--- name: capture-ladder-diagnosis description: Use this skill when a practice runs, ran, or is arguing about a Medicare between-visit program (chronic care management, transitional care, remote monitoring, APCM, behavioral health integration, and the like) that underperforms, "doesn't pay," or looks like it needs another hire, and wants to know where it actually fails and what to fix first. It builds the Capture Ladder from counts (Eligible, Offered, Enrolled, Engaged, Served, Billed, Paid, Helped), puts the partners' objections on rungs, values closing each gap, runs the four-level check and the book's evidence order before anyone proposes a fix, compares sites fairly, and names the one rung to fix first with the reasoning. Triggers on requests like "why doesn't our CCM program pay," "build our Capture Ladder," "do we need another nurse," "the billing company is losing claims," "patients won't use the devices," "why is one site better," "rebuild the ladder for the program we stopped," or "what do I bring to the partners' meeting." --- # Capture Ladder Diagnosis This skill is Labor as Code™ for the diagnostic method in *Getting Paid to Do the Right Thing*. Labor as Code is a skill that encapsulates a person's knowledge, judgment, taste, and skill in durable, inspectable, versioned form. The know-how is the habit of finding the rung where a between-visit program loses its patients, and the level where the cause lives, before anyone buys, hires, fires a vendor, or blames a person. The method spans five places in the book. The preface, "Medicare Already Pays for Much of the Care Between Visits," introduces the Attention Loop and the Capture Ladder with its gap table (Table P.1). "Physicians Pass on These Programs Because They’re Hard to Run" works the ladder in full: the twelve reasons practices give for leaving the money alone, each on its rung (Table 4.2), Riverbend's 2019 and 2026 ladders side by side, and the Practice Math box that values closing a quarter of each gap. "Medicare Is Mostly a Chronic-Care Business" gives the four levels to check before blaming anyone. "Small Independent Practices Can Run These Programs Well" gives the fair site comparison (Table 13.3). "Every Case Here Starts With the Wrong Diagnosis" gives the order to check evidence in. Read them for the argument and the Riverbend, Lakeshore, Cedar Valley, Northfield, and Summit cases. This skill runs the same method on the user's own program and never reuses the cases' numbers as the user's. **The skill works from counts, never charts.** Every rung is a count the practice already has or can pull. No patient names, dates of birth, or record numbers go into the session. Where the user needs to read records (Step 5), they read them; the skill takes their de-identified summary. **The skill never supplies a number.** Rung counts, denial reasons, payment per paid patient-month, minutes: each comes from the user, their billing company's remittances, or their own logs. A missing count reads "not counted," which is its own finding. The book's point about Helped applies to every rung: a rung nobody measures is a finding, not a zero. ## Before running 1. **Name one program and one month.** One ladder per program. If the user runs several programs, run the skill once for each. 2. **Get the current explanation.** Ask what the partners, the administrator, or the team currently believe is wrong, in the words they use ("we need another nurse," "the billing company loses claims," "patients won't use devices"). Write it down as the first diagnosis. The skill's job is to test it, not to adopt or dismiss it. 3. **Ask for the counts up front.** Eligible, offered, enrolled, engaged (any contact this month), served (met the program's monthly requirement), billed, paid, and whatever the practice measures for helped. Also ask for denial counts by reason, and for the collected amount per paid patient-month from remittances, not the fee schedule. 4. **Ask how each rung is defined.** Especially "engaged" and "served." A voicemail counted as contact inflates engaged. If sites or people will be compared, confirm they count the same way first. ## Step 1: Build the ladder and rank the gaps 1. Lay out the eight rungs in order: Eligible → Offered → Enrolled → Engaged → Served → Billed → Paid → Helped. 2. For each rung, record the count, the drop from the rung above as a number, and the drop as a percentage of the rung above. 3. Rank the gaps by patients lost. Circle the largest drop. That is the rung to work on first, whichever rung looks easiest to fix. 4. Mark each gap with what it usually means and where to look first, from Table P.1: | Gap | What it usually means | Where to look first | | --- | --- | --- | | Eligible but not offered | Nobody found the patient, or found them and never asked | The eligibility list and who owns working it | | Offered but not enrolled | The explanation, the trust, or the patient's coinsurance didn't land | The enrollment conversation and who has it | | Enrolled but not engaged | The patient stopped answering, stopped transmitting, or never saw the point | Outreach, devices, and whether the program fits the patient | | Engaged but not served | The required work didn't get done this month | Capacity, work queues, and how time or activity is tracked | | Served but not billed | The work happened but the record can't prove it | Documentation and charge capture | | Billed but not paid | A claim failed validation, eligibility, or a payer rule | Pre-claim checks and denial patterns | | Paid but not helped | The program pays and does nothing for the patient | Clinical design, and whether to keep the program at all | 5. If Helped is unmeasured, say so on its own line. It is the rung that should worry the practice most, and "nobody could tell me" is the finding. 