Chapter 33: Inpatient Facility Coding: MS-DRGs, Present on Admission, and the Hospital Revenue Cycle

"The hospital was not paid for the four days, the oxygen, or the bed. It was paid for the sentence that explained why they were needed." — constructed

Overview

Everything this book has priced so far was priced by the service. A code went on a line, the line had an allowed amount, and the claim's value was the sum of its lines. Chapter 23 built that arithmetic for the physician's office; Chapter 26 built the institutional claim that carries a hospital's charges.

This chapter is where that logic ends. When a Medicare patient is admitted to an acute care hospital, the hospital is not paid for what it did. It is paid for what was wrong with the patient — a single, predetermined amount for the entire stay, assigned by grouping the coded diagnoses and procedures into a Medicare Severity Diagnosis-Related Group (MS-DRG). The forty charge lines on the claim, the four days, the oxygen, the consultants: none of them changes the payment. The codes do.

That single design decision rearranges the entire revenue cycle. In the office, a better-documented record supports a better-supported line. In the hospital, one documented phrase can move the whole stay from one payment tier to another — and this chapter closes on an admission where the difference between "hypoxia" and "acute respiratory failure with hypoxia" is worth exactly \$1,867.44. Not because anyone upcoded anything. Because one sentence was written down and its alternative was not. If it isn't documented, it didn't happen has been this book's first theme since Chapter 4; in the inpatient hospital, the theme has a price with cents on it.

The chapter runs the whole machine once, honestly: why prospective payment exists (§33.1), the inpatient principal diagnosis rules Chapter 9 §9.4 deferred here (§33.2), the CC/MCC severity tiers (§33.3), how a grouper assigns a DRG (§33.4), the payment arithmetic (§33.5), the case mix index and what it cannot tell you (§33.6), present on admission and the hospital-acquired condition provision (§33.7), length of stay, outliers, and the transfer rule Chapter 26 §26.7 promised the math for (§33.8), and ICD-10-PCS at a structural level (§33.9). Then §33.10, the one word.

In this chapter, you will learn to:

• Explain what a prospective payment is and why Medicare adopted one
• Select an inpatient principal diagnosis, including the hard cases
• Use CC and MCC status correctly, including when it changes nothing
• Follow the grouper's logic from coded record to MS-DRG
• Compute a DRG payment and reconcile it to the dollar
• Read a case mix index without being fooled by it
• Assign POA indicators and apply the HAC provision
• Work the transfer rule's per-diem arithmetic
• Read an ICD-10-PCS code character by character

33.1 Why the hospital is paid for a stay, not a service

Start with the decision on the desk. You are the coder — perhaps a coder who has spent years in a practice like Northgate — and in front of you is your first inpatient record: a four-day admission, sixty pages, eleven charge lines already on the account from pharmacy and respiratory therapy and the laboratory, and a bill that will run to five figures in charges. The question you have been trained to ask is "which codes support which lines?"

The answer here is: none of them. On an inpatient Medicare claim, the diagnosis and procedure codes do not support the charges. They replace them as the basis of payment. The hospital will be paid one amount for this stay, and the codes you assign are what determine which amount.

Where the design came from

Until the early 1980s, Medicare paid hospitals on reasonable cost: the hospital spent money caring for beneficiaries, reported what it spent, and was reimbursed. The incentive problem is visible from the sentence. Every additional day, test, and supply generated additional reimbursement, hospital spending grew faster than anyone could fund, and no mechanism inside the payment system pushed back.

The replacement came out of research at Yale in the 1970s, where the diagnosis-related group was developed as a way of classifying hospital stays into groups that are clinically coherent and similar in resource consumption. New Jersey ran the idea as a payment pilot, and with the Social Security Amendments of 1983 the Congress made it Medicare's national method: the Inpatient Prospective Payment System (IPPS), effective for cost reporting periods beginning October 1, 1983. Case Study 1 tells that story properly, because it is the largest natural experiment in this book's subject matter and its results are the reason every later payment system looks the way it does.

Prospective is the load-bearing word. The price is set before the stay happens, per category of stay, by rule. The hospital that treats the patient efficiently keeps the difference; the hospital that spends more than the payment absorbs the loss. Payment stopped following cost and started following classification — and classification is built from codes, which is why this chapter is in a coding book rather than a finance one.

What IPPS pays, and what it does not

The system this chapter describes governs operating and capital payments to acute care hospitals for Medicare Part A inpatient stays. It is worth one paragraph of boundaries, because every boundary is a different payment system with its own chapter or its own book:

• The physicians are not in it. The hospitalist, the pulmonologist, and the surgeon bill their professional services separately, on professional claims, under Chapter 23's fee schedule logic. One admission, multiple claims — Chapter 16 §16.9's split, at full scale. The day-0 chest X-ray on Account 22-8891 makes the boundary concrete: the radiologist's interpretation goes out as 71046-26 on a professional claim (Chapter 19 §19.1's arrangements), while the technical resource — the machine, the technologist, the film — is simply part of the stay the DRG buys.
• The outpatient department is not in it. That is the Outpatient Prospective Payment System and Chapter 34.
• And several kinds of hospital are excluded — critical access hospitals, and inpatient psychiatric, rehabilitation, and long-term care facilities each have their own systems. Cite them by name if you work in one; this chapter's arithmetic does not transfer.

Commercial payers and state Medicaid programs frequently license or adapt the MS-DRG system for their own inpatient contracts — Chapter 2 §2.6's case rates are often exactly this — so the machinery below is not Medicare trivia even if you never bill Medicare.

The file this chapter works

The record on the desk is Account 22-8891, the inpatient admission this book introduced in its anchor set: Ridgeview Regional Medical Center (constructed), a 310-bed community hospital; a Medicare beneficiary, 71 years old, admitted through the emergency department (ED) in respiratory distress and discharged on day 4. The diagnoses, the coding, and the money all follow in this chapter, and every figure on the file is a constructed teaching figure.

One reminder before the rules start, because it prevents a common confusion: the UB-04 still carries all the charges. Revenue codes, units, dollars, the 0001 total line — everything Chapter 26 taught is present on the inpatient claim. The charges simply are not the payment basis. They still matter: the cost report is built on them, and §33.8's outlier computation reaches the payment through them. A charge structure nobody maintains corrupts both. The claim's form did not change when the payment logic did — what changed is which fields the money reads. On an inpatient claim the money reads FL 67 and FL 74: the diagnoses, the POA indicators, and the procedures.

And that is the chapter's thesis in one sentence: on an inpatient claim, the coding is the pricing. You code from the chart, but you get paid by the contract — and here the contract is a federal rule whose unit of purchase is the classified stay.

33.2 Principal diagnosis, inpatient rules

Chapter 9 §9.4 gave the definition and explicitly deferred the discipline of applying it to this section. Here is the definition again, because every word of it is load-bearing:

The principal diagnosis is the condition established after study to be chiefly responsible for occasioning the admission of the patient to the hospital for care.

That sentence comes from the Uniform Hospital Discharge Data Set (UHDDS) definitions, which govern inpatient reporting, and it is operationalized by Section II of the ICD-10-CM Official Guidelines for Coding and Reporting. Three phrases carry the weight:

• "After study." Not the presenting symptom, not the admitting impression — what the workup concluded. A patient admitted for chest pain whose workup finds a myocardial infarction has the infarction as principal. The UB-04 holds both honestly: FL 69 carries the admitting diagnosis (what was suspected), FL 67 carries the principal (what was established) — Chapter 26 §26.2, and the disagreement between them is the story of the admission, not an error.
• "Chiefly responsible." Not most severe, not most interesting, not most expensive. Responsible for the admission.
• "Occasioning the admission." The question is why this patient needed to be in a hospital bed — which is why a condition that developed on day 2, however serious, can never be the principal diagnosis. It did not occasion anything.

