NICAR Introduction 
September 2012

NICAR's latest version of the AERS data includes 1997 through the first two quarters of 2012.

The FDA Adverse Events Reporting System and its predecessor, the Spontaneous Reporting System, contain over 3 million reported medical problems associated with pharmaceutical products. AERS replaced the SRS on November 1, 1997; each uses a different set of tables and codes. This introduction discusses both SRS and AERS data. This update includes records from 1997 through the first quarter of 2012.

The FDA relies on adverse event reporting to flag safety issues and identify products for further epidemiological study, conducted by the Center for Drug Evaluation and Research (CDER) and the Center for Biologics Evaluation and Research (CBER).  It may ultimately prompt regulatory responses such as drug labeling changes, letters to health care professionals or market withdrawals.  See http://www.fda.gov/cder/aers/. Journalists who have used this data recommend using it to guide reporting on consumer medical issues, the FDA, the pharmaceutical industry or to flag potentially dangerous drugs on the market.

The agency requires product manufacturers and distributors to report adverse events regularly in accordance with 21 CFR 310.305 and 314.80.

According to a March, 2005, GAO report, the FDA has the authority to require that drug sponsors report adverse events to FDA with different reporting schedules based on the seriousness of the event, whether the event has been previously identified and the reactions is included in the drug�s label. Sponsors must report serious, unlabeled adverse events to FDA within 15 days of learning about them. Sponsors are required to report other adverse events quarterly for 3 years, then annually thereafter in the form of periodic adverse event reports. In addition, health care providers and patients can voluntarily submit adverse event reports to FDA through its MedWatch program. Adverse event reports become part of FDA�s computerized database known as the Adverse Event Reporting System (AERS).

FDA has the authority to withdraw the approval of a drug on the market for safety-related and other reasons, although it rarely does so. FDA does not have explicit authority to require that drug sponsors take other safety actions; however, when FDA identifies a potential problem, sponsors generally negotiate with FDA to develop a mutually agreeable remedy to avoid other regulatory action. For example, if FDA determines that an approved drug may produce adverse events not previously identified, FDA and the sponsor may negotiate on revised labeling for the drug, and then FDA may issue an accompanying Public Health Advisory for patients and health care providers that describe the safety information. FDA may also request that the sponsor restrict the distribution of the drug in order to minimize a significant risk associated with the drug. (The GAO report where this and other information can be found is in the documents folder on the CD.)

Adverse drug experiences are defined as: "Any adverse event associated with the use of a drug in humans, whether or not considered drug related, including the following: An adverse event occurring in the course of the use of a drug product in professional practice; an adverse event occurring from drug overdose whether accidental or intentional; an adverse event occurring from drug abuse; an adverse event occurring from drug withdrawal; and any failure of expected pharmacological action." (21 CFR 314.80)

****************************
IMPORTANT NOTES FOR THE USER
****************************

Read the FDA's readme files for IMPORTANT CAVEATS on interpreting the AERS data.  It is not a stand-alone source or "proof" that certain drugs are the most dangerous, etc.  The March_2005_GAO provides similar caveats for the use of AERS.  The database does not provide conclusive evidence for comparing the safety of one drug to another.  Do not assume that the drug was necessarily the cause of each adverse event.  Complicated medical conditions, multiple drug interactions and other variables may contribute.  Due to the voluntary nature of some reporting, this data does not represent ALL adverse drug events. Emphasis on reporting and efficiency of collecting reports has fluctuated, as seen by the relatively small number of records for the earlier years in this dataset and the incomplete entries throughout. Yet for reporters as well as epidemiologists, it can provide important insight for further investigation.  
The March_2005_GAO report notes that rare adverse events can provide a more likely link between a drug and the event.

Other FDA documents can save you lots of time and give you more accurate results by reviewing how to properly join tables with specific fields. Since a given patient can take more than one drug, one record in the table of demographics might result in multiple records in the drug12, reac12, indi12, sour12 and ther12 tables. Study carefully the use of the ISR and drugseq fields to make those records join properly in the Asc_nts.doc file.

Regulatory details on adverse event reporting are online at: http://www.fda.gov/medwatch/report/mfg.htm. For MedWatch guidelines for reporting, see http://www.fda.gov/Safety/MedWatch/HowToReport/default.htm. More information on specific drugs, FDA warning letters, etc. can be found at www.fda.gov

*****************************
DOCUMENTS INCLUDED
*****************************

The Documents (AERS_Docs folder):
 
Readfirst.txt: A NICAR document (the one you are reading now) that combines information from the FDA and NICAR's notes related to NICAR's data processing steps and/or advice to users.

Asc_nts.doc: Contains FIELD DEFINITIONS, CODES and tips for accurate joins.  Developed by the FDA.

