In a session at the Community Oncology Alliance (COA) Annual Meeting (April 24, 2020), Alti Rahman, MHA, MBA, practice administrator, Oncology Consultants (Houston, TX), and colleagues provided an explanation of the ability to improve practice workflows and prior authorization through robotic process automation (RPA).
Journal of Clinical Pathways sat down with Mr Rahman to further detail the usefulness of RPA in these processes and the impact such automation may have on clinical pathway interoperability with electronic medical records (EMRs).
Why is the time right for looking at opportunities to standardize and mitigate prior authorization? Is the prior authorization system on pace or lagging behind therapeutic advances in cancer care?
Mr Rahman: The prior authorization system is lagging behind therapeutic advances. The number of new treatments coming out in oncology is unprecedented. Ten years ago, the number of new treatments in a year would be around 15 or 20. Nowadays, that number is eclipsed in just a couple of months.
Payers are not accustomed to understanding utilization by specialty, or perhaps they do not have the bandwidth for it. They typically look at utilization by cost. In a cost-driven system, payers create their own customizations on what is or is not approved, or what they allow and therefore authorize.
The idea of “step therapy,” which allows payers to give preference over certain drugs, is becoming popular and often used, especially in oncology. As a practice or health institution that provides cancer services, there is no common denominator, so to speak. Each payer may have their own specific preference for what is approved or not approved. You have to navigate very unique technologies for each payer. Some may be online, some through a fax system, and even some through a telephone conversation.
In the transition to the digital space, there is no central clearing house for health care; there is no standardized medical record number or hub for authorization. If the region in which you provide cancer service has multiple payers, you are going to have many different websites or portals to log into to check preference.
This is a challenge and ultimately an added barrier to care. You want to be able to focus on providing care, but reducing some of the administrative burden in interacting with these softwares is needed.
In your COA session, you mentioned that there may be an opportunity to remove prior authorization altogether. Can you expand on this idea? How can this be done practically?
Mr Rahman: My COA session highlighted the utility of RPA in alleviating the prior authorization concerns on the part of the providers. RPA helps leverage more value-added time rather than figure out how to navigate different software and portals.
The prior authorization from a payer perspective will always be there, but the complexity of what practices have to do could be removed altogether through use of RPA.
If you have an RPA “bot” that you are training to understand the complexities of different prior authorizations, the bot is doing all of the work that would traditionally be done by humans. The bot takes the treatment, sends it to the payer, receives a response, and lets you now if the treatment is approved.
My comment in the COA session on removing prior authorization altogether was in reference to the process on the provider's end. In fact, prior authorization by payers will continue to grow because the process of providing cancer services and treatments is increasingly complex. First, there is physician preference. Then, there are clinical reviews. There are always updates to how many indications a specific treatment could possibly cover.
Therefore, it is important to distinguish the fact that there is a way to eliminate the prior authorization burden on practices, but not so much for payers. The answer is RPA.
What is the most notable source, if any, of pushback on RPA use to relieve prior authorization burden for providers?
Mr Rahman: Data “scraping” is often cited as a concern due to the inherent security risk involved. The act of “scraping” is moving data from one system to another. There is concern that data is not very secure during this process.
Security is a dynamic, ever-changing aspect of which we need to be mindful. As the RPA bots are trained, it is imperative for providers to ensure that the bots do not create errors in their own system or those of the payers. Thus, security is one of the items of which payers are concerned. What is the bot doing? How many actions is it undergoing? Could it create some latency issues? Could it bog down their systems because it is functioning both day and night?
RPA bots are not AI systems; they do not have their own way of learning and integrating knowledge. RPA bots are simply designed to automate redundant tasks, not to create any new knowledge from doing so.
What are some other areas in the workflow that providers are most interested in utilizing RPA?
Mr Rahman: Providers have internal systems that they utilize for patient care. They also have systems outside of health care institutions (eg, websites) from which they can pull or input information. There are two system buckets: internal and external.
Looking first at internal, institutions often have multiple systems – one for accounting, one for billing, one for clinical care, one for radiation care or radiology, and so on. Providers can invest in interfaces which are designed to synchronize important data elements across each of those systems (eg, patient demographics or clinical elements). Depending on the level of cooperation and how data tables or fields are structured (discrete data vs free form or narrative), some information may not be able to move across systems. This is where RPA can help; it can synchronize that piece to essentially move data from one system to the next, rather than rely on a data entry person to do the task.
When considering the external bucket, providers may need to find gloves, masks, and other PPEs during these times, but there are hundreds of websites with varying prices and levels of reliability. RPA bots can scour those websites, pull the prices, and offer providers a regular update on best value deals.
How do clinical pathway systems fit into this discussion? Can pathway systems integrate RPA for prior authorization or any other touch points?
Mr Rahman: Clinical pathway systems and EMRs are different systems. There are very few existing systems that have the two integrated into one, likely because the data across the systems are not fully synchronized. Providers have to repeat data entry from their clinical pathways into the EMR system because the primary patient care system is the EMR, whereas the pathways are often used as a provider-facing clinical decision-support tool.
If you are not able to replicate the information from the pathway that you navigate into the EMR, then you have a disconnect in the patient information. RPA can help synchronize the process. As you use a clinical pathway software, you can have RPA take that final decision and populate it into the EMR.
Without RPA, most health institutions have a dual system entry process in place, which opens the door for manual errors. RPA not only helps to lower the likelihood of these errors occurring, but also increases the amount of patient information included in the EMR.
Are there any other important points you would like to add to this discussion?
Mr Rahman: There is going to be a larger focus on administrative complexities in health care as “value” becomes more and more a priority. Providers will be looking towards using their resources on high-quality, low-cost care while trying to minimize administrative burden processes.
Value-based care will definitely drive more discussions around how to automate items that do not provide value, including data entry. Technology will keep paving the way for augmenting value.
