It’s clear the state of PA must change. Artificial intelligence has a role in streamlining administrative tasks.
In the latest American Medical Association (AMA) survey, 93% of physicians said prior authorizations delay patient care, and 82% said the process is so complicated that it causes patients to abandon treatment altogether. Prior authorization (PA) remains the top regulatory burden for most health care and medical professionals, often delaying or preventing access to essential medical care altogether.
In 2021 alone, Medicare Advantage insurers fully or partially denied more than 2 million prior authorization requests, according to data from the Kaiser Family Foundation. The arduous authorization process can have devastating impacts on patient care, with more than one-third of physicians reporting adverse effects for their patients, including hospitalization and even death.
If providers largely agree this process is so dangerous to patient care, why hasn’t it changed? The complications and delays caused by PA boil down to the exchange of information for providers and payers.
Prior authorization exists so insurers can reduce payments for medically unnecessary procedures and treatments. The argument against this cost-control process is that it burdens providers, plans, and patients unnecessarily. PA often delays or prevents access to care, and patients may spend more out-of-pocket in instances of improper denial.
Health plans are complex, and patients often don’t know what services require prior authorization, the criteria used for making a coverage decision, or if their doctors provide the necessary information to insurers to secure coverage. Today, payers have no standardized method for receiving and approving requests. It’s a largely manual, arduous, and time-consuming process that detracts from valuable resources and time spent on patient care. With health care staff forced to complete PA requests differently for every payer, it’s cumbersome, lengthy, and inefficient, creating slow processing and turnaround times.
Although 2022 data from the CAQH Index shows electronic authorization has increased from 12% to 28% over the last four years, the lack of standardization still needs to be addressed to ensure widespread adoption. It's problematic enough that the U.S. Centers for Medicare & Medicaid Services (CMS) recently proposed a rule that would require certain payers to implement an automated process, meet shorter time frames, and be more transparent about their decision-making.
The proposal from CMS is designed to relieve the administrative hassles of PA by creating a standard electronic model for exchanging data and could make it significantly easier to automate prior authorization. CAQH reports that one-third of prior authorizations are still completely manual, meaning they're completed over the phone, mail, fax, or email. More insurers provide web portals where medical staff can submit the necessary information. Still, the lack of automation coupled with the proliferation of web portals only adds to the volume of prior authorization methods that medical staff need to track.
Because prior authorization is based on data exchange, technology-driven improvements have tremendous potential and opportunity, especially with a standardized model. The biggest obstacle concerning introducing AI revolves around the need for accurate provider information to support the claims.
Insurers sift through clinical and administrative data to build a case for approval or denial. Payers need to pull information from the patient record and billing system while ensuring the administrative data regarding the provider, location, and office are all accurate. Treatment and procedures are typically denied because of poor documentation. Often requests lack the proper medical documentation as to why tests or procedures should be done. It could also be as small an error as the wrong medical code was included. The challenge of having accurate data fields highlights why the approval process is so cumbersome, especially when requests are created manually.
Today, robust exchange models are augmenting administrative efficiencies between providers and payers. The final step to solve the PA question requires a model that includes patient information. Without it, payers will always have incomplete data that prevents them from making proper decisions.
Last year, McKinsey estimated that artificial intelligence could automate 50% to 75% of the manual work involved with prior authorizations. More payers are recognizing how far AI could go to relieve the burden on providers by reducing administrative overhead, which currently accounts for 25% of health care spending.
From the patient's perspective, a streamlined, automated process has clear positive implications. Patients often need to wait a week for requests to be approved or denied, which can often cause them to forgo treatment entirely. According to McKinsey, AI can reduce that window to days, if not hours.
Physicians agree that the state of prior authorization has tangibly negative implications on patient health. When you take the risk to patients and combine it with the financial obligation it places on providers dealing with the administrative side, it’s clear the state of PA must change. If automation can enhance efficiency and reduce provider dissatisfaction while improving patient health, there’s a strong case for introducing automation into the mix.
Eric Demers is the CEO of Madaket Health. He believes we can transform healthcare delivery through the power of data and interoperability. With more than 25 years of global health care experience, Eric has built and scaled leading technology and service companies, from early stage to Fortune 100.