Chapter 38 — Exercises

How to use these. Section D is the one that matters most and the one certification exams test hardest: write the queries out in full, on paper, and check each draft against §38.3's seven elements before you look at anything else. Section F is the engine-output work, and it is best done the way you would do it on the job — cover the "comment" column, decide, and only then compare. Items marked † have worked solutions in the answers appendix. No answers appear in this file.

All records, engine outputs, dashboards, and evaluation figures below are constructed teaching examples. No vendor, product, benchmark, or performance figure is asserted. Verify every code in the current year's code set: ICD-10-CM changes October 1, CPT January 1, HCPCS Level II and the NCCI edits quarterly.

Section A — What CDI is and what it is not (items 1–4)

A.1 Define clinical documentation integrity in one sentence, then name the three clauses of your definition that each exclude something the discipline is accused of being.

A.2 † A hospital's chief financial officer describes the CDI program as "a revenue program that pays for itself." Give the two-part response: what is accurate in that description, and what specific operational failure it will eventually produce. Name the section of this chapter that owns the failure.

A.3 Distinguish CDI from auditing (Chapter 37) in one sentence. Then state the single structural difference that makes them different instruments rather than the same instrument at different times.

A.4 A CDI director reports that in two years the program has never issued a query that resulted in a diagnosis being removed. Give three explanations that do not require anyone to be acting in bad faith, and say which one a reviewer computing the program's claims data will assume.

Section B — Copy-forward and documented negatives (items 5–9)

B.5 † Write the copy-forward policy. In no more than six lines, state which sections of a progress note may carry forward and which may not, and give the one-sentence principle that decides the list.

B.6 Chapter 15 §15.12 called copied-forward text "the single most damaging thing in a modern medical record." Explain, in three sentences, why a clinician should care about that independently of any billing consequence, and why leading the conversation with the clinical version works better.

B.7 † Design the measurement. Describe, in five steps, how you would measure copy-forward across a five-physician practice without opening a single chart for review purposes — what you compare, against what, and what the output looks like. Name the chapter and section that calls this class of analysis by its name.

B.8 The March 14 procedure note (Chapter 4 §4.10) contains "no aspirate obtained" and "no imaging guidance used." For each clause, state (a) the coding question it closes, (b) the specific code it excludes or confirms, and (c) what a coder would have to do if the clause were absent.

B.9 † A practice manager proposes adding "no imaging guidance used" as a default value in the joint-injection template, "since we almost never use ultrasound." Give the two-sentence refusal, name the mechanism this book has watched produce a dozen failures, and state where the proposed template would have to be recorded if it were built anyway.

Section C — The concurrent review (items 10–13)

C.10 Define concurrent review, working DRG, and the reconciliation between a working DRG and a final coded DRG. State which of the three is never a bill and why that matters.

C.11 Give the four things concurrency buys that a retrospective review cannot, and name the chapter and section that establishes each one.

C.12 † Build a concurrent-review worklist. Name four selection criteria, state what each is good at finding, and identify which one exists specifically to keep the other three honest. Then state what happens to the program if that one is dropped for capacity reasons.

C.13 A practice wants to move CDI into the professional-services setting. Distinguish the pre-visit review from the post-visit, pre-bill review: what each is for, which rung of Chapter 37 §37.10's durability ladder each occupies, and where the hard boundary sits between surfacing a condition and harvesting one.

Section D — The compliant query (items 14–20)

D.14 List the seven elements of a compliant query from §38.3. For each, name the specific failure that occurs when it is missing.

D.15 † Rewrite the leading query. Here is a query as drafted [constructed]:

Dr. —, the patient's albumin has been low all week and she's had a dietitian consult. Please document severe protein-calorie malnutrition in your progress note so this admission groups correctly. Let me know if you have any questions — thanks!

(a) Identify every one of §38.3's five markers of a leading query that this draft trips, quoting the words that trip each. (b) Rewrite it as a compliant query, in full, including an option set. (c) State what should happen if the original has already been sent.

D.16 A coder sends a query; the physician answers "no change, the assessment as written is what I intend." Two days later the coder sends a second query on the same encounter, worded differently. State whether the second query is compliant, and explain your answer without reference to its wording.

D.17 † Chapter 21 §21.8 established that a query cannot ask a surgeon to remember something eighteen months later. State the general rule this chapter derives from that limit, then apply it: for each of the following, say whether a retrospective query is available and why. (a) An operative report documents "debridement performed" with no anatomic region stated. (b) A discharge summary documents a transfusion, an estimated blood loss, and an assessment reading only "blood loss." (c) A record contains no mention of a condition, and the physician recalls treating it.

D.18 Chapter 22 §22.6 found six weeks of conservative therapy documented in the history of present illness rather than in the assessment. (a) Explain why a coder may not move it. (b) Write the query that is available. (c) Then make the argument — in two sentences — that the query is the weaker of the two available fixes, and name the stronger one.

D.19 † The retention question. State the two defensible models for where a query lives after it is answered, and then give the three rules that hold under either model. For each rule, name what goes wrong when an organization violates it.

D.20 Construct a query option set for a genuinely ambiguous respiratory record that contains five options and is nonetheless leading. Then fix it with a single addition, and state the test that detects this failure in any option list.

Section E — Metrics (items 21–25)

E.21 Name the two CDI metrics that corrupt a program when they become targets, and state in one sentence each how the target is met without anyone improving anything.

E.22 † The selection argument. Explain, in a paragraph a director would accept, how a CDI program can become a leading program without sending a single non-compliant query. Name the chapter and section whose grouper "what-if" discussion predicted this, and state the sentence that draws the line.

