Challenges in Medical Coding - Concomitant Medication Therapy Data in Clinical Trials

Clinical trials are intended to prove or disprove a hypothesis set in the study protocol. For this reason, the quality of captured data plays an important role in the study outcome. From a Data Management (DM) perspective, the aim of the clinical study is to achieve clean, complete and consistent clinical data. To reach this aim, DM teams perform collection, cleaning and data management activities. The primary, and most important objective of all data management tasks is to ensure the high-quality of data by keeping the error and missing data rate at the lowest possible level.

One important part of data cleaning is medical coding (or encoding). In the Case Report Forms (CRF), Concomitant Medications and Adverse Events data can be presented in various ways depending on the country, therapeutic area or investigators’ preferences. Additionally, the method of providing this information into EDC can be different and investigators have multiple options for documenting this critical data. Medical encoding is used to standardize the quality of this data in the system.

The aim of this paper is to give an answer to the question: how to capture and encode concomitant therapies and medications in order to keep clinical trial data clean, complete and consistent.

Specificity of Concomitant Medications Encoding

Someone could ask why encoding is so important in the clinical data collection and analysis process. The answer is simple-data collected in the CRF and clinical database have to be:

• absolutely accurate

• suitable to easily perform statistical analyses

• meet all the protocol-specified parameters and requirements.

However, because of the large number of possibilities how to capture and encode the concomitant medications data, this is a complex activity.

Encoding can be complicated, especially if a study protocol assumes capturing not only concomitant medications for each condition, but also all diet supplements. The more data are captured, the more there is a need for structured medical encoding to ensure data is captured in the same consistent way across the study.

A multitude of encoding methods and rules could be a problem for later statistical data analyses because of the potential lack of consistency. This means that there is a necessity to develop standardized procedures regarding the encoding process.

Drugs can be encoded to trade names or generic names (active ingredients). It would be wise to keep the same standard for all coded drugs within the study. The most common issue when encoding concomitant medications is a situation where the recorded drug or supplement cannot be referenced in the used dictionary. Moreover, there are a lot of cases where diet supplements can cause difficulties in coding because of limited access to data regarding the composition of the total ingredients.

The process of concomitant medication and therapy coding depends on four factors (Figure 1):

• investigator’s knowledge regarding the used therapy

• quality of data entered into CRF

• coders’ knowledge and skills

• an internal quality check (QC)

Figure 1. Quality of Medical Coding Elements (Source: KCR)

Besides the need to code terms which are difficult to classify, coders can come across situations like combined data terms, spelling mistakes, synonyms and abbreviations. Additionally, in some cases it is necessary to revise the clinical history context for additional information.

All these factors can actively influence the quality of the concomitant medications coding (Figure 2).

Figure 2. Medical Coding process and results (Source: KCR)

Commonly used dictionaries

The most commonly used dictionary for drug coding is the WHO Drug Dictionary (WHO-DD) developed and lead by the World Health Organization (WHO). The dictionary is maintained and updated by Uppsala Monitoring Center. Currently there are three dictionary types:

• WHO Drug Dictionary

• WHO Drug Dictionary Enhanced

• WHO Herbal Dictionary

These dictionaries are very comprehensive as they contain information not only about drugs, but also about blood products, homeopathic remedies, herbal products, vaccines, radio-pharmaceuticals and dietary supplements.

Besides medical product names, also an ATC classification is an integral part of the dictionary. Because of this fact not only drug names can be coded, but also an indication for each drug can be added.

Virtues and vices of data collection approaches

Concomitant medications are mainly entered in two ways:

• trade name; market name of the drug

• generic name; name of the active substance

Which one has to be chosen for encoding when both are given? The Medical Dictionary WHO-DD does not have a preference as to the form the term is to be coded. Nevertheless, it is considered good practice to code in the name of the active substance.

Example 1:

• “Drug A (paracetamol)” should be encoded to “Paracetamol”

• “Drug B (ibuprofen)” should be encoded to “Ibuprofen”

While there are no problematic cases with monospecimens, it can be the opposite with many of the combined formulations. In order to maintain high quality, it is recommended to code all the components of the drug to one common name which in this case is the trade name of the drug.

Example 2:

• “Paracetamol / caffeine / pseudoephedrine” should be encoded to “Drug C”

• “Paracetamol / ibuprofen” should be encoded to “Drug D”

Sometimes one trade name is available with different active ingredients. That could be a potential problem and one of the disadvantages of using a trade name for medical encoding.