The rules the definition does not settle

The definition sounds self-executing. It is not, and Section II exists because of the cases it does not settle. The ones you will meet in the first month:

Two or more conditions that each meet the definition. A patient admitted with both a COPD exacerbation and acute heart failure, both present on admission, both worked up, both treated. Where the Guidelines' sequencing instructions, the Tabular's conventions, and the circumstances of admission do not establish one over the other, either may be sequenced first. Read that rule with respect: it is the only place in this chapter where the coder holds a genuinely discretionary choice with money on both sides of it, and §33.10's compliance discussion returns to what disciplined use of that discretion looks like. The discretion is real but narrow — it exists only when the conditions genuinely co-occasioned the admission, and the record has to support that reading.

A symptom followed by contrasting or comparative diagnoses. The workup ends without a winner — "syncope, possibly arrhythmic, possibly orthostatic." The Guidelines direct how these are sequenced and reported, and the current text is the authority; the point to carry is that the record's uncertainty is reportable, not a defect to be coded around.

The uncertain diagnosis at discharge. Chapter 9 §9.5 owns this rule and its outpatient opposite: inpatient, a condition documented at discharge as "probable," "suspected," "likely," or "still to be ruled out" is coded as if established. This is the single most consequential setting difference in diagnosis coding, and coders who move between settings must consciously switch. What Chapter 9 could not show you is the stakes: in the office, the uncertain-diagnosis rule changed a code. Here it can change the DRG.

The original treatment plan not carried out. Admitted for a scheduled procedure, the procedure canceled — the condition that occasioned the admission remains principal even though nothing was done about it, and the record explains the rest.

A complication of care occasioning the admission. When a complication of surgery or other care is what brought the patient in, the complication code is principal, with the Tabular's instructions governing the specifics — and Chapter 4 §4.7's discipline applies with full force: complication is a provider's word. A coder may not promote an outcome into a complication because it looks like one.

Admission from outpatient surgery. Same-day surgery goes wrong, or recovery does not go as planned, and the patient is admitted. The Guidelines direct the selection — the complication if there is one; the reason for the surgery if no complication is documented — and the current text governs the details.

📋 Read the Chart

text FIGURE 33.1 — "The coding summary, day 4" [Account 22-8891] THE DOCUMENT The coder's diagnosis worksheet for a four-day inpatient stay, built from the ED record, the H&P, three days of progress notes, and the discharge summary. THE CONTEXT Ridgeview Regional Medical Center (constructed), Medicare beneficiary, 71, admitted through the ED in respiratory distress on a background of COPD. Discharged home day 4. WHAT IT SHOWS Principal: J44.1, COPD with acute exacerbation. The H&P states the patient was admitted for treatment of the exacerbation. Secondary: J96.01, acute respiratory failure with hypoxia — documented in the ED record and the H&P, POA = Y. Also on the record: I50.32 (chronic diastolic heart failure), E11.22 (type 2 diabetes with diabetic CKD), N18.31 (CKD stage 3a), Z79.4 (long-term insulin), each documented as evaluated or treated during the stay. A stage 3 sacral pressure ulcer, L89.153, first documented on day 2. POA = N. WHAT IT DOESN'T It does not say the respiratory failure occasioned the admission by itself; the physician framed the admission as the exacerbation, with the respiratory failure as its acute consequence. And it does not yet say what the stay will be paid — that takes §33.4 and §33.5. THE DECISION Sequence J44.1 as principal. Report J96.01 as a secondary diagnosis with POA = Y, and the chronic conditions per Section III. Report L89.153 with POA = N — accurately, whatever it does to the payment. THE LESSON An inpatient record is coded from the whole stay, "after study" — but sequenced from the admission. The two ends of the stay answer different questions.

Why J44.1 and not J96.01 first? Because this is the exact situation Chapter 11 §11.5 flagged as a Chapter 33 problem. Both conditions were present on admission. The respiratory failure is the more alarming diagnosis, and the instinct to lead with the most severe thing is strong and wrong — the test is chiefly responsible for occasioning the admission, and this record's physician documented the admission as treatment of the exacerbation, with the failure as its manifestation. The Guidelines' chapter-specific instructions on respiratory failure sequenced with another acute condition govern; on this record they support the exacerbation as principal. On a different record — a patient brought in in respiratory failure, intubated in the ED, with COPD as background — the failure itself may be principal, and the DRG would come from a different family entirely, one built around respiratory failure rather than COPD. Same two codes, opposite sequencing, different money — decided by what the record says occasioned the admission.

🔢 Code It

The documentation, condensed from a discharge summary (constructed teaching example): "78-year-old admitted from the ED with productive cough, fever, and infiltrate on chest X-ray. Treated with IV antibiotics for community-acquired pneumonia, organism not identified. History of COPD, maintained on home inhalers, stable throughout the stay and continued on home regimen. Discharged day 3."

The path: the condition established after study, chiefly responsible for the admission, is the pneumonia. Organism not identified → J18.9 (pneumonia, unspecified organism) as principal. The COPD is reportable as a secondary diagnosis under Section III — it was evaluated and its treatment continued — as J44.9 (COPD, unspecified), because this record documents no exacerbation. Verify both in the Tabular; the COPD categories carry instructional notes about associated conditions that must be read on every encounter, and the pneumonia/COPD combination in particular is governed by the current Tabular instructions — read them, because the answer is convention-driven, not intuitive.

The plausible wrong answer: J44.1 as principal, on the reasoning that "pneumonia in a COPD patient is an exacerbation." It is not, unless the provider says it is. The exacerbation is a clinical characterization that belongs to the physician; a coder who infers it has made a clinical judgment (Chapter 4 §4.7) — and, as §33.4 will show, has also moved the stay from the pneumonia DRG family (193–195) into the COPD family (190–192). A sequencing inference here is not a style error. It is a different payment, reached without documentation — the same shape as the I10/J44.1 claim in Chapter 26 §26.10's figure, where an implausible principal diagnosis was the first thing a careful reader questioned.

One more thing belongs in this section because the inpatient setting changes it: the secondary diagnoses have their own admission test. Section III governs which additional conditions are reportable — those that were clinically evaluated, treated, worked up, or that extended the stay or increased nursing care. That is the inpatient cousin of the "addressed" discipline Chapter 9 §9.7 taught: a problem-list entry is not reportable because it exists; it is reportable because the stay engaged with it. On Account 22-8891 the heart failure, the diabetes with its kidney disease, and the insulin status all pass that test — the record documents their evaluation and management during the stay. Report what the stay engaged; nothing else. §33.3 explains why that list is suddenly worth real money.

33.3 Comorbidities and complications: CC and MCC

In the office, secondary diagnoses on a claim mostly told the payer a story that supported medical necessity. On an inpatient claim, secondary diagnoses are payment inputs, because the MS-DRG system prices severity — and severity is measured by what else is wrong with the patient.

A CC — complication or comorbidity — is a secondary diagnosis that CMS has designated as meaningfully increasing the resources a stay consumes. A comorbidity is a condition the patient brought to the hospital; a complication is one that arose during the stay. The designation does not care which — the C and the C are one payment concept, which is why the abbreviation gets to be ambiguous.

An MCC — major complication or comorbidity — is the higher tier: a secondary diagnosis designated as a major driver of resource consumption. Acute respiratory failure, acute renal failure, sepsis — conditions of that gravity.

The Centers for Medicare & Medicaid Services (CMS) maintains the CC and MCC designations as lists, published with the MS-DRG Definitions Manual and revised annually in the IPPS final rule — codes move on, off, and between tiers every year. A textbook that told you "this code is a CC" as a permanent fact would be wrong within a rulemaking cycle, so this book tells you the structural facts instead and they are enough:

• Every secondary diagnosis is either an MCC, a CC, or neither, per the current year's list.
• Many DRG families are split into three severity tiers — with MCC, with CC, without CC/MCC — and some into two or one; the split exists only where the data showed the severity levels genuinely cost different amounts.
• The highest-severity secondary on the record decides the tier. One MCC puts the stay in the "with MCC" DRG. No MCC but at least one CC: the middle tier. Neither: the base tier.

