After CAC Go-Live: Copy-Forward, an Outdated Grouper, and a Problem List That Codes Itself
Three months after a computer-assisted coding (CAC) go-live at a 300-bed hospital, the tool's suggestion acceptance rate is 94%. An internal audit of 30 inpatient records finds: 6 records in which an active diagnosis appears only in text copied forward from a prior admission; 3 records grouped with last year's DRG grouper because the encoder update was deferred; 2 records in which a problem-list entry of 'history of COPD' was suggested by CAC and accepted as active COPD. The HIM director wants a corrective action plan before the next billing cycle and asks how CAC should be used going forward.
1. For each of the three findings, state the coding/integrity impact and the immediate action required.
Read the worked answer
1. Copy-forward (6 records). Impact: a diagnosis supported only by text duplicated from an earlier admission is not supported for this encounter; where the text crossed into a different encounter, it may also be a privacy issue. Action: re-review each record against the current encounter's assessments and plan; remove unsupported codes; flag the note to its author; correct any claim not yet submitted. Where already paid, evaluate corrected claims.
2. Deferred grouper update (3 records). Impact: the grouper determines the MS-DRG, so an outdated version can produce an incorrect DRG and incorrect payment in either direction - this is a reimbursement and compliance issue, not a cosmetic one. Action: install and validate the current grouper immediately (updates follow the federal schedules, including the October 1 and April 1 cycles), re-run the affected records, identify every claim processed since the effective date, and correct them.
3. 'History of COPD' accepted as active (2 records). Impact: the problem list is a workflow tool, not a source document for code assignment; a historical entry does not become an active condition because software surfaced it. Action: remove the code; if the provider's current documentation is ambiguous, query rather than infer.
2. Who is accountable for the final code set, and how should that change how CAC and coder productivity are managed?
Read the worked answer
The coder remains accountable for every code that reaches the claim. CAC produces suggestions from text pattern matching; it does not read the record, weigh clinical indicators, apply sequencing rules, or detect copy-forward. A 94% acceptance rate is a statistic about a tool, not evidence about accuracy - and in this audit, the errors are concentrated exactly where a human review should have intervened.
Practically, that means:
- Every CAC-suggested diagnosis (not a sampled subset) is validated against the provider's documentation for the current encounter before the claim is released.
- Acceptance rate must never be a productivity target or an individual performance measure. If coders are measured on accepting suggestions, the incentive runs directly against the validation the tool depends on, and 94% stops being reassuring.
- Productivity standards must budget review time for record review, not keystrokes; a coder clearing more charts by trusting the tool is a risk, not a top performer.
- The query process remains the coder's tool for genuine ambiguity; CAC output is never a substitute for a query, and a coder cannot let software manufacture a diagnosis the physician did not document.
3. Design the corrective action plan: what controls and metrics do you put in place before the next billing cycle?
Read the worked answer
Before the next billing cycle (immediate):
- Suspend release of the 11 affected records; re-review, recode, and correct.
- Validate encoder and grouper version against the current federal update; document the version in use.
- Re-run all claims processed since the grouper effective date; correct claims within timely filing.
Standing controls: 4. Pre-bill review of all records with a CAC suggestion that changes the principal diagnosis, adds a CC/MCC, or is supported only by text matching in a section with copied content. 5. Focused audits - at minimum monthly for the first two quarters, then quarterly - with a defined error-rate threshold that triggers re-education rather than only rework. 6. Automated copy/paste and copy-forward detection where the EHR supports it, plus physician education on documentation hygiene and on not carrying forward prior assessments. 7. Problem-list governance - a rule that problem-list entries are never coded without current documentation, and a periodic reconciliation of the list. 8. Version control check for grouper, encoder, and chargemaster on a scheduled calendar keyed to the federal update dates. 9. Metrics that pair volume with accuracy: error rates by coder, by service line, and by CAC-suggested code type; acceptance rate reported only alongside audit accuracy. 10. Audit trail review - confirm the EHR log captures who changed what and when, so that post-submission amendments can be traced and corrected claims justified.