Inferring pregnancy episodes and outcomes within a network of observational databases.

Administrative claims and electronic health records are valuable resources for evaluating pharmaceutical effects during pregnancy. However, direct measures of gestational age are generally not available. Establishing a reliable approach to infer the duration and outcome of a pregnancy could improve pharmacovigilance activities. We developed and applied an algorithm to define pregnancy episodes in four observational databases: three US-based claims databases: Truven MarketScan® Commercial Claims and Encounters (CCAE), Truven MarketScan® Multi-state Medicaid (MDCD), and the Optum ClinFormatics® (Optum) database and one non-US database, the United Kingdom (UK) based Clinical Practice Research Datalink (CPRD). Pregnancy outcomes were classified as live births, stillbirths, abortions and ectopic pregnancies. Start dates were estimated using a derived hierarchy of available pregnancy markers, including records such as last menstrual period and nuchal ultrasound dates. Validation included clinical adjudication of 700 electronic Optum and CPRD pregnancy episode profiles to assess the operating characteristics of the algorithm, and a comparison of the algorithm's Optum pregnancy start estimates to starts based on dates of assisted conception procedures. Distributions of pregnancy outcome types were similar across all four data sources and pregnancy episode lengths found were as expected for all outcomes, excepting term lengths in episodes that used amenorrhea and urine pregnancy tests for start estimation. Validation survey results found highest agreement between reviewer chosen and algorithm operating characteristics for questions assessing pregnancy status and accuracy of outcome category with 99-100% agreement for Optum and CPRD. Outcome date agreement within seven days in either direction ranged from 95-100%, while start date agreement within seven days in either direction ranged from 90-97%. In Optum validation sensitivity analysis, a total of 73% of algorithm estimated starts for live births were in agreement with fertility procedure estimated starts within two weeks in either direction; ectopic pregnancy 77%, stillbirth 47%, and abortion 36%. An algorithm to infer live birth and ectopic pregnancy episodes and outcomes can be applied to multiple observational databases with acceptable accuracy for further epidemiologic research. Less accuracy was found for start date estimations in stillbirth and abortion outcomes in our sensitivity analysis, which may be expected given the nature of the outcomes.

Channeling in the Use of Nonprescription Paracetamol and Ibuprofen in an Electronic Medical Records Database: Evidence and Implications.

INTRODUCTION: Over-the-counter analgesics such as paracetamol and ibuprofen are among the most widely used, and having a good understanding of their safety profile is important to public health. Prior observational studies estimating the risks associated with paracetamol use acknowledge the inherent limitations of these studies. One threat to the validity of observational studies is channeling bias, i.e. the notion that patients are systematically exposed to one drug or the other, based on current and past comorbidities, in a manner that affects estimated relative risk. OBJECTIVES: The aim of this study was to examine whether evidence of channeling bias exists in observational studies that compare paracetamol with ibuprofen, and, if so, the extent to which confounding adjustment can mitigate this bias. STUDY DESIGN AND SETTING: In a cohort of 140,770 patients, we examined whether those who received any paracetamol (including concomitant users) were more likely to have prior diagnoses of gastrointestinal (GI) bleeding, myocardial infarction (MI), stroke, or renal disease than those who received ibuprofen alone. We compared propensity score distributions between drugs, and examined the degree to which channeling bias could be controlled using a combination of negative control disease outcome models and large-scale propensity score matching. Analyses were conducted using the Clinical Practice Research Datalink. RESULTS: The proportions of prior MI, GI bleeding, renal disease, and stroke were significantly higher in those prescribed any paracetamol versus ibuprofen alone, after adjusting for sex and age. We were not able to adequately remove selection bias using a selected set of covariates for propensity score adjustment; however, when we fit the propensity score model using a substantially larger number of covariates, evidence of residual bias was attenuated. CONCLUSIONS: Although using selected covariates for propensity score adjustment may not sufficiently reduce bias, large-scale propensity score matching offers a novel approach to consider to mitigate the effects of channeling bias.

Feasibility and utility of applications of the common data model to multiple, disparate observational health databases.

OBJECTIVES: To evaluate the utility of applying the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) across multiple observational databases within an organization and to apply standardized analytics tools for conducting observational research. MATERIALS AND METHODS: Six deidentified patient-level datasets were transformed to the OMOP CDM. We evaluated the extent of information loss that occurred through the standardization process. We developed a standardized analytic tool to replicate the cohort construction process from a published epidemiology protocol and applied the analysis to all 6 databases to assess time-to-execution and comparability of results. RESULTS: Transformation to the CDM resulted in minimal information loss across all 6 databases. Patients and observations excluded were due to identified data quality issues in the source system, 96% to 99% of condition records and 90% to 99% of drug records were successfully mapped into the CDM using the standard vocabulary. The full cohort replication and descriptive baseline summary was executed for 2 cohorts in 6 databases in less than 1 hour. DISCUSSION: The standardization process improved data quality, increased efficiency, and facilitated cross-database comparisons to support a more systematic approach to observational research. Comparisons across data sources showed consistency in the impact of inclusion criteria, using the protocol and identified differences in patient characteristics and coding practices across databases. CONCLUSION: Standardizing data structure (through a CDM), content (through a standard vocabulary with source code mappings), and analytics can enable an institution to apply a network-based approach to observational research across multiple, disparate observational health databases.