Non-unique trade names

A unique trade name is a name which is used for the same product in all countries where the drug is approved. This is the opposite when compared to non-unique trade names. Non-unique trade names are the result of cases, where the same trade name is used in different countries with different sets of ingredients, or the same trade name is used in different pharmaceutical forms which contain different sets of active ingredients. The last example can occur when a product is changed in its composition without a change to its trade name.

Example 3:

• “Drug E” can be encoded to one of the three options:
• “Benzalkonium chloride”

• “Domiphen bromide”

• “Hexylresorcinol”

• “Drug F” can be encoded to one of the four options:
• “Aufanofin”

• “Amoxicillin sodium / Clavulanate potassium”

• “Amoxicillin / Clavulanic acid”

• “Amozicillin trihydrate / Clavulanate potassium”

With cases like these there can be difficulties with identifying the medicinal product by using the reported trade name only, therefore additional information about the drug is necessary in order to code the term properly.

If medication cannot be referenced based on the data entered into the CRF, a data query should be raised to confirm the active substance. If this will not help to reach a conclusion, the coder can use “differentiators” – additional information provided in the encoding dictionary about each specific case (Figure 3). That kind of additional information is helpful and can be used as a second resource during coding. Thanks to this, the coder can usually solve the issues with non-unique trade names easily, it just requires additional time and investigations.

Figure 3. The differentiators applicable for medicinal product encoding (Source: KCR)

Trade name or generic?

“Trade name or generic?” is a question often asked by data managers during the development of data entry guidelines for capturing medications’ data. It is a complex question; so advantages and disadvantages for both options should be considered.

The main benefit of entering medication as a trade name is the ease of entry and further encoding of combined drugs. This would not be as easy, if we decided to collect generic drug names. We will have to answer several questions, for example: how to capture dose in case of combined drugs components?, etc.

However, a generic name collection seems to be more valuable and beneficial for the future data analysis, despite some difficulties with combined drugs. The main disadvantage of this approach is related to issues and difficulties in coding resulting from non-unique terms of trade names (generic name is a 100% match with the referenced term) or possible questions regarding non-referenced terms, when the trade name is not available in the particular dictionary version given.

Conclusion

In a CRF data regarding Concomitant Medication therapies can be presented in multiple ways. Medical coding is a common task to ensure the consistence of the captured data. The process of encoding depends on some critical factors, which largely influence the quality of the data. Medications entered into the system can be provided as trade or generic names and the coder has to choose the best of the available coding scenario to ensure the consistency between entered and coded data. If any of the drugs cannot be referenced in the dictionary, we advise to remember differentiators which can be used as an additional source of information in coding dictionaries.

References

1. Jackson T. Data quality: not selling ourselves short. HIM J. 2014;43(2):4-6.
2. Schroll JB, Maund E, Gøtzsche PC. Challenges in coding adverse events in clinical trials: a systematic review. PLoS One. 2012;7(7):e41174.
3. Babre D. Medical coding in clinical trials. Perspect Clin Res. 2010 Jan;1(1):29-32.
4. Bolton P, Warren J. Toward a model for the evaluation of clinical coding systems. Aust Fam Physician. 2001 Aug;30(8):814-6.
5. Lee DH, Lau FY, Quan H. A method for encoding clinical datasets with SNOMED CT. BMC Med Inform Decis Mak. 2010 Sep 17;10:53
6. Krishnankutty B, Bellary S, Kumar NB, Moodahadu LS. Data management in clinical research: An overview. Indian J Pharmacol. 2012 Mar;44(2):168-72.
7. Nair GJ. Ensuring quality in the coding process: A key differentiator for the accurate interpretation of safety data. Perspect Clin Res. 2013 Jul;4(3):181-5.
8. Maund E, Tendal B, Hróbjartsson A, Lundh A, Gøtzsche PC. Coding of adverse events of suicidality in clinical study reports of duloxetine for the treatment of major depressive disorder: descriptive study. BMJ. 2014 Jun 4;348:g3555
9. WHO Drug dictionary Best Practices® v. 3.0; UMC

Magdalena Dziedziurko is a Clinical Data Associate in the Biometrics and Clinical Trial Data Execution Systems Department at KCR. She can be reached at info@kcrcro.com