When Does a Treatment Become Medically Necessary?

According to many insurance companies, a medically necessary treatment is “necessary for the symptoms and diagnosis or treatment of the condition, illness, or injury.” However, the phrase “not medically necessary” is often what patients hear when an insurance company simply doesn’t want to pay for something. For decades, most insurers have employed medical review boards that assess new medical procedures and technologies in order to determine whether they are “sufficiently accepted within the medical community” to meet their requirements for coverage. But without widespread instances of efficacious use cases as well as proof that the new technology improves the bottom line, it can be next to impossible to get insurance companies to approve and reimburse trailblazing technologies.

Up until recently, medical AI solutions languished in this limbo. But for the first time ever, in September of 2020, the U.S. government authorized a reimbursement for a new AI technology. Under the auspices of a government incentive to use new FDA-approved technology, Medicare and Medicaid Services (CMS) will now approve the use of AI systems for identifying and prioritizing specific abnormalities in the brain scans of stroke patients.

This is a huge development for both the field of radiology and AI in medicine in general. Ideally, stroke brain scans need to be reviewed by an interventional radiologist as part of a specific group of care protocols followed used to identify and treat strokes.  However, many smaller hospitals do not have an interventional radiologist on staff. ContaCT’s AI solution can quickly flag images showing clots typical of ischemic stroke with large vessel occlusion. This is significant, because according to the CDC, 87% of strokes are ischemic, meaning blood flow to the brain is blocked.

According to Dr. Hugh Harvey, Directory of Radiology at a leading cancer research institute in Europe, “the quicker you can cure or treat a stroke, the less likely it is that they’re going to have long-term paralysis or neurological problem [and] may reduce death rates.” Dr. Harvey explained what a recent health economics analysis used to measure the efficacy of the new care pathway which includes the essential AI technology was attempting to determine: “Can improve timed diagnosis? Can it therefore improve downstream morbidity and mortality and quality of life of patients?”

Fortunately for patients, the answer seems to be yes; the incorporation of the AI solution significantly improved the time it took to identify and expedite treatment emergency room patients with intercranial hemorrhage.

Typically, the use of AI solutions is generally paid for by individual providers using a subscription system, with the cost then passed on to patients. Thus, it is very difficult to determine when a new technology is “medically necessary.” A provider can strongly encourage the use of a cutting-edge technology, but if isn’t covered by a patient’s insurance, the cost is often prohibitive.

However, with the incentive of the aforementioned New Technology Add-on Payment (NTAP), the U.S. government is setting a precedent that it’s willing to foot the bill for promising new solutions that seem likely to deliver a return on investment. This is a game-changer for private health insurers and could spur a boost in investment in companies working on medical AI products. 

While there seems to be consensus that medical AI solutions are the future, the recent past has been a long, drawn-out affair as startup companies wait for their validation studies, FDA approvals, and finally (finally!) acceptance into standard medical protocols.

The bottom line: When does something become “medically necessary”? When a government says it will reimburse health care providers and insurers for its use.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>