Frequently Asked Questions - MedDRA

1. What is MedDRA?

MedDRA is the Medical Dictionary for Regulatory Activities. It is a globally accepted, clinically validated medical terminology used within all phases of the drug development process, including classification of medical events for clinical trials and drug safety.

MedDRA development was started by the International Conference on Harmonisation (ICH) in the early 1990s with the goal of creating a global standard medical terminology to supersede all other terminologies. MedDRA 1.0, the alpha version, was released for international review and evaluation in November, 1994. The first official version, MedDRA 2.0, was released in July 1997.

MedDRA is owned by the International Federation of Pharmaceutical Manufacturers and Associations (IFPMA) acting as trustee for the ICH steering committee. It is maintained and distributed by the MSSO (Maintenance and Support Services Organization).

Return to top

2. How do I get a copy of MedDRA?

You should contact the MedDRA MSSO (Maintenance and Support Services Organization) to get a copy of MedDRA. Click here to go to our Useful Links page where you will find a link for the MSSO web site.

Return to top

3. If you are doing our MedDRA coding, do I still need a MedDRA license?

Yes. The MedDRA license agreement prevents us from giving coded data to another company unless that company also holds a current MedDRA license. So even if ThesIS is doing your coding and/or versioning, you will still need to obtain a MedDRA license. The good news is that the MedDRA license fees have been significantly reduced (especially for smaller companies), so the cost should no longer be an issue.

Return to top

4. Where can I get more information about MedDRA?

General information about MedDRA is available from the MedDRA MSSO web site (accessible via our Useful Links page).

You can also contact us with MedDRA questions. In addition, we offer training classes on MedDRA.

Return to top

5. How long have you coded to MedDRA?

This initial use of MedDRA distinguishes us from most other early adopters of MedDRA who used it only within drug safety. By using it early on for coding clinical trials as well as drug safety, we have had many years of experience developing and refining practical techniques for effective MedDRA coding and versioning on a large scale.

Return to top

6. Why is it important to be medically qualified in order to code in MedDRA?

MedDRA is much more specific than legacy dictionaries such as COSTART. With more than 60,000 lowest level terms (LLTs), MedDRA offers many choices for the same and/or similar medical concepts. The dramatic increase in the level of specificity with MedDRA means your coders must be medically knowledgeable in order to fully understand the medical concept or condition described within a reported term and to accurately distinguish among the many like choices that MedDRA offers.

Go to the next question below for some examples of mistakes commonly made by medically unqualified coders.

Return to top

7. Do you have any examples of mistakes with MedDRA coding due to inexperience with the dictionary and/or lack of proper medical qualifications?

• coding to a general (less specific) LLT when a much more specific LLT is available
• coding to an LLT that is contains more information than is contained in the reported term (i.e. adding information to the reported term)
• coding to an LLT that ignores unique information contained within the reported term
• coding to an LLT that does not accurately reflect the medical concept(s) contained in the reported term

It is not often apparent to an inexperienced MedDRA coder and/or reviewer that the wrong LLT has been chosen. Quite often, a chosen LLT may seem to be reasonable for a given reported term, especially to someone without adequate medical knowledge, when in fact there is a much better choice available.

Examples of each of these common mistakes are below.

Coding to a general (less specific) LLT when a much more specific LLT is available

Coders inexperienced with MedDRA and its high level of specificity will often choose a general (less specific) LLT even when a much more specific LLT is available. This problem can be exacerbated by thesaurus systems that fail to present to the coder a truly complete list of candidate LLTs (due to inadequate search capabilities, ill-conceived and/or implemented autoencoders, etc.).

As a simple but easy to understand example, an inexperienced coder might code the reported term of:

          INCREASE OF DIASTOLIC BLOOD PRESSURE

to the LLT of:

          BLOOD PRESSURE INCREASED

This choice may seem reasonable and is certainly not egregiously wrong, but MedDRA has an even more specific -- and therefore better -- LLT available:

          BLOOD PRESSURE DIASTOLIC INCREASED

While it may seem excessive to worry about such seemingly subtle differences, failure to code to the most specific LLT available can have a significant impact on how your data is summarized. (Click here for an example.)

