Eighteen months after the cadence went live, Jordan sits across a conference table from an Optum network manager who has just said the sentence that changes the practice's economics: "We are moving behavioral health toward value-based arrangements, and practices that can demonstrate outcomes will see enhanced rates." Jordan's practice can demonstrate outcomes; it has eight quarters of verified PHQ-9 and GAD-7 trajectories in structured fields. What it lacks is the document that converts those trajectories into a payer-credible quarterly report: aggregates, response rates, a clinician narrative, and an honest accounting that survives the question every payer analyst is trained to ask: "what happened to the clients who did not improve?" This lesson teaches the reporting layer of measurement-based care: quarterly outcome reports for Aetna, Optum, or CCBHC value-based-care contracts, mapped to the CCBHC quality measures (DEP-REM-12, ASC, SRA-BH-C, SUB), built to avoid the cherry-pick trap, with the clinician's narrative, the part AI cannot write, at the center. By the end you will have the Quarterly Payer Outcome Report Skeleton with Clinician Narrative, the artifact that turns measurement based care therapy into contract leverage.

The Annual Report, Not the Brochure

The controlling analogy for this lesson is the difference between a company's annual report and its marketing brochure. A brochure shows the flattering photographs; an annual report shows audited numbers, including the losses, with management's narrative explaining what the numbers mean and what is being done about the weak lines. Investors trust annual reports precisely because the losses are in them; a company that reported only winning quarters would be committing fraud, and everyone reading would know it. Your quarterly outcome report to a payer is an annual report, not a brochure. It contains response rates and non-response rates, improved trajectories and flat ones, denominators stated plainly, and a clinician narrative that explains the numbers the way a credible CFO explains margins: what moved, what did not, why, and what is changing. Payers have analysts who read these documents for a living, and the fastest way to lose a value-based negotiation is to hand them a brochure.

Understand what the payer is buying before you write a line. In a value-based care contract, the payer shifts some payment from volume (sessions delivered) toward value (outcomes achieved), and the contract's machinery runs on measurement: defined instruments, response thresholds, reporting periods, denominators. An Aetna or Optum behavioral health arrangement specifies which measures count, what improvement means, and how often you report; a CCBHC operates under prospective payment (PPS) with required quality measures attached. In every case, the report is the contract's sensory organ: it is how the payer perceives your clinical work. The report's integrity is therefore not compliance overhead; it is the practice's reputation rendered quarterly, and the cadence and trajectory disciplines from the last two lessons are what make an honest one possible.

The Anatomy of the Quarterly Report

A payer-credible quarterly outcome report has five sections, ordered the way an analyst reads. Section one, the denominator statement: how many clients were in the measured population, defined how (for example, all clients with a depressive disorder diagnosis active for at least eight weeks of the quarter, on the PHQ-9 cadence), with the measurement completion rate (what percentage of prescribed administrations were actually completed). The denominator comes first because it is the first thing an analyst checks and the first place cherry-picking hides. Section two, the aggregates: baseline severity distribution, mean and median change from baseline, and the dated reporting period. Section three, the response analysis: the percentage of the measured population achieving clinically meaningful change (the 5-point-or-greater PHQ-9 improvement from the last lesson), the percentage flat, the percentage deteriorating, all against the same stated denominator. Section four, the clinician narrative, which the next section of this lesson treats at length because it is the section that cannot be delegated. Section five, the actions: what the practice changed in response to its own data, plan revisions, frequency adjustments, added measures, training, because a payer reading "we found non-response and revised treatment plans" is reading a functioning clinical system, which is what enhanced rates are buying.

AI's role in this anatomy is the assembly of sections one through three, with the boundary this chapter has drawn twice already: the model aggregates verified trajectory data, computes rates against the stated denominator, and formats tables; it does not select which clients count, does not characterize results, and does not write the narrative. The working prompt: "From the attached verified trajectory worksheet data for Q2, produce: (1) denominator statement using exactly this population definition: [your definition]; (2) measurement completion rate; (3) baseline severity distribution and mean/median change; (4) response analysis as percentages of the full stated denominator achieving 5-point-or-greater PHQ-9 improvement, no meaningful change, and 5-point-or-greater deterioration; (5) a blank section headed Clinician Narrative; (6) a blank section headed Actions Taken. Include every client in the population definition. Do not exclude any client for any reason. Do not interpret, characterize, or editorialize any figure. Flag any client whose data is incomplete rather than omitting them." The prohibitions and the no-exclusion clause are the anti-cherry-pick architecture; a model asked casually to "summarize our outcomes" will helpfully foreground the good news, which is the trap with a machine accelerant.

