On 17 March Medicare began paying for a video visit at the same rate as one in the room, and six weeks later it repriced a telephone call from about $14 to about $110. This is what an industry built once the rules it had blamed were suspended, which was a video call sitting next to the chart rather than inside it. The record never moved, and every shortcut rests on a waiver renewed ninety days at a time.
14 October 2020·8 min read·healthcareregulation
The standing explanation for why healthcare could not adopt software quickly had three parts, and all three were checkable. Privacy law forbade the convenient tools, so the consultation had to happen on something a hospital had procured rather than something a patient already had. Payment rules made the remote version of a visit worth a fraction of the room version, or worth nothing, depending on where the patient was sitting. Licensure stopped at a state line. Every one of those was true. None of them had ever been tested, because until this year nobody had a mechanism for switching them off.
Between the sixth of March and the thirtieth of April, all three were switched off.
The sequence is worth having exactly, because the order tells you which constraint was doing the work. Congress moved first: section 102 of the supplemental appropriations act signed on 6 March added a waiver authority that let the Secretary reach the geographic and originating-site restrictions on Medicare telehealth at all. Eleven days later CMS used it. The same day, the HHS Office for Civil Rights said it would not penalize the good faith use of ordinary consumer video tools. On 30 April a second round widened it again.
I want to be careful about the register, because the reason all of this happened is that hospitals were filling and people were dying, and the exercise I am about to run on the evidence is not a comment on that. But it is the closest thing to a controlled experiment this industry will get. A sector had spent two decades naming its obstacles; they were removed, under conditions nobody chose, and its answer arrived within weeks. Both camps come out of that partly right and neither comes out comfortable.
CMS made the remote visit worth the same as the room. The fact sheet of 17 March is short and one line in it does almost all of the work: these visits "are considered the same as in-person visits and are paid at the same rate as regular, in-person visits", effective for services from 6 March, anywhere in the country, including the patient's home. Before that, Medicare generally paid only when the patient was in a rural area and had traveled to an approved facility to sit in front of the camera.
OCR suspended the privacy enforcement posture, and the contract with it. The notification of enforcement discretion, also 17 March, said OCR would not impose penalties for noncompliance with the HIPAA Rules in connection with the good faith provision of telehealth over any non-public facing product, and named FaceTime, Zoom, Skype and others. It also, in a sentence that got far less attention, said it would not impose penalties for the lack of a business associate agreement with the video vendor. That is the contract layer, not the technology layer, and it matters more.
Licensure moved least, and moved in two places at once. The federal waiver removed the Medicare payment rule that a practitioner be licensed in the state where the service is furnished. It did not and could not remove state licensure law, which belongs to the states; the 17 March fact sheet lists who may bill and then says, parenthetically, "subject to state law". Practitioners could cross a line for Medicare's purposes and still not for their board's, unless their state acted separately, and fifty bodies acting separately is not a suspension. It is fifty suspensions with fifty end dates.
The telephone settles it. There is no version of the story in which the telephone was waiting on a technical advance. It has been in clinics for a century, every practice has one, and calling a patient has never required a procurement cycle or an integration. What it required was that the call be worth something. When CMS repriced telephone evaluation and management from roughly $14 to $41 up to roughly $46 to $110, it changed nothing whatsoever about what a telephone can do and changed everything about whether a practice could afford to use one for a visit.
The volume followed the price. Writing in Health Affairs on 15 July, the CMS Administrator reported that before the emergency about 13,000 fee-for-service beneficiaries a week received a telemedicine service, and that in the last week of April the figure was close to 1.7 million. Over 9 million beneficiaries received one between mid-March and mid-June. More than 3 million of them, close to a third, used an ordinary telephone. Take those numbers for what they are: internal claims analysis, preliminary because providers have a year to submit, published by the head of the agency describing her own agency's actions.
So here is the sentence the rest of this turns on. When a technology's adoption tracks a payment rule rather than a product release, the constraint everyone described as technical or regulatory was economic, and it was economic the whole time. That reframing is not a small one. It moves the question from what the industry can build to what the industry gets paid for having built, and those two questions have different answers and different people responsible for them.
It was a video layer, sitting outside the record. The list OCR published is the tell. It names general-purpose meeting products and a handful of telehealth-specific ones, and what they have in common is that none of them is a clinical system. They are a call, placed next to a chart rather than inside it. That is exactly the right thing to reach for when you have days rather than quarters, and it is also a description of a capability that is not attached to anything.
