Chapter 24: Patient Access: Registration, Eligibility, Prior Authorization, and the Front-End Fixes

"Everything the business office does after the visit is either collecting money or repairing something that happened in the first four minutes." — constructed

Overview

Part IV was about what a claim has to survive. Part V is about building one, in the order a claim actually travels — and it starts before the patient is in the room.

This chapter is about the four minutes at check-in, and it makes a claim that sounds like front-office boosterism until you do the arithmetic:

The largest single category of preventable denial in most organizations is created at registration, by people who are not paid to think about claims, in less time than it takes to read this paragraph.

And it is the cheapest place in the entire revenue cycle to fix anything.

Chapter 1 §1.6 introduced the idea. Chapters 14, 18, 19, 20, and 23 each found a failure that could have been prevented earlier and cheaper than it was found. This chapter is where "earlier and cheaper" bottoms out — there is nothing before registration.

One structural note before you start. The people who do this work are frequently the lowest-paid, least-trained, and most-interrupted staff in the organization, and they are asked to capture the information on which every downstream dollar depends while a person stands in front of them wanting to be seen. Any account of front-end failure that does not begin there is not an honest account.

In this chapter, you will learn to:

• Do the arithmetic on where a claim is cheapest to fix
• Name the registration fields that cause denials
• Run an eligibility transaction and read what comes back
• Tell a referral from an authorization from a precertification
• Document an authorization so it survives
• Administer a Medicare Secondary Payer questionnaire
• Ask for money without making it worse
• Build an estimate and state its limits
• Measure the front end

And you will rewind Account 10-4471 to check-in — to see what the front desk did right, and the one thing it missed.

24.1 The cheapest place to fix a claim

Take one error — a transposed digit in a member identification number — and follow it.

   WHERE IT IS CAUGHT              WHAT IT COSTS

   At registration ............... ~30 seconds. Ask again, retype.
                                   No claim exists yet.

   At the scrubber ............... a few minutes. The claim is held,
                                   someone corrects it, it goes out.
                                   Still no payer involvement.

   At the clearinghouse .......... a rejection. Someone works a report,
   (front-end rejection)           corrects, resubmits. Days of lag.
                                   ► and only if somebody reads the report

   At adjudication ............... a DENIAL. Work queue, research,
                                   correction, resubmission. Weeks.
                                   ► and the timely filing clock has
                                     been running the entire time

   After timely filing ........... the money is gone.

Four things to take from that diagram.

The cost rises by roughly an order of magnitude at each step. Thirty seconds, then minutes, then a work queue, then a lost claim.

The information gets harder to obtain as it gets later. At registration the patient is standing there. At adjudication, six weeks later, somebody has to telephone them — and Chapter 22's Case Study 1 showed what happens when the thing you need to ask about happened months ago.

The rejection step is invisible if nobody reads the report. Chapter 27 §27.7 is emphatic about this, and it is the failure that turns a thirty-second fix into a total loss without anyone deciding anything.

And the last row is not hypothetical. Chapter 1 §1.3's timely filing limits are measured in months, and Chapter 18's Case Study 1 was a practice that ran out of them on care it had genuinely provided.

🧮 Run the Numbers

What a front-end error actually costs — in the currency the front end is staffed in: MINUTES.

Where caught	Staff minutes
Registration — ask again, retype	0.5
Scrubber — hold, correct, release	3
Clearinghouse rejection — read the report, correct, resubmit	12
Denial — research, correct, resubmit, and follow up	commonly 30–60, across several touches
After timely filing	—

Half a minute at the front desk. Something on the order of forty-five at the back.

A ratio of roughly ninety to one, before anyone converts it to money.

And that is the good outcome, because it assumes the denial was worked and won. Chapter 29 covers what fraction are not — and the last row has no time at all, because there is no work to do.

Two notes on why this table is in minutes.

Minutes are what a staffing decision is actually made in, and Case Study 2 is about a staffing decision. Converting to dollars requires a fully loaded labor rate, which Chapter 31 §31.7 publishes and which this chapter deliberately does not borrow — the arithmetic that turns Account 10-4471's own touches into a dollar figure belongs to Chapter 40, and doing it here would spend the book's closing argument early.

And the ratio is the point regardless of the rate. Whatever your loaded rate is, the same number multiplies both sides, and ninety to one survives it.

The uncomfortable implication: an organization that will not fund thirty seconds of front-desk time is buying roughly ninety times as much back-office time instead — and it is buying it on a budget line that is not visibly connected to the decision.

24.2 Registration: the fields that cause denials

Not all fields are equal. A short list produces most of the damage, and it is worth knowing it cold.

The patient's name

It must match the payer's record exactly, including hyphens, suffixes, and the difference between a legal name and the name the patient uses.

A patient who introduces herself as "Beth" may be "Elizabeth" on the policy. A hyphenated surname may be stored unhyphenated. A junior may be recorded without the suffix.