That third rule does more work than it appears to. Read its consequences off slowly, because they are where new inpatient coders' intuitions fail:

Severity does not accumulate. A record with one MCC and a record with one MCC plus six CCs group to the same DRG and pay the same amount. Once the tier is set, additional CCs change nothing about this claim's payment — though they still belong on the claim when Section III supports them, because §33.6's case mix index, quality measurement, and Chapter 36's risk adjustment all read the full record.

The principal diagnosis cannot be its own CC. The severity tiers are built from secondary diagnoses. And a handful of related exclusions follow the same logic: a secondary too close to the principal to represent additional burden is not counted against it — the grouper applies exclusion logic from the Definitions Manual, which is one of several reasons §33.4 will tell you to let the software do the lookup and reserve your judgment for what feeds it.

And POA can switch a designation off. §33.7: a condition that would be an MCC does not function as one for payment if it was acquired in the hospital and sits on the HAC list. Severity the hospital caused is not severity the hospital is paid for.

Account 22-8891, tier by tier

On the anchor record, the secondary diagnoses are J96.01, I50.32, E11.22, N18.31, and Z79.4 — plus the day-2 pressure ulcer, which §33.7 handles. J96.01, acute respiratory failure with hypoxia, is an MCC, and that single designation is the whole story: the stay groups to the "with MCC" tier of the COPD family regardless of what the other secondaries are designated. The chronic conditions — established heart failure and diabetes with kidney disease — are exactly the kind of documented, managed comorbidities that populate CC lists, and on a record without the respiratory failure they would be what stood between the middle tier and the bottom one. Whether a specific chronic code carries CC status this year is a lookup, not a memory: verify against the current list, every year, without exception.

⚠️ Where Claims Die

The reversed loss. Most of this book's failure stories are about claims that overstated something. Inpatient severity coding fails in the other direction more often: the record supports a CC or an MCC and the claim never receives it, because the condition lives in a flowsheet, a lab value, or a consultant's note that the discharge summary never absorbed — or because it was treated and resolved on day 1 and nobody wrote it in a codable form.

Chapter 5 §5.8 said upcoding and downcoding are both errors, and here is that principle with a facility's rent attached: a hospital that systematically under-captures documented severity is misdescribing its patients, understating its case mix index, distorting the quality data computed from its claims — and being underpaid for care it actually delivered. The disciplined response is not "code conservatively." It is: code what the record supports, all of it, and only that.

The gap between clinically present and documented in codable language is not yours to close by inference. It is the clinical documentation integrity (CDI) profession's entire reason to exist, and §33.10 shows what one instance of that gap costs. Chapter 38 §38.3 owns the query that closes it compliantly.

33.4 How a DRG is assigned

The assignment is performed by software — the grouper, which Chapter 6 §6.3 put in your toolkit — and no human assigns DRGs by hand in production. You still have to know the logic, for the same reason Chapter 6 gave about encoders: the software decides nothing; it executes decisions that were made in the coding. When a DRG looks wrong, the error is almost never in the grouper. It is in what the coder fed it.

The pipeline, for the ordinary case:

THE GROUPER'S LOGIC — from coded record to MS-DRG

  THE CODED RECORD
    principal dx (FL 67, sequenced first) · secondary dx + POA · procedures (FL 74)
        │
        ▼
  [0]  PRE-MDC CHECK
       A short list of resource-dominant cases — transplants,
       tracheostomy with prolonged ventilation — group on the
       PROCEDURE before any diagnosis logic runs.        │ not one? ▼
        │
        ▼
  [1]  MAJOR DIAGNOSTIC CATEGORY (MDC)
       The PRINCIPAL DIAGNOSIS maps the stay into one of 25
       MDCs — broadly, one per body system or major cause.
       J44.1 → the respiratory MDC.
        │
        ▼
  [2]  THE SURGICAL PARTITION
       Is there an OR PROCEDURE on the record (an ICD-10-PCS
       code CMS designates as operating-room-defining)?
         YES → a SURGICAL DRG family within the MDC,
               selected by the procedure's hierarchy
         NO  → a MEDICAL DRG family, selected by the
               principal diagnosis
        │
        ▼
  [3]  THE SEVERITY SPLIT
       Within the family: does any secondary diagnosis that
       counts (POA logic applied, exclusions applied) reach
         MCC?        → "with MCC" tier
         CC only?    → "with CC" tier
         neither?    → "without CC/MCC" tier
        │
        ▼
  ONE MS-DRG                                  [schematic — not to scale]

Walk Account 22-8891 through it. No pre-MDC condition. J44.1 lands the stay in the respiratory MDC. No OR procedure was performed — oxygen, nebulizers, and steroids do not open the surgical partition — so the stay takes the medical path, and within it the principal diagnosis selects the COPD family. The severity split reads the secondaries, finds J96.01 (MCC, POA = Y, no exclusion), and assigns:

MS-DRG 190 — chronic obstructive pulmonary disease with MCC.

Had the record carried a CC but no MCC: 191, COPD with CC. Neither: 192, COPD without CC/MCC. The pneumonia patient from §33.2's Code It walks the identical pipeline to a different family — 193/194/195, simple pneumonia and pleurisy, with MCC / with CC / without — and a heart-failure principal reaches 291/292/293. One pipeline, every admission in the country.

Three structural facts complete the picture:

The count and content of the groups change annually. There are roughly seven hundred fifty MS-DRGs — the number moves as CMS adds, deletes, and re-splits groups in the IPPS final rule, so verify the current count and never memorize it. The system itself was last restructured in 2008, when Medicare replaced the older DRGs with the severity-split MS-DRGs this chapter teaches — a change with its own cautionary tale, told in Case Study 2.

The surgical partition outranks intuition. A surgical DRG is defined by the presence of an OR-designated procedure, not by whether anyone would call the patient "a surgical patient" — and the designation belongs to CMS's tables, not to the operating schedule. Some bedside procedures carry OR designation; some procedures done in an OR do not. §33.9's PCS structure is what feeds this step, and a procedure code error that crosses the partition produces some of the largest single-claim payment errors in the hospital.

And the grouper's version matters. Groupers are versioned to the federal fiscal year, which begins October 1 — the same date the ICD-10-CM and ICD-10-PCS code sets update. A claim grouped under the wrong year's logic is a claim priced under the wrong year's rules. This is this book's standing update-cycle warning wearing hospital clothes: ICD-10-CM and ICD-10-PCS change every October 1, CPT every January 1, HCPCS Level II quarterly — and the MS-DRG definitions, the CC/MCC lists, and every relative weight in §33.5 change every year in the IPPS final rule. Code and price from the current year's tools, never from a textbook, including this one.

🔍 Check Your Understanding

1. A patient is admitted for a COPD exacerbation; on day 2 she falls and fractures a hip, which is surgically repaired. Which step of the pipeline does the fracture repair change, and can the fracture be the principal diagnosis?
2. A record carries one MCC. The coder finds two more well-documented CCs. What does adding them do to the DRG? Why might they still belong on the claim?
3. Same clinical facts, two records: one documents "admitted in acute respiratory failure due to COPD exacerbation," the other "admitted for COPD exacerbation; hypoxic on arrival." Which grouper inputs differ?

