Prenatal Exposure to Valproic Acid Across Various Indications for Use.

Importance: Teratogenic outcomes associated with valproic acid use represent a substantial concern for persons of childbearing age. Regulatory agencies worldwide have enhanced warnings or implemented risk minimization programs to reduce exposure during pregnancy. Objectives: To determine pregnancy rates during valproic acid use and concomitant contraception use across indications. Design, Setting, and Participants: This retrospective cohort study used data from the Merative MarketScan commercial claims databases from January 1, 2005, to December 31, 2020, to identify female patients aged 12 to 44 years who initiated valproic acid treatment and had continuous insurance enrollment 6 months before initiation and 9 months after treatment end. A treatment episode included consecutive prescription fills that occurred within 7 days from the end of the days' supply of the previous dispensing. Data were analyzed from March 1 to September 10, 2023. Main Outcomes and Measures: Treatment episodes were categorized by inferred indication using diagnoses preceding treatment initiation, including epilepsy, migraine or headache, mood disorders, and unknown or off-label uses. Pregnancy incidence rate ratios (IRRs) were calculated and were adjusted for age and calendar year. Contraceptive use (prescription contraceptives, intrauterine devices, and implants) during treatment was examined. Results: The cohort included 165 772 valproic acid treatment episodes among 69 390 women (mean [SD] age, 29.8 [10.0] years). Mood disorders (42.5%) were the most common indication, followed by migraine or headache (20.1%), with epilepsy playing a minor role (14.9%). Pregnancy incidence rates during valproic acid use remained unchanged, with a rate of 1.74 (95% CI, 1.14-2.53) per 100 person-years in 2005 and a rate of 1.90 (95% CI, 1.16-3.12) per 100 person-years in 2019. Compared with epilepsy, pregnancy rates were more than double for mood disorder (IRR, 2.16 [95% CI, 1.93-2.42]) and migraine or headache (IRR, 2.01 [95% CI, 1.92-2.09]). Few treatment episodes coincided with contraceptive use (37 012 [22.3%]), and oral dosage forms were the most common (27 069 [73.1%]). Conclusions and Relevance: In this cohort study of patients of childbearing age who used valproic acid, pregnancy rates during valproic acid use did not decrease despite enhanced US Food and Drug Administration safety communications, and contraception use remained low. Patients with migraine and mood disorders accounted for the largest proportion of valproic acid use and had the highest pregnancy rates, while patients with epilepsy had the lowest. These findings suggest a need to enhance efforts to mitigate prenatal exposure to valproic acid, especially for indications where the risk of use during pregnancy outweighs the benefit.

Prenatal Care Initiation and Exposure to Teratogenic Medications.

Importance: With new legal abortion restrictions, timing of prenatal care initiation is critical to allow for discussion of reproductive options among pregnancies exposed to teratogenic medications. Objective: To investigate the prevalence of prenatal exposure to teratogenic medications and prenatal care initiation across gestational weeks. Design, Setting, and Participants: This descriptive, population-based cross-sectional study used health encounter data from a national sample of individuals with employer-sponsored health insurance. A validated algorithm identified pregnancies among persons identifying as female that ended with a live or nonlive outcome between January 2017 and December 2019. Data were analyzed from December 2022 to December 2023. Exposures: Prenatal exposure to any of 137 teratogenic medications, measured via pharmacy and medical claims. Measurement of prenatal care initiation was adapted from the Children's Health Care Quality Measures. Main Outcomes and Measures: Prevalence of prenatal exposure to teratogens and prenatal care initiation by gestational week. Timing of prenatal teratogenic exposure was compared with timing of prenatal care initiation and legal abortion cutoffs. Results: Among 639 994 pregnancies, 472 472 (73.8%; 95% CI, 73.7%-73.9%) had a live delivery (mean [SD] age, 30.9 [5.4] years) and 167 522 (26.2%; 95% CI, 26.1%-26.3%) had a nonlive outcome (mean [SD] age, 31.6 [6.4] years). Of pregnancies with live deliveries, 5.8% (95% CI, 5.7%-5.8%) were exposed to teratogenic medications compared with 3.1% (95% CI, 3.0%-3.2%) with nonlive outcomes. Median time to prenatal care was 56 days (IQR, 44-70 days). By 6 weeks' gestation, 8186 pregnancies had been exposed to teratogenic medications (25.2% [95% CI, 24.7%-25.7%] of pregnancies exposed at any time during gestation; 1.3% [95% CI, 1.3%-1.3%] of all pregnancies); in 6877 (84.0%; 95% CI, 83.2%-84.8%), prenatal care was initiated after 6 weeks or not at all. By 15 weeks, teratogenic exposures had occurred for 48.9% (95% CI, 48.4%-49.5%) of all teratogen-exposed pregnancies (2.5% [2.4-2.5] of all pregnancies); prenatal care initiation occurred after 15 weeks for 1810 (16.8%; 95% CI, 16.1%-17.5%) with live deliveries and 2975 (58.3%; 95% CI, 56.9%-59.6%) with nonlive outcomes. Teratogenic medications most used within the first 15 gestational weeks among live deliveries included antiinfectives (eg, fluconazole), anticonvulsants (eg, valproate), antihypertensives (eg, lisinopril), and immunomodulators (eg, mycophenolate). For nonlive deliveries, most antihypertensives were replaced by vitamin A derivatives. Conclusions and Relevance: In this cross-sectional study, most exposures to teratogenic medications occurred in early pregnancy and before prenatal care initiation, precluding prenatal risk-benefit assessments. Prenatal care commonly occurred after strict legal abortion cutoffs, prohibiting consideration of pregnancy termination if concerns about teratogenic effects arose.

