 and January 2024 with a BMI [&ge;] 27, a negative history of GLP-1 RA use, and a follow-up [&ge;]60 days later. Patients were stratified by Medicare age eligibility (60-64 vs. 65-69) and the proportions prescribed and subsequently dispensed AOM GLP-1 RA were compared. ResultsIn total, 413,833 AOM eligible older adults were included in our cohort, with 208,067 (50.3%) Medicare-aged adults and 205,766 (49.7%) adults aged 60-64. Among eligible patients, 0.2% of Medicare-aged patients and 0.4% of patients aged 60-64 were prescribed AOM GLP-1 RA, a significant difference (p < 0.01). Among those prescribed AOM GLP-1 RA, 15.2% of Medicare-aged patients and 22.7% of patients aged 60-64 were dispensed AOM GLP-1 RA within 60 days, a significant difference (p<0.01). In both age groups, prescribing and dispensing were more common for females and those with higher BMI. ConclusionsFewer than 1% of older adults with overweight or obesity and without T2D were prescribed an AOM GLP-1 RA. New use of GLP-1 RA was significantly lower for Medicare-aged adults, compared to 60-64-year-olds, with differences occurring at both medication prescribing and dispensing stages. While coverage of AOMs is limited by many insurers, Medicares unique prohibition on AOM coverage may contribute to differentially lower use among Medicare-aged adults.

Background and AimsThe glucagon-like peptide-1 receptor agonist (GLP-1 RA) semaglutide has demonstrated efficacy for the secondary prevention of cardiovascular disease among patients with overweight/obesity without diabetes mellitus. However, the comparative effectiveness of GLP-1 RA versus other antiobesity medications (e.g. phentermine-topiramate) not been evaluated. MethodsThis was a retrospective, observational, cohort study using target trial emulation methodology using the Truveta electronic health record database of more than 120 million patients. Adult patients with a body mass index (BMI) >=27 kg/m2, a history of cardiovascular disease (prior ischemic stroke, transient ischemic attack, or myocardial infarction, or known coronary artery disease, heart failure, or peripheral artery disease) without diabetes mellitus were included in the study. The primary endpoint was time to first major adverse cardiovascular or cerebrovascular event (MACCE, defined as stroke or myocardial infarction). ResultsIn total, 35,240 were included in the bupropion-naltrexone versus GLP-1 RA comparison, and 27,051 were included in the phentermine-topiramate versus GLP-1 RA comparison. In the pre-weighting cohort, GLP-1 RA use was associated with decreased hazard of MACCE compared to bupropion-naltrexone (HR 0.50 [95% confidence interval (CI) 0.36-0.69]) and phentermine-topiramate (HR 0.43 [95% CI 0.30-0.60]). In the propensity score-overlap weighted cohort, GLP-1 RA prescription was not associated with a lower hazard of MACCE than bupropion-naltrexone (aHR 0.69 [95% CI 0.47-1.00]) but was associated with a lower hazard compared to phentermine-topiramate (aHR 0.61 [95% CI 0.41-0.91]; adjusted absolute rate difference 0.98 per 1000 person-years). ConclusionsPrescription of a GLP-1 RA was associated with a lower risk of subsequent MACCE than phentermine-topiramate.

