Can a Medical Coding Backlog Delay Reimbursement Even When Coding Is Accurate?

Amanda McGurk

13 min read · November 19, 2026

Photograph of a man looking at a long shelf packed with finished unmarked folders in a quiet office, representing completed charts that have not yet been billed

Yes. An encounter cannot reach a payer until it is coded, so unfinished medical coding delays the earliest possible submission date. Accuracy reports and denial reports only cover claims already submitted, which means a growing backlog of correctly coded charts stays invisible in both.

HFMA’s MAP Keys divide the pre-billing holdups into four measures. None of them isolates medical coding. PB-1 is where coding holds land, mixed in with every other reason an account has not been billed.

MeasureMAP KeyWhat it means in practice
Days in Total Discharged Not Final Billed (DNFB)PB-1Gross dollars for accounts discharged but not yet final billed, for any reason. Medical coding is one contributor among several
Days in Final Billed Not Submitted to Payer (FBNS)PB-2Gross dollars for claims that are final billed but held by edits in the claims processing tool
Days in Total Discharged Not Submitted to Payer (DNSP)PB-3PB-1 and PB-2 combined, as a trending indicator of claim generation and submission
Total Charge Lag DaysPB-4Time between date of service and charge posting

All four definitions come from HFMA’s MAP Keys , which describes PB-1 as a month-end snapshot in gross dollars and a "trending indicator of claims generation process."

A note on vocabulary. DNFB and "discharged" are hospital revenue cycle terms, and these four measures are built around an inpatient discharge event. If you run an ASC or a physician group, the equivalent question is the same one in different words: how many encounters are unbilled, and what is your days-to-bill. Read "discharged not final billed" below as "completed but not yet billable," and the rest of this article applies to your setting without change.

What a medical coding backlog is

A medical coding backlog is the population of encounters that have had their documentation completed but have not yet had diagnosis and procedure codes assigned. It is measured in charts and in days of age, not in dollars, because the dollar value depends on what each chart is expected to collect rather than what it was charged.

That distinction matters for everything below. A backlog counted in gross charges will always look larger than the cash it represents.

Why the cash gap shows up with no coding errors to point at

Here is the situation that sends a revenue cycle leader looking for an explanation.

Cash is behind forecast. The denial rate is flat or improving. The medical coding audit scores came back strong. Nobody can name a coding error that caused the variance, because there is no coding error to name.

The encounters are sitting at step two of a six-step sequence:

  1. The encounter happens and documentation is completed and signed.
  2. Medical coders assign diagnosis and procedure codes.
  3. The claim is built and runs through billing edits.
  4. The claim goes to the payer.
  5. The payer adjudicates.
  6. Payment posts, or a denial comes back.

An encounter held at step two has not reached steps three through six. The payer has never seen it, so it generates no denial, no adjustment, and no remittance. It also generates no entry in the reports a coding manager reviews each week.

This is the gap that makes the problem hard to see. Accuracy describes claims that were submitted. Denials describe claims that were adjudicated. Both metrics are calculated on a denominator that excludes the charts causing the cash variance. A team running at 98% accuracy and nine days behind has nine days of correct, unbilled work, and the accuracy figure will not flag any of it.

Revenue cycle teams tend to describe it in a sentence like "we’re spending more, but the queue isn’t moving," or, in an ASC, "the surgery is done, so why isn’t the case ready to bill?"

What the measures actually say

Two published sources set the boundaries here, and neither says what teams often assume it says.

HFMA on pre-billing. The MAP Keys define PB-1 as gross dollars for accounts "discharged but not yet final billed for the reporting month." The definition carries no hold reason. Unsigned notes, open queries, late charge posting, registration corrections, and late-arriving ancillary results all land in the same number that medical coding delays land in. HFMA keeps claims held by processing-tool edits in a separate measure, PB-2, and excludes each population from the other so the two do not overlap.

CMS on filing deadlines. For Medicare fee-for-service, CMS states that you must submit claims to the correct MAC "no later than [1 CY] after the date of service," and that claims arriving later are denied, "with a few exceptions" that the page does not enumerate. The same page notes that a late-filing denial is not an initial determination and cannot be appealed.