Core concepts in pharmacoepidemiology: Measurement of medication exposure in routinely collected healthcare data for causal inference studies in pharmacoepidemiology.

BACKGROUND: Observational designs can complement evidence from randomized controlled trials not only in situations when randomization is not feasible, but also by evaluating drug effects in real-world, considering a broader spectrum of users and clinical scenarios. However, use of such real-world scenarios captured in routinely collected clinical or administrative data also comes with specific challenges. Unlike in trials, medication use is not protocol based. Instead, exposure is determined by a multitude of factors involving patients, providers, healthcare access, and other policies. Accurate measurement of medication exposure relies on a similar broad set of factors which, if not understood and appropriately addressed, can lead to exposure misclassification and bias. AIM: To describe core considerations for measurement of medication exposure in routinely collected healthcare data. METHODS: We describe the strengths and weaknesses of the two main types of routinely collected healthcare data (electronic health records and administrative claims) used in pharmacoepidemiologic research. We introduce key elements in those data sources and issues in the curation process that should be considered when developing exposure definitions. We present challenges in exposure measurement such as the appropriate determination of exposure time windows or the delineation of concomitant medication use versus switching of therapy, and related implications for bias. RESULTS: We note that true exposure patterns are typically unknown when using routinely collected healthcare data and that an in-depth understanding of healthcare delivery, patient and provider decision-making, data documentation and governance, as well as pharmacology are needed to ensure unbiased approaches to measuring exposure. CONCLUSIONS: Various assumptions are made with the goal that the chosen exposure definition can approximate true exposure. However, the possibility of exposure misclassification remains, and sensitivity analyses that can test the impact of such assumptions on the robustness of estimated medication effects are necessary to support causal inferences.

Validation of an ICD-9-Based Algorithm to Identify Stillbirth Episodes from Medicaid Claims Data.

INTRODUCTION: In administrative data, accurate timing of exposure relative to gestation is critical for determining the effect of potential teratogen exposure on pregnancy outcomes. OBJECTIVE: To develop an algorithm for identifying stillbirth episodes in the ICD-9-CM era using national Medicaid claims data (1999-2014). METHODS: Unique stillbirth episodes were identified from clusters of medical claims using a hierarchy that identified the encounter with the highest potential of including the actual stillbirth delivery and that delineated subsequent pregnancy episodes. Each episode was validated using clinical detail on retrieved medical records as the gold standard. RESULTS: Among 220 retrieved records, 197 were usable for validation of 1417 stillbirth episodes identified by the algorithm. The positive predictive value (PPV) was 64.0% (57.3-70.7%) overall, 80.4% (73.8-87.1%) for inpatient episodes, 28.2% (14.1-42.3%) for outpatient-only episodes, and 20.0% (2.5-37.5%) for outpatient episodes with overlapping hospitalizations. The absolute difference between the dates of the algorithm-specified stillbirth delivery and the medical record-based event was 4.2 ± 24.3 days overall, 1.7 ± 7.7 days for inpatient episodes, 14.3 ± 51.4 days for outpatient-only episodes, and 1.0 ± 2.0 days for outpatient episodes that overlapped with a hospitalization. Excluding all outpatient episodes, as well as pregnancies involving multiple births, the PPV increased to 82.7% (76.8-89.8%). CONCLUSIONS: Our algorithm to identify stillbirths from administrative claims data had a moderately high PPV. Positive predictive value was substantially increased by restricting the setting to inpatient episodes and using only input diagnostic codes for singleton stillbirths.

Risk of fetal or neonatal death or neonatal intensive care unit admission associated with gadolinium magnetic resonance imaging exposure during pregnancy.