AJOG at a glanceA. Why was this study conducted? This study aimed to evaluate the risks and benefits of discontinuing maternal prescriptions for chronic hypertension and hypothyroidism during pregnancy. B. What are the key findings? Discontinuing beta-blockers and thyroid hormones was associated with some improved maternal outcomes, such as a reduced risk of miscarriage. We found limited evidence of an association between the discontinuation of multiple drugs subclasses such as renin-angiotensin system drugs and differential risk of maternal or neonatal outcomes. C. What does this study add to what is already known? This study provides observational evidence that discontinuing certain medications may not be strongly associated with differential risk of maternal or neonatal outcomes and highlights the need for further research due to limitations such as confounding by indication and potential data misclassification. Exclusion of pregnant women from clinical trials, due to ethical concerns, has limited evidence on medication safety during pregnancy, resulting in conservative guidance. Yet, rising prevalence of chronic conditions in reproductive-age women has increased medication use. This study evaluates risks and benefits of discontinuing drug prescriptions for chronic hypertension, and hypothyroidism during pregnancy using linked primary care records from a longitudinal intergenerational database. Using UK Clinical Practice Research Datalink (CPRD) GOLD, we conducted multivariable regression models, adjusted for covariates, to assess maternal treatment discontinuation on various outcomes. Cohorts of 3,232 and 3,334 pregnancies with chronic hypertension and hypothyroidism respectively were derived from the CPRD. Discontinuing vasodilator antihypertensive drugs for hypertension was associated with increased gestational age (mean difference: 3.98 weeks, 95% CI: 1.61, 6.35). Estimated associations between other antihypertensives (calcium-channel blockers, diuretics or renin-angiotensin system drugs) and any study outcome crossed the null. Discontinuing thyroid hormones for hypothyroidism were associated with reduced the odds of miscarriage (OR: 0.29, 95% CI: 0.15, 0.54) and increased gestational age (mean difference 1.84 weeks, 95% CI: 0.12, 3.57). Results were robust in sensitivity analysis. This study reports a reassuring lack of association between many drug subclasses and adverse offspring outcomes. This evidence of potential risk associated with treatment discontinuation may guide clinical decision-making for treating chronic hypertension and hypothyroidism during pregnancy in similar populations.

2OBJECTIVEThe glucagon-like peptide-1 (GLP1) and sodium-glucose co-transporter-2 (SGLT2) classes are increasingly being used around the globe to treat type 2 diabetes and its comorbidities. Decision-analytical models (DAMs) are used to evaluate the cost-effectiveness of the different products within these classes, but their assumptions regarding the treatment pathways and time until insulin may not reflect real-world practice. This study compares the model assumptions found in a recent review with registry data. RESEARCH DESIGN AND METHODSReal-world practice was represented by a nationwide registry-based cohort of 62,238 people with type 2 diabetes included in an early (2012 to 2018), and a late period (2019 to 2021). Treatment pathway assumptions were compared using counts and proportions. Time until insulin initiation was compared among groups starting on the comparators dipeptidyl peptidase-4 inhibitor (DPP4), GLP1, or SGLT2. Incidence rates, hazard ratios, and absolute risks were utilized. The latter were derived from cumulative incidence functions calculated using the Aalen-Johansen estimator, accounting for competing risks. RESULTSThe treatment pathway assumptions of initiating insulin after a short time on the comparators did not correspond with our data. Neither did the assumptions surrounding time until insulin, as the time differed between comparator group, and inclusion period, with SGLT2 having the lowest risk. Further, time until insulin was observed to be much longer in real-world practice than in the model assumptions. CONCLUSIONSKey model assumptions used to inform decision-makers on the cost-effectiveness of expensive drug classes like GLP1 and SGLT2, are flawed and need to be updated.

PurposeReal-world evidence (RWE) is essential for post-licensure monitoring of vaccine safety, as pre-licensure trials are often limited in sample size, population diversity, and duration of follow-up. The Vaccine Monitoring Collaboration for Europe (VAC4EU) has developed a structured validation pipeline that operationalizes Brighton Collaboration (BC) case definitions for real-world data (RWD) in a harmonized, scalable and reusable manner. MethodsVAC4EU developed a systematic, stepwise approach to validate vaccine safety outcomes utilizing BC case definitions where available. The approach involves: 1) Critical review of BC definition and adaptation to RWD by clinical and RWD experts; 2) Creation of dummy cases based on published reports; 3) Creation of REDCap electronic data collection forms (eDCF) incorporating decision logic to assigned levels of certainty (LOC); 4) Iterative testing of decision logic; and 5) comprehensive training of abstractors with real-time feedback. A dedicated task force assigned reference LOCs for dummy cases. Inter-rater reliability was measured using Fleiss kappa ({kappa}) by comparing abstractor LOCs to the reference standard. ResultsThe validation pipeline was applied to 16 COVID-19 vaccine safety outcomes, 13 of them had existing BC definitions. Eleven eDCFs required adaptation to accommodate missing or unstructured information common in RWD. In total, 78 dummy cases were developed across the 16 outcomes, and 15 REDCap eDCFs were created. Myocarditis and pericarditis shared a single form. Across 33 trained abstractors, 747 individual case abstractions were completed. Agreement analysis showed 93 discrepancies (12.4%) and moderate overall concordance ({kappa} = 0.55), with the lowest for Thrombosis with Thrombocytopenia Syndrome ({kappa} = -0.05). ConclusionThe VAC4EU validation pipeline provides a standardized framework for training and validating vaccine safety outcome validation using RWD. By adapting BC case definitions and emphasizing targeted abstractor training with ongoing feedback, this approach improves the reliability of post-marketing surveillance studies.