The claim will reject or deny as "patient not found" or "subscriber not matched," and the fix is a transcription correction that takes seconds and could have been avoided by asking one question: "How does your name appear on your insurance card?"

The date of birth

One transposed digit and the patient does not exist. This is the single most common cause of a front-end rejection in most organizations, and it has no clinical consequence and no clinical warning — the patient in front of you is unmistakably themselves.

The member identification number

Copy it from the card, not from memory, not from the last visit, and not from what the patient says. Cards get reissued with new numbers. Plans change identifiers on renewal.

And the alphanumeric confusions are predictable: O and 0, I and 1, S and 5, B and 8. A staff member who knows this list reads the ambiguous characters back.

The subscriber relationship

Who holds the policy? The patient, a spouse, a parent, someone else. Getting this wrong sends the claim to a policy the patient is not on, and it is the field most often filled in by assumption — an adult patient is assumed to be their own subscriber, and frequently is not.

The group number and the plan

A payer may have dozens of products. The card carries the information; the registration record frequently carries only the payer name. "Blue Cross" is not a plan, and a claim routed on that much information is being routed on a guess.

Address, and the guarantor

Statements go here. A wrong address produces no denial at all — it produces a patient who never receives a bill, an account that ages, and eventually a collection action against someone who was never told. Chapter 32 §32.6 is about statements; this is where they go wrong.

Accident and injury information

"Was this related to an accident? Auto? Work?" If yes, another payer may be primary — auto insurance, workers' compensation, liability. Chapter 2 §2.9 covered coordination of benefits; this question is how the front desk finds out.

Identity — the field that is not about billing

One front-desk duty is not a billing task at all, and it is the one with the worst consequences when it fails.

Verify that the person in front of you is the person on the coverage.

Medical identity theft — the use of another person's identity to obtain care or coverage — is a real and under-discussed problem, and it is different in kind from every other error in this chapter.

Three consequences, in ascending order of seriousness:

A claim for services the coverage holder did not receive, which becomes a billing dispute.

A permanent contamination of a medical record. The impostor's blood type, allergies, diagnoses, and medications enter the victim's chart. Unwinding it is difficult, sometimes impossible, and the victim frequently discovers it during care they need.

And a patient harmed by information that is not theirs, which is the reason this is a patient safety matter before it is a revenue matter.

The practical measure is modest: ask for identification, compare it to the coverage, and treat a mismatch as a question rather than an accusation. Names change, cards are old, and most mismatches are ordinary. But a practice with no identity step has no way to notice the one that is not.

This is also the front-desk task that most obviously is not about money, which is worth saying to a class and to a staff member: the reason for it is the chart, not the claim.

⚠️ Where Claims Die

The most expensive registration field is the one nobody thinks of as a field: "has anything changed?"

A returning patient's record is copied forward. Nobody asks. And then:

• The patient changed jobs and has new coverage — Chapter 18's Case Study 1, where an obstetric package was billed to a plan that had not covered most of it
• The patient turned 65 and Medicare is now primary or secondary
• A divorce changed the subscriber
• A child aged off a parent's policy
• The employer changed carriers on January 1 and the card in the file is last year's

None of these produce a warning. The record looks complete because it is complete — for a situation that no longer exists.

The fix is a scripted question at every visit, not just new ones: "Has anything changed with your insurance, address, or phone since we saw you?" Ten seconds, and it catches the entire category.

And January is the month. Plan years turn over, employers change carriers, and deductibles reset. A practice that verifies more aggressively in January and February is spending effort exactly where the errors are.

24.3 Insurance verification and the 270/271

Asking the patient is not verification. Asking the payer is.

The 270 is an eligibility and benefit inquiry — a standardized electronic question to a payer.

The 271 is the response.

These are HIPAA standard transactions — Chapter 27 §27.2 covers the full set — and they are the mechanism by which a practice finds out, before the visit, whether coverage exists.

What the transaction can tell you

• Is the patient active with this payer, on this date?
• What plan, and what group?
• Is there a copayment, and for what service types?
• What is the deductible, and how much has been met?
• What is the coinsurance?
• Is there other coverage on file? — the coordination of benefits question
• Sometimes: whether a referral or authorization is required

What it cannot tell you

And this list matters more, because it is the source of every "but we verified" conversation.

It is a snapshot, not a guarantee. Eligibility on Tuesday is not eligibility on Thursday. Coverage terminates retroactively. Chapter 22's Case Study 1 is what happens when something moves between verification and service.

Deductible amounts lag. The deductible figure reflects claims the payer has processed — not claims the patient has incurred. A patient who had surgery last week may have met their deductible in fact and not yet in the payer's system. An estimate built on a lagging deductible will be wrong, and it will be wrong in the direction of asking the patient for too much.

It rarely tells you about medical necessity. Chapter 22's entire apparatus — coverage determinations, supporting diagnoses, frequency limitations — is not in an eligibility response. "The patient is eligible" and "this service is covered for this patient" are different statements, and the transaction answers only the first.

