Tailoring CDS Tool Improves Biomarker Documentation in Cancer Treatment

Daniel Rubin, PharmD, BCOP

A clinical decision support (CDS) tool can be optimized to improve the documentation of crucial biomarker results in cancer care, specifically for adjuvant and metastatic non–small cell lung cancer (NSCLC) and metastatic castration-resistant prostate cancer (mCRPC), according to a study by Daniel Rubin, PharmD, BCOP, McKesson.¹

Previously, a significant number of biomarker results were documented as "unknown" within the CDS tool, likely due to delays in result availability. To address this, researchers redesigned the tool's prompts to better align with the typical timelines for receiving biomarker results across different cancer types. The tool was then programmed to re-prompt for unknown biomarkers during subsequent therapy selections. 

This approach led to significant improvements. In NSCLC, the percentage of documented known biomarkers increased substantially, from 69% to nearly 78%. Similar improvements were observed in mCRPC, where known biomarkers increased from 34% to 61%.

These findings show the importance of tailoring CDS tools to the specific nuances of each cancer type. The researchers plan to implement these adjustments across a broader range of cancer types within the CDS tool.

In an interview with Targeted Oncology^TM, Rubin further discussed the study, showing how a CDS tool can improve the documentation of biomarker results in cancer care.

Targeted Oncology: What is the background of this study?

Rubin: We work with a clinical decision support tool that helps guide treatment decisions at a disease- and patient-specific level. Upon looking at some of our data, we noticed an increased number of unknown biomarkers, which is a little surprising given the explosion of precision medicine and how many treatments are now dictated by the presence of biomarkers.

We wanted to see if we could tweak our decision support tool, called Clear Value Plus, and modify how we prompt for the documentation of these biomarkers. We aimed to determine whether this increase in unknown biomarkers is an actual trend or if we can significantly reduce this number through these adjustments.

What were the goals of the analysis?

The goal was to better understand whether the documented unknowns of these biomarkers were more so related to the timing of the results, the biomarker prompts themselves, because we do have a belief that most of our providers are testing in an appropriate manner, especially as this is a big increase in the field.

As far as our methodology went, it went with a pilot program to start in the non–small cell lung cancer space, specifically adjuvant, and to determine how we can approach and change our approach to the biomarker prompts in the tool. And to go from there, look at some initial results and tweak it to other indications, specifically metastatic non–small cell and then eventually metastatic castration-resistant prostate cancer.

Can you summarize your findings?

We were able to see an improvement, resulting in a lowered documentation of unknown biomarkers. We also noticed an improvement in the usability of the decision support tool itself. This translated to a better experience for providers, reducing “click fatigue” by minimizing the number of questions that needed to be answered. It also enhanced the tool's alignment with clinical workflow. For instance, the tool now better accommodates the typical sequence of biomarker testing: often starting with tissue analysis, followed by next-generation sequencing and then blood tests. This allows the decision support tool to seamlessly integrate with this clinical workflow.

What do you consider to be the key takeaways?

I think the key takeaways are that decision support tools can really aid in helping to navigate the treatment landscape as it greatly changes with the increase in precision medicine. But that doesn't have to come with a burden of click fatigue and a challenge of using another electronic tool. It can be done in a way that fits the clinical flow; it can be designed with a careful attention to the questions themselves, and so that this is really meant to be a tool and not another computer program run awry.

Until there's a situation setup for more of an automatic interoperable flow of data from all our vendors, and everywhere throughout the health system, that we can navigate guideline treatments and help have a better understanding of what documentation has been done, and what values are available to us, especially from the biomarker standpoint. And from pharmacists, it is really important that documentation is well kept together, because as far as the validation and the verification of any sort of orders that are done, having that available in a discreet, well-met, well-captured area is really important.

What do you see as the next steps for this research? How can these findings start to be implemented?

We are going to try to apply these methods, which have been successful in these specific diseases, to the other 30+ diseases for which our tool provides decision support. The ultimate goal is to achieve more automated documentation behind the scenes. Ideally, biomarker results would be captured directly and seamlessly, not as PDFs, but in a structured, discrete format. This would allow providers to focus on patient care without the burden of manually entering biomarker data.

However, due to the challenges presented by diverse technologies and varying data standards, we are currently focusing on improving prompt curation. We aim to make the decision tool a more user-friendly experience, recognizing that providers already navigate numerous digital tools and interfaces. This is a significant consideration, as we understand the demands placed on providers.

REFERENCE:

Rubin D, Handy V, Gilmore S, et al. Impact on biomarker documentation in community oncology by optimizing clinical decision support. J Clin Oncol. 2024;42(suppl 16);abstr 11182. doi:10.1200/JCO.2024.42.16_suppl.11182