ObjectiveTo analyze the safety and efficacy of different direct oral anticoagulant agents (DOACs) compared to warfarin in patients with concomitant atrial fibrillation (AF) and valvular disease or concomitant AF and chronic kidney disease (CKD). MethodsWe conducted literature searches in MEDLINE, Embase, and EBM Reviews to examine randomized-controlled trials (RCTs) and non-RCTs that included the aforementioned patient populations treated with warfarin or DOAC (rivaroxaban, dabigatran, apixaban, or edoxaban) and assessed outcomes of bleeding, stroke, or systemic/arterial thromboembolism. Meta-analysis was performed for eligible studies using the Mantel-Haenszel method random-effects model. Results3,172 studies were screened and 154 studies were selected after two levels of screening. Meta-analysis showed that, in patients with concomitant AF and CKD, DOAC was associated with reduced bleeding in non-RCTs (OR 0.65, 95% Cl [0.49, 0.86], p=0.003), particularly in more severe CKD (eGFR < 60mL/min/1.73m2). Apixaban in particular was associated with reduced bleeding (OR 0.52, 95% Cl [0.44, 0.63], p<0.00001) and stroke incidence (OR 0.60, 95% Cl [0.41, 0.87], p=0.007). In patients with concomitant AF and valvular disease, DOAC was associated with reduced bleeding (OR 0.75, 95% CI [0.57, 0.97], p=0.03) and stroke incidence (OR 0.66, 95% CI [0.47, 0.93], p=0.02) in non-RCTs. ConclusionOur study studied populations that are typically excluded from large-scale anticoagulation studies and our findings suggest that DOACs may be superior to warfarin both in the prevention of thromboembolic event and in the reduction of bleeding risks in patients with concomitant CKD or valvular disease.

ObjectiveAngiotensin-II Receptor Blockers (ARBs) are commonly prescribed; however, their adverse events may prompt new drug prescription(s), known as prescribing cascades. We aimed to identify potential ARB-induced prescribing cascades using high-throughput sequence symmetry analysis. MethodsUsing claims data from a national sample of Medicare beneficiaries (2011-2020), we identified new ARB users aged [&ge;]66 years with continuous enrollment [&ge;]360 days before and [&ge;]180 days after ARB initiation. We screened for initiation of 446 other (non-antihypertensive) marker drug classes within {+/-}90 days of ARB initiation, generating sequence ratios (SRs) reflecting proportions of ARB users starting the marker class after versus before ARB initiation. Adjusted SRs (aSRs) accounted for prescribing trends over time, and for significant aSRs, we calculated the naturalistic number needed to harm (NNTH); significant signals were reviewed by clinical experts for plausibility. ResultsWe identified 320,663 ARB initiators (mean {+/-} SD age 76.0 {+/-} 7.2 years; 62.5% female; 91.5% with hypertension). Of the 446 marker classes evaluated, 17 signals were significant, and three (18%) were classified as potential prescribing cascades after clinical review. The strongest signals ranked by the lowest NNTH included benzodiazepine derivatives (NNTH 2130, 95% CI 1437-4525), adrenergics in combination with anticholinergics, including triple combinations with corticosteroids (NNTH 2656, 95% CI 1585-10074), and other antianemic preparations (NNTH 9416, 95% CI 6606-23784). The strongest signals ranked by highest aSR included other antianemic preparations (aSR 1.7, 95% CI 1.19-2.41), benzodiazepine derivatives (aSR 1.18, 95% CI 1.08-1.3), and adrenergics in combination with anticholinergics, including triple combinations with corticosteroids (aSR 1.12, 95% CI 1.03-1.22). ConclusionThe identified prescribing cascade signals reflected known and possibly under-recognized ARB adverse events in this Medicare cohort. These hypothesis-generating findings require further investigation to determine the extent and impact of these prescribing cascades on patient outcomes.