And benefit detail varies enormously by payer. Some return rich, service-type-specific information. Some return little more than "active." The transaction is standardized; what payers choose to populate is not.

When to run it

"We verify eligibility" is not a process. When you verify is the process, and the answer is not once.

   AT SCHEDULING ........ catches: no coverage at all, wrong payer,
                          a plan the practice is not contracted with,
                          an authorization requirement
                          ► the only point at which the encounter can
                            still be REARRANGED

   48-72 HOURS BEFORE ... catches: terminations, plan changes,
                          a deductible figure fresh enough to
                          estimate from
                          ► BATCH. Run the whole schedule at once.

   DAY OF SERVICE ....... catches: the termination that happened
                          yesterday
                          ► REAL TIME, at check-in, for the
                            individual patient

Two operational facts.

Batch and real-time are different transactions with different economics. A batch inquiry submits the whole schedule and returns a file; a real-time inquiry answers one patient in seconds. Most practices should do both — batch for the pre-visit sweep, real time at the desk.

And the 48-hour check is the one that pays for itself. It is early enough that a problem can be addressed and late enough that the answer is still true. A practice that verifies only at check-in finds problems when it can do nothing but proceed.

🎓 Exam Watch

The transaction pairs are worth memorizing as pairs, because exams ask for the number and because the numbering is not intuitive:

270 / 271	eligibility and benefit inquiry / response
276 / 277	claim status inquiry / response
278	services review — prior authorization request and response
837	the claim (P professional, I institutional, D dental)
835	the remittance advice
834	benefit enrollment
820	premium payment

The trap: 278 is the authorization transaction, and candidates reach for 276/277 because "status" sounds right. 276/277 asks where a submitted claim is. 278 asks permission for a service that has not happened.

24.4 Reading an eligibility response

A 271 is not designed to be read by a person, and most practices see it through their system's rendering of it. Read that rendering skeptically, because the summarization is where information is lost.

📋 Read the Chart

Source: eligibility response, as rendered by a practice management system Encounter: established patient, office visit scheduled What it says:

text Subscriber .......... [name], active Plan ................ Northfield Mutual PPO Group ............... [number] Effective ........... [date] Office visit copay .. $30.00 Deductible .......... $1,500.00 individual / $700.00 remaining Coinsurance ......... 20% Out of pocket max ... $4,000.00 / $1,150.00 remaining Other coverage ...... none indicated

What it means: the patient is covered, and a \$30.00 copay applies to an office visit.

What it does NOT mean, and here is where the money is:

The \$700.00 remaining deductible is as of the payer's last processed claim. If the patient had laboratory work two weeks ago that has not adjudicated, the real figure is lower. An estimate built on \$700.00 may ask for money the patient does not owe.

"Other coverage: none indicated" is not "there is no other coverage." It is the payer has none on file, which is a statement about the payer's records. §24.7's questionnaire exists because that sentence is unreliable.

The copay applies to an office visit. It says nothing about a procedure, a laboratory service, or an injection performed at the same encounter — all of which are on Account 10-4471's claim, and all of which fall under the 20% coinsurance, not the copay.

What to do about it: collect the copay, and do not tell the patient it is what they will owe. §24.9 is about the difference.

Where it appears: on every eligibility response, every day, in every practice — and the third item is the one that produces the phone call four weeks later.

The three questions to ask of any eligibility response

1. As of what date? Verify for the date of service, not the date you looked.

2. What is this response silent about? Silence is not a negative. "No authorization required" and "we did not return authorization information" render identically in most systems.

3. What did the payer say versus what did my system display? A rendering that shows five fields from a response containing forty has made editorial choices you did not make.

24.5 Referral versus authorization versus precertification

Three words used interchangeably in conversation and meaning three different things, and getting them confused produces denials that are painful precisely because everyone believed the matter was handled.

REFERRAL — a primary care physician's direction of a patient to a specialist. It is about who the patient sees. In plans that require one, the specialist's claim denies without it. The referral is generally the PCP's responsibility to issue.

PRIOR AUTHORIZATION — a payer's approval of a specific service before it is furnished, based on clinical information. It is about what is done. Chapter 22 §22.10.

PRECERTIFICATION — in most usage, a synonym for prior authorization; in some plans, a distinct notification requirement for admissions. When a payer uses the word, ask what it means in their contract.

A fourth term worth knowing because it is different in kind:

PREDETERMINATION — a non-binding advance review of whether a service would be covered. It is not an authorization and it does not guarantee payment. Some payers offer it for services that do not require authorization, and it is genuinely useful for expensive elective procedures.

   WHICH ONE DO I NEED?

   Is the patient going to a specialist under a plan
   that requires PCP direction?              ► REFERRAL

   Is a specific service on the payer's list
   of services requiring approval?           ► PRIOR AUTHORIZATION

   Is this an admission the payer wants
   notice of?                                ► whatever they call
                                               precertification

   Do you want a non-binding opinion in
   advance?                                  ► PREDETERMINATION

📞 On the Phone

"We have a referral on file, so we don't need an authorization, right?"