Answers: (1) The repair is an OR procedure — the stay crosses to the surgical partition and a surgical DRG. The fracture cannot be principal: it did not occasion the admission — it happened after it, and its POA indicator will say so. (2) Nothing — the tier is already set by the MCC; they belong on the claim because Section III supports them and because the CMI, quality data, and risk adjustment read the whole record. (3) The second record supports no J96.- code at all — "hypoxic" is a finding, not the diagnosis — so the severity split may run without an MCC. That gap is §33.10.

33.5 Relative weight, base rate, and the payment arithmetic

The DRG is a classification. The payment attaches to it through two numbers, and the arithmetic is one multiplication long:

PAYMENT = the DRG's RELATIVE WEIGHT × the hospital's BASE RATE

The relative weight expresses how resource-intensive a DRG's average stay is compared with the average of all stays, which is scaled to sit near 1.0. A weight of 1.1015 says "about ten percent more costly than the national average case"; the biggest surgical DRGs carry weights many times that. CMS recalibrates every weight annually from national claims data, in the IPPS final rule.

The base rate is what one average-weight case is worth at this hospital. It starts from a national standardized amount, which is split into a labor-related share and a non-labor share; the labor share is multiplied by the wage index of the hospital's labor market, so a case in a high-wage city is funded at that city's cost of staffing it. Operating and capital amounts are computed separately in practice; this book folds them into one illustrative figure. Onto the product of weight × base rate, the rule then layers the adjustments that recognize what an averaged national price cannot: percentage add-ons for teaching hospitals (indirect medical education) and for hospitals serving a disproportionate share of low-income patients, uncompensated-care payments, new-technology add-ons, penalties and bonuses from the quality programs — and §33.8's outlier payments for the cases the average badly misprices. Every one of those is real money and none of them changes the core multiplication, so this book computes the core and names the rest.

For Ridgeview, this book freezes one illustrative figure and reuses it everywhere: base rate \$6,200.00 [constructed — verify current amounts, shares, and your hospital's wage index in the annual IPPS final rule], built as a labor share of \$4,216.00 × wage index 1.0000 + non-labor \$1,984.00 = \$6,200.00. (Check: 4,216.00 + 1,984.00 = 6,200.00 ✓ — a 68% labor share, near the real structure; the actual shares and amounts change annually.)

🧮 Run the Numbers

The COPD family at Ridgeview [constructed — verify current weights in the annual IPPS final rule and the hospital's own base rate]:

MS-DRG	Description	Relative weight	× base rate \$6,200.00
190	COPD with MCC	1.1015	\$6,829.30
191	COPD with CC	0.8003	\$4,961.86
192	COPD without CC/MCC	0.6555	\$4,064.10

Checks: 6,200.00 × 1.1015 = 6,829.30 ✓ · 6,200.00 × 0.8003 = 4,961.86 ✓ · 6,200.00 × 0.6555 = 4,064.10 ✓

The differences between the tiers are the point of the table:

text 190 vs 191 6,829.30 − 4,961.86 = 1,867.44 the MCC's value 191 vs 192 4,961.86 − 4,064.10 = 897.76 the CC's value 190 vs 192 6,829.30 − 4,064.10 = 2,765.20 the full spread

Checks: 1,867.44 + 897.76 = 2,765.20 ✓

Interpretation. Account 22-8891 groups to DRG 190 and Ridgeview is paid \$6,829.30 for the stay — whether the patient stayed three days or six, whether charges totaled nine thousand dollars or nineteen. The severity tier is decided by documentation and coding, which means the \$1,867.44 between 190 and 191 is decided there too. Hold that number; §33.10 is about it.

Two habits to build from the arithmetic:

Never quote a weight or a rate as current. Every number above is a teaching figure. The real weights are published annually in the IPPS final rule's tables, the real base-rate components in the same rulemaking, and the hospital's finance office knows its actual blended rate. The structure — weight × base rate, labor share × wage index — is what transfers; the values expire every October 1.

And notice what stopped mattering. In Chapter 23 you reconciled payment to the line: units, modifiers, fee schedule lookups. Here, line-level charge detail has no direct payment role — a fact that reorganizes where the revenue risk lives. The inpatient hospital's exposure is not "a line denied"; it is the whole stay priced one tier wrong, in either direction, times every stay where the same documentation gap repeats.

🎓 Exam Watch

Facility credentials — the Certified Coding Specialist (CCS) and the Certified Inpatient Coder (CIC) — test this chapter directly; the Certified Professional Coder (CPC) touches it lightly. The arithmetic questions are gifts: weight × base rate, read the table, done. The traps are conceptual: (1) a question that adds a second CC to a record that already has an MCC and asks what changes (nothing — tiers do not stack); (2) a stem where the "most severe" condition developed after admission and the answer choices invite you to sequence it as principal (it cannot be); (3) "which of these changes the payment?" with the length of stay among the choices (within the ordinary range, it does not — that is the whole meaning of "prospective"). And every exam expects you to know that weights, CC/MCC lists, and DRG definitions change annually, effective October 1.

33.6 Case mix index and what it does and does not say

One number summarizes a hospital's whole inpatient book of business, and because this chapter's readers will someday sit in meetings where it is used carelessly, the book is going to be precise about it.

The case mix index (CMI) is the average relative weight of a hospital's discharges over a period. Multiply the CMI by the base rate and you have the average payment per case.

For a toy hospital that discharged exactly the three COPD tiers [constructed]:

   CMI = (1.1015 + 0.8003 + 0.6555) ÷ 3  =  2.5573 ÷ 3  =  0.8524

   average payment = 0.8524... × $6,200.00 ≈ $5,285.09
   cross-check: (6,829.30 + 4,961.86 + 4,064.10) ÷ 3 = 15,855.26 ÷ 3 = $5,285.09 ✓

Real hospitals compute it across thousands of discharges, and finance watches it the way a practice watches Chapter 31's days in AR, because at a fixed base rate, CMI is revenue per case. A CMI that drifts down two points is a budget problem with a named owner by Friday.

What the CMI honestly says: how resource-intensive the hospital's coded and grouped caseload was. Compare a tertiary referral center's CMI with a community hospital's and the difference is real — the transplant programs and the trauma service are in the number.

What it does not say, and this is the section's reason to exist: the CMI cannot distinguish between "our patients got sicker" and "our documentation and coding got better." Both raise it, identically, because the CMI does not measure patients. It measures records — the severity that survived the trip from bedside to code. A hospital that starts a CDI program and captures the acute respiratory failure it had been leaving as "hypoxia" will watch its CMI rise with no change whatever in who came through the door. That rise is legitimate — the earlier number was wrong, understated by documentation failure — but notice what it means for the metric: the CMI is a measure of sickness that is also a measure of documentation capture, and the number alone cannot tell you which one moved. Medicare learned this at national scale the year the MS-DRG system launched, and paid for the lesson in the billions; Case Study 2 tells it.

It follows that CMI is a diagnostic, not a target. Used as a diagnostic, it is excellent: a CMI falling while the clinical service mix holds steady is exactly how a hospital discovers a severity-capture problem — the §33.3 reversed loss showing up on a dashboard. Used as a target ("raise the CMI"), it becomes the most dangerous instruction in this chapter, because the coding department controls one path to it and that path runs through §33.10's compliance line. The book's own denominator discipline from Chapter 29 §29.7 applies here too: a CMI moved by one high-weight transplant, by a change in what the surgical service does, or by a documentation initiative are three different events wearing one number.

📞 On the Phone

The chief financial officer's office, quarterly review. "Our CMI is up four points over last year. Great work by the CDI team."

The honest answer, from the coding manager: "Some of it is ours. I can show you which part — we track the DRG shifts that followed a physician query, and those are documented severity we used to miss. But the cardiac surgery program also ramped up this year, and those weights move the average by themselves. Before we take credit, let me split the change into service-mix and capture-rate components, because if we call it all CDI this year, then the year the program's gains plateau, this same number will say we failed."