6. Read denials before blaming whoever sends the claims. Another practice billing the same patient that month, or a patient who moved to a Medicare Advantage plan, is a pre-claim check the practice didn't run. Neither is a billing-company problem. Every cause the skill writes names a step in the process. Delete any cause that blames patients in general ("patients aren't interested") or staff in general ("the team needs to work harder"). Those are guesses. ## Step 2: Put the objections on the ladder If the user has reasons the practice gave for not running the program, or for stopping it, place each one on the rung it hits, using the words people actually said. Use the book's twelve reasons as the reference (Table 4.2): | Reason | Rung it hits | What actually fixes it | | --- | --- | --- | | Uncertainty and changing rules | Before the ladder | Build the operation underneath the codes; review rule changes once a year on a calendar | | Confusing rules | Engaged → Served; Served → Billed | One plain page of rules per program for the staff who do the work, checked by the compliance adviser | | Patient cost sharing | Offered → Enrolled | An honest consent conversation that names the cost, after checking secondary coverage | | Staffing | Engaged → Served | Measure where staff minutes go before hiring anybody | | Supervision | Before the ladder; Served → Billed | Learn what general supervision allows | | Incompatible billing | Billed → Paid | A pre-claim check for other billers, the current plan, and same-month combinations | | Documentation | Served → Billed | Record the work as it happens, in a structure that answers the payer's questions | | Device logistics | Enrolled → Engaged | One owner for devices, a setup call, and a follow-up when a device goes quiet | | Workflow fragmentation | Eligible → Offered | One owner and one work queue per program | | Audit fear | Before the ladder; Served → Billed | Build audit evidence as you go and hold the program to the Honest Standard | | Bad vendor experiences | Before the ladder; Paid → Helped | Keep the clinical relationship inside the practice; never pay anyone per enrollment | | Low trust that payment will last | Before the ladder | Build a capability that survives code changes; size staff to what the program earns | Each placement needs a mechanism, not just a rung. "Staffing affects everything" is a shrug. "Staffing means enrolled patients don't get their required minutes this month" names a rung. Group the objections by rung and report which rung has the most and which has none. The rungs with the most objections are where the partners will want evidence. Cost sharing is the one patient-side reason. It explains a gap at Offered → Enrolled. It can't explain why a practice never offers the program at all. ## Step 3: Value closing each gap, then check capacity Only when the user supplied a collected amount per paid patient-month: 1. For each gap, close a quarter of it and assume the moved patients convert at the practice's own current rates on every rung below. Count the extra paid patient-months and multiply by the user's collected amount. Show the arithmetic. 2. Then check capacity for the top gap. The extra paid patients at the top need extra served patients, and each served patient needs the program's monthly minutes. Multiply and compare with the minutes the team has. If the team already can't serve patients it has (a large Engaged → Served gap), say so. The book's rule governs the sequence: don't pour patients into the top of the ladder while the middle is full. Pushing offers without fixing capacity moves patients from "never asked" to "enrolled and not served," which is worse, because now the patient has been promised something. If the user has no collected amount, skip the dollars and rank by patients lost; say that the dollar comparison waits on remittances. ## Step 4: Find the level before anyone names a fix For the widest gap, and for any specific miss the user describes (a late follow-up, a patient who fell off), check four levels in this order: 1. **Policy.** Does the payer's rule make the right work hard or impossible? 2. **Practice system.** Do the tools, data flows, and templates make the right work the default? 3. **Team.** Does a named role own this work, with the time to do it? 4. **Individual.** Only now: did a person with the right tools, clear ownership, and a clear expectation fail to do the job? Clear the system and team levels before reaching the individual. Propose fixes only at the highest level where the evidence shows a real cause. If the answer drifts to "retrain staff," ask which system change would have made the retraining unnecessary. Size changes where the level usually sits (Table 4.4). In a solo or small practice the ladder usually breaks at Engaged → Served because the program lives in one nurse's hours; check what happens in a month when that nurse is out. In a small group, sites get very different results because each runs the program its own way. In a large group or health system, Eligible → Offered and Enrolled → Engaged break because a central team calls patients who don't know them. ## Step 5: Test the first diagnosis in the book's evidence order Take the first diagnosis from "Before running this" and check evidence in this order, stopping when one step settles it: 1. **Find the rung.