No, and this is a conversation worth having carefully, because the person asking is usually a scheduler doing their best with two words that sound like the same thing.

What works: "They're two separate things and a visit can need both. The referral is your primary doctor saying it's okay for you to see us — that's about who. The authorization is your insurance approving the specific procedure — that's about what. We have the first one. I'm checking whether the plan requires the second for this procedure, and I'll call you back today either way."

Three things that does. It separates the concepts in one sentence each. It does not blame anyone for the confusion, which is genuine. And it commits to a callback with a time, which is the difference between a patient who trusts the process and one who calls back angry.

And the internal version of the same failure: a scheduling note reading "auth on file" that refers to a referral. Two different fields, in most systems, and a note is not either of them.

24.6 Obtaining and documenting a prior authorization

Chapter 22 §22.10 covered what an authorization is and Chapter 22's Case Study 1 covered how one drifts. This section is the mechanics.

Obtaining it

Determine whether one is required. Payers publish lists. They change. A service that did not require authorization last year may this year, and the practice finds out through a denial unless somebody is watching the updates.

Submit the clinical information the policy asks for, not a generic packet. Chapter 22 §22.5's procedure applies: read the policy, find the required facts, and supply them. An authorization request that omits a required element is denied for the omission and re-worked, which is Chapter 22's Case Study 1 shape happening a step earlier.

Submit through the payer's preferred channel — a portal, the 278 transaction, a fax, a telephone line. The channel affects the turnaround and some channels produce a documented reference number automatically and some do not.

Documenting it

This is the part practices do badly, and it is the part that matters at audit.

Record all of it, in the account, not in a note:

• The authorization number
• The date obtained
• The name of the person or the system that issued it
• The approved CODE or codes
• The approved DATE RANGE
• The approved number of units, visits, or sessions
• The approved FACILITY, if it is site-specific
• The approved RENDERING PROVIDER, if it is provider-specific

Chapter 22's Case Study 1 was four drifts — date, facility, code, and physician — and three of them had straightforward answers nobody knew to ask about, because the approval had been filed rather than read. This list is the reading.

⚖️ Compliance Check

An authorization number recorded without its scope is a number, not a record.

When a claim denies as unauthorized and the practice responds "we have an authorization," the payer's next question is for what, on what date, at what site, by whom — and a practice holding only a number cannot answer.

Three failures that follow, all documented in audits of this area:

Authorizations obtained for one thing and used for another, without anyone noticing the mismatch, because nobody recorded what was approved.

Authorizations that had expired, used because the number was still in the field.

And authorization numbers entered on claims that were never actually obtained — which is a different thing entirely, and is a false statement on a claim. This happens, it happens under production pressure, and it is the reason the scope record matters as much as the number: a practice that records scope has a process; a practice that records a bare number has a field somebody can type into.

24.7 The Medicare Secondary Payer questionnaire

A questionnaire that exists because of a statute, and one of the few places in this book where a front-desk task is directly mandated.

Medicare is, in defined circumstances, the SECONDARY payer — meaning another plan pays first. The provider is obligated to determine whether that is the case, and the Medicare Secondary Payer questionnaire (MSPQ) is the standard instrument.

The circumstances are specific and they are the questions:

• Is the patient covered by a group health plan through their own or a spouse's current employment? (Size-of-employer rules apply.)
• Is the treatment for a work-related injury? → workers' compensation
• Is it for an injury in an accident? → auto, no-fault, or liability insurance
• Does the patient have black lung benefits, or is this a Veterans Affairs–authorized service?
• Is the patient in the coordination period following end-stage renal disease entitlement?

Two things about this that surprise people.

Chapter 24 §24.3's eligibility response is not sufficient. A 271 reporting "no other coverage" reflects the payer's records. The questionnaire asks the patient, and it exists precisely because the records are frequently wrong.

And billing Medicare as primary when another payer is primary is an overpayment. Chapter 5 §5.1's sixty-day rule attaches on identification, and Chapter 31 §31.9 covers the recoupment. A questionnaire that was not administered is not a defense.

The practical consequence: the MSPQ is not paperwork. It is the mechanism by which a practice discovers that it is about to bill the wrong payer, and it is administered by the person with the least time and the least training in the building. That is worth naming, and §24.10 is about measuring whether it is happening at all.

🔍 Check Your Understanding

A 68-year-old patient presents. She has Medicare and also has coverage through her husband's current employer, which has 900 employees. She is being seen for back pain that began after she fell at her part-time job.

Which payer is primary, and what has to happen?

Answer: Neither of the ones you first thought of.

The fall at work makes this a workers' compensation matter, and workers' compensation is generally primary for treatment of a work-related injury. That question comes before the Medicare-versus-group-health analysis entirely.