What that answer models: never accept praise from a number that cannot distinguish the thing you did from the thing that happened anyway. The follow-up question an auditor would ask — "show me the queries behind the shift" — is one the CDI team should be able to answer from its own records, which is Chapter 38 §38.4's subject.

33.7 Present on admission and hospital-acquired conditions

Every diagnosis on an inpatient Medicare claim carries one more character of information: the POA — present on admission — indicator, reported for the principal and every secondary diagnosis. The question it answers is exactly what it says: was this condition present when the order for inpatient admission occurred? Conditions that developed during an outpatient encounter — in the ED, in observation, in outpatient surgery — before the admission order count as present on admission, a detail that matters constantly in real records and on exams.

The values:

The asymmetry between U and W is deliberate and teaches the whole design: U penalizes the record, W respects the clinician. If the documentation could not support an answer, Medicare resolves the doubt against the hospital — insufficient documentation is the hospital's own failure, and U on a consequential diagnosis should be rare and should trigger a query, not a shrug. If the clinician genuinely cannot determine timing — a condition that declares itself on day 2 but may well have been brewing on arrival — W is the honest answer and it is not penalized.

Why the indicator exists: the HAC provision

The POA indicator is not documentation trivia; it is the sensor for a payment policy. In the Deficit Reduction Act of 2005, Congress directed Medicare to stop paying extra for certain conditions acquired during the stay, and the mechanism is precise:

For a defined list of hospital-acquired conditions (HACs) — conditions judged reasonably preventable through evidence-based care, including stage 3 and 4 pressure ulcers, certain falls with injury, certain surgical site infections, catheter-associated urinary tract infections, vascular catheter-associated infections, and retained foreign objects — a code reported with POA = N (or U) does not count as a CC or MCC for the severity split. The condition is still coded, still on the claim, still in the data. It simply cannot raise the payment tier.

Read the mechanism's edges carefully, because they are where the understanding gets tested:

It removes severity credit; it does not deny the claim. The stay still pays — at the tier the other secondaries support. If some other MCC is present with POA = Y, the HAC changes nothing about this claim's payment at all.

It is a different thing from the HAC Reduction Program, the quality program (created later, under the Affordable Care Act) that reduces overall payment for hospitals in the worst-performing quartile on hospital-acquired-condition measures. Same acronym family, two mechanisms — one works claim by claim through POA, one works hospital by hospital through a percentage. Verify the current parameters of both; quality-program details change by rulemaking.

And its dollar effect on any one claim depends entirely on the rest of the record — which is exactly what Account 22-8891 demonstrates.

📋 Read the Chart

text FIGURE 33.2 — "The ulcer on day 2" [Account 22-8891] THE DOCUMENT Nursing skin assessments: the head-to-toe on admission (day 0), documenting intact skin, Braden score recorded; and the day-2 wound care consult documenting a stage 3 sacral pressure ulcer. Physician documentation of the ulcer follows on the day-2 progress note. THE CONTEXT The same four-day COPD admission. The coder is assigning POA indicators for every diagnosis on the claim. WHAT IT SHOWS L89.153 (pressure ulcer of sacral region, stage 3), first documented day 2, with an admission assessment affirmatively documenting intact skin. POA = N is not a judgment call on this record; the admission assessment settles it. WHAT IT DOESN'T It does not show a payment change on THIS claim. Stage 3 pressure ulcers are on the HAC list, so with POA = N the code cannot act as severity — but J96.01 (POA = Y) already holds the stay at DRG 190. The HAC provision fires and the payment does not move. THE DECISION Report L89.153, POA = N. Do not omit it, do not soften it, and do not treat "it doesn't change the DRG" as a reason to skip it — it feeds the quality data, the internal incident review, and the HAC Reduction Program's measures either way. THE LESSON Run the counterfactual to see what the policy protects: on a version of this record WITHOUT the respiratory failure, the ulcer would have been the only MCC-tier condition — and with POA = N it could not rescue the payment. The stay would group as if the ulcer were not there: with a documented CC, DRG 191; the difference is §33.5's $1,867.44. Severity the hospital caused is severity the hospital absorbs.

⚖️ Compliance Check

POA reporting is an attestation about clinical chronology, and it is audited as one — the Office of Inspector General (OIG) has kept hospital-acquired conditions and POA accuracy on its work plans, and the indicator is fully visible in claims data, which means the pattern-detection methods Chapter 21 §21.10 described work here too: a hospital whose POA = Y rate on HAC-list conditions sits far above its peers is answering a question nobody asked yet.

The bright lines: POA comes from the record — the admission assessments, the H&P, the timeline the clinicians documented — never from the payment consequence. Changing an N to a Y to preserve severity credit is a false statement on a federal claim; so is a query constructed to invite the provider to backdate a condition's onset. The compliant query when timing is genuinely unclear — and timing frequently is genuinely unclear — asks the open question and accepts the answer, including "cannot be determined," which is what W exists to report. Chapter 38 §38.3 owns the query's form; Chapter 31 §31.9's sixty-day overpayment rule owns what happens when an audit finds the pattern after payment. As always: requirements evolve, payer and state specifics vary, and your compliance officer and the current CMS guidance are the authorities — verify.

33.8 Length of stay, outliers, and the transfer rule

A prospective payment is a bet on an average, and this section is about the three places the average meets reality: the expected stay, the catastrophic stay, and the stay that ends somewhere else.

Two means, and why the geometric one runs the payment rules

CMS publishes two length-of-stay statistics for every DRG, recalculated annually from national claims:

• the arithmetic mean length of stay (AMLOS) — the ordinary average: total days ÷ total cases;
• the geometric mean length of stay (GMLOS) — the average computed multiplicatively (the nth root of the product of the stays), which systematically discounts extreme values.

Why keep two? Because inpatient stays are skewed: most cluster near a few days, and a small number run very long. A handful of sixty-day outliers can drag an arithmetic mean far above where the typical patient actually sits, while barely moving the geometric mean. Watch it happen [constructed]:

   five stays: 2, 3, 4, 5, and 26 days

   AMLOS = (2 + 3 + 4 + 5 + 26) ÷ 5        = 40 ÷ 5      = 8.0 days
   GMLOS = (2 × 3 × 4 × 5 × 26) ^ (1/5)    = 3,120^0.2   ≈ 5.0 days

One long stay pushed the arithmetic mean to 8.0 — a number that describes no patient in the set — while the geometric mean stayed at 5.0, next to the cluster where the typical case lives. That is why the payment rules key on the GMLOS: it is the honest "expected stay" for the ordinary case. The AMLOS is still published and still useful — utilization review leans on it — but when a rule in this section says "the mean," it means the geometric one.

Understand also what length of stay does not do: within the ordinary range, it does not change the payment. Discharge the COPD patient on day 3 or day 5 and DRG 190 pays \$6,829.30 either way — that is the prospective design working, and it is the pressure behind every utilization-review conversation about discharge planning. The exceptions are the two rules that follow.

The outlier: when the average would be absurd

For the rare stay whose costs run catastrophically past the payment — the sixty-day ICU course in a DRG priced for four days — IPPS pays an outlier payment: an additional amount that begins only after the hospital's estimated costs exceed the DRG payment plus a fixed-loss threshold set annually by rule, and then covers only a percentage of the costs beyond that point. Estimated costs are derived from the claim's charges through the hospital's cost-to-charge ratio — which is the promised answer to §33.1: this is where the charge lines on an inpatient claim touch payment, and why a neglected chargemaster distorts even DRG reimbursement. The structure to remember: the hospital always absorbs a substantial loss before the first outlier dollar arrives, and continues sharing costs after. Outlier policy trims catastrophe; it never makes a long stay profitable. Parameters change every year; verify them in the current rule.