FDA_Readme: This is the original readme text file included with the most recent AERS data. Most of the file descriptions in this document apply to the original text files processed by NICAR, however, there are some notes relevant to how the data has been maintained or updated over time. 

Layout.xls. This Excel file gives a side-by-side listing of the FDA's original field name and the NICAR version. It contains the final table layouts; disregard the table design notes in the FDA readme.  It also lists any NICAR added fields. 

Reporting forms: There are two of these.  The FDA forms used for AERS reporting are FDA 3500A (Mandatory reporting) and FDA 3500 (Voluntary reporting).


March_2005_GAO:  This GAO report discusses how the FDA uses AERS, listing several case studies that may help show the value and limitations of the data.


The dataset:
Folders are arranged by dates corresponding to the dates the FDA received the report, not the date that the adverse events happened. The FDA compiles the AERS data quarterly, on a standard calendar year. 

AERS -- Folders 9798, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011 AND 2012 contain the new format. Each contains at least the following 6 tables: DEMO, DRUG, REAC, THER, SOUR, OUT. Note that the table FOLL has been eliminated from the older tables, and THER is new. In 2002, two tables are added: INVALID AND INDI. There was no INVALID table in 2003 through 2012. REfer to ASC_nts.doc for a list of tables for each year.
	
For older data: The transition between the old SRS and new AERS system means that some 1997-1998 data is in the old format, some in the new. The FDA estimates that 119 1997-98 records are in the SRS data, also available from the NICAR Database Library.

Stats: The FDA includes record counts per quarter per year. These documents are contained in the STATS folder. Years before 2006 included counts of null records per field per table per quarter as well as counts for gender, patients, etc. Please note that NICAR has found that the record counts in these documents do not always match up with the total records found in the raw data. We have asked the agency about such discrepancies. A review of raw text files has found that the NICAR counts are correct. Because of this, we've included a file in the STATS folder called NICAR_RECORDCOUNTS.xls. This file contains record counst for 2008, 2009, 2010, 2011 and the first quarter of 2012. Please contact NICAR with any questions.


To open tables with Microsoft Access: Save the files to your computer. Open a blank database in Access, name it, and save it to your computer. In this new database, go to File menu, select "GET EXTERNAL DATA" and from that menu choose "IMPORT." Select a table to import; Access should automatically create the desired table in your database with no need to specify settings. Only one table can be imported at a time.

General tips for using the data:

ISR FIELD
Note the FDA indicates ISR numbers are unique for identifying a report. Each ISR should be listed only once in the DEMO tables. However, 286 ISR numbers appear twice in the DEMO06 table and the problem exists for more current data as well. This is at least partly due to a FDA updating of certain records (for example, to change the Mfr field name) during which the older records were not deleted. Some of these duplicate records differ from the original ones due to the date. In these circumstances, the older ones should be removed. NICAR has made inquiries with the FDA about dupicate records and ISR numbers in the 2010, 2011 and 2012 data. As of September 21, 2012 the FDA has not responded.

AGE FIELDS 
In the old SRS format, ages measured by month were preceded by the letter M (M11, M06). All other ages in the field are assumed to be years.  All ages are in the original character format because converting it to a number field would make "M" entries 0.
 
In the new data, there is a column age and agecode (year, month, etc.). The age field is also character data since it holds years, months, days, etc. all together. If you need to work with ages, be sure to use the agecode in the query.   

DEATH DATE
Beginning with the 2010 data, the FDA is no longer providing the date of death in the DEMO tables. THe field exists, but it remains empty. According to the FDA readme file: "It has been decided to withhold death date from the extract as this information may identify the patient or person using the product. The outcome code of death will still be included so death reports will remain readily identifiable."

MFR NAMES  
Always search with wildcards.  For example, in some tables there are at least 25 ways to say "Pfizer" and the ingenuity grows with each year (very few involving proper spelling). The difference may be as subtle as PFIZER INC. and PFIZER INC (no period).  
Search for individual parts of compound names. Pharmaceutical companies love to merge, but AERS may contain listings from the days when Pharmacia and Upjohn were still single.  

Popular variants include company research or quality control divisions. Examples (sic):  BRISTOL MYERS SQUIBB CO PHARMACEUTICAL RESEARCH INSTITUTE,  JANSSEN RESEARCH FDN DIV JOHNSON AND JOHNSON,  PFIZER REGULATORY SAFETY PFIZER PHARMACEUTICALS 

DRUG NAMES  
Be sure to look up alternative names (generics, chemical compounds, abbreviations, etc.) if tracking a specific product. Names may begin with numbers, which denote doses, concentrations, or chemical names. 
	
A portion of the entries are enclosed in parenthesis or single quotes. This occurs when the drug name entry could not be validated with a medical reference and was entered verbatim -- quote marks, parentheses and all -- from the original report.