And the group health question would also have made Medicare secondary, independently: a large-employer group health plan covering a beneficiary through a spouse's current employment is generally primary to Medicare. Two independent reasons Medicare is not primary here, and the questionnaire asks about both.

What has to happen: the workers' compensation carrier is identified and billed, and Medicare is not billed as primary. If the compensation claim is denied or contested, there is a defined process — and it is not "bill Medicare and sort it out later."

The point of the item: the MSPQ is a sequence of questions, not a form. Stopping at the first "yes" is right; skipping to the question you expected is not.

24.8 Point-of-service collection and how to ask

The patient became a major payer — Chapter 32 §32.1 makes the argument at length — and money collected before or at the visit is collected at a fraction of the cost of money chased afterward.

Chapter 32 owns patient financial responsibility. This section owns the moment at the desk.

Why it is collected here

Three reasons, and only one of them is about efficiency.

Collection rates fall sharply once the patient leaves. A balance billed and mailed is substantially harder to collect than the same amount asked for in person, and the difference grows with every statement cycle.

The cost of collecting rises. Statements, phone calls, and eventually a collection agency's fee. Chapter 31 §31.7's arithmetic of when to stop chasing exists because that cost is real.

And — the one that matters most and is least discussed — the patient prefers to know. A person who is told at check-in that they owe \$30.00 has been informed. A person who receives a bill for \$47.58 six weeks later has been surprised, and surprise is the thing patients report as the grievance, more than amount.

How to ask

The scripting matters, and it is learnable.

What works: "Your copay for today is \$30.00. How would you like to take care of that?"

Why: it states the amount as a fact, it does not ask permission, and the question is about method rather than about whether.

What does not work: "Would you like to pay your copay today?" — which invites "no," and to which "no" is a complete and reasonable answer.

Three things that make the conversation worse:

Asking in front of other patients. A financial conversation at a counter within earshot is a dignity problem before it is a collection problem.

Insisting when someone says they cannot. They have told you something true and difficult. Chapter 32 §32.7's payment plans and §32.8's financial assistance exist for this, and a front desk that knows they exist can offer something instead of pressing.

And quoting a number you cannot stand behind. §24.9.

⚖️ Compliance Check

What a practice may and may not do about patient cost-sharing.

May: collect copayments, coinsurance, and deductibles at the time of service. Offer payment plans. Offer financial assistance under a written policy. Offer a prompt-pay or self-pay discount under a written policy — Chapter 23 §23.7.

May not: routinely waive cost-sharing.

The reasoning is worth understanding rather than memorizing. Cost-sharing exists in the benefit design to make the patient a participant in the decision. A provider who routinely waives it has removed that participation — and has, in effect, offered the patient something of value to choose them, which is why the anti-kickback analysis reaches it (Chapter 5 §5.9). It also means the amount actually charged is not the amount represented on the claim.

The word doing the work is "routinely." A waiver based on a documented, individualized determination of financial hardship is a different thing, and it is what a financial assistance policy is for. The difference between the two is a written policy and a record — which is the same answer this book has given about ABNs, about waste, and about queries.

24.9 Financial clearance and the pre-service estimate

Financial clearance is the front end's completion check: before the service, is everything in place for this to be billable?

   FINANCIALLY CLEARED means ALL of:

   ☐ Identity and demographics verified
   ☐ Coverage verified FOR THE DATE OF SERVICE
   ☐ Coordination of benefits resolved (MSPQ where applicable)
   ☐ Referral obtained, if required
   ☐ Authorization obtained AND ITS SCOPE RECORDED, if required
   ☐ Medical necessity screened, where a policy governs
   ☐ Patient responsibility ESTIMATED
   ☐ Patient INFORMED of the estimate
   ☐ Collection attempted or arrangement made

Two notes on that list.

"Medical necessity screened" is Chapter 22's apparatus applied in advance — checking, for services governed by a coverage policy, whether the ordered diagnosis is on the supporting list. This is where a Chapter 22 §22.11 problem gets caught before the service rather than after, and it is the single highest-value front-end check that almost nobody does.

And "informed" is not the same as "collected." A patient who cannot pay today has still been informed, and the encounter is still cleared.

The estimate, and its honest limits

Chapter 32 §32.2 covers what a good estimate contains. What belongs here is what an estimate cannot do, because the front desk is where the promise gets made.

It depends on the deductible figure, which lags — §24.3.

It depends on codes that do not exist yet. At check-in nobody knows what will be performed. Account 10-4471's patient arrived for a follow-up and received an injection, and no estimate built at the desk that morning could have included it.

It depends on the claim adjudicating as expected, which Chapter 21 established it may not — the day-17 denial on line 1 was correct coding meeting a payer's own edit policy.

Which is why the sentence matters more than the number.

What works: "Based on what we know today, your share should be around \$X. That's an estimate — if the doctor finds something that needs treating, or your deductible has changed, the final amount can be different. You'll get a statement that shows exactly how it was calculated."

What does not work: "You'll owe \$X." You do not know that, and a patient who was told a number and receives a different one has been given a reason to distrust everything else you said.

