Advera Health Analytics is a global leader in pharmacovigilance software, analytics, and data at the leading edge of drug safety science. Founded in 2010 by a group of healthcare entrepreneurs, Advera Health's mission is to mitigate risk in the healthcare system by improving the transparency and actionability of drug safety data through the curation and aggregation of large disparate datasets and the application of proprietary analytics.
Advera Health's FDA Adverse Event Reporting System (FAERS) is the gold standard for cleaned, deduplicated, and dynamically linked FAERS data. The FAERS database is a centralized, computerized, information dataset that is broadly used by FDA and other pharmacovigilance experts for post-marketing drug safety surveillance. Although FAERS data is made publicly available as a quarterly download on the FDA's website, the raw data can present multiple complications. These include the inability to collect all relevant data under a single drug record, assigning terms to appropriate adverse event naming conventions, deduplicating records, and mapping to other datasets. In order to diminish these complications in FAERS data and obtain accurate insight, Advera has developed a methodology which applies extensive data cleansing and normalization.
Asset Class: Equities
Data Frequency: Quarterly
Delivery Frequency: Quarterly
History: Data available back to 1997
Advera Health Analytics Evidex platform has been used to extract over 14 million Adverse Effect (AE) case reports from the publicly available FAERS database and the previous Legacy AERS (LAERS). The FAERS/LAERS ASCII quarterly data extract files for each year and quarter are parsed into seven files containing patient demographic information, drugs taken, adverse events, patient outcomes, diagnosed patient indications, report sources, and detailed drug therapy information.
Data validation steps are run to verify that all case reports had key identification fields filled in. These include the Primary ID/ISR, Case ID, drug sequence identification, and MedDRA® AE terms. Case reports missing or containing malformed key identification fields are discarded. AE information is coded according to the latest MedDRA version.
Drug Mapping Steps
Automated and manual steps are combined when mapping the raw drug data collected from each quarterly extract. The automation process matches drug details from each case report based on four data fields (i.e., the FAERS raw drug name, route of administration, dose form, and the application number of a manually curated master reference set created and maintained by Evidex).
Entries in the reference set are composed of the same four fields linked to an Evidex drug identifier (ID). Cases within a quarterly extract that had drug details which exactly match those in the reference set are automatically mapped to Evidex drug IDs.
Drug details in a case report within a quarterly extract that are not exact matches are sent to an analyst team for manual review and classification (manual curation). The analysts compare the raw drug name, route, dose form, and NDA number fields in question against available prescribing information and links them to an existing or new drug ID. Completed, manually assigned drug mapping pairs are then added back to the "Reference Set" to improve future automation.
One of the major hurdles to working with the FAERS database is the substantial number of "duplicate" reports. Advera has developed a sophisticated multi-step process for deduplicating data. Every report has a Case ID, which is the main ID for individual patients with one or more reports. In addition, each Case ID can have multiple Primary IDs (or ISR in LAERS) that represent different case versions. Typically, the first case report for a patient is tagged as the initial report with subsequent reports tagged as follow ups. When there is more than one Primary ID for the same Case ID, each Case ID is only counted once, with the latest report version retained.
The second deduplication step helps account for scenarios where the duplicated case versions are not linked by the FDA processing logic to the same Case ID. Advera applies an externally validated methodology of deduplicating the FAERS database called "Adverse Event Open Learning through Universal Standardization (AEOLUS)" to account for duplicate events that may be recorded under different Case IDs. This process captures duplicate cases that may have been listed with different Case IDs by evaluating four "key" demographic fields (i.e., event date, age, gender, reporter country), a concatenated alphabetic ordered list of raw drug names, and a concatenated alphabetic list of raw AEs (i.e., preferred terms).
These additional duplicate cases are counted only once in all subsequent analyses. When the deduplication process is complete, the final standardized dataset is consolidated into a clean format to allow for analysis.
Pharmaceutical Investors – Risk Mitigation
Pre-approval drug clinical trial programs often fail to uncover serious and life-threatening side effects of drugs. The Advera FAERS DataFeed provides detailed information based on a broader universe of patient's experiences. This allows investors to develop an understanding of which drugs are at risk of facing FDA action based on previously undiscovered AEs.
Monitoring long-term trends in AEs can help investors predict labelling changes and other FDA action.
The first quarter after a drug is launched commercially can also be telling. Broad adoption may uncover previously unseen severe AEs.