The Cherry-Pick Trap

Name the trap precisely, because it rarely looks like fraud from the inside. Cherry-picking is any denominator manipulation that makes the response rate look better than the caseload's truth: reporting only clients who completed treatment (survivorship bias, because dropouts skew non-response), only clients with complete measurement (which rewards measuring your improvers), only the clinicians with good numbers, or excluding "complex cases" as outliers. Each exclusion has a plausible-sounding rationale in the moment, and each one converts an annual report into a brochure. The exposure is severe: a payer auditing a value-based contract reconciles your reported denominators against claims data it already holds, and a practice whose report says forty measured clients while claims show ninety depression-diagnosis clients in active treatment has not made a presentation choice; it has misrepresented performance under a contract that pays on performance. That conversation does not end with a rate adjustment.

The defense is structural, not moral effort. First, the population definition is written once, in the contract's terms or your own stated terms, and applied mechanically every quarter; clients enter and exit the denominator by rule, never by result. Second, the completion rate is reported alongside the response rate, because a 90 percent response rate on 40 percent measurement completion is a measurement problem wearing a success costume, and stating it yourself beats the analyst discovering it. Third, dropouts appear in the report as what they are, with retention efforts noted, because attrition is clinical information, not noise to discard. Fourth, the no-exclusion clause stays in the AI prompt permanently. And fifth, the honest paradox from the trajectory lesson scales up: the report that includes its non-responders is more persuasive, not less, because it proves the measurement is real. A payer analyst who sees 58 percent response, 31 percent flat with documented plan revisions, and 11 percent deteriorated with documented clinical escalation is reading a practice that finds its problems and acts; a report showing 96 percent response is reading either a miracle or a denominator, and analysts do not believe in miracles.

The report that includes its non-responders is the report that gets believed. A payer analyst reading 96 percent response is not impressed; they are reaching for your denominator.

The Clinician Narrative: The Section AI Cannot Write

Section four is the heart of the report and the hard boundary of this lesson: the clinician narrative is composed by a clinician, in the clinician's reasoning, every quarter, AI permitted only to format it afterward. The narrative does three things no aggregate can. It interprets: "Response rates in the depression population held at 58 percent; the flat cohort concentrates in clients with documented housing instability, consistent with external maintenance factors rather than treatment failure, and we have added case-management coordination for that cohort." It contextualizes: "The deterioration figure includes two clients whose scores rose during planned trauma-processing phases, an expected and clinically monitored pattern, both with documented re-assessment." And it commits: "Next quarter we are piloting behavioral activation training for the three clinicians whose caseloads show clustered early non-response." That is a forecaster writing, and a payer's clinical reviewers, many of them licensed clinicians themselves, can tell within a paragraph whether the narrative was reasoned or generated.

Here is the worked example of the verifiable detail AI cannot supply, carried up from the trajectory lesson to the contract table. In Jordan's Q2 report, the response section includes de-identified client-level support per the contract's data terms, and the anchor case is the one you already know: a PHQ-9 falling from 18 to 9 across twelve dated administrations, a delta of minus 9, nearly twice the 5-point clinically meaningful threshold. In the narrative, that delta does work no adjective and no model can do: it is the verified, field-traceable evidence that weekly 90834 treatment with behavioral activation produced measurable response in a moderate-severity presentation, and it justifies, in the contract's own currency, both the medical necessity of the completed course and the practice's claim to the enhanced rate tier. AI assembled the series; the clinician verified the scores against the EHR fields, confirmed the sustainment, made the continuation decision the trajectory licensed, and wrote the sentence. If an auditor pulls that chart, every number in the report resolves to a dated field entry and a signed note. That resolution, number to field to signature, is what "audit-proof" means, and no language model can manufacture it.