BACKGROUND: Concerns have been raised about prenatal exposure to magnetic resonance imaging with gadolinium-based contrast agents because of nonclinical findings of gadolinium retention in fetal tissue and 1 population-based study reporting an association with adverse pregnancy outcomes. OBJECTIVE: This study aimed to evaluate the association between prenatal magnetic resonance imaging exposure with and without gadolinium-based contrast agents and fetal and neonatal death and neonatal intensive care unit admission. STUDY DESIGN: We constructed a retrospective cohort of >11 million Medicaid-covered pregnancies between 1999 and 2014 to evaluate the association between prenatal magnetic resonance imaging exposure with and without gadolinium-based contrast agents and fetal and neonatal death (primary endpoint) and neonatal intensive care unit admissions (secondary endpoint). Medicaid claims data were linked to medical records, Florida birth and fetal death records, and the National Death Index to validate the outcomes and gestational age estimates. Pregnancies with multiples, concurrent cancer, teratogenic drug exposure, magnetic resonance imaging focused on fetal or pelvic evaluation, undetermined gadolinium-based contrast agent use, or those preceded by or contemporaneous with congenital anomaly diagnoses were excluded. We adjusted for potential confounders with standardized mortality ratio weighting using propensity scores. RESULTS: Among 5991 qualifying pregnancies, we found 11 fetal or neonatal deaths in the gadolinium-based contrast agent magnetic resonance imaging group (1.4%) and 73 in the non-gadolinium-based contrast agent magnetic resonance imaging group (1.4%) with an adjusted relative risk of 0.73 (95% confidence interval, 0.34-1.55); the neonatal intensive care unit admission adjusted relative risk was 1.03 (0.76-1.39). Sensitivity analyses investigating the timing of magnetic resonance imaging or repeat magnetic resonance imaging exposure during pregnancy and simulating the impact of exposure misclassification corroborated these results. CONCLUSION: This study addressed the safety concerns related to prenatal exposure to gadolinium-based contrast agents used in magnetic resonance imaging and the risk thereof on fetal and neonatal death or the need for neonatal intensive care unit admission. Although the results on fatal or severe acute effects are reassuring, the impact on subacute outcomes was not evaluated.

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

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

An evaluation of co-use of chloroquine or hydroxychloroquine plus azithromycin on cardiac outcomes: A pharmacoepidemiological study to inform use during the COVID19 pandemic.

Background: Chloroquine or hydroxychloroquine (chloroquine) plus azithromycin is considered as therapy for COVID-19. With benefit evaluations underway, safety concerns due to potential additive effects on QTc prolongation should be addressed. Objective: We compared risk of cardiac adverse events between combinations of chloroquine and azithromycin and chloroquine and amoxicillin. Methods: We conducted a retrospective cohort study using the IBM MarketScan Commercial Claims and Medicare Supplemental Databases, 2005-2018. We included autoimmune disease patients aged ≥18 years initiating azithromycin or amoxicillin for ≥5 days during chloroquine treatment. Patients had continuous insurance coverage ≥6 months before combination use until 5 days thereafter or inpatient death. Two outcomes were sudden cardiac arrest/ventricular arrhythmias (SCA/VA) and cardiac symptoms. We followed patients for up to 5 days to estimate hazard ratios (HR). Covariates were adjusted using stabilized inverse probability treatment weighting. Results: We identified two SVC/VA events among >145,000 combination users. The adjusted incidence of cardiac symptoms among azithromycin and amoxicillin users was 276 vs 254 per 10,000 person-years with an adjusted HR of 1.10 (95%CI, 0.62-1.95). Conclusion: Combination use of chloroquine and azithromycin at routine doses did not show pronounced increases in arrhythmias in this real-world population, though small sample size and outcome rates limit conclusions.

Updating the Cost Effectiveness of Oral Anticoagulants for Patients with Atrial Fibrillation Based on Varying Stroke and Bleed Risk Profiles.

