There has never been a time in the pharmaceutical industry when the coding of adverse events was more pivotal. The industry is not quite to the point of knowing that three identical events will trigger a product withdrawal. But it’s close.

In that context, Phase Forward last week updated its Central Coding software. It is designed to simplify the coding of verbatim reports using the industry’s standard terminological behemoths, the World Health Organization’s WHO Drug Dictionary (WHO-DD) and the Medical Dictionary for Regulatory Affairs (MedDRA). The software also works with internal, proprietary company-specific dictionaries.

Just as a spell checker can automatically correct “teh,” Cental Coding can automatically apply an appropriate standardized code for “head pain.” You’re welcome to do that by hand, of course. Many sponsors do. But that leads to its own complications if “head pain” receives a new code. Or several. Or if there is a new edition of your dictionary. We’re probably oversimplifying to suggest there is a one-to-one relationship, because the situation with some codes is more complex and hierarchical.

Simpler Synonyms

But for the sake of illustration, suppose “head pain” is of interest in a trial of toenail fungus but too general in a migraine study. Even with companies using software to handle the coding, the repetitive nature of matching verbatims to specific codes, not to mention the sheer volume of the work, has many sponsors on a hunt for faster, more accurate and less tedious solutions.

“Without these strong tools, there are a lot of terms that don’t get matched and don’t have any recommendations. The coder is forced to manually go figure out what the term should be and apply that mapping,” says Phase Forward senior product marketing manager Rob Quinn.

Quinn refers readers to a recent webinar by the Society For Clinical Data Management on the topic of coding and dictionary management. Those sessions noted the difficulty of re-coding as new dictionaries emerge, a standard feature of the Phase Forward product.

Over Your Shoulder

Quinn adds that the Central Coding software is dynamically watching and learning from what the coder does. It can take over and make certain that a particular term is automatically coded in a particular manner.

“If you start to code a certain way, it notices that and starts to do it in that way,” says Quinn. The rate at which the system picks the right automatically-generated code can rise over time. Terms like “and” and “in” can be stripped from the verbatim, allowing the software to focus on the words that matter.

The granular level of control by term or dictionary has been well-received by the 40-some customers currently using the software. People in the trenches seem to like it, Quinn reports. “Before Central Coding 2.0, we didn’t have a solution that provided review and control capabilities before they clicked ‘accept,’” he says.

New Dictionaries

Some of the new functionality is designed to help sponsors and contract research organizations manage dictionaries that may be updated quarterly or annually. The dictionaries themselves can help with such updates to some degree, but mostly by comparing a new dictionary to its immediate predecessor. The task is more complicated if you’re comparing a new dictionary with one that is several editions older.

Central Coding software is also designed to help users understand the ramifications of a new dictionary, perhaps signaling that a new version will have no effect on an oncology trial but significant consequences for studies of the central nervous system.

Batch Mode

Whenever a new version of any dictionary appears, Central Coding can generate an impact report about a subset of updated terms that may affect one clinical trial or a group of them. The key is that manual re-coding of terms is kept to a minimum, and that the system can re-code terms with perfect consistency on its own once humans have told it what to do. 

The older codes can be automatically updated in batch mode. “If there was only one selection for ‘head pain,’ ” says Quinn, “we will recode your original verbatims with the new code. Once you update your synonym list, you can apply that across all your trials in the therapeutic area.”

EDC Linkage

This is not to suggest that some sort of robot will embark on an unauthorized recoding of your data. Rather, Quinn says, the system is set up to allow companies to monitor and review how terms are being handled. “We’re providing customers the ability of having complete control and review of what’s being suggested,” says Quinn.

Formally, the product is known as Central Coding for InForm 2.0, a reference to Phase Forward’s software for electronic data capture (EDC). Barely two years old, the coding program is an indication of the importance Phase Forward places on drug safety, for which it has an entirely separate suite of products, but also its desire to make sure its EDC franchise is well-defended.

Looking Across Multiple Trials

One EDC competitor Quinn cites allows coding changes to be applied to a single case report form. But Phase Forward’s system, he says, can rely on a centralized store of terms and reach into multiple trials across a portfolio of studies. “It’s integrated with the InForm environment to allow the coding group to code across multiple trials,” says Quinn.

In some cases, the system can consolidate multiple terms into one, so that what’s really happening in the data can be more easily understood. Being able to do that across more than one study, Quinn believes, is an advantage. “If you don’t have consistency in coding analysis across trials, how can you do cross-trial analysis?” he asks. It’s a good question, and one that points to a surprising lack of visibility across trials, even at the largest companies and regulatory bodies.