🧮 Run the Numbers

Building the estimate, with everything this book has established.

A patient is scheduled for a procedure. You need four inputs, and you have three of them.

1. What will be done? Say the schedule says 20610, a large joint injection, plus the drug.

2. What is the allowed amount? Not the charge. Chapter 23 §23.1: the patient's share is computed from the allowed amount, and the allowed amount is in the contract. For Northfield Mutual: 20610 = \$78.60**, **J1030 = \$6.28.

3. What is the benefit design? From the 271: deductible met, 20% coinsurance on services other than the office visit.

4. What has the patient already paid this year? This is the input you do not have — §24.3's lag.

text 20610 ....... allowed $78.60 × 20% = $15.72 J1030 ....... allowed $ 6.28 × 20% = $ 1.26 ─────── ESTIMATED PATIENT SHARE ............. $16.98

Now the three ways that number is wrong, and a good estimate says so:

The deductible may not actually be met. The 271 said it was, as of the payer's last processed claim. If it is not, the patient owes the full allowed amount, not 20% of it — \$84.88 instead of \$16.98, a five-fold difference produced by a lagging field.

The service may change. Chapter 22 §22.11's re-excision, Chapter 18 §18.5's converted-to-open — medicine changes what happens, and correctly.

And the out-of-pocket maximum may be reached, in which case the patient owes nothing.

The honest estimate is therefore a number and a sentence, and the sentence is not boilerplate — it names the specific thing that could move. "That assumes your deductible is met, which our check this morning says it is. If a claim from another provider is still processing, this could be higher, and the statement will show you why."

24.9a Who obtains the authorization

One operational question this chapter has been walking past, and it causes more internal conflict than anything else in patient access.

When a service ordered by one practice is performed by another, who obtains the authorization?

There is no universal answer, which is precisely why it produces fights.

The general shape:

Three principles that resolve most disputes.

The party who knows the CODES is best placed to obtain it. An authorization is granted for specific codes (§24.6), and the ordering office frequently does not know what will be coded.

The party who does not get paid without it has the incentive. Chapter 22's Case Study 1 shows the consequence landing on the performing practice, which is a reason it should not rely on someone else.

And whoever obtains it must communicate the SCOPE to whoever bills it. A number passed along without its date range, units, and codes reproduces §24.6's failure across two organizations.

The practical instruction: write down who does it, per payer, per service type, and give the document to scheduling. A referring office and a performing office that each believe the other is handling it produces an unauthorized service and two organizations who are each certain they were right. Chapter 17's Case Study 2 was the same shape with fracture care, and it also took months.

24.10 Measuring the front end

Everything above is unmeasurable in the ordinary reports, which is why front-end work is chronically under-resourced: its output is claims that did not have problems, and nothing counts those.

Five measures, and the last two are the ones nobody has.

Clean claim rate — the percentage of claims accepted on first submission without rejection or denial. A blunt instrument — it mixes front-end and coding causes — but it moves when the front end improves.

Front-end denial rate — denials attributable to eligibility, registration, authorization, or coordination of benefits. Chapter 29 §29.4's root-cause classification is what makes this computable, and a practice without that classification cannot produce this number at all.

Eligibility verification rate — the percentage of scheduled encounters with a verification on file for the date of service. Not "at some point." For the date.

Authorization capture rate — of encounters requiring authorization, how many had one, with scope recorded, before the service.

And point-of-service collection rate — dollars collected at or before service as a percentage of dollars owed by patients at that point.

The measure nobody has, and should

A sixth, which no benchmark publishes and which is the most diagnostic of the set:

Of the encounters that generated a front-end denial, what percentage had a completed financial clearance checklist?

If the answer is high, your checklist is missing an item. The process ran and the claim denied anyway, which means the process does not cover the failure.

If the answer is low, your checklist is not being completed — which is a staffing, workload, or training problem, and is a completely different fix.

Most organizations cannot compute it at all, because clearance is not recorded as a discrete event. Recording it is one field, and it is the difference between knowing which of those two problems you have and guessing.

⚠️ Where Claims Die

The front end is measured by the back end's failures, and it is staffed by the front end's budget.

That sentence is the structural problem in one line, and it explains behavior that otherwise looks irrational.

A denial caused at registration appears in the business office's denial report. It is worked by business office staff, on business office time, against a business office productivity standard. Nothing about it appears on any report the front desk sees, and nothing about the cost appears in the department that caused it.

So the organization experiences a front-end problem as a back-end volume problem — and responds by adding back-end staff, which works, and which costs roughly ninety times as much per instance.

The fix is not a new report. It is routing the existing one. Front-end denials, classified by cause, sent weekly to the people who cause them, with the specific accounts attached. Not as discipline — as feedback, which is the thing this job is structurally denied.

Chapter 29 §29.9 is about making this argument to an organization that does not want to hear it, and the reason it needs its own section is that the argument fails on the first three attempts.