Coding to an LLT that is contains more information than is contained in the reported term (i.e. adding information to the reported term)

          ABDOMINAL PAIN, INCREASED SERUM AMYLASE, INCREASED SERUM LIPASE

to the LLT of:

          PANCREATITIS

It can also occur if the coder makes an assumption about the reported term, e.g. coding the reported term of:

          CRUSHING CHEST PAIN

to the LLT of:

          MYOCARDIAL INFARCTION

because the event was reported within a cardiac trial, rather than coding to:

          CHEST PAIN

Coding to an LLT that ignores unique information contained within the reported term

Inexperienced coders will often fail to capture all of the unique information contained within a reported term, e.g. coding the reported term of:

          CORONARY ATHEROSCLEROSIS WITH RECURRENT ANGINA PECTORIS

to the LLT of:

          CORONARY ATHEROSCLEROSIS

rather than requesting the reported term be split so that both medical concepts are coded:

          CORONARY ATHEROSCLEROSIS -> CORONARY ATHEROSCLEROSIS
          RECURRENT ANGINA PECTORIS -> ANGINA PECTORIS

When a reported term includes a list of signs and/or symptoms, a coder without sufficient medical training may have difficulty determining how many medical concepts are presented in the reported term.

For example, a reported term of:

          ANAPHYLACTIC REACTION, RASH, DYSPNEA, HYPOTENSION, LARYNGOSPASM

can be coded solely to:

because rash, dyspnea, hypotension, and laryngospasm are all characteristic signs/symptoms of anaphylactic reaction.

On the other hand, a reported term of:

          MYOCARDIAL INFARCTION, CHEST PAIN, DYSPNEA, DIAPHORESIS, ECG
          CHANGES, JAUNDICE, ICTERUS

would need to be split into two terms so that both medical concepts can be coded:

          MYOCARDIAL INFARCTION, CHEST PAIN, DYSPNEA, DIAPHORESIS, ECG
          CHANGES -> MYOCARDIAL INFARCTION

          JAUNDICE, ICTERUS -> JAUNDICE

Coding to an LLT that does not accurately reflect the medical concept(s) contained in the reported term

Coders who lack sufficient medical knowledge may often choose an LLT that is more specific than what has been reported, resulting in an LLT that does not accurately represent the reported term. For example, a medically unqualified coder might code the reported term of:

          PNEUMONIA LEFT LOBE

to the LLT of:

          LOBAR PNEUMONIA

under the mistaken belief that LOBE and LOBAR are synonymous. Lobar pneumonia, however, is a specific type of pneumonia:

          Per Dorland's: a type of acute bacterial pneumonia with abundant
          edema, usually limited to just one lobe of a lung; the most common
          kind is pneumococcal pneumonia

As another example, an unqualified coder might think the reported term of:

          EXCESSIVE SUDATION

was a misspelling of SEDATION and code it erroneously to the LLT of:

          SEDATION EXCESSIVE

In fact, sudation is defined in Dorland's as "sweating," meaning the reported term should be coded instead to the LLT of:

          EXCESS SWEATING

Return to top

8. Does it really matter if I fail to choose the best possible LLT? Will such seemingly subtle differences in MedDRA coding have any impact on how my data is summarized?

Failure to code to the most specific LLT available can significantly impact how your data is summarized. Take, for example, the reported term of:

          BLOOD PRESSURE INCREASING, NO RESPONSE TO MEDS

MedDRA provides an LLT of:

          CHANGE IN BLOOD PRESSURE

that might seem to be an appropriate choice. However, the LLT of:

is a better, more specific choice because it captures the direction of the change in blood pressure. But the LLT of:

          BLOOD PRESSURE INCREASED REFRACTORY

is the best choice because it is the most specific, capturing not only the direction of the change but also the resistance to treatment as indicated in the reported term.