The CCBHC Layer: PPS Reporting and the Quality Measures

If you work in or contract with a Certified Community Behavioral Health Clinic, the reporting obligation is not a negotiation, it is a condition of the model: CCBHCs are paid under a prospective payment system (PPS) and report required quality measures, several of which are exactly the MBC machinery this chapter built. Map your cadence to four of them by name. DEP-REM-12, depression remission at twelve months: the PHQ-9 cadence playing the long game, because demonstrating remission at twelve months requires baseline and follow-up administrations only a maintained cadence produces; a clinic that measures sporadically simply cannot report it. ASC, alcohol screening with brief counseling where indicated: a screening-and-response measure your intake and cadence workflows must capture as structured data. SRA-BH-C, suicide risk assessment: the measure documents that risk assessment occurred, and the program's hard rule does not bend for a quality measure: the assessment is clinician-conducted, the CSSRS is clinician-administered, AI never scores it and never assigns risk, and what the reporting layer captures is that the clinician's assessment happened and was documented, never an AI-generated risk value. SUB, the substance use disorder measure set, which intersects with 42 CFR Part 2 handling rules from earlier in this level: SUD reporting flows through the consent and segmentation architecture Part 2 requires, and aggregate reporting must not become a side door around redisclosure rules.

The operational lesson of the CCBHC layer generalizes to every payer: quality measures are cadence specifications in disguise. DEP-REM-12 tells you the PHQ-9 must be administered at defined points across a year; SRA-BH-C tells you risk assessment must be documented as a discrete, retrievable event; ASC tells you screening must land in structured fields. A practice that reads its contract's measures first and designs the cadence to produce them, rather than discovering at quarter's end that the data does not exist, is doing the measurement-based care version of reading the exam before studying. AI helps here as a compliance assembler: feed it the measure specifications and your cadence calendar, and ask it to flag the gaps ("DEP-REM-12 requires a twelve-month follow-up administration; the current cadence has no trigger past week 26"), then fix the cadence, because the gap report is the weather station finding a missing sensor, and only the clinician decides what to install.

The Contract Table: Using the Report as Leverage

Return to Jordan's conference room, because the report's highest use is prospective. The practice with eight quarters of honest reports negotiates from evidence: here is our measured population definition, our completion rate, our response rate against the full denominator, our documented non-response management, and the trend across quarters. That practice can ask the informed questions that protect it: Which instruments and thresholds does the contract count, and do they match our cadence? What is the attribution rule (which clients count as ours)? What case-mix adjustment exists, because a practice that takes complex, high-acuity clients will show lower raw response rates than one that screens them out, and an unadjusted contract quietly punishes the practices doing the hardest work? What are the reporting periods, the submission format, and the audit rights? And what happens at the downside: is the arrangement upside-only bonus, or is base rate at risk?

The same evidence discipline protects against the contract's temptations. A practice paid on response rates faces a new cherry-pick trap at the front door: the quiet incentive to select clients likely to improve and refer out the ones who are not. Name that incentive in your internal policy and neutralize it: admission decisions remain clinical, case-mix is documented, and the narrative reports acuity accurately, because the durable position in value-based care is the practice whose numbers are smaller and real rather than larger and curated. One more guardrail belongs in writing before any contract is signed: outcome data flows to payers as aggregates and contract-specified elements, under the contract's data terms, never as a raw feed of session-level clinical material; the report is the interface, and clients are not reduced to scores in a payer's database any more than in your worksheet. What the payer needs is evidence the system works; what the client is owed is that their trajectory remains clinical information first.

The Applied Problem: The Quarterly Payer Outcome Report Skeleton

Your artifact is the Quarterly Payer Outcome Report Skeleton with Clinician Narrative, a reusable template you populate every quarter from the verified trajectory worksheet. Build the skeleton with these fixed sections. Header: practice, contract or payer, reporting period, date, preparer, and the population definition quoted verbatim from the contract or your standing definition. Section one: denominator statement and measurement completion rate. Section two: aggregates (baseline severity distribution, mean and median change, dated period). Section three: response analysis against the full denominator: percent achieving 5-point-or-greater PHQ-9 improvement, percent without meaningful change, percent deteriorating, attrition stated as attrition. Section four: Clinician Narrative, a fixed three-paragraph frame (interpretation, context, commitments), marked "clinician-composed" in the template so no quarter ships without it. Section five: Actions Taken in response to the data. Appendix: the CCBHC measure map where applicable (DEP-REM-12, ASC, SRA-BH-C, SUB, each with its data source in your cadence), and the de-identified client-level support per the contract's data terms.

