The manual prior authorization process stands as a notorious drain on both time and money. For ASCs, this aspect of the revenue cycle curtails efficiency but also strains staff resources, jeopardizing cash flow and overall financial success.
The smallest oversight during the authorization process, even unexpected changes in payer rules, can often lead to claim denials, further compounding these challenges. While automation offers clear financial benefits by reducing manual tasks and associated costs, streamlining prior authorizations requires a balance between emerging technology and human expertise.
Major payers in the health insurance industry have embraced automation and leveraged machine learning (ML) to expedite the prior authorization review process. However, prioritizing speed over accuracy can have severe consequences for patients. Recent class action lawsuits against payers, with Cigna receiving the latest in March, highlight the risks of simply relying on algorithms to process claims without adequate human review. In Cigna’s case, its algorithm allegedly automatically denied more than 300,000 requests without medical experts confirming if the denials were valid and without checking the patients' medical history.
The allure of relying on AI to complete manual, labor-intensive tasks can be appealing, especially for ASCs grappling with high turnover rates or inefficient revenue cycle workflows. However, when deploying ML in revenue cycle processes, the efficacy of algorithms hinges on the quality of human instruction. One could argue that it’s tough for a machine to learn effective prior authorization management if humans don’t have the time and expertise to teach it.
To enhance the speed of the prior authorization process, ASCs are adopting various automation tools. These tools should be used in conjunction with human expertise to ensure accuracy:
- Electronic Prior Authorization (ePA) or Automated Claim Submissions: Submitting authorization requests electronically though a web portal or clearinghouse streamlines the process, minimizing manual efforts and delays associated with faxing and phone calls.
- Auto Posting ERAs: Automatically post Electronic Remittance Advice (ERA) files to billing systems or practice management software. ERAs are electronic documents sent by insurance companies to explain the adjudication of claims, eliminating the need for manual intervention in the posting process, as the ERA data is automatically imported and matched with corresponding claims, speeding up the reconciliation process and reducing errors in posting payments and adjustments.
- Automated Claims Statusing (Autostatusing): Software systems automatically query the payer's system to obtain real-time updates on the status of each claim, including whether it has been received, processed, paid, or denied. Autostatusing helps ASCs monitor the progress of their claims without manual intervention, allowing them to identify and address any issues promptly. This automation improves workflow efficiency, reduces administrative burden, and helps expedite reimbursement.
There are also more advanced forms of RCM automation that include AI processes that will likely play a more pivotal role in the future:
- Robotic Process Automation (RPA): Software robots automate repetitive tasks like data entry and document retrieval, enhancing efficiency.
- AI and ML Algorithms: Machine learning (ML) can be added to revenue cycle workflows to automate tasks such as verifying insurance eligibility and benefits in real time. ML-workflows identify patients requiring authorization and automatically submit requests to payers.
- Computer Assisted Medical Coding (CAMC): Utilizes natural language processing (NLP) and machine learning (ML) algorithms to automatically extract relevant information from electronic health records (EHRs) and other clinical documents to suggest appropriate codes for each procedure, reducing the time and effort required to manually review and assign codes.
While streamlined processes and automation can alleviate the burden of repetitive tasks and mitigate the risk of human error, the indispensable role of human expertise in the PA process cannot be overstated. Technology doesn’t guarantee complete immunity from claim denials. Specialized knowledge of the revenue cycle enables timely incorporation of critical information, such as unexpected changes in payer guidelines for prior authorizations. For example, since payer guidelines change frequently and authorization requirements can be subject to change without notice, if you have a list of surgeries that require authorization from a particular payer, it’s still crucial to verify procedures that aren’t on the list as they might still require authorization.
Another prior authorization best practice is getting approval upfront for a broad scope of codes to reduce your chances of a claim denial. For instance, if a patient is approved for a shoulder arthroscopy but during the procedure the surgeon discovers adding debridement is necessary, this can impact your ASC’s ability to get paid if debridement codes weren’t approved ahead of time. Computer Assisted Medica Coding (CAMC) can expedite the process of compiling a broad range of codes for each procedure; however, coders should work alongside computer systems to verify suggested codes and ensure accuracy.
In a recent CAMC experiment, nimble’s coders evaluated the potential of using the ChatGPT algorithm to generate a comprehensive list of relevant CPT codes for commonly performed ASC surgeries, including arthroscopic shoulder debridement, which can be complex to code. While AI showed promise in suggesting potential codes depending on surgical outcomes, there was a range of specificity and inaccurate responses, underscoring the irreplaceable role of knowledgeable, certified coders in verifying the codes for compliance and revenue capture.
Additionally, with the right ML-workflows, you can pull a range of codes based on your ASC’s past cases, including CPT codes you might’ve forgotten to apply previously for prior authorizations. ML-workflows can also help track which payers approve a small set of codes for a given procedure, so you don’t spend time submitting for codes that aren’t likely to get approved.
On the AI horizon is also Voice RPA technology to record phone conversations between staff and payer representatives and offer real-time “scripts” of what human staff members can say to overturn a denial based on previous conversations and aggregated data.
This technology is becoming capable of having conversational voice calls with payers to effectively overturn denials without human intervention. However, human expertise and oversight is still warranted since historical data may no longer be applicable if new payer rules are introduced or if your facility is interacting with a payer for the first time.
In conclusion, AI technology learns from data to make predictions or decisions but putting processes on “autopilot” without human expertise is unwise. Since the reimbursement landscape is prone to frequent change, it’s fair to anticipate that AI will continue to learn, adapt, and require human oversight to ensure accuracy and compliance. When navigating the complex landscape of payer requirements for prior authorizations, establishing a balance between technology and human insight is the most effective approach to decreasing errors and time-intensive, repetitive tasks. In turn, knowledgeable RCM experts can redirect their focus by solving novel problems that impact revenue capture, ultimately driving organizational success.
Visit nimble solutions online at nimblercm.com or at booth #502 during the Becker’s 21st Annual Spine, Ortho, Pain Management Driven ASC + the Future of Spine Conference on June 19 – 22, 2024.