Physical therapy patient acquisition: how patients arrive

Somebody did measure how physical therapy patients arrive. Pendergast and colleagues took five years of claims from a Midwest insurer, sorted 62,707 outpatient PT episodes by whether a plausible referring physician had billed anything in the 30 days before the initial evaluation, and got a clean split. Their 2012 paper in Health Services Research reports 17,497 episodes self-referred, 28 percent, and 45,210 physician-referred, 72 percent.

The claims run from 2003 to 2007. Nothing published since has measured the split again.

So the plan built on “most of our patients come from doctors” and the plan built on “patients shop for us now” are both guesses off the same missing number. Here is what the evidence does support.

Do PT patients arrive by referral or on their own?

In the one US measurement available, a bit more than a quarter arrive on their own, and two details make that 28% more useful than it looks. The authors excluded fractures and traumatic joint disorders, because those patients see a physician anyway, so 28% is the self-referral share among conditions where self-referral was plausible at all. And self-referred episodes were not worse business: after adjustment they ran 86 percent of the physician-referred visit count at $0.87 of allowable per dollar, with no difference in health care use once the episode ended.

The trend line has to come from somewhere else. In the Netherlands, where direct access opened in 2006, Scheele and colleagues tracked back pain episodes in a national physical therapist registry and reported in Physical Therapy that the direct access share climbed from 28.9% in 2006 to 52.1% in 2009. Three years from a quarter to a majority, once the law and the payers both said yes. Different country, different insurance system. Read it as a trajectory, not a translation.

Whether a specific patient can skip the referral is a separate question with a real answer: the state rules, Medicare’s certification requirement and the payer side are all in our direct access guide. This piece is about the demand side.

Where does a musculoskeletal episode actually start?

Almost never at a PT clinic. Harwood and colleagues used the Health Care Cost Institute’s 2015 and 2016 claims, roughly 50 million insured people a year across four private insurers, to identify 3,799,593 adults with a new low back pain diagnosis and assign each one to the first provider they saw. In BMC Health Services Research they report physical therapists first for 109,480 of them, or 2.9%. Chiropractors were first for 942,925, or 24.8%. Primary care took 25.2%, orthopedists 5.1%.

Read the caveat before you quote it. That counts the first billed provider for one diagnosis in a commercially insured population, not the referral route into a PT clinic, and the patient who walks in with a script from last month’s ortho visit sits in the orthopedist’s column. The shape is still hard to argue with: for the most common diagnosis physical therapists see, chiropractors are the first stop nearly nine times as often.

What actually decides which clinic a patient picks?

Distance, cost and continuity, in roughly that order, and the research is more specific about it than the advice usually is.

Sarigiovannis and colleagues ran a discrete choice experiment with 382 musculoskeletal physiotherapy patients, asking them to pick between service descriptions that varied on who treats you, the wait to the first follow-up, the number and format of follow-ups, distance and parking. Converted into miles in BMC Health Services Research, the patients being treated by physiotherapists would drive an extra 11.72 miles to keep the same clinician for follow-ups, 3.53 miles for a clinic with ample parking, and 0.86 miles for every week shaved off the wait to the first follow-up. Continuity outweighed everything else on the list. This was a UK service, so read the ordering rather than the absolute values.

Harwood’s team got there from the other direction. Correcting for selection bias meant finding something that predicts which provider a patient walks into but has nothing to do with the outcome. The copay on that first visit worked, and so did differential distance, the gap between how far the chosen provider sat and how far the nearest alternative sat. Both cleared the F-statistic threshold for a strong instrument. Price and proximity predict provider choice well enough to build a model on.

The broader literature backs that unromantic version. Victoor and colleagues, in a 2012 scoping review for BMC Health Services Research, concluded that “there is no such thing as the typical patient” and that comparative information “seems to have a relatively limited influence on the choices made by many patients.” Potappel and colleagues then videotaped Dutch GP consultations across 2015 and 2016, and in the 117 that ended in a referral, coders judged that 56% of patients seemed to have some, or a lot of, input into where they were sent. Round that off: at the moment of referral, close to half of patients are not choosing anything.