Populate it once, now, with your real last quarter. Run the assembly prompt against your verified worksheet data, with the no-exclusion clause and both prohibitions intact. Then perform the verification pass that is the clinician's alone: reconcile the denominator against your actual active caseload for the period (the check a payer's claims-matching will eventually perform for you, less kindly); re-compute the response percentage by hand from the worksheet rows; trace your anchor case, the minus 9, from the report's figure to the EHR fields to the signed notes, end to end, once, so you know the chain holds; then write the three narrative paragraphs yourself, longhand if that is what it takes to keep the reasoning yours. Finally, stress-test it with the analyst's question: hand the draft to a colleague and ask, "What happened to the clients who did not improve?" If the report already answers, it is an annual report. If the answer requires a conversation, revise before any payer sees it.

"Done" looks like this: the skeleton is saved as a dated template in your governance file; one real quarter is fully populated, verified, and narrative-complete; the population definition is written and frozen; the CCBHC map is filled or marked not applicable; and the next quarter's population pull is a calendar event, not a memory. When the network manager's email arrives proposing the value-based amendment, the practice that has this document does not scramble. It forwards last quarter's report and asks about case-mix adjustment.

Key Takeaways

• The quarterly outcome report is an annual report, not a brochure: response rates and non-response rates, stated denominators, and a clinician narrative that explains the weak lines and what is changing. Payers staff analysts who read these for a living, and the report including its non-responders is the one that gets believed.
• The five-section anatomy mirrors how an analyst reads: denominator statement with measurement completion rate first, then aggregates, then response analysis against the full denominator (5-point-or-greater PHQ-9 improvement as the response threshold), then the clinician narrative, then actions taken. AI assembles sections one through three under a no-exclusion clause; it never selects clients, characterizes results, or writes the narrative.
• The cherry-pick trap is denominator manipulation with a plausible rationale: completers-only, complete-measurement-only, best-clinician-only, or "complex case" exclusions. The defense is structural: a frozen population definition applied by rule, completion rate reported beside response rate, attrition reported as attrition, and the no-exclusion clause permanent in the prompt. Payers reconcile your denominators against claims data they already hold.
• The clinician narrative interprets, contextualizes, and commits, and it is composed by a clinician every quarter; payer clinical reviewers can tell within a paragraph whether it was reasoned or generated. The worked anchor: a verified PHQ-9 delta of minus 9 (18 to 9 across twelve dated administrations) is the field-traceable evidence that justifies both medical necessity and the enhanced rate tier, and every number must resolve to a dated field entry and a signed note.
• CCBHC reporting maps the cadence to named quality measures: DEP-REM-12 (depression remission at twelve months, requiring a maintained PHQ-9 cadence), ASC (alcohol screening), SRA-BH-C (suicide risk assessment, clinician-conducted; AI never scores the CSSRS or assigns risk, for a quality measure or anything else), and SUB (substance use, respecting 42 CFR Part 2 consent and segmentation). Quality measures are cadence specifications in disguise: read the contract's measures first and design the cadence to produce them.
• At the contract table, eight quarters of honest reports are negotiating leverage: ask which instruments and thresholds count, what the attribution rule is, what case-mix adjustment exists (unadjusted contracts punish practices treating high-acuity clients), and whether downside risk touches base rates. Neutralize the front-door cherry-pick incentive in policy: admissions stay clinical, acuity is reported accurately, and data flows to payers as aggregates, never as raw clinical feeds.
• The artifact is the Quarterly Payer Outcome Report Skeleton with Clinician Narrative: frozen population definition, denominator and completion rate, aggregates, full-denominator response analysis, the three-paragraph clinician-composed narrative, actions taken, and the CCBHC measure appendix. Verify by reconciling the denominator against the active caseload, re-computing response by hand, and tracing the anchor delta from report to field to signature before any payer sees it.