The MedDRA hierarchy for each of these LLTs is different, so the reported term will be summarized differently depending upon which LLT you choose:

In general, MedDRA is highly specific, but the level of specificity varies throughout the dictionary. As a result, a coder needs to be particularly diligent to ensure that the most specific LLT has been chosen and that reported terms are accurately summarized.

Because MedDRA is so highly specific, there are many LLTs that seemingly describe the same medical concept within the same degree of specificity. However, these equally specific, seemingly similar LLTs are not always grouped into the same preferred term (PT) and/or system organ class (SOC). Which LLT you choose can therefore affect how your data is summarized.

For example, MedDRA provides the following two LLTs:

Coding reported terms such as:

to the LLT of:

          BLOOD PRESSURE HIGH

will cause them all to be summarized as instances of hypertension, whereas coding them to:

          BLOOD PRESSURE INCREASED

will cause them to be classified simply as investigations.

Return to top

9. What is the difference between accuracy and consistency?

Please see the answer to this question in our Services FAQ.

Return to top

10. Do you use any MedDRA autoencoders?

We do not use any autoencoders for MedDRA coding. We do not believe there are any autoencoders available today that are capable of coding to highly specific dictionaries such as MedDRA.

Click here to go to our Services FAQ to learn about our general opinion of autoencoders.

Return to top

11. How does your MedDRA versioning service work?

We then go through a innovative process to migrate your data to the newer MedDRA version. Please refer to our MedDRA versioning page for details on the steps performed to update your data. This updating process is done entirely within our system.

Once this versioning process is done, we provide to you a file containing all of your reported terms along with the coding as updated to the newer version. You can load this file into your clinical trials and/or drug safety systems. You can also convert this file into SAS and use it directly to re-analyze your data. The decision of when and how to apply the new version is therefore entirely within your control.

We also provide reports that show you the differences between the original coding and the coding to the newer version, including those instances where the term is still coded to the same LLT, but the hierarchy for that LLT has changed within the new version.

Return to top

12. To which level in MedDRA do you code?

We always code to the lowest level term (LLT) as this is the most specific level available in MedDRA.

Return to top

13. Which levels of the MedDRA hierarchy will you return with my coded data?

You can specify the levels of the MedDRA hierarchy you would like returned with your coded data, and also whether you would like the numeric codes and/or the text. Most of our clients prefer to receive all levels of the hierarchy (LLT, PT, HLT, HLGT, and SOC), including both codes and text. Some clients request only the LLT, PT and SOC (codes and text).

Return to top

14. I understand that MedDRA is multi-axial, meaning that a preferred term may have more than one system organ class (SOC) assigned to it. Which system organ class will you return with my coded data?

Although we will return secondary SOCs if requested, we do caution against their use. Using the primary SOCs results in consistent representation of specific events between companies and regulators. If you want to use secondary SOCs, we suggest you first discuss it with us to ensure that your use of them makes sense.

Return to top

15. How does your MedDRA versioning service differ from the versioning services from other companies?

Our MedDRA versioning service is superior to our competitors' offerings in a number of ways:

• We have proprietary versioning software, developed from our long-term and extensive use of MedDRA, to facilitate the versioning. Our coders utilize this special software to quickly and precisely pinpoint only those reported terms that need to be re-coded. You get the effect of completely re-coded data without our having to actually re-code all your data.

• You get more than just reports. You get actual data completely migrated to the new version.

• Your data is completely versioned, including instances where a newly added LLT is a better choice for your previously coded data.

• The versioning is first done within our systems, so there is no impact on your ongoing operations.

• We can version between any two versions.

• ThesIS's versioning techniques and proprietary tools have been developed and refined from our long-term use of MedDRA. We have many years of experience versioning in MedDRA and can therefore achieve better results than anyone else.

Return to top

16. Do you offer any training on MedDRA?

We offer several training classes covering MedDRA. In addition, we offer one-on-one MedDRA training intended to supplement "standard" MedDRA coding classes. Training one-on-one with a ThesIS expert enables your coder(s) to get a better and more specific understanding of MedDRA than is achievable through a class setting.

Return to top