E.23 Distinguish these three activities and say which are defensible: (a) prioritizing which records get reviewed by financial materiality; (b) deciding whether to send a query you have already concluded is clinically warranted, based on what the answer would be worth; (c) deciding whether to send it based on the likely direction of the answer.

E.24 † Using the two constructed programs in §38.4's 🧮, recompute every rate shown and verify that the response categories sum to the responses in both programs. Then write the paragraph a compliance officer would write recommending Program B's design, using only numbers from the table.

E.25 An auditor asks a coding manager, "show me the queries behind the shift." List, field by field, exactly what a defensible program produces in response — and state what the inability to produce it says about the case mix index that prompted the question.

Section F — The engine: computer-assisted coding and NLP (items 26–31)

F.26 Name the seven steps of the CAC pipeline in order, and state which step produces the output this chapter calls the most valuable and least discussed.

F.27 † Audit the engine's output. Here is an engine's suggestion panel for a constructed inpatient record [constructed]:

  CODE     CONF   SOURCE TEXT MATCHED
  -------  ----   -------------------------------------------------
  J18.9    0.88   "chest x-ray concerning for right lower lobe
                   infiltrate; treat empirically for pneumonia"
  I21.9    0.72   "father died of an MI at 58"
  R06.02   0.91   "shortness of breath on admission"
  J44.1    0.94   "known COPD, acute exacerbation"
  A41.9    0.55   "sepsis ruled out; blood cultures negative x2"
  E11.9    0.90   "type 2 diabetes"

For each suggestion: accept, reject, or query — and name which of §38.6's four qualifiers (negation, experiencer, temporality, certainty) is at issue where one is. Then state which two suggestions together would produce a claim that is internally inconsistent, and why.

F.28 Explain why an engine is more section-sensitive than a human reader, not less — and state what that does to Chapter 22 §22.6's finding about the right fact in the wrong section.

F.29 † Complete the claim. Using Account 10-4471's four service lines and four diagnoses, write out the diagnosis pointers as they belong on the claim. Then state what an engine that assigns pointers by first-listed default would produce, name the specific clinical assertion that error makes, and cite the chapter and section that established the correct pointer.

F.30 "An engine cannot flag the absence of a sentence nobody wrote." State the structural reason, then give the honest qualification: what can software surface, how it does it, and why that capability carries §38.4's problem into a rule set.

F.31 Your CDI worklist is generated by twelve clinical-indicator rules. Write the one question to ask about the set as a whole, describe what a good answer looks like, and propose two rules that fire in the direction most rule sets never do.

Section G — Autonomy, the loop, and measurement (items 32–35)

G.32 † Apply the four-property autonomy test from §38.7 to each of the following, property by property, and give a verdict: (a) screening mammography reporting; (b) established-patient office visit leveling; (c) shoulder arthroscopy coding from an operative report; (d) venipuncture and routine laboratory panels. For the one that fails on property 4 in a way the others do not, explain what that failure teaches a system over time.

G.33 Work the denial. A practice implements a scrubber rule that appends modifier 25 whenever an office visit and a minor procedure appear on the same date. Denials fall. Eighteen months later a payer's chart review requests forty encounters. (a) Name the case study in this book this repeats and the one difference in this version. (b) State what the practice can and cannot fix now. (c) Name the register field that would have caught it and what the evidence test would have consisted of.

G.34 † Audit this claim population. Design the quality program for a coder-in-the-loop workflow: name the population you sample from, the standard you score against, how you report the result, and the one thing you must not score against. Then explain why sampling the coder's rejections measures the wrong thing.

G.35 Name the three design decisions that determine whether a review loop is real, state who in a typical organization makes each one, and identify which of the three is a staffing decision wearing a technology decision's clothes.

Section H — Judgment, the exam, and the Encounter (items 36–40)

H.36 † Using §38.9's constructed evaluation, compute precision and recall from the raw counts, then compute post-review accuracy and post-review recall after the loop. Verify all four sums. Then answer in two sentences: which single number in the table carries a False Claims Act exposure, and why is it the smallest one?

H.37 The cost asymmetry, stated for a physician who thinks coding low is the safe choice. Write the two-paragraph explanation: what a missed code costs, what an unsupported code costs, and why the two are not symmetric — citing the chapter and section that settled the downcoding question for this book.

H.38 The ethics dilemma. A vendor proposes a CDI module with a rule set that surfaces "records grouping without a major complication or comorbidity where clinical indicators for one are present," priced on a share of the incremental payment the queries produce. Identify what is legitimate in the proposal, identify every feature that is not, and rewrite the engagement in one paragraph as something a compliance officer could sign. Name the two features of your rewrite that do the actual work.

H.39 Certification-style. Which of the following queries is compliant? (A) "The record documents an oxygen saturation of 84% on room air and 4 L by nasal cannula. Can you document acute respiratory failure so the admission groups correctly?" · (B) "The record documents an oxygen saturation of 84% on room air and 4 L by nasal cannula titrated overnight; the assessment reads 'COPD exacerbation, hypoxic on arrival.' Based on your clinical judgment, can the respiratory status be further specified? Options: acute respiratory failure with hypoxia / hypoxemia without respiratory failure / assessment as written is intended / clinically undetermined / other." · (C) "Please confirm the patient had acute respiratory failure." · (D) "Two similar admissions last month were documented as acute respiratory failure. Would you like to document it here as well?" Choose, and say in one sentence each why the other three fail and which of the five markers each one trips.

H.40 † The Encounter extension. Take §38.5's engine output on Account 10-4471's March 14 note and write the coder's disposition memo: for each of the eight suggestions, accept or reject with a one-line reason; then list the four findings the engine could not produce, state for each whether the limitation is structural or a configuration failure, and name the chapter and section that owns each finding. Close with the two sentences you would say to a manager who asks whether the practice still needs a coder on this chart.