Do online reviews change where patients go?

They move a minority, and the best US survey on it is old. Hanauer and colleagues surveyed a national panel of 2,137 adults in September 2012 and reported in JAMA that 65% were aware of online physician rating sites, below the 87% aware of ratings for cars. Among respondents who had sought physician ratings in the past year, 35% said they picked a physician because of good ratings and 37% said they avoided one because of bad ratings. Across all respondents, 59% called the sites “somewhat important” or “very important” when choosing a physician, though only 19% picked “very important.” Among those who had never looked, 43% gave a lack of trust in the information as the reason.

The experimental evidence is thinner still. Martino and colleagues randomized 1,347 new health plan members who had to pick a primary care physician, nudging half of them by letter and phone call to read the plan’s online quality report. The nudged group looked at it 28% of the time against 22% for the control, and while they did end up choosing physicians with higher patient experience ratings, the authors reported in Medical Care that reading the report did not explain the difference.

Reviews are worth having and worth answering. They are not the number one thing, and anyone who tells you the size of the effect is quoting something they cannot show you.

What does Google actually say about local ranking?

Three things, on its own help pages, and everything else you have read about it is inference. Google’s guidance for Business Profiles says local results are ranked on relevance, distance and prominence: relevance is “how well a Business Profile matches what someone is searching for,” distance is “how far each business is from the customer who’s searching,” and prominence “means how well-known a business is.” The same page says “more reviews and positive ratings can help your business’s local ranking,” and, flatly, that “there’s no way to request or pay for a better local ranking on Google.”

One of those three is a fact about where the searcher is standing, which no clinic can change and the claims researchers used as an instrument. What is left in your hands is dull and finite: an accurate profile, correct hours, the right category, photos, replies to reviews.

Who self-refers, and what that tells you

People who have had physical therapy before and liked it. The Dutch registry study found the odds of arriving by direct access rose with previous physical therapy care (odds ratio 1.7), recurrent back pain (1.7), higher education (2.0) and, most strongly, back pain lasting less than seven days (4.2). Patients over 55 were less likely to self-refer (0.6).

The interview evidence agrees. Alshareef and colleagues interviewed 32 people whose employer plan gave them a financial incentive to self-refer, between August 2017 and March 2018, and reported in BMC Health Services Research that what separated the 15 who used it from the 17 who went through a provider was knowing the program existed, what they believed about medication and surgery, and prior positive experience with physical therapy. A PLOS One review of 26 direct access musculoskeletal studies adds the demographic sketch, with its own qualifier attached: those patients look largely similar to GP-referred ones, though often younger, slightly more educated and better off.

So the highest-probability self-referrer in your market is a former patient with a recurrence, inside the first week, who remembers that it worked. That person is already on your discharge list. Nobody has to go find them.

They do have to be able to book. In Orion, patients request, reschedule and cancel from the patient portal inside the rules the clinic sets, so a Tuesday-night flare-up does not have to survive until the phones open, and eligibility runs before the visit on both plans, so the coverage answer arrives before the patient does instead of as a bill afterwards. Whether patients use an after-hours booking window is a settled question, and the numbers are in our self-scheduling piece. Why they stop coming once they have started is a different body of evidence.

The field nobody fills in

Every study above is somebody else’s patients, in somebody else’s market, mostly a decade or two ago. The number that would settle where your attention belongs is sitting in your own intake: how this patient came to be here. Not “referring provider,” which the biller fills in for the claim, but how they found you, asked once, in the patient’s own words, and read monthly.

Most practices cannot answer it. They can name their top referring physicians to the visit, because referring provider is a billing field and that report writes itself, and they infer the rest. That asymmetry is a good part of why physician relationships get the attention they do: it is the one channel that arrives with a report attached.

Adding the question costs one field. Next year you would be arguing from your own data instead of from a Midwest insurer’s 2007.