When Health Care Is Precise About Prices but Careless About People

What if the most consequential failures in American health care are not failures of medicine, but failures of administration?

A drug can be priced to six decimal places under the 340B program, while a person can lose health coverage because a letter was sent to an old address. One system is engineered for microscopic financial precision. The other is so brittle that a missing phone number can outweigh an ongoing medical need.

That contrast reveals a deeper problem: health care has become highly exact where money is measured, but remarkably imprecise where human eligibility is concerned.

The central question is not simply whether public programs spend too much or whether pharmaceutical prices are too high. It is this: What does a health system owe people when its rules are exact, but the information used to apply those rules is incomplete?

The strange asymmetry of precision

The 340B drug pricing program is built around a formula. For a covered outpatient drug, the ceiling price is derived from the Average Manufacturer Price from the preceding calendar quarter, measured at the smallest unit of measure, minus the Unit Rebate Amount. The calculation uses six decimal places, even though the price ultimately published by the Health Resources and Services Administration is rounded to two.

This is not needless mathematical theater. Small pricing differences, multiplied across enormous volumes of medication, can become significant. A fraction of a cent can matter when a hospital, clinic, or manufacturer handles thousands or millions of units. The system therefore treats precision as a form of fairness. If the price is governed by a formula, the inputs and calculations must be disciplined.

Now compare that design with Medicaid eligibility redeterminations. During the nationwide review of enrollment, large numbers of people lost coverage for procedural reasons. In some states, more than 80 percent of those removed from the rolls were disenrolled because of administrative issues, including outdated contact information, rather than a confirmed finding that they were no longer eligible.

The contrast is more than ironic. It exposes a hierarchy of institutional attention.

When a price is at stake, the system asks: What is the correct number? Which quarter applies? What is the smallest unit? How should the result be rounded?

When a person is at stake, the system may ask only: Was the form returned? Was the address current? Did the notice reach anyone?

A system can be financially exact and socially careless at the same time.

This is the hidden tension connecting drug pricing and Medicaid administration. Both are attempts to convert a complicated social reality into a manageable rule. But one is designed to protect the integrity of a transaction, while the other often treats an administrative record as a proxy for a human being.

The proxy problem: when paperwork stands in for reality

Every large public program needs proxies. Agencies cannot personally investigate every household, every prescription, or every change in circumstance. They use forms, databases, deadlines, formulas, and notices to stand in for facts that are otherwise expensive to collect.

The danger begins when the proxy is mistaken for the reality.

An outdated address is not evidence that a person is ineligible. It is evidence that the system does not know how to reach the person. A missing document is not necessarily evidence that a household has too much income. It may indicate confusion, limited internet access, unstable housing, language barriers, illness, or a notice that arrived after the deadline.

Yet procedural systems often collapse these distinctions. They convert “we do not have enough information” into “the person does not qualify.” This is an epistemic error, not merely a customer service flaw. The system is claiming knowledge it does not possess.

The same issue appears in pricing, though in a different form. A 340B ceiling price is not the market value of a drug in every setting. It is a legally defined result generated from specified inputs. Its legitimacy depends on maintaining the connection between the formula and the purpose of the statute: limiting the amount paid by certain covered entities for covered outpatient drugs.

In both cases, the rules are only as trustworthy as the relationship between the measurement and the thing being measured.

For drug pricing, the relevant question is whether the formula accurately produces the legally intended ceiling price. For eligibility, the relevant question is whether a procedural event accurately tells us that a person no longer qualifies. The first system has been designed to interrogate its inputs. The second often treats weak inputs as final answers.

This suggests a useful framework for evaluating administrative systems: the evidence burden should match the consequence.

If a minor data discrepancy changes a payment by a few cents, a recalculation may be sufficient. If a procedural discrepancy removes a person’s access to insulin, prenatal care, psychiatric treatment, or a specialist, the system should require stronger confirmation and provide multiple chances to correct the record.