The record itself did not move. Core clinical systems are certified against configurations, integrated through interfaces negotiated pair by pair, and changed on schedules measured in quarters for reasons that are mostly good. None of that changed this spring, and nobody attempted it. The suspension touched what a visit was worth and what tool it could happen on. It did not touch what the visit was written into, and nothing in it made the record easier to change.
Nobody re-platforms on a rule with an end date. This is the part I would defend against the accusation of timidity. Given a flexibility whose duration is unknown and whose renewal is at the discretion of a cabinet secretary, building the cheap reversible thing at the edge is the correct engineering decision, not the cowardly one. The mistake is not making it. The mistake is forgetting you made it, and then treating the result as though it were architecture.
Start with the contract. A business associate agreement is not paperwork; it is the mechanism by which a covered entity's obligations reach a vendor holding its data, and what makes the vendor's breach the vendor's problem too. OCR said it would not penalize the absence of one during the emergency. It did not say the exposure went away, and could not, because enforcement discretion binds a regulator rather than a plaintiff. Every relationship stood up in that window has a gap where a contract normally sits, and closing it later is a negotiation entered with no leverage.
The documentation flows back by hand, and this is the seam that never announces itself. The visit happens in one system and the note lands in another, so somebody retypes it, and the retyping is invisible because it is distributed across every clinician doing it. Then the second interface gets written. The vendor's integration existed for one record system, the organization runs three, and the piece that reconciles them is written twice by different people against different assumptions about which identifier is authoritative.
The emergency is renewed ninety days at a time. The public health emergency was determined on 31 January, backdated to the 27th, and has been renewed three times: on 21 April, on 23 July, and on 2 October, effective the 23rd. Ninety days is the whole planning horizon. Every waiver above sits on top of that determination, so the honest statement of the architecture's lifetime is that it is guaranteed until late January and after that it is a decision somebody has not made yet.
The proposal to keep some of it is still a proposal. An executive order signed on 3 August directed the Secretary to review the temporary measures and propose a regulation extending them as appropriate. Two weeks later the CY 2021 fee schedule proposal did that. It is a proposed rule in a comment period, which is to say it is a stated intention and not yet a rule, and anybody building against it today is building against a draft.
There is one question to ask of each piece. Would this still be worth running if the payment reverted tomorrow? For a triage questionnaire the answer is probably yes, because it saves the organization something regardless of billing. For a video product bought at list price on the strength of parity payment, the answer is that you are holding a subscription whose business case is a waiver. Every item in the estate is one or the other, and the list is short enough to write down before anybody has to.
The privacy rules exist because medical information is the category of personal data that most reliably costs people their job, their insurance, their custody arrangement or their safety when it escapes. Nothing about the past seven months has made that less true, and a consultation carried on a consumer platform under a suspended penalty regime is a consultation with a weaker guarantee attached. That is not an abstraction to be traded away in a sentence about velocity. It is somebody's diagnosis moving through a company that never signed anything about it.
The right way to read the suspension is that a trade was made, urgently, under conditions where it was obviously the correct trade, and made by a regulator rather than by each organization for itself. The failure mode I would worry about now is not that the trade was wrong. It is that it was never scored, so the industry has an experience of speed and no measurement of what the speed cost, and a year from now the only remembered half will be the half that felt good.
What I would actually claim, then, is narrower than the title. The industry did not discover that it could move fast. It discovered that a specific and previously unexamined constraint, the price of a remote encounter, had been holding back a specific and comparatively easy class of change, and that the harder classes did not move even with everything relaxed. That is a smaller finding. It is also the only one the evidence carries.
I should be plain that none of this was an experiment anyone would have chosen to run, and that the emergency it is embedded in is still going on as I write this, which is exactly why I would rather write it down now than remember it later. The findings are provisional. The claims data are preliminary by the agency's own statement, the fee schedule is a proposal, and the emergency determination has a date on it three months out. Anyone quoting this in a year should check every one of those.
What I expect to survive is the shape rather than the numbers. Somewhere in every organization that moved this spring there is a video call and a booking page and a person retyping a note, and all of it is standing on a waiver that was renewed in October for ninety days. The rules will end on a date, and the systems built against them will find out afterward, one interface at a time, which is a slower way to learn it than reading the expiry off the paperwork today.