Do not generalize the Medicare rule to every payer. Commercial and Medicaid managed care filing windows are set by contract and by state rule, and many are shorter than a year. Build the aging report against the deadline that applies to each account.

The distinction readers get wrong

Is DNFB the same as a medical coding backlog? No. DNFB is a bucket; a medical coding backlog is one thing in the bucket. PB-1 counts every account that has not been final billed regardless of why, so coding-held charts, documentation-held charts, and charge-capture delays all contribute to the same figure.

Two consequences follow, and both cut against intuition.

A rising DNFB number does not prove you have a medical coding problem. The movement may be entirely documentation or charge posting.

A flat DNFB number does not prove you lack one. Growth in coding-held charts can hide behind improvement somewhere else in the same bucket, and the net stays level while your coding queue ages underneath it.

To attribute DNFB movement to medical coding, you need the hold reason recorded on each account. Without it, the measure tells you something is wrong upstream of billing and nothing about what.

Is coding accuracy the same as coding timeliness? No. Accuracy asks whether the codes on a submitted claim were correct. Timeliness asks how long the encounter waited before anyone assigned them. A team can score well on the first and poorly on the second in the same month, and the reporting most teams run will surface only the first.

Practical guidance

Sort the held charts before you price them

Group the coding-held population into four buckets. They have different owners and different fixes.

  • Ready to code. Documentation is complete and signed. Nothing blocks a medical coder except available coding capacity. This bucket responds to capacity and queue design.
  • Waiting on documentation. A note is unsigned, a query is open, or a required report has not posted. Adding medical coders does not move these. They belong to the provider and CDI workflows.
  • Needs specialist judgment. Complex operative cases, unusual payer rules, specialty coverage gaps. These wait on a specific person.
  • Coded, held downstream. Medical coding is finished and a billing edit or another check is holding the claim. These have already left the coding queue.

Only the first bucket is a coding capacity problem. Pricing the whole population as one makes the number large and the diagnosis useless. Our guide to reducing a medical coding backlog covers the operational cleanup for each bucket.

Estimate the cash tied to coding-held encounters

How do you estimate the cash tied to coding-held encounters? Multiply daily encounter volume by expected collectible value per encounter by the extra days of medical coding delay.

Hypothetical illustration. Your own figures, not MediCodio AI data and not a performance claim.

200 encounters per business day × $150 average expected collectible amount × 5 additional business days of medical coding delay = $150,000 associated with delayed submission.

State the assumptions when you bring this to finance:

  • $150 is an average expected collectible amount, net of contractual allowances and expected denials. Gross charges are not cash and will inflate the figure.
  • Five days is the delay above your normal coding turnaround time, not total turnaround.
  • 200 per business day is a steady state. A queue that is growing or shrinking changes the arithmetic.

Read the result correctly. The $150,000 is revenue whose submission is delayed, and most of it will still be collected on the normal payer cycle. It is also not cash that arrives the day the queue clears, because each released encounter still has to pass billing edits, reach the payer, and complete adjudication. Treat it as a timing figure for cash forecasting and a sizing figure for deciding how much the delay is worth fixing.

Measure these eight things together with finance

MeasureWhat it tells you
Count of coding-held encountersSize of the queue, independent of dollars
Age distribution of held encountersWhether the queue is aging or simply large
Hold reason on each encounterWhich of the four buckets each chart sits in
Estimated collectible value of the held populationCash timing exposure, on expected collections
Documentation-ready to coding-complete timeThe interval medical coding controls
Coding-complete to claim-submission timeWhether finished coding is reaching the payer
Coding quality on released chartsWhether speed cost you accuracy
Payment timing on released chartsWhether earlier submission produced earlier cash

The last two catch the failure mode. Clearing a queue fast and generating denials moves the delay downstream instead of removing it.

Measure the submission date rather than the coding completion date. Coding completion is an internal milestone. Submission is the event the payer sees.

