Five years ago, finding a therapist meant a months-long search for someone who took your insurance, paying out of pocket, or going without. That lack of access was the defining problem in mental health, and it is why Headway was founded. Working with clinicians and health plans, we have built in-network capacity that now includes more than 85,000 clinicians serving millions of patients covered by more than 100 health plans.
That work is not finished (particularly for prescribers, pediatric care, rural care, and specialty care), but access was always a means to something else. A patient who finds a clinician in three days instead of three months has gotten a better start, not a better outcome. What matters is that they get the right level of care that actually helps them get better.
What we need to drive toward
The measure of success is simple. Are patients getting better?
We can be specific about what that looks like. Patients can find the care they need. They stay in treatment long enough to see improvement. Symptoms and functioning measurably change. Crises and hospitalizations go down. Patients reach a point where treatment has done its job and can end.
None of that is controversial. It is what patients have always wanted, what clinicians have always been trying to deliver, and what health plans want to be paying for. What has been missing is the ability to see it happening across a population rather than one chart at a time.
Why it’s been so hard: Nobody can see the whole picture
If the goal is this obvious, it is worth asking why the field has struggled for so long to say what good care is.
Part of the answer is that quality already means three different things depending on who you ask, and all three answers are reasonable. Health plans define it through access, network adequacy, HEDIS benchmarks, and cost trend, which tells them care happened but not whether it worked. Clinicians define it through the clinical picture, the therapeutic alliance, and change over time, and that information usually stays inside treatment notes. Patients define it in more simple terms: I want to feel better, and I need care I can afford. No shared instrument produces all three views, so none of them can be checked against the others.
Underneath that sit three structural problems:
Measurement is still rare in practice. Fewer than one in five mental health practitioners use measurement-based care at all, and as few as 5% use it at the interval the evidence recommends. That is partly a training problem and partly a workflow problem. A clinician in solo practice who wants to administer a PHQ-9 every four weeks has to source the instrument, remember the schedule, score it by hand, chart it, and interpret change over time, all outside the tools she uses to run her practice. Measurement has to cost the clinician close to nothing, which means building it into the visit itself. At Headway we have put PROMs into the platform and paired that with Measurement Informed Care learning pathways. Patients on Headway’s platform have completed more than 4 million PROMs this year alone, and more than two-thirds achieved clinically significant improvement within 90 days.
The delivery model is fragmented by design. Mental health is delivered by hundreds of thousands of independent clinicians in solo and small practices. There is no shared record, no common intake, and no way to see across the system. A health plan can tell you a member attended eight sessions. It cannot tell you what the clinician assessed, what the treatment plan was, or whether the member improved. Shared infrastructure changes the unit of measurement from the practice to the population. When tens of thousands of independent clinicians work on common systems, you can see who is seeking care and why, what was assessed and documented, how symptoms change across a population, and which clinicians tend to do well with which patients.
The field has not agreed on what to measure. PHQ-9 and GAD-7 are the closest thing to a shared instrument, and they are narrow by design. They describe depression and anxiety symptoms, which is one part of what patients tell us matters. Working, sleeping, parenting, staying in a relationship: those are the outcomes people come to care for, and the field has no standardized way to record a patient’s goals at intake, track attainment against them, and report them at scale.
What not knowing costs us
This is more than a clinical problem. It has already reshaped the economics of mental health.
PwC’s medical cost trend survey names rising mental health use as one of five forces pushing employer health spending up by roughly 9%, the steepest climb in nearly two decades. Mental health visits are up 62% since 2018, and that’s driven by more people getting more care rather than by higher prices. Some of that increase is exactly what a decade of work to improve access looks like.
But how many sessions people use, and what those sessions cost, are the only things most of the system can measure. So they’ve quietly become the stand-in for success. Everyone can agree that’s the wrong test. Eight sessions is not an outcome, and a rising session count tells you nothing about whether the people in those sessions got better.
When the data can’t answer that question, the only lever left is to hold down the number of visits. Access becomes the thing that gets squeezed, for reasons that have nothing to do with whether the care was worth it.
The way clinicians get paid reinforces it. Payment is per session, not per patient who improves, and nobody can be rewarded for results nobody can see. That’s a big part of why payment tied to outcomes has stayed rare in mental health while it has spread in other areas of medicine.
Get measurement right and the whole conversation changes. When plans, clinicians, and patients can all see that care is working, access stops being a cost to manage and becomes something everyone has a reason to protect.
So what would we be measuring? Very little of what follows is in dispute. The field largely agrees on what good care is. It hasn’t yet built the ability to prove it.
What effective care should look like
Those are real obstacles, but they are not disagreements about the destination. When I step back, none of the following is controversial, and all of it is achievable if the field aligns.
Good care is built on the foundation of patient safety. Risk is assessed and documented, and when a patient is in crisis the clinician has a path to support they can use in the moment. This is important; in our peer-reviewed analysis of patients arriving for care through Headway, nearly one in five reported thoughts of suicide at intake.
Good care sets goals and tracks progress. Treatment that starts with clear goals - established by the patient and clinician in partnership - ensures alignment on what is most important and defines what success looks like. Ongoing assessment - by PROMs or some other measure of goal attainment - provides feedback on whether care is working. Measurement-informed establishes where a patient is starting, agreeing on what improvement would look like, checking whether it is happening, and changing course when it is not. That loop includes knowing when treatment is a success.
Good care treats the whole person, including medical comorbidity and social context. A person’s physical condition, their mental health, and the circumstances they live in — housing, income, isolation, safety — move together, each one shaping the trajectory of the others. Depression makes diabetes harder to manage; unstable housing makes both harder to treat. In Milliman’s analysis of 21 million commercially insured lives, the 27% of people with a mental health condition accounted for 56.5% of total healthcare spending — and the overwhelming majority of that spend was for medical and surgical care, not mental health, which was just 4.4% of the total. In our own data, suicidal ideation tracked closely with social vulnerability.
Good care is affordable and efficient. Getting the right patients to the right level of care, and recognizing when they are ready to step down or finish, is both a clinical obligation and the reason the model can scale.
Good care is rewarded. Right now the system pays the same whether a patient improves or stalls, which means the clinicians doing the hardest and most effective work carry the cost of doing it well. Once outcomes are visible, they can be recognized: in how clinicians are paid, in how patients are matched to them, and in which models plans choose to grow. Measurement without reward is paperwork. Reward without measurement is guesswork.
We’re on our way, but there’s more to do
In July, I joined leaders from a broad group of health plans, medical societies, and provider organizations and national mental health organizations in signing the National Behavioral Health Quality and Best Practices Voluntary pledge signed a pledge, which focuses on advancing evidence-based clinical pathways in mental health. That level of alignment would have been difficult to imagine five years ago, and what was clear in the room is that agreeing on a standard is the easy part. Verifying that care actually meets these standards, across a field this fragmented, will require data, workflows, and cooperation that no single health plan, clinician organization, or company holds on its own.
Health plans hold the claims. Clinicians hold the clinical picture. Platforms like ours hold the infrastructure that can connect the two. No one piece is enough by itself, and the patient is the one who pays for the gaps between them.
Headway will keep building toward this, and sharing what we learn along the way.
