Bias in electronic health records impact clinical trial results – report

Science and Health

The results of clinical trials may not be as accurate as they seem due to biases in electronic health records, a recent article suggested.

The report, which was published in the academic journal Contemporary Clinical Trials but which has not been peer-reviewed, argued that a person’s race or socioeconomic status could impact the outcome of a clinical trial and render its results not broadly applicable among the general population.

And it is the potential bias in electronic health records that could influence this the most.

Bias in health care and science: How prejudices in electronic health records could make health inequality worse

The article centered around embedded pragmatic clinical trials (ePCTs). These are, essentially, randomized control trials, which are used to collect and analyze data in the healthcare system in order to learn more about different treatments and their risks and benefits. However, these types of tests can be very expensive and time-consuming to carry out. In order to make it easier to collect all this data, experts created pragmatic clinical trials, which are embedded as part of routine health care in order to gather data from electronic health records. 

The goal behind such a system is to show the real-world effectiveness of treatments on a broad population.

An illustrative image of medical records. (credit: PXFUEL)

However, using electronic health records has its own problems.

Firstly, not everyone even has access to electronic health records since not everyone has access to health care in the first place. This is something that is evident among certain demographics that either can’t afford health care, are unable to allocate enough travel time to receive healthcare or have a high distrust of the medical system in general.

In other words, relying on electronic health records has the inherent consequence of completely excluding significant percentages of the population who don’t have access to healthcare in the first place. 

Another issue is that even for those who do have electronic health records, they may not be complete or accurate.

Certain health record data rely on patient-reported outcomes. However, not all patients are able to report their outcomes, either because they don’t have access to the means to do so or because they lack medical literacy.

Another problem has to do with race. As the article notes, there is a historic tendency in the medical field to gather evidence from trials with mostly white participants. This historic trend is included in electronic health record tools, which can come with the same exclusionary results as trials that only examined white people.

In traditional clinical trials, these issues may often be overcome by having stricter rules, sterile laboratory conditions, and being able to exclude people with underlying conditions.

But relying on electronic health records, the article argued, creates a self-perpetuating cycle of exclusion from trials, thereby placing them at an even bigger disadvantage. And with the medical system transitioning to artificial intelligence algorithms, the problem can become even worse in the future if it isn’t addressed.

While methods were suggested to help overcome it, such as instituting greater transparency for sources, input, and data for electronic health records, more work is still needed to reduce health inequality.