How MediCodio AI addresses it

If the sort above shows that coding capacity or chart-level visibility is what holds your encounters, this is where autonomous AI medical coding changes the interval.

The platform runs autonomous and assistive coding together. Charts that the documentation fully supports are coded autonomously. Complex or flagged charts get AI-assisted coding, where a certified medical coder reviews and finalizes. That maps onto the buckets above: the autonomous path works the ready-to-code volume without competing for medical coder hours, and the assistive path keeps specialist-judgment cases with a credentialed person instead of letting them block the routine queue. Medical coders handle the complex charts; the platform handles the volume.

The eligible share depends on how your documentation arrives, so prove it on your own charts first with a chart trial rather than taking a rate on trust.

Four published capabilities address the measures above:

  • Charts coded within 24 hours of encounter close. A known coding turnaround time turns the documentation-ready to coding-complete interval into something you can plan a cash forecast around.
  • Documentation Quality Gates. The note is scored against what it supports and read for contradictions, and a chart that fails is flagged with the reason named. A documentation dependency enters the query workflow early instead of surfacing after a medical coder opens the chart.
  • Chart Queue and Lifecycle. Every chart sits in one lifecycle bucket, imported, processing, reviewed or executed, so the backlog is a number on a screen rather than a reconstruction.
  • SLA Tracking. Each chart shows time since arrival, target turnaround and time remaining, with thresholds set per client. That is the age data the measures table requires.

Coding runs at 98%+ accuracy with a 100% code-level audit trail, so released charts hold up when speed increases. Codes return ready for the claim.

Scope matters here, because the measures above come from a hospital vocabulary. MediCodio AI does medical coding only, across 35+ specialties, outpatient and professional fee. That coverage includes emergency medicine, radiology and pathology on the hospital outpatient side, and the ASC specialties of GI, orthopaedics, pain and ophthalmology. It does not include inpatient coding, and it does not include billing or claim submission.

Explore CODIO AI’s coding capabilities .

Frequently asked questions

Can accurate medical coding still delay payment?

Yes. Accuracy describes claims that were submitted. An encounter held at medical coding has never been submitted, so it appears in neither accuracy nor denial reporting. A team can score well on accuracy while adding days of coding turnaround time, and the cash variance shows up before any report does.

Is DNFB the same as a medical coding backlog?

No. HFMA’s PB-1 measure covers every account discharged but not yet final billed, for any reason, including documentation, charge capture and registration holds. Medical coding is one contributor. You need the hold reason recorded on each account before you can attribute any DNFB movement to your coding queue.

How do you estimate the cash tied to coding-held encounters?

Multiply daily encounter volume by average expected collectible amount per encounter by the number of extra days of medical coding delay. Use expected collections rather than gross charges, and read the result as delayed timing rather than lost revenue or cash available the moment the queue clears.

Does clearing a medical coding backlog create new revenue?

No. It changes when existing revenue reaches the payer and, if downstream conditions hold, when it is paid. Revenue still depends on payer adjudication and your contracted rates. Treat backlog reduction as a cash timing improvement, and measure it on submission date and payment timing.

What filing deadline should a backlog aging report use?

Use the deadline that applies to each account. CMS sets Medicare fee-for-service at one calendar year from the date of service, with exceptions. Commercial and Medicaid managed care windows come from your contracts and state rules, and many are shorter. Verify each before you build the report.

Book a demo

Sources

SourceURL
HFMA, MAP Keys (PB-1 DNFB, PB-2 FBNS, PB-3 DNSP, PB-4 Charge Lag Days)https://www.hfma.org/data-and-insights/map-initiative/map-keys/
CMS, MLN Web-Based Training, CMS-1500 claim, timely filinghttps://www.cms.gov/Outreach-and-Education/MLN/WBT/MLN4462429-MLN-WBT-1500/1500/lesson01/08/index.html

See it in action

Ready to transform your medical coding?

See how MediCodio's AI platform delivers 98%+ accuracy across deployments since 2023, with turnaround under 24 hours at enterprise volume, inclusive of coder review, across 35+ specialties.

Loading posts...