Hospital Revenue Integrity · Chapter 2

Six Leaks, Part Two: When Does Fixing the Leak Pay for Itself?

If your hospital invests one million baht in AI that checks coding before claims go out, how much has to come back each year for it to pay for itself? This piece gives you a formula you can check with a calculator, and shows how to find your hospital's real number: run the same AI over records that were already billed and paid.

Infozense · Hospital CFO · Hospital Director · ~7 minutes

The first piece in this series explained where the six leaks come from, but not how much your hospital loses to them. We cannot tell you that number directly either, because every hospital's data is different, and a number made up before seeing your data is the first thing a finance office should distrust. What this piece gives you instead is a way to estimate it: a payback formula with a worked example, and how to find your own hospital's real number, to answer the question the finance office actually needs answered: when does fixing the leak pay for itself?

The formula, per one million baht invested

Forget package prices. For every one million baht of one-time investment in an AI coding-integrity system, assume roughly 20 percent of that, two hundred thousand baht per year, in running costs: maintenance, updates, the care any production system needs.

Then the payback clock depends on exactly one number: R, the annual revenue your hospital recovers per million invested. That number is measured, not promised. The formula is simple arithmetic:

payback in months = 12 x 1,000,000 / (R minus 200,000)
Measured recovery R (per ฿1M invested, per year) Net after running cost Payback
฿500,000 ฿300,000 ~40 months
฿1,000,000 ฿800,000 ~15 months
฿1,200,000 ฿1,000,000 ~12 months
฿2,000,000 ฿1,800,000 ~7 months
฿4,000,000 ฿3,800,000 ~3 months

Read the third row twice, because it is the threshold: if your measured recovery reaches 1.2 times your investment per year, the system pays for itself in twelve months. Below that line, the case weakens month by month. Above it, every additional baht of measured leak shortens the clock.

Every cell in that table is checkable with a calculator. None of it claims to know your hospital.

Or skip the table and put your own numbers in below.

Payback calculator
Net annual benefit
฿1,000,000
Payback
12.0 months
Above the threshold: pays back within a year.
0 12 months (threshold) 36+
Cumulative money in and out, month by month
Recovered revenue, cumulative Paid out, cumulative (investment + running cost)

Show the numbers as a table
MonthRecoveredPaid outNet

Where R comes from

Not from us. From a look-back audit of your own records. Records that were already coded and paid are too many for anyone to re-check, so nobody knows whether revenue leaked. The audit uses AI to check several hundred inpatient records, comparing the codes actually billed with what the record contains, such as recorded diagnoses, lab results and drugs given, under the published TDRG and ICD-10-TM rules. The AI flags each gap with the part of the record behind it, and your hospital's own coder confirms every one, counting both codes that were missed and codes the record does not support. Price the net difference in adjRW at your own payer rates. That measured delta, annualized, is R. R measured this way is a floor, because it counts only what the record's data can confirm. If the floor pays back, the investment does. This is the pilot we described in part one, and it is the only source of R we will accept in our own business case.

What we can say, with citations, is that when others have looked, they found discrepancies:

SourceSampleWhat it found
BMC Health Services Research, 2011 (reporting the NHSO's 2008 audit) The NHSO's 2008 Summary and Coding Audit: 57,828 records from 931 hospitals, chosen because their data looked abnormal Errors in 42 percent of abstractions from medical record to discharge summary; the most common error was the secondary diagnosis, at 28 percent
Healthcare Informatics Research, 2017 118,971 records from the NHSO's 2014 audit results, selected by abnormal-data criteria For the 20 most common principal diagnoses, discharge summaries captured only 7.3 to 37.9 percent of the cases the auditors found
Integrated Social Science Journal, Mahidol University, 2014 1,107 inpatient records at Mahidol University's Hospital for Tropical Diseases Discrepancies in 322 records; the corrections raised adjRW by 182.84 and, by the authors' own calculation, reimbursement by 2,420,928.35 baht (about 2.4 million)

These numbers come from published research. They do not mean your hospital has the same problem; the only way to know is to check your own data.

Side benefits that are not in the payback table

Recovered revenue is not the only thing that improves when claims are coded correctly the first time. Four more benefits follow. They are not in the payback table, because we will not put a number on them for you, but each one can be measured from your own data, before and after the AI is in use.

Claims coded correctly the first time
Money arrives sooner

Fewer claims are denied, so fewer wait for another round. Less money sits unpaid; it is cash freed up, not new income.

Measure: average days from sending a claim to being paid

Less rework

Coders spend less time fixing and resending denied claims, and have more time for other work.

Measure: claims fixed and resent per month

Less risk of clawback

AI flags codes the record does not support, not only codes that were missed, so fewer claims are exposed to clawback when the payer audits them later.

Measure: number and value of clawed-back claims

Ready for audit

Every flag points to the part of the record and the rule behind it, so an auditor can see where each code came from.

Measure: time spent preparing documents for an audit

The payback table counts recovered revenue only; these four come on top.

Invest once, use it for many jobs

Everything above counts only one job: checking that diseases are coded correctly. It is the part the look-back audit can measure.

The same investment is built to carry more. The governed knowledge base that checks codes against charts is designed so that a different rulebook, once loaded, applies the same method elsewhere: audit readiness, record completeness, drug to diagnosis matching, regulatory reporting. Each of those arrives when its rulebook is built and loaded, not before. That first job pays for the investment on its own arithmetic. The other jobs are what the same machine can take on next, with no extra hardware to buy.

One machine · one knowledge base · one way of checking
First job · pays back on its own
Checking disease coding

Do the codes billed match what the record supports?

Next · when its rulebook is loaded
Audit readiness

Are the documents and evidence an auditor asks for complete?

Next · when its rulebook is loaded
Record completeness

Does the record hold everything it needs to?

Next · when its rulebook is loaded
Drug to diagnosis matching

Does every drug given have a diagnosis that justifies it?

Next · when its rulebook is loaded
Regulatory reporting

Is the data owed to regulators complete and correct?

Every job runs on the same machine, with no extra hardware to buy. Each one still needs its own rulebook built first, and the cost of building each rulebook depends on the scope of that job.

How to decide, step by step

The decision takes four steps, and each one uses your hospital's own numbers, not anyone's promise:

  1. Run the look-back audit on your own past records to find R.
  2. Put R into the formula: 12 × investment ÷ (R − yearly running cost). The result is the payback in months.
  3. If it pays back within 12 months, size the system to the number you measured, not to a sales brochure.
  4. If it does not, you have spent only the cost of the audit, and you know where you stand before the larger investment.

We start with the look-back audit on your hospital's real cases, showing each item in baht, and the results belong to the hospital whatever you decide. If the numbers pay back, we implement the AI system that helps your coders check every claim before it goes out, so leaks are stopped at the source instead of found in an occasional look-back. If you already have your R, bring it and we will work it through the formula together.

Hospital Revenue Integrity

Want to know your own R?

We run the look-back audit on your own records, showing each item in baht, and the results are yours whatever you decide. If the numbers pay back, we also implement the AI system that checks coding before claims go out.

Let's talk →

contact@infozense.com  |  +66-82-242-4008  |  Bangkok, Thailand