** Steps 1–3 above. A staffing fix for a finding problem wastes a salary. 2. **Find the level.** Step 4 above. 3. **Find the minutes.** Where does the human time go, care minutes against admin minutes? If nobody has measured, the answer is the two-week study the `administrative-denominator-reduction` skill runs. Don't accept "we need more hours" until someone has measured what the hours now buy. 4. **Read ten records.** The user picks ten patients at random and reads what happened to them this month. Aggregates hide mechanisms ten charts show in an hour. 5. **Follow the money and the incentives.** The vendor contract, the staff pay plan, the pricing. People and vendors do what they're paid to do. For the first diagnosis, write the result that would prove it wrong. A diagnosis nobody could prove wrong is an opinion. The book's eight cases show the pattern: "we need another nurse" was a missing eligibility list; "patients won't use devices" was a cohort chosen by eligibility, not clinical reason; "we picked a bad vendor" was a program nobody owned; "she's just better" was a better queue; "we need twelve more nurses" was a 38-field template; "patients don't tell us" was a fax inbox nobody checked. Use these as patterns to test for, never as the user's answer. ## Step 6: Compare sites or people fairly, if the user has more than one When two or more sites, teams, or care managers run the same program: 1. Calculate each rung as a percentage of the rung above, per site. Flag differences over ten percentage points on the same rung. 2. Before calling any site better, rule out each factor from Table 13.3: the same definition of each rung, similar patients, similar payer mix, the same time period and program maturity, caseload per person, and incentives. Say which the user's data rules in or out, and what count would settle each one it can't. 3. Reject "better culture" or "better staff" as a conclusion. A star performer is often running a process nobody has written down. 4. Treat the best site's practice as a hypothesis until it works at a second site. Recommend testing it at one more site, not copying it everywhere. ## Step 7: Conclude with the rung to fix first Close with a short statement, not a ladder with no verdict. It has five parts: - **The rung to fix first,** with its count, and the level where its cause lives. - **The first diagnosis, confirmed or rejected,** with the evidence step that settled it, or the step still needed. - **The fix,** at the level the evidence points to, with an owner and a date. One fix at a time, because two fixes at once make it impossible to tell which one worked. - **What not to do yet.** Typically: don't hire before the minutes are measured; don't push offers while Engaged → Served is wide; don't switch vendors before the practice's own gaps have owners. - **The ask for the partners.** Keep it small: a measurement, a cohort, a decision date. The book's rule: take the program to the partners as the care the practice already gives for free, and let the ladder show the money. Never lead with a revenue number. Where two readings stay live (the ladder can't say whether an Engaged → Served gap needs more hours or better-spent hours; one site's lead survives some checks but not others), present both side by side, each with its reasoning and the measurement that would settle it. Where the counts are too thin to name a rung (several rungs "not counted"), say so plainly, and make counting them the first recommendation. For a program the practice will keep running, point to the weekly form of this diagnosis: the Monday Morning Page in "The Care Delivery Professional's New Operating Model" puts the Capture Ladder by program first, admin minutes second, overdue escalations third, then denials, patients who fell off, and one thing to fix. ## Checks before calling the work done - **Every count traces to the user.** Delete any number the skill supplied. - **Every drop is shown as a number and a percentage of the rung above,** and the arithmetic checks. - **Helped has a line,** even if the line says "not measured." - **No cause blames patients or staff in general.** - **The first diagnosis has a disproving test next to it.** - **The conclusion names one rung, one level, one fix, and one owner.** ## What this skill does not do It doesn't read patient records, and it doesn't take identifiable data. It doesn't decide whether to hire, fire a vendor, or stop a program; the partners do, with the ladder in hand. It doesn't borrow Riverbend's counts or the book's illustrative $60 as the user's. It doesn't answer billing, coding, or compliance questions; those go to the practice's compliance adviser. And it doesn't round a wide gap at the top up to "a revenue opportunity." When the counts are missing, the correct output is "count these rungs first," not a confident diagnosis built to fill the silence.