Nutritional supplement names are notably dirty due to alternate names and multiple listings, including multi-vitamin combinations. While the FDA does not regulate these products as pharmaceuticals, they are present in AERS, as in cases where a patient was taking them along with other drugs.     

DATES
Date fields were modified to a MO/DA/YEAR format. It may be useful to view the full four-digit year to identify outliers.  Event dates and manufacturers' dates contain several distant past and impossible future dates. An FDA data analyst noted that, while some dates are simply errors, it is possible for someone to have been taking a drug for many decades, or to have started taking the drug before the first SRS system began in 1969. 

If you need to search dates and want to omit blanks, build a query using the field that contains the original FDA date information and the true date field. For example: To get all non-blank event dates after 1990, use a query such as: Select * from table WHERE not empty (Eventdt) and (Evdate > {12/31/1989}).  EVDATE is NICAR date field and EVENTDT is the FDA original.

DRUG TABLE
Every event is identified by a unique ISR (AERS data) or CONTROL number (SRS data). Many of these appear multiple times in the DRUG table, indicating records for patients taking more than one product.  The drug table contains a second identifying field, DRUGSEQ. Used with the ISR, it joins the drug table to the therapy table, which shows the period that the patient was using a product. If one ISR involves many drugs, the DRUGSEQ matches Drug A to the proper dates of A's use, Drug B to its dates, etc.  Note that the THER table does not contain information for every ISR.  (See NOTE 6 in the FDA readme for more on this join.)
	
MEDICAL TERMINOLOGY
The reaction tables use two sets of standardized medical terminology to describe the problems observed in patients. The MedDRA system is a new and widely adopted standard, but it's also a proprietary service making it difficult to find a public look-up source.  However, individual terms can be found by searching available medical dictionaries.  The AERS field DOSEVBM � the dose as listed on the reporting form � also contains many unfamiliar abbreviations (and a significant amount of dirty data).  For help with the jargon: www.pharma-lexicon.com offers free simple searches (and inexpensive advanced searches) on abbreviations, drug names and companies. www.medterms.com provides clear info for non-experts and does a cross-search for the term in media articles. www.libraryspot.com/dictionaries/medicaldictionaries.htm offers a selection of searchable medical 
dictionaries.dmoz.org/Reference/Dictionaries/By_Subject/Medicine/ links to diverse resources ranging from dictionaries of the terms used on ER, to spelling and pharmaceutical references.

The IRE Resource Center offers a variety of stories and tip sheets related to the FDA and pharmaceuticals. Search the entire collection at http://www.ire.org/resourcecenter.  To order, call (573) 882-3364.   

Tipsheets:
#2407 In this tipsheet, David McKie first explains his own CAR investigations into health care, and then lists ten strategies that helped him get the story. The strategies include everything from "Be willing to negotiate" to "Explain to your audience what the database can't tell you." McKie includes a detailed explanation for each tip.

#2697 David Evans offers a number of clarifications of confusing medical terms and issues, along with other helpful hints that can aid reporters who are investigating the medical world and/or pharmaceutical industry. He gives information on institutional review boards, informed consent, among others, and also provides a list of helpful Web sites.

#2232 Jack Dolan lists and explains a number of Web sites and databases valuable to reporters covering health care or the health beat.

#2161 Adams lists sources of pharmaceutical data that include the National Ambulatory Medical Care Survey as well as data from the FDA. The background, data description and pitfalls are described for each dataset.

#1284 compiles information on searching for pharmaceutical Web sites, based on a New York 2000 Conference panel of medical reporters.



Stories:
#22238 CBC investigated the drugs that as many as 1.5 million Canadian seniors are taking. These drugs, such as Ativan and Amiodarone, have been found to be dangerous for seniors. These drugs are listed on a peer-reviewed list called BEERS in the U.S., but not in Canada. The major finding of this investigation was that one out of twenty deaths among seniors that were ascribed to adverse drug reactions involved a drug on the BEERS list.

#24488 In Connecticut, the "adverse-event" law is supposed to ensure that hospitals report medical accidents that cause harm or death to patients to the state Department of Public Health. The law was revised in 2004 and since then the number of reported adverse-event cases has dropped "dramatically," suggesting that the medical mishaps are not being shared with the public and the state.

#17368  The Los Angeles Times follows seven drugs approved since 1993, despite misgivings inside the FDA. The drugs were later withdrawn, and the article links them to roles in 1,000 deaths.

#18827 60 Minutes investigates the lack of research into drugs to cure tropical disease, which are often neglected in favor of developing drugs for wealthier markets in developed nations.

#17401 CBS news tells the story of Relenza and Rezulin, and internal FDA debates over their safety and efficacy.
	
#18243 Over 100,000 patients die each year due to adverse drug reactions, according to the American Prospect. The story examines deregulation trends and pharmaceutical company�s political contributions.