How to Start a Collaborative Care Program
To start a collaborative care program, a primary care practice builds five things: a care team (a behavioral health care manager plus a consulting psychiatrist), a patient registry, a measurement-based workflow using validated tools, a consent and initiating-visit process, and billing setup for the collaborative care codes. Sequence them in that order and start small.
Collaborative care (the Collaborative Care Model, or CoCM) is one of the most-studied approaches in behavioral health, with more than 90 randomized controlled trials behind it. But the evidence describes the *model*, not the *rollout*. What follows is the operator's version: the building blocks, a realistic sequence, and the places practices get stuck.
Five components have to be in place before the first patient is enrolled. Skip one and the model quietly stops working.
Start with the care manager, because everything else routes through that role. Practices staff it in different ways — a licensed clinical social worker, an RN, or another behavioral health clinician — but the person needs to be comfortable with measurement-based follow-up and with a caseload rather than a scheduled therapy panel. Then line up psychiatric consultation, usually a fractional arrangement since one consultant can cover a large caseload by reviewing cases indirectly.
Define the handoffs explicitly: who introduces the program to the patient and when, how the PCP flags a patient for enrollment, how the care manager escalates a stuck case to the consultant, and how the consultant's recommendations get back to the PCP for orders. Write these down. The most common early failure isn't clinical — it's ambiguity about who does what after enrollment.
The registry is what separates collaborative care from ordinary referral. It is a single view of the whole caseload: every enrolled patient, their most recent PHQ-9 or GAD-7 score, the trend since enrollment, and their next follow-up date.
It does two jobs. First, it drives *proactive* follow-up — the care manager works the list and reaches out to people who are due, instead of waiting for patients to book. Second, it surfaces patients who aren't improving, so the team changes the plan rather than assume one referral solved the problem.
You do not need custom software to start. A structured spreadsheet or an EHR-based registry can carry the first cohort. What matters is that it tracks scores over time and flags who is due and who is stuck.
Collaborative care is measurement-based care by definition. The loop:
1. Screen and baseline. Use a validated instrument — PHQ-9 for depression, GAD-7 for anxiety — to establish a starting score. 2. Enroll and set a target. A meaningful improvement, such as a 50% score reduction or remission, becomes the goal. 3. Follow up on a cadence. The care manager checks in regularly, re-administers the tool, and delivers brief interventions. 4. Review at caseload rounds. The consulting psychiatrist reviews cases — especially non-responders — and recommends adjustments. 5. Adjust or step up. Change the approach for patients who aren't improving; graduate those who are.
The discipline is in step 5. A program that measures but never changes course is measurement *theater*. The whole point is to treat to target.
Collaborative care requires patient consent, because it involves a specific set of billable services and typically a monthly cost-share. Consent can be verbal but must be documented, and patients should be told the program involves a care team, ongoing outreach, and possible cost-sharing.
There's usually an initiating visit — a face-to-face or telehealth encounter with the PCP where the program is introduced and the patient agrees. Make consent documentation part of the enrollment step so it never gets skipped.
Collaborative care is reimbursed through a dedicated set of time-based codes. Medicare established the CoCM codes, and many state Medicaid programs and commercial plans cover them as well — coverage and rates vary by payer, so confirm before you rely on them.
The mechanics that trip practices up:
Verify current codes and payer rules with your billing team and each plan before launch — this is the component most exposed to policy changes.
Don't try to stand up all five components at once. A workable order:
1. Confirm the economics and coverage for your payer mix. 2. Hire or assign the care manager and secure psychiatric consultation. 3. Stand up the registry and write down the workflow and handoffs. 4. Set consent, the initiating visit, and billing processes. 5. Pilot with one or two engaged PCPs and a small cohort — enough to find the friction before you scale. 6. Review data, fix the workflow, then expand.
Some practices build all of this in-house; others partner with an organization that supplies the care management, psychiatric consultation, registry, and billing infrastructure. Either path can work — the trade-off is upfront build time and hiring risk versus a faster, supported start. What doesn't work is launching without one of the five building blocks.
Most of the timeline is hiring the care manager and securing psychiatric consultation. Once the team, registry, consent process, and billing setup are in place, a small pilot cohort can begin fairly quickly. Building the workflow deliberately before scaling matters more than launching fast.
A full-time care manager can support a substantial active caseload because follow-ups are structured and brief, and the psychiatric consultant reviews cases indirectly. Start with a smaller panel during the pilot and grow it as the workflow stabilizes.
No. A structured spreadsheet or an EHR-based tracker can carry the first cohort. It just has to track validated scores over time and flag who is due for follow-up and who isn't improving.
Yes. Medicare established time-based CoCM codes, and many Medicaid and commercial plans cover them, though rates and rules vary by payer. Reimbursement is monthly and based on cumulative care-management minutes, so accurate time tracking is essential. Confirm current codes and coverage with your billing team and each plan.