24.11 🗂️ The Encounter — rewind to check-in

Account 10-4471, Tuesday March 14, before the visit.

(Constructed, and consistent with everything frozen since Chapter 6.)

What the front desk did

Verified eligibility for the date of service. The 271 returned the response in §24.4: active, Northfield Mutual PPO, \$30.00 office visit copay**, **\$1,500.00 deductible met February 28, 20% coinsurance, \$4,000.00 out-of-pocket maximum.

Asked whether anything had changed. Nothing had.

Determined that no referral was required. Northfield Mutual's PPO does not require PCP direction, and the patient was seeing her own primary care physician in any case.

Determined that no prior authorization was required for an established-patient office visit.

Collected the \$30.00 copay at check-in.

All of that is correct, and it is better than a great many practices manage. The deductible was met, which is the fact that makes the rest of this encounter's arithmetic work — Chapter 2's four numbers depend on it.

The one thing it missed

Nothing on the schedule said a procedure might happen.

The appointment was booked as a follow-up for chronic conditions. The knee complaint was six weeks old and worsening, and the patient intended to raise it — but nobody asked, and the scheduling note said nothing.

So no check was made for whether 20610 required prior authorization under Northfield Mutual's policy. (It did not. That is luck, not process.)

And the copay collected was \$30.00, which is what an office visit costs. The patient's actual responsibility for the encounter was \$47.58 — the copay plus 20% coinsurance on the injection, the drug, and the venipuncture.

   COLLECTED AT CHECK-IN ........... $30.00
   ACTUAL PATIENT RESPONSIBILITY ... $47.58
                                     ────────
   BILLED AFTER THE FACT ........... $17.58

Which generated a patient statement on day 70 — the account's 100-day timeline, mapped in Chapter 1 — showing \$47.58 of responsibility, the \$30.00 already collected, and a balance due of \$17.58 — and, after the appeal resolved on day 66, a balance the patient paid in full on day 100. Chapter 28 §28.11 posts it.

What a better front end would have done

Not "collected \$47.58." Nobody could have known the injection would happen.

Three things, all cheap:

Asked what the visit is for. "Anything else you'd like the doctor to look at today?" — at scheduling or at check-in. The knee would have surfaced, and a screening question about procedures becomes possible.

Screened for authorization on the procedure that might result. Ten seconds against a published list. It would have returned "not required," and the practice would have known that rather than assumed it.

And framed the copay as a copay. "Your copay today is \$30.00. If the doctor does anything beyond the visit itself, there may be additional coinsurance and you'll see that on your statement." One sentence, and the day-70 statement is expected rather than a surprise.

The front desk did nothing wrong and the encounter still produced a surprise.

That is the honest finding, and it is the chapter's argument in miniature: the front end's failures are usually not errors. They are questions that were not asked, and the reason they were not asked is that nobody had written them down.

Summary

The front end is the cheapest place to fix a claim, by roughly two orders of magnitude: half a minute at registration versus something on the order of forty-five at the back — roughly ninety to one — and the entire allowed amount once timely filing runs out.

The registration fields that cause denials: name exactly as on the card · date of birth · member identification number (copy it; watch O/0, I/1, S/5, B/8) · subscriber relationship · group and plan · address and guarantor (a wrong one produces no denial and an uninformed patient) · accident and injury information.

And the most expensive field is the one nobody thinks of as a field: "has anything changed?" A returning patient's record is complete — for a situation that no longer exists. January is the month.

And one front-desk duty is not about billing at all: verify that the person is the person on the coverage. Medical identity theft contaminates a chart permanently, and the reason for the identity step is the record, not the claim.

The 270/271 answers eligibility. It cannot tell you: that coverage will still exist on the date of service · a current deductible (the figure lags processed claims) · whether the service is medically necessary · anything a payer chose not to populate.

Verify three times, not once: at scheduling (the only point the encounter can still be rearranged) · 48–72 hours before, in batch (early enough to act, late enough to still be true) · day of service, in real time.

"Other coverage: none indicated" is a statement about the payer's records, which is why §24.7's questionnaire exists.

Referral = who the patient sees, issued by the PCP. Prior authorization = a payer approving a specific service. Precertification = usually a synonym; ask what the payer means. Predetermination = non-binding, and not an authorization.

RECORD AN AUTHORIZATION'S SCOPE, NOT JUST ITS NUMBER.

Number · date · issuer · codes · date range · units · facility · rendering provider. An authorization number recorded without its scope is a number, not a record.

The MSPQ determines whether Medicare is secondary — employment-based coverage, workers' compensation, auto or liability, black lung, VA, ESRD coordination. The eligibility response is not sufficient, and billing Medicare as primary when it is not is an overpayment.

Point-of-service collection: ask about method, not permission. Never in front of other patients, never insist when someone cannot, never quote a number you cannot stand behind. Routine waiver of cost-sharing is prohibited; individualized hardship determination under a written policy is not.