BACKGROUND: Previous investigations into the cost effectiveness of direct oral anticoagulants only considered individual stroke risk but not bleed risk even though bleeding is an important and potentially fatal side effect for anticoagulated patients. OBJECTIVE: This study aimed to evaluate the cost effectiveness of dabigatran, rivaroxaban, apixaban, and edoxaban vs warfarin in patients with atrial fibrillation with varying stroke/bleed risk profiles over a lifetime horizon. METHODS: A Markov micro-simulation was adapted to examine the lifetime costs and quality-adjusted survival of five anticoagulants from a US private payer's perspective. The study hypothetical cohort consisted of 10,000 patients with atrial fibrillation with age, CHA2DS2-VASc, and HAS-BLED scores similar to a commercially insured patient with atrial fibrillation cohort. Model input parameters including the efficacy and safety of each strategy, utilities, and cost were estimated from public sources, published literature, and analysis conducted in the IBM MarketScan database. Lifetime cost, quality-adjusted life-years, and incremental cost-effectiveness ratios were assessed for each treatment strategy. Subgroup analyses stratified by age, stroke risk score alone, bleed risk score alone and both were performed. Uncertainty was assessed by a deterministic sensitivity analysis and a probabilistic sensitivity analysis. RESULTS: The base-case analysis suggested dabigatran was the optimal treatment with an incremental cost-effectiveness ratio of $35,055 per quality-adjusted life-year relative to warfarin. Subgroup analyses stratified by age, stroke risk score, and bleed risk score alone were largely consistent with the base-case analysis. Subgroup analyses stratified by both stroke and bleed risk score showed edoxaban was the preferred treatment in patients with a low stroke and a low or medium bleed risk, and patients with a high stroke and low bleed risk. Apixaban was the preferred treatment in patients with a medium stroke and high bleed risk. Results of the deterministic sensitivity analysis indicate the model results were most sensitive to the drug cost and hazard ratio for stroke and bleeding event. Results of the probability sensitivity analysis showed dabigatran is cost effective vs. other treatments in 32.8% and 42.4% of iterations at a willingness to pay of $50,000/quality-adjusted life-year and a willingness to pay of $100,000/quality-adjusted life year, respectively. CONCLUSIONS: From a US private payer's perspective, dabigatran appears cost effective compared with other anticoagulants. This study indicated risk stratification especially considering both stroke and bleed risk simultaneously is important not only in clinical practice but also in health technology assessment exercises among patients with atrial fibrillation.

Developing and validating the Personal Risk of Oncogenic Human Papillomavirus infection score in US Women.

BACKGROUND: Determining risk scores for genital high-risk human papillomavirus (HRHPV) infection in women will support more efficient cervical cancer screening strategies. OBJECTIVE: We developed and validated point scores to predict the likelihood of any genital HRHPV infection in women. METHODS: We conducted the cross-sectional analysis in 2017 and used data from the 2005-14 US National Health and Nutrition Examination Survey (7337 women aged 25-59 years; 6300 women aged 30-59 years). Predictors were reproductive health practices, risk behaviors and demographic variables. The outcome was a positive result for any of the 21 genital HRHPV genotypes. The 2005-12 cohorts were used as training and testing sets to develop scores that best classified women into three risk groups: low risk (<20%), average risk (20-30%) and high risk (>30%). The 2013-14 cohort was used to validate the final scores. RESULTS: Two-point scores with six self-reported variables were created to predict any HRHPV risks for the two age groups: the Personal Risk of Oncogenic HPV (PRO-HPV25) for women aged 25-59 years old and PRO-HPV30 for women aged 30-59 years old. The scores were successfully prospectively validated, with good calibration with regards to the predicted and observed rates of HRHPV infection. The scores had fair discrimination (c-statistics: 0.67-0.68). CONCLUSION: The PRO-HPV risk scores can identify groups at low, average and high risk of genital HRHPV infection. This information can be used to prioritize women for cervical cancer screening in low-resource settings or to personalize screening intervals.

Signs and symptoms of Group A versus Non-Group A strep throat: A meta-analysis.

Introduction: Both non-Group A streptococcal (non-GAS) pharyngitis and Group A streptococcal (GAS) pharyngitis are commonly found in patients with sore throat. It is not known whether or not they present with similar signs and symptoms compared to patients with non-streptococcal pharyngitis. Methods: MEDLINE was searched for prospective studies that reported throat culture for both GAS and non-GAS as a reference standard, and reported at least one sign, symptom, or the Centor score. Summary estimates of sensitivity, specificity, likelihood ratios (LR+ and LR-), and diagnostic odds ratios (DOR) were calculated using a bivariate random effects model. Summary receiver operating characteristic (ROC) curves were created for key signs and symptoms. Results: Eight studies met our inclusion criteria. Tonsillar exudate had the highest LR+ for both GAS and non-GAS pharyngitis (1.53 versus 1.71). The confidence intervals of sensitivity, LR+, LR-, and DOR for all signs, symptoms, and the Centor score between two groups overlapped, with the relative difference between sensitivities within 15% for arthralgia or myalgia, fever, injected throat, tonsillar enlargement, and tonsillar exudate. Larger differences in sensitivities were observed for sore throat, cervical adenopathy, and lack of a cough, although the difference for lack of a cough largely due to a single outlier. Discussion: Signs and symptoms of patients with GAS and non-GAS pharyngitis are generally similar. No signs or symptoms clearly distinguish GAS from non-GAS infection. Further work is needed to determine whether Group C streptococcus is a pathogen that should be treated.