The transfer rule: the arithmetic Chapter 26 promised

Chapter 26 §26.7 stated the rule from the claim's side — discharge status codes, FL 17, and a two-digit field that decides more money than any other small field on the form — and deferred the arithmetic here. The logic first:

A full DRG payment buys a complete stay. A hospital that transfers the patient to another acute care hospital partway through has not delivered a complete stay — the receiving hospital will be paid for finishing it, and paying both in full would pay for the admission roughly twice. So for a transfer (discharge status 02) with a length of stay at least one day short of the DRG's GMLOS, the transferring hospital is paid per diem: the full DRG payment divided by the GMLOS, doubled on the first day (front-loaded, because admission days genuinely cost more), for each day of the stay — capped at the full DRG amount.

🧮 Run the Numbers

Account 22-8891's counterfactual transfer [constructed — the weights and base rate are this book's teaching figures; the GMLOS is illustrative; verify all of them in the current IPPS tables]. Suppose the respiratory failure had not stabilized, and on day 2 Ridgeview transferred the patient to a tertiary center (discharge status 02). DRG 190; full payment \$6,829.30; assume GMLOS 4.0 days; length of stay 2 days — at least one day below the GMLOS, so the rule applies.

```text per diem 6,829.30 ÷ 4.0 = 1,707.325 → $1,707.33 per day day 1 (×2) 1,707.325 × 2 = $3,414.65 day 2 1,707.325 = $1,707.33 ───────── TRANSFER PAYMENT $5,121.98

versus the full DRG $6,829.30 the difference $1,707.32 stays with the system, funding the receiving hospital's claim ```

Checks: 3,414.65 + 1,707.33 = 5,121.98 ✓ · 6,829.30 − 5,121.98 = 1,707.32 ✓

And the cap, one line: had the stay run 3 days — 3,414.65 + 1,707.33 + 1,707.33 = 6,829.31 — the per-diem total would already exceed the full DRG by rounding, and the payment is the lesser: \$6,829.30. By the GMLOS, transfer and discharge pay the same, which is the rule's design: a hospital is never paid more for transferring than for finishing.

Now the parts a coder has to carry:

The rule reads FL 17, and only FL 17. The payer does not know a transfer happened except from the discharge status code — which is what makes Chapter 26's Case Study 1 this rule's cautionary tale. A unit clerk defaulting to 01 (discharged home) on actual transfers overstates every affected claim by exactly this section's arithmetic, claim after claim, with no denial, no signal, and no way for the person entering the code to see the consequence — until an auditor, who can compute a discharge-status distribution from claims data without opening a single chart, extrapolates the pattern into a repayment. Chapter 37 §37.6 owes you the extrapolation arithmetic; this section is the money it multiplies.

The post-acute version reaches further than the acute version. For a defined list of DRGs — revised annually — the same per-diem logic applies when the patient goes on, within the rule's conditions, to post-acute settings: a skilled nursing facility (status 03), home health beginning within the window (06), inpatient rehabilitation (62), and related dispositions. For some DRGs on the list the rule softens to a blend — roughly half the full payment plus per diems — recognizing front-loaded costs. The list, the qualifying statuses, and the blend formula live in the current rule and the Medicare Claims Processing Manual; what does not change is the operational lesson: whether the post-acute transfer policy applies to a claim is decided by a two-digit code entered far from the coding department, and somebody in the building has to be reconciling that field against the discharge summary. Chapter 26 §26.7 gave you the check; it costs a sample and a schedule.

Both directions are wrong. Status 01 on a transfer overpays; a transfer status on a genuine discharge underpays. Chapter 5 §5.8's symmetry, in a new field: accuracy is the obligation, and "conservative" miscoding is still miscoding.

33.9 ICD-10-PCS at a structural level

Inpatient hospital procedures are not coded in CPT. They are coded in ICD-10-PCS — the Procedure Coding System — a code set used only for hospital inpatient procedure reporting, maintained by CMS, updated every October 1 alongside ICD-10-CM. The physician performing the surgery still bills CPT on the professional claim; the hospital reports the same event in PCS on the institutional claim, in FL 74. One operation, two code sets, two claims — Chapter 16 §16.9's split again, now with different vocabularies on each side.

This book teaches PCS at the structural level: enough to read a code, understand what feeds §33.4's surgical partition, and know what you do not yet know. Facility coders train on PCS for months; a section is not that training and will not pretend to be.

The structure is unlike anything else in this book. A PCS code is not looked up so much as built: seven characters, each position a defined axis, each value chosen from a table.

THE ANATOMY OF AN ICD-10-PCS CODE (Medical and Surgical section)

    0    D    T    J    4    Z    Z
    │    │    │    │    │    │    │
    │    │    │    │    │    │    └── 7  QUALIFIER — additional attribute; Z = none
    │    │    │    │    │    └─────── 6  DEVICE — anything left behind; Z = none
    │    │    │    │    └──────────── 5  APPROACH — how the site was reached;
    │    │    │    │                     4 = percutaneous endoscopic
    │    │    │    └───────────────── 4  BODY PART — J = appendix
    │    │    └────────────────────── 3  ROOT OPERATION — the objective;
    │    │                               T = Resection (cutting out ALL of a body part)
    │    └─────────────────────────── 2  BODY SYSTEM — D = gastrointestinal
    └──────────────────────────────── 1  SECTION — 0 = Medical and Surgical

That is 0DTJ4ZZ — a laparoscopic appendectomy, the classic first example in PCS instruction [teaching example — verify in the current year's PCS tables before any real use]. Read it back: Medical and Surgical section, gastrointestinal system, Resection, appendix, percutaneous endoscopic approach, no device, no qualifier. Every character answers one question; no character is decorative. Two mechanical facts: values run 0–9 and A–Z but never the letters I or O (too close to 1 and 0), and there are no decimals — a PCS code is always exactly seven characters.

The root operation, which is the concept worth owning

Character 3 — the root operation — is the intellectual center of the system and the one term from PCS this chapter puts in your permanent vocabulary. The Medical and Surgical section defines 31 root operations, each with a precise definition, and the coder's job is to match the objective of the procedure as documented to the definition — not the procedure's name, not the specialty's jargon, the objective. A sampler, to show the precision:

The Excision/Resection line is the canonical example of why documentation reads differently here: a "lobectomy" is Resection of the lobe — because PCS defines each lung lobe as its own body part and all of it came out — while removing part of a lobe is Excision. The word the surgeon used does not decide; the anatomy and the objective do. PCS has its own Official Guidelines governing these selections, free from CMS, updated annually, and its own conventions — including a strict rule familiar in spirit from Chapter 8: the Index points you toward a table, but the code is always built from the table, never lifted from the Index.

Why a diagnosis-coding book teaches you this much and no more: because §33.4's surgical partition runs on PCS. Whether a stay groups medical or surgical — one of the largest single payment forks in the system — is decided by whether the record carries an OR-designated PCS code, and the designation hangs on exactly these characters. A wrong approach value or a Resection coded as an Excision is not a nuance; it can be a different DRG. When this chapter's Encounter checkpoint sends Account 10-4471's patient to the operating room, the procedure side of that story is a PCS story — told, deliberately, at the structural level only. Where this chapter needs a procedure it cannot verify, it describes the structure generically — the discipline this book has applied to every code set since Chapter 8's two-step rule: a category described honestly beats a code invented confidently.

🎓 Exam Watch

PCS lives on the facility exams — the CCS and CIC expect real fluency; the CPC does not test it. The reliably tested distinctions are the root-operation boundaries: Excision versus Resection (portion versus all of a body part — and remember the body-part definitions do the work, as with the lung lobe), Extirpation versus Excision (taking out foreign solid matter versus cutting out the body part itself), Insertion versus Replacement (does the device take the body part's place?). The other reliable stem gives you a procedure name from the operative report's header and documentation that contradicts it — and the credential is testing whether you code the objective documented, not the title. Same discipline as every chapter since 4: the note, not the label.