Fidelity assessment of a clinical practice research datalink conversion to the OMOP common data model.

BACKGROUND: The unique structure and coding of the Clinical Practice Research Datalink (CPRD) presents challenges for epidemiologic analysis and for comparisons with other databases. To address this limitation we sought to transform CPRD into the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). METHODS: An extraction, transformation and loading process was developed, which detailed source code mappings, Read code domain classification, an imputation algorithm for drug duration and special handling of lifestyle/clinical data. Completeness and accuracy of the above elements were assessed. A final validation exercise involved replication of a published case-control study that examined use of nonsteroidal anti-inflammatory drugs (NSAIDs) and the risk of first-time acute myocardial infarction (AMI) in raw CPRD data and the CPRD CDM. FINDINGS: All elements of the CPRD CDM transformation were assessed to be of high quality. 99.9 % of database condition records and 89.7 % of database drug records were mapped (majority unmapped drugs were devices and over-the-counter products); 3.1 % of duration imputations were deemed possibly erroneous and prevalences for selected conditions and drugs across CPRD raw and CDM data were equivalent. Results between the replication raw data and CDM study agreed for conditions, demographics and lifestyle data with slight NSAID exposure data loss owing to unmapped drugs. CONCLUSION: CPRD can be accurately transformed into the OMOP CDM with acceptable information loss across drugs, conditions and observations. We determined that for a particular use, case CDM structure was adequate and mappings could be improved but did not substantially change the results of our analysis.

Analysis of patient claims data to determine the prevalence of hidradenitis suppurativa in the United States.

BACKGROUND: Recent prevalence estimates for hidradenitis suppurativa (HS), a chronic inflammatory skin condition, are limited by timeliness, population size, and generalizability. OBJECTIVE: We sought to develop prevalence estimates for HS in the United States using large health care claims databases. METHODS: A retrospective analysis used PharMetrics Integrated Database to gather health care claims information for HS among patients with 12 or more months of continuous enrollment in a commercial health care plan throughout 2007. Included patients had: 1 or more diagnoses with International Classification of Diseases, Ninth Revision, Clinical Modification code 705.83 for HS during 2007 without a Current Procedural Terminology code for HS; 1 or more Current Procedural Terminology codes of 11450, 11451, 11462, 11463, 11470, or 11471 during 2007 without International Classification of Diseases, Ninth Revision, Clinical Modification code 705.83; or both. Age- and gender-specific prevalence projections were calculated. RESULTS: Among included patients (n = 7927), mean age (SD) was 38.2 (14.73) years, and 5834 (74%) were women. Most patients (n = 5205; 66%) were aged 30 to 64 years. The overall prevalence estimate was 0.053% (95% confidence interval 0.051-0.054). When adjusted for gender and age, prevalence rates were 0.052% and 0.051%, respectively. The most common procedures for HS were excision of skin and subcutaneous tissue axillary/inguinal simple or intermediate repair. LIMITATIONS: Limitations were a health insured-only population; 12-month enrollment period for 2007; HS-specific procedural codes; and possible HS misclassifications. CONCLUSION: We found a low rate of clinically detected HS (0.053%; approximately 146,000-162,000 patients in the United States in 2007), with affected persons almost 3 times as likely to be female and the highest prevalence in those aged 18 to 44 years.

Dose patterns in commercially insured subjects chronically exposed to opioids: a large cohort study in the United States.

BACKGROUND: Little data exist on how opioid doses vary with the length of exposure among chronic opioid users. METHODS: To characterize the change in the dosage of opioids over time, a retrospective cohort study using the PharMetrics database for the years 1999 through 2008 was conducted. Individuals exposed to opioids in 2000 who had 2 opioid dispensings at least 6 months apart and were opioid naive (did not receive any opioid 6 month before their exposure in 2000) were included. The date of the first dispensing in 2000 was defined as the index date and the dispensing had to be for a strong and full agonist opioid. All opioid doses were converted to oral morphine equivalent doses. Exposure was classified as continuous or intermittent. Mean, median, interquartile range, and 95th percentile of opioid dose over 6-month periods, as well as the percentage of subjects who ever received a high or very high opioid dose, were calculated. RESULTS: Among the 48,986 subjects, the mean age was 44.5 years and 54.5% were women. Intermittent exposure was observed in 99% of subjects; continuous exposure was observed in 1% of subjects. The mean duration of exposure for the subjects who were continuously exposed to opioids was 477 days. In subjects with no cancer diagnosis who were continuously exposed to opioids, the mean, 25th, 50th, and 75th percentile of dose was stable during the first 2 years of use, but the 95th percentile increased. Seven percent of them were exposed to doses of 180 mg or more of morphine at some point. CONCLUSIONS: Dose escalation is uncommon in subjects with intermittent exposure to opioids. For subjects with continuous exposure to opioids who have cancer, doses rise substantially with time. For those without cancer, doses remain relatively stable for the first 2 years of use, but subsequently increase. Seven percent of subjects with no cancer diagnosis will be exposed to daily doses of 180 mg or more of morphine equivalent at some point.