Why does the number of results change when I search for YAZ in the field patient.drug.medicinalproduct vs. patient.drug.openfda.brand_name?

Question

Depending on how I search for drugs in the openFDA drug adverse event API, I get different counts (numbers of results).

Here are two searches for the drug YAZ, between January 1, 2004, and September 30, 2012:

• https://api.fda.gov/drug/event.json?search=receivedate:[20040101+TO+20120930]+AND+patient.drug.medicinalproduct:"YAZ" (Results: 20,033)
• https://api.fda.gov/drug/event.json?search=receivedate:[20040101+TO+20120930]+AND+patient.drug.openfda.brand_name:"YAZ" (Results: 22,028)

Why are the result counts different depending on which field I search in? Isn’t brand name the same as medicinal product?

There has been some confusion about this, including in these posts about openFDA in "IBM Data Magazine" Part 1 Part 2 Part 3

Answer 1

patient.drug.medicinalproduct and patient.drug.openfda.brand_name are not the same.

First, patient.drug is a list of multiple drugs from the adverse event report. For each drug, there is always a medicinalproduct (name), and sometimes an openfda section.

• medicinalproduct is the drug name as written in the original adverse event report (FAERS record), imported into the openFDA database. It could be a brand name, or a generic name. It might even be misspelled. (cf. IBUPROPHEN which has ~59 results, instead of IBUPROFEN which has ~40k results.)
• brand_name is part of the openfda section. It’s an annotation, like a “sticky note,” added to each item in patient.drug. It has additional possible names and codes that can be used to identify the drug in medicinalproduct. It’s only added if the drug name in medicinalproduct was spelled correctly, and the name matched in certain other FDA and NLM datasets; openFDA calls the process of adding this section “harmonization.”

Why do the searches have different result counts? Four ways to search, with explanations.

In this case, YAZ is one of the many brand names for a certain generic drug, DROSPIRENONE AND ETHINYL ESTRADIOL. In fact, many brands contain the same ingredients, but in different quantities and forms: GIANVI, SYEDA, YAZ, OCELLA, LORYNA, YASMIN, VESTURA, ZARAH, etc. And this Mayo Clinic article lists even more brand names that contain the same generic drug. Here’s how different searches work, and what the result counts mean. (The diagram may help and is followed by specific examples.)

1. Find exact matches on the drug name (brand or generic) in the original report.

Criterion: The original report listed at least one drug with the exact name YAZ.

Use patient.drug.medicinalproduct.exact.

https://api.fda.gov/drug/event.json?search=receivedate:[20040101+TO+20120930]+AND+patient.drug.medicinalproduct.exact:"YAZ"

Results: 19,946.

2. Find “fuzzy” matches on drug name (brand or generic) in the original report.

Criterion: The original report listed at least one drug with a name like YAZ, YAZ (24), YAZ 28, etc.

Use the field patient.drug.medicinalproduct.

https://api.fda.gov/drug/event.json?search=receivedate:[20040101+TO+20120930]+AND+patient.drug.medicinalproduct:"YAZ"

Results: 20,033. As expected, this is larger than the exact match, because it included more possible names.

3. Find reports for a drug, when you know its brand name—whether the original report used the brand name or generic name.

Criterion: The report has an openfda annotation with the brand name YAZ.

Use the field openfda.brand_name.

This will match when:

• The original report listed at least one drug with a name like YAZ, YAZ (24), YAZ 28, etc. (in the medicinalproduct field).
• The original report listed at least one drug with the generic name DROSPIRENONE AND ETHINYL ESTRADIOL (in the medicinalproduct field).

https://api.fda.gov/drug/event.json?search=receivedate:[20040101+TO+20120930]+AND+patient.drug.openfda.brand_name:"YAZ"

Results: 22,028. As expected, this is even larger than the previous search, because it included even more possible names.

4. Find reports for a drug, when you know its generic name—whether the original report used the generic name or any of the known brand names.

For drug adverse events, it’s often important to find all the reports for a drug—like IBUPROFEN, no matter what brand name was reported (ADVIL, MOTRIN, etc.). However, it’s not always appropriate—and may not be for a hormone therapy for contraception like DROSPIRENONE AND ETHINYL ESTRADIOL, where the different brand names may represent drug products with different strengths and dosage schedules.

Criterion: The report has an openfda annotation with the generic name DROSPIRENONE AND ETHINYL ESTRADIOL.

Use the field openfda.generic_name.

This will match when:

• The original report listed at least one drug with a name like YAZ, YAZ (24), YAZ 28, or YASMIN, or OCELLA, or any other brand names that the generic drug is marketed as.
• The original report listed at least one drug with the generic name DROSPIRENONE AND ETHINYL ESTRADIOL.

https://api.fda.gov/drug/event.json?search=receivedate:[20040101+TO+20120930]+AND+patient.drug.openfda.generic_name:"DROSPIRENONE+AND+ETHINYL+ESTRADIOL"

Results: 30,459. As expected, this is the largest result set, since it includes the largest possible set of drug names.

(The result counts above are valid as of the openFDA “last updated date” of 2014-08-06. In the future, when this date changes, the data may be slightly different.)