33.10 One word, \$1,867.44

Everything in this chapter now assembles into the single example this book has been pointing toward since Chapter 26 cited it by name.

Return to Account 22-8891's emergency department, day 0, around two in the morning. The patient is 71, known COPD, struggling to breathe. Oxygen saturation is low and documented; supplemental oxygen is started and titrated; the patient is admitted. Every clinical fact is real and recorded. The question is one sentence in the H&P, and there are two ways the physician might write it:

📋 Read the Chart

text FIGURE 33.3 — "Two sentences, one patient" [Account 22-8891 — the B version is constructed as the counterfactual] THE DOCUMENT The assessment line of the admitting H&P, in two versions. A — what this record actually says: "Acute exacerbation of COPD with acute hypoxemic respiratory failure; admit, oxygen, steroids, bronchodilators." B — the counterfactual: "COPD exacerbation, hypoxic on arrival; admit, oxygen, steroids, bronchodilators." THE CONTEXT The same patient, the same vitals, the same orders, the same four days. The clinical care under A and B is identical in every respect. WHAT IT SHOWS A documents a diagnosis: acute respiratory failure with hypoxia — codable, J96.01, an MCC, POA = Y. DRG 190; $6,829.30. WHAT IT DOESN'T B documents a finding. "Hypoxic" supports a symptom- level code at most, and no coder may infer respiratory failure from saturation values, oxygen requirements, or clinical gestalt — that inference is a clinical diagnosis, and it belongs to the physician (Ch. 4 §4.7). Under B the record supports no MCC: DRG 191; $4,961.86. THE DECISION Under A: code it. Under B: do not code what you cannot read — and do not let it end there. B is the exact occasion for a CDI query, because the clinical evidence for the unwritten diagnosis is sitting in the vitals and the orders. Chapter 38 §38.3 owns the query's compliant form. THE LESSON A − B = $1,867.44, and nothing clinical differs. The difference is one sentence, written or unwritten by an exhausted physician at 2 a.m. — which is why the CDI profession exists, and why it must work both directions.

Version B is not wrong. It is unwritten. This book built that distinction on Account 10-4471, where Chapter 14 §14.4 found that the office note never states the decision to inject was made that visit — strongly implied, clinically obvious, and absent as a sentence. The finding was not that the record was false; it was that the record was incomplete in a way that has consequences, and the appeal in Chapter 30 had to be constructed from surrounding evidence instead of quoted. Version B is the same species of gap wearing hospital clothes, and it is worth being precise about why the coder's hands are tied: the physician saw respiratory failure — the treatment proves it — but treatment is not documentation of a diagnosis, and a coder who assigns J96.01 from an oxygen flow rate has diagnosed the patient. Everything this book said in Chapter 4 lands here at maximum stakes: the coder does not know what the provider did or thought; the coder knows what the provider wrote.

The difference is \$1,867.44 — per occurrence. Now scale it the way a hospital must: respiratory failure phrasing, the specificity of heart failure (Chapter 11 §11.3's four axes), CKD staging (§11.8), the diabetes linkages (§9.7) — every one is a documented-versus-unwritten gap with a severity tier attached, across thousands of admissions a year. That aggregate is the business case for clinical documentation integrity, and it is why CDI reviews records concurrently — while the patient is still in the bed and the physician can still be asked — rather than after discharge, when Chapter 4 §4.5's amendment rules make every correction slower and weaker. Chapter 38 owns that whole apparatus: the concurrent review, the compliant query, and the leading query that ends careers. What this chapter owns is the payment consequence, and its honest statement cuts both ways:

The same number that measures the documentation gap measures the temptation. A query that asks the open question — "the record documents an oxygen requirement and saturation of X%; can the respiratory status be further specified?" with the full menu of answers including "no" — closes a real gap. A query engineered to harvest the MCC, sent only when the answer would raise payment, is a severity-capture program indistinguishable from the schemes Chapter 5 taught you to name; "DRG creep" has been the term of art since 1981, Case Study 2 shows what it did at national scale, and CC/MCC capture patterns are a standing subject of federal audit scrutiny. The defensible program queries in both directions — including the query that removes an unsupported MCC — and can show an auditor its query rate, its agreement rate, and its physician-response text without flinching. The accurate record and the defensible record are the same record. That has been this book's second theme since Chapter 1, and DRG 190 versus 191 is the cleanest proof it will ever get: both DRGs describe the same patient; only one describes the chart.

And notice, finally, that the same \$1,867.44 stands behind Chapter 26 §26.10's implausible facility claim — hypertension sequenced principal over a COPD exacerbation — because a sequencing error and a documentation gap cash out through the identical machinery: the DRG reads the record, and the record is only as good as its weakest deliberate act. One word, one sequence choice, one two-digit discharge status: in this payment system, small acts of accuracy are denominated in four figures.

🗂️ The Encounter — the same patient, admitted

Account 10-4471 does not become an inpatient file — the knee was injected in an office and the claim was paid, appealed and all, by day 100. This chapter's checkpoint is a lens: what if the same patient's knee had instead been a broken hip? Run the counterfactual carefully, because it converts everything this chapter taught into the vocabulary of a patient you already know. (The real file is unchanged; nothing below alters it.)

The setup (constructed): the same 58-year-old — diabetes, hypertension, hyperlipidemia, the CKD 3a problem-list entry — falls at home, fractures the right hip, is admitted through the ED, and undergoes surgical repair the next morning.

What this chapter contributes — the whole record, re-read under inpatient rules:

• Principal diagnosis: the fracture — a code from the S72.- category for the documented site, seventh character A (Chapter 12 §12.4's initial encounter, now on an inpatient claim), established after study and unambiguously the reason for admission. The external cause story rides along: W19.XXXA, place of occurrence Y92.009 — the Chapter 12 codes that had no place on the office claim now describing how an admission began.
• Secondary diagnoses, Section III applied: E11.9, I10, and E78.5 are reportable — an inpatient stay evaluates and manages all three (home medications continued, glucose monitored perioperatively). The CKD 3a problem-list entry faces the same test it faced in Chapter 9 §9.7: reportable only if this stay engages it — and, per the canon this book has kept since Chapter 4, no diabetes–CKD linkage may be assumed by anyone but Chapter 36. POA = Y across the board; the fracture existed before the admission order.
• The procedure: the repair is reported by the hospital in ICD-10-PCS — structurally, a Medical and Surgical section code whose root operation depends on what was actually done: a prosthetic replacement is Replacement; internal fixation of the fracture is a different root operation with its own table. Either way it carries OR designation, and that is the fact that matters here.
• The grouping: the PCS code flips §33.4's surgical partition; the stay grows toward the lower- extremity surgical families. With her secondaries — chronic, stable, and nothing approaching MCC gravity — the severity split lands without MCC: the shape of MS-DRG 470, major hip and knee joint replacement without MCC, if the repair is a replacement. One admission, one number, one payment — where the office version of this patient generated four lines, three code sets, and a modifier dispute.

What the lens settles: that you can now do this. Principal selection, Section III, POA, the partition, the split — the same skills, patient by patient, that §33.2 through §33.5 taught on Account 22-8891.

What it does not settle: the dollars. She is 58 and commercially insured through Northfield Mutual — MS-DRGs are Medicare's system, and whether her admission pays by DRG, per diem, or a negotiated case rate is a term of Northfield's hospital contract (Chapter 2 §2.6), not a fact a coder can derive. And DRG 470's current weight is a lookup this book deliberately does not fake: the classification is the coder's; the price is the contract's and the current year's table's. Code from the chart; get paid by the contract.

