Codes come from the chart: the authorization you must not get

3 min read

There is a version of this job where the instruction is "get the authorization, whatever it takes": nudge the CPT to a code the payer approves, pick the diagnosis that fits the policy instead of the chart, add the modifier that worked last time. People are put under real pressure to work this way. It does not work, and this article is about why - mechanically, not morally.

#Why a forced authorization collapses

An authorization binds a specific payer decision to specific facts: this member, this diagnosis, this procedure code, this place of service, this window. At the claim, the payer compares what was billed - and the records behind it - against what was authorized. The moment the billed reality and the authorized story diverge, the authorization stops protecting the claim.

• A CPT changed to get approval, then billed as what was actually done: the codes do not match, and the service performed is unauthorized. The denial reads no authorization for billed service, and it is correct.
• A diagnosis chosen for the policy rather than the chart: the clinical record submitted with the claim, or pulled in an audit, does not support the indication. The approval was obtained on facts that were never true, which payers name invalid or incorrectly obtained authorization.
• A modifier added to make it go through: a modifier asserts something about the service, and the documentation must support that assertion. A modifier the notes do not back is the same divergence in a smaller container.

The pattern version is worse than the case version. Payers profile submission patterns, and a practice whose approvals repeatedly rest on facts its own records do not support has converted an administrative problem into an audit problem that touches every claim.

#Where each code actually comes from

Diagnosis codes

From the clinical documentation of the encounter: what the physician recorded about this patient. Authorization staff transcribe them; they do not choose them.

Procedure codes

From what is planned, and later from what was performed. When the plan changes intraoperatively, that is a known workflow with its own path; see When the service changes.

Modifiers

From the documentation, usually at the coding team's direction. "It worked on the last case" is a description of the last case's chart, not of this one's.

#What to do instead when the criteria are not met

1. Document what is true, better
   Most requests that look short are missing evidence that exists: the duration of conservative therapy, the failed alternative, the objective score. See Clinical documentation that survives review.
2. Read the policy and argue inside it
   If the patient genuinely meets criteria, show it against the policy's own elements; see Medical necessity criteria.
3. Escalate to the clinical conversation
   If the patient does not meet the standard and the physician believes the service is still right, that argument belongs in a peer-to-peer, made honestly, not in a code field.
4. Price the honest uncertainty
   Where coverage is genuinely ambiguous, a predetermination puts the payer's expectation in writing before anyone commits.