Seats from $1,500 a month · one month minimumUS business hours, your time zoneops@softhomeglobal.comCost calculator
Soft Home Global
A clinician at a keyboard talks with a seated patient while a nurse takes notes on a clipboard

Operations

How medical coding works in a practice

Coding is the one back-office function where being fast and wrong is worse than being slow. This is how to set up an outsourced coding seat so the record leads and the query process actually works.

5 minute read

Coding is the one back-office function where speed is not the primary virtue. Every other queue in the revenue cycle gets better as it gets faster. This one gets worse, because the failure modes are not delay — they are a claim that says something the record does not support, in either direction.

That makes an outsourced coding arrangement different from an outsourced posting or eligibility arrangement. What has to be designed is not throughput. It is the path a question takes when the documentation and the code do not line up.

The record leads, always

A coder codes from the documentation. Not from the appointment type, not from the fee ticket, not from what the same provider billed for a similar patient last month, and not from a list of codes the practice would like to use more often. Every one of those shortcuts produces claims that cannot be defended if anybody asks.

This sounds obvious and is violated constantly, usually under time pressure and usually with good intentions. It is worth stating explicitly in the working agreement, because it is the standard everything else depends on.

The query process is the whole thing

A coding seat will encounter documentation that is incomplete, ambiguous or internally inconsistent. What happens next determines whether the arrangement works.

A good query is specific, names the element in question, does not suggest the answer, and reaches a clinician who can respond within a reasonable period. A bad query process — a general inbox, no response expectation, no escalation — leads directly to one of two outcomes. Either the coder guesses, which is a compliance risk, or the coder codes conservatively every time, which is a slow revenue leak nobody ever sees.

So the query route needs deciding before the first chart is touched:

  • How a query reaches a clinician, in what system.
  • What a query must contain, and what it must not.
  • The response period expected, and who follows up when it passes.
  • What the coder does with an unanswered query — which should be to code what the record supports and flag the encounter, never to hold it indefinitely.
  • Who reviews recurring query patterns, because a clinician generating the same query every week is a training conversation, not a coding problem.

Under-coding is not the safe option

When a coder is uncertain and has no way to ask, the instinct is to code down. It feels prudent. It is not: it is a decision to bill less than the service, on every affected encounter, with no denial to reveal it and no report that counts it.

The honest position is that both directions are coding failures. The answer to both is identical — code what the documentation supports, and fix the documentation where it does not support what happened.

The audit trail

Every coding decision should be reconstructable: what was assigned, from which documentation, by whom, on what date, and whether anything was queried. This is not bureaucracy. It is what makes a payer audit answerable, and it is what lets a practice tell the difference between a coder’s error and a documentation gap when a pattern of denials appears.

It also protects the coder, which matters if the arrangement is to last. A coding seat with no record of its own reasoning is the first thing blamed when something goes wrong.

Audit, sample, feedback

An outsourced coding arrangement needs an agreed review sample from the beginning — a defined number of encounters per period, reviewed by whoever the practice trusts to review coding, with the findings returned to the coder as specific corrections rather than as a score.

The point of the sample is not surveillance. It is that coding drifts, in both directions, and drift is only visible against a review. A practice that reviews nothing for a year and then discovers a systematic error has a repayment problem that a monthly sample would have caught in week three.

What to agree before starting

Scope by service line

Which specialties, which encounter types, and which are excluded. Coding competence is specialty-specific, and “all of it” is not a scope.

Code-set responsibilities

Who maintains the code and edit updates in the system, and who is responsible for checking that annual changes have been applied. This is a frequent gap because it belongs to nobody by default.

What the practice reserves

Some practices reserve certain decisions entirely — level of service for particular encounter types, for example. That is a reasonable choice and needs to be written down rather than discovered.

The denial route

What happens when a coded claim denies. The route should be back to documentation review, never straight to a code change.

Access and record handling

Named accounts inside the practice’s own system under its own controls, with access limited to what the work requires. Charts should be read where they live, not copied out.

How to assess it

  • What did the review sample find, and are the same findings recurring?
  • How many encounters were queried, and how many queries were answered within the expected period?
  • Are coding-related denials falling, and which reasons remain?
  • Is there a documentation pattern behind the queries, and has anybody raised it with the clinicians concerned?
  • Is the turnaround from documentation complete to coded within the agreed window?
  • Can a randomly chosen claim from last month be traced back to the note that supports it?

That last question is the one worth asking first, and the one most arrangements fail.

How Soft Home Global staffs this

Soft Home Global provides trained full-time back-office seats from Rawalpindi, Pakistan. A certified coder seat is published at $2,500 per seat per month, billed per seat with a one month minimum.

The seat codes inside your system on US hours, from your documentation, to the scope and the reserved decisions you set. Queries go to the clinicians on the route you define, in your format. What was coded, from what, and what was queried is recorded so that any claim can be traced back to the record behind it.

Where a coded claim denies, the account goes to documentation review rather than to a code change, and where the denial reflects an authorization or eligibility problem rather than a coding one, it is routed to those queues — seats Soft Home Global also staffs.

The first thing to set up

Not the coding. The query route. Decide how a question about a note reaches the clinician who wrote it, what response period is expected, and who chases when that period passes.

An outsourced coding arrangement with a working query route improves documentation as a side effect. One without a query route quietly teaches a coder to guess, and the practice will not find out for a year.

Questions people ask about this

What does an outsourced medical coding seat do?
It reads the clinical documentation for each encounter and assigns the procedure, diagnosis and modifier set the record supports, queries the clinician where the documentation is unclear or incomplete, and records what was coded and from what — so that every claim can be traced back to the note behind it.
Can medical coding be outsourced to Pakistan?
Yes. It is done widely, and the practical requirements are the same as for any coding arrangement: access to the record system, a working query route to the clinicians, an agreed audit sample, and clarity about which coding decisions the practice reserves to itself.
How much does an outsourced coding seat cost?
Soft Home Global publishes $2,500 per seat per month for a certified coder seat, billed monthly with a one month minimum. It is priced per seat rather than per chart, so the rate does not change with encounter volume.
What happens when documentation does not support the service?
The coder queries the clinician; they do not code what they believe was intended. A query names the specific element that is missing and asks for clarification without suggesting an answer. If the query is not answered, the code assigned is the one the record actually supports.
Is under-coding safer than over-coding?
It is not safe, it is just quieter. Over-coding creates audit and repayment exposure; under-coding creates an unbillable loss on every affected encounter and never triggers a denial, so nobody finds it. Both are coding failures, and the answer to both is the same: code what the documentation supports.
Should coders change codes when a payer denies a claim?
No. A denial is a reason to re-read the documentation and decide whether the original code was correct. If it was, the route is an appeal. Changing a code to whatever the payer will pay, without a documentation basis, is a compliance problem regardless of who does it.

More from the same desk

Next step

One seat. One month. Cancel any time.

Twenty minutes on a call is enough to tell whether this fits. If it does not, I will say so.

Or write to ops@softhomeglobal.com