The open questions: all six of the file's questions stand where Parts IV and V left them — only Q4, whether the denial could have been prevented, remains open, and it belongs to Chapter 40. The counterfactual adds none, which is itself the point: an inpatient file is not a more mysterious object than the office claim you have balanced to the penny since Chapter 2. It is the same record, read by a different machine.

Summary

The inpatient hospital is paid for the classified stay, not the itemized services. Since 1983, IPPS has paid one predetermined amount per admission, set by the MS-DRG the coded record groups to. The UB-04 still carries every charge; the money reads the codes — which makes inpatient coding and inpatient pricing the same act.

The principal diagnosis is the condition established after study to be chiefly responsible for occasioning the admission — not the presenting symptom, not the scariest diagnosis, not anything that developed after admission. The hard cases have rules: two conditions equally responsible (either may lead, and the discretion is real but narrow), symptoms with contrasting diagnoses, plans not carried out, complications, admissions from outpatient surgery — and Chapter 9 §9.5's inpatient uncertain-diagnosis rule codes "probable" at discharge as established.

Secondary diagnoses are payment inputs. The CC/MCC lists, revised annually, tier each DRG family; the single highest-severity secondary sets the tier, and severity does not stack. On Account 22-8891, J96.01 — acute respiratory failure with hypoxia, POA = Y — is the MCC that makes the stay DRG 190.

The grouper runs one pipeline: pre-MDC check, principal diagnosis → MDC, the surgical partition (driven by OR-designated ICD-10-PCS codes), then the severity split. Payment = relative weight × base rate (labor share × wage index + non-labor): at Ridgeview's constructed \$6,200.00, the COPD family pays \$6,829.30 / \$4,961.86 / \$4,064.10 — a \$1,867.44 step for the MCC and \$897.76 for the CC. Every weight, list, and rate changes each October 1 in the IPPS final rule; code and price only from current tables.

The case mix index is the average weight of discharges — revenue per case at a fixed base rate — and it cannot distinguish sicker patients from better documentation. Use it as a diagnostic, never as a target.

POA indicators report clinical chronology — Y, N, U (resolved against the hospital), W (respected), exempt — and they arm the HAC provision: a HAC-list condition with POA = N cannot serve as a CC/MCC. On 22-8891 the day-2 stage 3 pressure ulcer (L89.153, POA = N) changes nothing because J96.01 already holds the tier; without the respiratory failure, the ulcer could not have rescued it.

Length of stay does not change the payment — except at the edges. Outlier payments begin only past a fixed-loss threshold and share costs thereafter. The transfer rule pays a transferring hospital per diem — full DRG ÷ GMLOS, doubled on day 1, capped at the full DRG — when the stay runs at least a day short of the geometric mean; the post-acute version extends the logic to a defined DRG list. The rule reads FL 17, which is why Chapter 26's discharge-status discipline is a payment control.

ICD-10-PCS builds seven-character codes from tables — section, body system, root operation, body part, approach, device, qualifier — and the root operation (Excision, Resection, Replacement, Extirpation…) encodes the procedure's objective as documented, not its name. PCS feeds the surgical partition; this book teaches its structure and sends you to the current tables for everything else.

And one word is worth \$1,867.44. "Acute respiratory failure with hypoxia" is a diagnosis and an MCC; "hypoxic" is a finding and nothing. The record that says the second when the first was true is not wrong — it is unwritten, and the compliant answer is the CDI query (Chapter 38 §38.3), asked in both directions, because the accurate record and the defensible record are the same record.

Key Terms

Inpatient Prospective Payment System (IPPS) — Medicare's payment system for acute care hospital inpatient stays: one predetermined payment per discharge, set by classification, effective since October 1, 1983. (Ch.33)

MS-DRG (Medicare Severity Diagnosis-Related Group) — the classification unit of IPPS: a group of clinically coherent, similarly resourced stays, severity-split by CC/MCC, to which the payment weight attaches. (Ch.33)

Principal diagnosis (inpatient) — the condition established after study to be chiefly responsible for occasioning the admission; defined by the UHDDS, governed by Guidelines Section II, reported in FL 67. (Ch.33)

CC (complication or comorbidity) — a secondary diagnosis designated by CMS as meaningfully increasing a stay's resource use; supports the middle severity tier. (Ch.33)

MCC (major complication or comorbidity) — the higher designation; a single MCC among the secondaries places the stay in the "with MCC" tier. (Ch.33)

Major diagnostic category (MDC) — one of 25 broad, mostly body-system groupings into which the principal diagnosis maps a stay before DRG selection. (Ch.33)

Grouper logic — the sequence pre-MDC → MDC → surgical partition → severity split by which software assigns the MS-DRG from the coded record. (Ch.33)

Relative weight — the annually recalibrated factor expressing a DRG's average resource intensity against the all-DRG average; multiplied by the base rate to price the stay. (Ch.33)

Base rate — the hospital-specific value of one average-weight case: a national standardized amount split into a labor share (adjusted by the wage index) and a non-labor share, plus policy adjustments. (Ch.33)

Case mix index (CMI) — the average relative weight of a hospital's discharges; revenue per case at a fixed base rate; cannot distinguish patient acuity from documentation capture. (Ch.33)

POA indicator — the present-on-admission character (Y, N, U, W, exempt) reported with each inpatient diagnosis; U resolves against the hospital, W is honored. (Ch.33)

Hospital-acquired condition (HAC) — a condition on the payment-provision list that, when reported with POA = N or U, cannot serve as a CC/MCC; distinct from the ACA's HAC Reduction Program. (Ch.33)

Geometric mean length of stay (GMLOS) — the multiplicative average stay for a DRG, resistant to outliers; the length-of-stay statistic the payment rules key on, versus the arithmetic mean (AMLOS). (Ch.33)

Outlier payment — the additional payment for extraordinarily costly stays, beginning only past a fixed-loss threshold and sharing (not covering) costs beyond it. (Ch.33)

Transfer DRG rule — per-diem payment (full DRG ÷ GMLOS, day 1 doubled, capped at the full DRG) to a hospital transferring a patient with a stay at least one day below the GMLOS; extended to post-acute settings for a defined DRG list. (Ch.33)

ICD-10-PCS — the seven-character, table-built code set for hospital inpatient procedures; updated every October 1; feeds the surgical partition. (Ch.33)

Root operation — the third PCS character: the precisely defined objective of the procedure (Excision, Resection, Replacement, Extirpation…), coded from what was documented rather than what the procedure is called. (Ch.33)

Spaced Review

1. Ridgeview's base rate is \$6,200.00 and a DRG's relative weight is 0.8003. Compute the payment. The coder then validates a documented MCC that was initially missed, moving the stay to a weight of 1.1015. Compute the new payment and the difference — and state what document had to contain what kind of sentence for that move to be legitimate.

2. A condition is first documented on day 3 of a stay. Can it be the principal diagnosis? Can it be an MCC for this stay's payment if it is on the HAC list? If it is not on the HAC list? Name the indicator that carries the answer.

3. (Chapter 9) A discharge summary reads "probable viral meningitis." Outpatient, the same phrase appears in an office note. What is coded in each setting, and which sections of the Official Guidelines govern each answer?

4. (Chapter 11) State the respiratory-failure sequencing question in one sentence, and explain how the same two codes — J44.1 and J96.01 — can legitimately produce different principal diagnoses on different records. What, in this chapter's terms, is riding on it?

5. (Chapter 32) The hospital's patient-facing obligations did not pause for this chapter's payment machinery: what does a good faith estimate owe a scheduled inpatient admission's patient, and why is "the DRG decides what Medicare pays" not an answer to "what will I owe?"

Next: Chapter 34. The same building, one floor down and paid by an entirely different machine — the hospital outpatient department, where the unit of payment is neither the service nor the stay but the ambulatory payment classification, and where the \$318.00 supply line from Chapter 1's emergency department bill finally gets its explanation.