Every release and every piece of in-flight work for Medical Auto Coding. Filter by status, kind, or tag, then read the full notes below.
| Status | Kind | What changed | Version | Ticket |
|---|---|---|---|---|
| Fix merged | Improvement | Excludes1 diagnosis pairs are resolved by the book on anesthesia, surgery, E/M and radiology claims, plus obstetric modifier and validation refinements | No version | No ticket |
| Fix merged | Improvement | Supervised, one-to-one directed and relief anesthesia claims keep every provider line | No version | No ticket |
| Fix merged | Improvement | The second validation check on an anesthesia claim keeps the diagnoses that the diagnosis check decided | No version | No ticket |
| Fix merged | Improvement | Automated validation alerts now say why, and ten new alerts carry the questions validation could not answer on its own | No version | No ticket |
Work that is planned, in progress, or merged and waiting for a production deploy. It is not released yet.
Diagnosis pairs the classification says cannot be coded together are now resolved by the book, and you see a note instead of a question.
An Excludes1 note in the ICD-10-CM Tabular List says that two conditions are not coded together. For example, the note at K83.1 (obstruction of bile duct) excludes "obstruction of bile duct with cholelithiasis (K80.-)". The autocoder now reads these notes for the fiscal year of the date of service. It gives the diagnosis check each conflicting pair, the note, and any combination code that the classification offers. Then it verifies the result on the finished claim. This applies to anesthesia, surgery, E/M and radiology claims.
Each pair ends in one of these outcomes:
Automated claim validation (ClaimCleaner) now runs two times on an anesthesia claim. The second run checks the finished claim, after the diagnosis check and the Excludes1 resolution. So a combination code that the autocoder applies is checked before the claim reaches you. ClaimCleaner questions come from the second run only, so a question never names a diagnosis that has already left the claim. If the first run fails, the claim carries one failure alert, not two.
See the Anesthesia Alert Reference and Diagnosis Coding for the triggers and what to verify.
Some anesthesia claims bill one anesthesia code on two lines on purpose, one line for each provider:
Automated claim validation (ClaimCleaner) can read these two lines as a duplicate. The second validation check on an anesthesia claim now keeps both lines. Every provider line stays billed, and the claim records the removal that the autocoder did not apply. The kept line also receives the diagnosis and modifier corrections that validation made on the other line of the same anesthesia code. The payment modifiers stay as they are, because the two lines differ there on purpose.
For example: an anesthesiologist supervises five concurrent rooms, and a CRNA gives the anesthetic in one of them. The CRNA line with QX stays on the claim beside the anesthesiologist line with AD.
Automated claim validation (ClaimCleaner) checks an anesthesia claim two times. The second check runs on the finished claim, after the claim-wide diagnosis check. The second check now keeps every diagnosis that the diagnosis check decided from the record.
For example: the record documents a left knee replacement, and the diagnosis check keeps Z96.652 (presence of left artificial knee joint). The second check no longer swaps that code for the right side.
Every finished claim runs through automated claim validation before it reaches you. Validation checks NCCI procedure pairs, add-on code rules, Medicare payment rules, coverage articles, code validity for the date of service, and diagnosis conventions. This release changes what you see from it in two ways.
Validation alerts now name the rule. Before this release, an alert read "Modifier 59 added by ClaimCleaner" and stopped there. You could see the change but not the rule behind it, so confirming the change meant looking the rule up yourself. Each of these alerts now ends with the rule, in the form "Reason: [the rule's own explanation] [the rule id]". Read that sentence first. It names the payment rule or coding convention that produced the change. The alerts themselves are unchanged: same names, same types, same severities.
The alerts that gained the reason are Anesthesia Code Changed, Diagnosis Added, Modifier Added To An Anesthesia Code, Modifier Added To A Surgical Code, Line Left With No Code, and Line Removed By Validation.
Ten new alerts carry the questions validation could not answer. Validation also reports problems it will not correct on its own, because the answer needs the record, not a rule. Until now none of those reports reached a coder. Ten new alerts carry the ones worth your time:
Each alert reads "ClaimCleaner:" then the rule's explanation and its rule id. The type and the severity come from the rule. Every rule in this set classifies as Compliance today. The alert carries the rule's own classification, so a future validation rule could arrive as Financial. One exception: the units alert for a single line over an absolute limit is Compliance, severity 5, set by the autocoder, because validation marks that finding Informational after reducing the units in its own copy of the claim.
What to look out for. Expect a small number of new alerts, not a flood. An alert appears only when five conditions are true.
The rule is on the autocoder's review list. The rule's own classification is not Informational. Validation did not fix the problem itself. The line is still billable on the finished claim. The diagnosis problem that the finding names is still on the finished claim.
For an Excludes1 pair, the alert stays while the diagnosis that carries the Excludes1 note, or another code under the same note, is still billed together with a diagnosis from the note's family. When the autocoder cannot read the note from the finding, it checks the two named diagnoses instead.
The autocoder removes an alert again when it later moves the line out of billing, or when the diagnosis problem that the finding names is no longer on the finished claim. If the claim-wide diagnosis check replaces the diagnosis that carries the note with another code under the same note, or replaces the other diagnosis with another code from the note's family, the alert stays, because the conflict is still on the claim. The alert then also names the pair that the claim bills now. When the diagnosis check leaves two findings on the same pair of billed diagnoses, or two conflicts land on one line, you get one alert that names each pair. Conflicts on different lines raise one alert each. For the other diagnosis findings, and for an Excludes1 finding whose note the autocoder cannot read, a replacement or removal of a named diagnosis removes the alert. A finding about a diagnosis that is missing, for example no diagnosis for the side billed or no diagnosis that a coverage article covers, stays, because the autocoder cannot tell whether a diagnosis added later qualifies.
Start with the rule text in the alert, then check the record for the fact the rule needs: the side of service, the patient demographics, the covered diagnosis, the distinct service, or the units of service.
If a family of these questions does not fit your workflow, your organization can silence it or route it to a different reviewer role. Ask your Hank representative. A silenced alert still appears as an informational note.
See the Anesthesia Alert Reference for the full list, the trigger for each alert, and what to verify.