The care-gap closure playbook for community health centers.

What a care gap actually is

A care gap is the distance between the care a patient should have received by now and the care your records show they received. A woman due for a mammogram who hasn’t had one. A patient with diabetes who hasn’t had an A1c in a year. A two-year-old behind on well-child visits. Each open gap is simultaneously three things: a clinical risk for the patient, a lower rate on a quality measure, and — under payer quality programs and value-based contracts — money your health center earns or forfeits.

Closing a gap requires a chain of events to go right: you know the gap exists, you reach the patient, the patient agrees, a visit or test happens, and the evidence lands back in your records as structured data. Most gap-closure programs fail at a specific link in that chain, and the failures cluster at the end of the year.

Why the year-end sprint fails

The classic pattern: quality reviews the rates in October, leadership notices the distance to target, and staff spend November and December working a call list. It underperforms every time, for reasons that compound:

• Contact information decays. Phone numbers and addresses go stale fastest for exactly the patients least connected to care — the ones most likely to have open gaps. A list built from January registration data is partly fiction by November.
• The calendar is against you. Holidays mean travel, competing family demands, and the highest no-show rates of the year. You’re asking patients to book the year’s least convenient appointments.
• Capacity crunches. Every other health center is running the same sprint against the same December 31 deadline, while your own schedule fills with flu season and staff PTO. Even willing patients can’t always find a slot.
• Evidence misses the deadline. A colonoscopy in late December may not produce documentation in your system until January — clinically closed, statistically missed.

The sprint isn’t wrong because outreach is wrong. It’s wrong because it compresses twelve months of contact attempts, scheduling, and documentation into the eight worst weeks of the year.

The playbook

1. Build one source of truth for who’s due. If the EHR dashboard, the payer registry, and a spreadsheet each have their own list, your team spends energy arguing with the data instead of working it. Merge every source — EHR, claims, payer files, registry feeds — into a single deduplicated list per measure, and resolve conflicts with explicit rules. Patients who already closed the gap elsewhere shouldn’t get a call; patients only the payer knows about shouldn’t be invisible.
2. Prioritize by deadline and clinical urgency, not the alphabet. A worklist sorted by name treats a patient whose measure window closes Friday the same as one with six months of runway. Rank by days-to-deadline first, clinical severity second — the uncontrolled A1c above the routine screening — so the highest-stakes patients get the freshest attempts.
3. Reach patients in a real conversation, in their language. One-way reminder blasts get ignored because they can’t answer the patient’s actual questions: what is this test, do I need to fast, can I come Saturday, can you talk to my daughter instead? Outreach that converts is two-way — it answers, reschedules, and switches language when the patient does. For many health center populations, outreach that can’t hold a conversation in Spanish isn’t outreach at all.
4. Catch patients at visits already on the books. The cheapest gap to close belongs to the patient already coming in tomorrow. Morning huddle prep — every scheduled patient, every open gap, on one sheet — turns existing visits into closure opportunities without a single outbound call. A blood pressure check, a depression screen, a fluoride varnish: done in the room, because someone looked the night before.
5. Log outcomes per measure, so the list shrinks truthfully. “Called — left voicemail” is not an outcome. Track per-gap dispositions: scheduled, refused, deceased, moved, transferred care, evidence received. Refusals and exclusions documented in structured fields clean your denominator honestly; unreachable patients route to address-verification instead of getting the same dead number redialed monthly.
6. Watch the funnel weekly. Attempts → reached → agreed → scheduled → completed → documented. Each transition is a rate you can manage. If reached-but-not-agreed is high, your messaging or offer is the problem. If scheduled-but-not-completed is high, you have a no-show or access problem. If completed-but-not-documented is high, you have a data problem masquerading as a clinical one. A weekly funnel review tells you which lever to pull while there’s still time to pull it.

The staffing math nobody writes down

Be honest about the arithmetic of manual outreach. A meaningful outreach attempt — dial, wait, voicemail or conversation, documentation — costs minutes of staff time, and most attempts don’t connect on the first try. Multiply attempts-per-closure by minutes-per-attempt by thousands of open gaps, and the real cost of a serious manual program is a full-time job or several, usually absorbed invisibly by staff who were hired to do something else.

This is the layer AI outreach is built to absorb: the routine attempts, the reminders, the rebooking, the first conversation in either language — with human staff handling the cases that need judgment. At our first deployment — a Los Angeles community health center — automating that routine layer returned about 2,025 staff hours per year to the team, alongside a 15% improvement in care-gap closure within three months. And because agent capacity grows by configuration rather than hiring, a mid-size clinic’s entire year of routine outreach fits comfortably inside it.

Measured outcomes at our first deployment — a Los Angeles community health center. Read the case study.

Measure-specific tactics that work

Colorectal cancer screening: lower the barrier to the floor

For many patients the obstacle isn’t willingness — it’s that “colonoscopy” means prep, sedation, a ride home, and a day off work. Stool-based at-home kits remove nearly all of that. Offer the kit in the outreach conversation itself (“we can mail one to your home this week”), confirm the mailing address while you have the patient, and build a follow-up loop for unreturned kits at two and four weeks. Track kit returns as their own funnel stage; an unreturned kit is a warm lead, not a closed case.

Diabetes control: sequence the lab before the visit

An A1c drawn three days before the appointment turns the visit into a medication-adjustment conversation; an A1c ordered at the visit produces a result nobody acts on until the next one. Outreach for diabetes measures should book two things in one conversation — the lab first, then the visit — and treat a completed lab with no follow-up visit as its own gap to chase.

Well-child visits: batch the family

Parents juggling work and transportation won’t make three separate trips for three children. When outreach reaches a household, check every child’s status and offer same-day, back-to-back slots for all of them — and pair the visit with whatever else is due, from immunizations to dental fluoride. One trip, several gaps, one grateful parent.

The system, not the sprint

Nothing in this playbook is exotic: one trustworthy list, deadline-first priorities, real conversations, huddle prep, honest logging, weekly funnel review. What makes it work is that it runs in March with the same intensity as November — which is exactly what a platform should make easy. That’s what we built AI outreach and deadline-ranked worklists for: the discipline above, running every day, without asking your staff to become a call center.