Financial clearance is a nine-item checklist, and "medical necessity screened" is the highest-value item almost nobody does.

An estimate is built from the ALLOWED AMOUNT, not the charge — Chapter 23 §23.1 — and it cannot know the deductible exactly, the codes that do not exist yet, or how the claim will adjudicate. The sentence matters more than the number, and a good sentence names the specific thing that could move.

When one practice orders and another performs, write down who obtains the authorization — per payer, per service type — and whoever obtains it must pass on the SCOPE, not the number. Two offices each believing the other is handling it produces an unauthorized service and two organizations certain they were right.

Measure: clean claim rate · front-end denial rate (which requires Chapter 29 §29.4's classification) · eligibility verification rate for the date of service · authorization capture rate with scope · point-of-service collection rate.

The front end is measured by the back end's failures and staffed by the front end's budget.

The fix is not a new report — it is routing the existing one, weekly, with accounts attached, as feedback rather than discipline.

Account 10-4471's front desk verified eligibility, asked what had changed, correctly determined no referral or authorization was needed, and collected the \$30.00 copay. It did not ask what the visit was for. The knee was not on the schedule, no procedure screening was possible, and the patient's actual responsibility was \$47.58** — leaving **\$17.58 to be billed and a statement to arrive on day 70.

The front desk did nothing wrong and the encounter still produced a surprise. The front end's failures are usually not errors — they are questions nobody had written down.

Key Terms

Patient access — the front-end function comprising scheduling, registration, verification, authorization, and financial clearance. (Ch.24)

Registration — capture of the patient's demographic, coverage, and guarantor information. (Ch.24)

Eligibility verification — confirming with the payer that coverage is active for the date of service. (Ch.24)

270/271 transaction — the standard electronic eligibility and benefit inquiry and its response. (Ch.24)

Coordination of benefits — the determination of which payer is primary when more than one covers the patient. (Ch.24)

Prior authorization — a payer's advance approval of a specific service, based on clinical information. (Ch.24)

Referral — a primary care physician's direction of a patient to a specialist; about who, not what. (Ch.24)

Precertification — generally a synonym for prior authorization; in some plans a distinct admission notification requirement. (Ch.24)

Predetermination — a non-binding advance review of whether a service would be covered. Not an authorization. (Ch.24)

Medical necessity screening — checking, before the service, whether the ordered diagnosis supports the service under the governing coverage policy. (Ch.24)

Point-of-service collection — collection of patient responsibility at or before the encounter. (Ch.24)

Financial clearance — the determination that all front-end requirements for a billable encounter have been satisfied. (Ch.24)

Medicare Secondary Payer questionnaire (MSPQ) — the standard instrument for determining whether another payer is primary to Medicare. (Ch.24)

Clean claim rate — the percentage of claims accepted on first submission without rejection or denial. (Ch.24)

Front-end denial rate — denials attributable to eligibility, registration, authorization, or coordination of benefits. (Ch.24)

Predetermination — a non-binding advance review of whether a service would be covered. (Ch.24)

Batch versus real-time eligibility — the whole schedule submitted at once and returned as a file, versus a single patient answered in seconds at the desk. Most practices need both. (Ch.24)

Medical identity theft — use of another person's identity to obtain care or coverage; contaminates the victim's medical record and is a patient safety matter before it is a revenue matter. (Ch.24)

Spaced Review

1. A transposed digit in a member identification number is caught at four different points. Give the approximate cost at each, and name what is lost at the fifth.

2. Name six registration fields that cause denials. Which one produces no denial and what does it produce instead?

3. What is the most expensive registration field, and why does it produce no warning? Which month matters most?

4. Name four things a 270/271 cannot tell you. Which one produces estimates that ask patients for too much?

5. "Other coverage: none indicated." What does that sentence actually assert, and which section exists because of it?

6. Distinguish a referral, a prior authorization, a precertification, and a predetermination.

7. (Chapter 22) Name the eight things to record about an authorization. Which four correspond to the four drifts in Chapter 22's Case Study 1?

8. Why is an eligibility response not sufficient for coordination of benefits? What is the consequence of billing Medicare as primary when it is not?

9. Rewrite "Would you like to pay your copay today?" and explain why your version works.

10. What may a practice do about patient cost-sharing, and what may it not? What word is doing the work in the prohibition?

11. Account 10-4471's front desk did five things correctly and missed one. Name the miss, and say why the chapter calls it a question rather than an error.

12. Name the three points at which eligibility should be verified and what each one catches. Which is the only point at which the encounter can still be rearranged?

13. Why is identity verification a patient safety matter before it is a revenue matter? Name the consequence that cannot be unwound.

14. Build an estimate: 20610 allowed \$78.60, J1030 allowed \$6.28, deductible met, 20% coinsurance. Then state what the figure becomes if the deductible is not actually met, and explain why the eligibility response might have said it was.

15. A PCP orders advanced imaging performed at an imaging center. Name three principles that decide who obtains the authorization, and say what must be communicated along with the number.