The larger the human consequence, the less defensible it is to rely on a fragile proxy.

Why outsourcing is tempting, and why it is not enough

One proposed response to administrative failures is to contract with large health insurers and managed care companies. These firms already operate claims systems, call centers, provider networks, and eligibility technologies. At first glance, outsourcing appears practical. Why build a new administrative infrastructure when companies with enormous scale already have one?

The temptation is understandable, but scale is not the same as alignment.

A large insurer may be capable of processing millions of transactions. That does not prove it will design the process around preserving eligible people’s coverage. Administrative efficiency can mean faster resolutions, but it can also mean faster termination, fewer human reviews, and more aggressive automation. The central question is not who can process the most cases. It is which party bears the cost when the system is wrong.

This is an incentive problem.

For a manufacturer, a lower 340B ceiling price may reduce revenue on a covered transaction, even as the program serves a statutory goal. For a state administering Medicaid, reducing enrollment can lower immediate program spending, even when some disenrollments are temporary, erroneous, or caused by paperwork. For an insurer paid through a contract, the financial consequences may depend on the terms of the agreement, performance measures, and how responsibility is assigned.

If the contractor is rewarded primarily for reducing administrative cost, then “successful” automation may produce fewer cases, not better decisions. If it is rewarded for accurate continuity of coverage, rapid reinstatement, and successful contact with hard to reach households, the same technological capacity can be used very differently.

The lesson is not that private contractors can never help. The lesson is that administrative capacity must be governed by public purpose.

A contractor should not be judged only by call volume, processing speed, or the number of files closed. It should also be evaluated by measures such as:

• The share of people reached through more than one communication channel.
• The percentage of procedural terminations reversed after a person responds.
• The time required to restore coverage when eligibility is confirmed.
• The number of people who experience an avoidable gap in treatment.
• The rate at which automated decisions receive meaningful human review.

These measures change the definition of efficiency. A process that closes a case quickly but leaves a patient without medication is not efficient from the patient’s perspective. It has merely transferred the cost to an emergency department, a clinic, a pharmacy, or the patient’s health.

Build a system that distinguishes uncertainty from ineligibility

The most important redesign principle is simple: do not treat administrative silence as a substantive answer.

A person who fails to return a form should enter an uncertainty pathway, not automatically an ineligibility pathway. That pathway could include text messages, phone calls, outreach through trusted clinics, pharmacy notifications, community organizations, and real time opportunities to update an address. It should be easy to correct a record before coverage ends, and even easier to restore coverage when the person is found eligible.

This resembles a safety system in aviation or engineering. A warning light does not automatically mean the aircraft has failed. It means the system has detected an anomaly that requires investigation. Health coverage administration should adopt the same logic.

An outdated address is an anomaly. A missing form is an anomaly. A mismatch between databases is an anomaly. None should be treated as conclusive proof that the underlying person has changed status.

A practical model would classify cases into three levels:

1. Confirmed eligibility: The available evidence supports continued coverage.
2. Confirmed ineligibility: Reliable evidence shows that the person no longer meets the rules.
3. Unresolved status: The system lacks enough information to decide.

The third category is where current systems often fail. They erase it by forcing unresolved cases into the second category. That may simplify a dashboard, but it corrupts the meaning of the data.

The 340B framework offers a useful institutional lesson here. Its calculation specifies inputs, units, timing, and rounding. Administrative eligibility systems should be equally explicit about uncertainty. They should record not only the decision, but also the strength of the evidence behind it. A termination based on verified income is not equivalent to a termination based on an undeliverable letter, and the system should never pretend otherwise.

This is also where continuity of care matters. Coverage is not merely a legal status that begins and ends at a database timestamp. It is an infrastructure that allows a patient to keep taking medication, return to a clinician, manage a chronic condition, and seek help before a crisis. A short interruption can create costs that exceed any administrative savings, especially for people with diabetes, cancer, kidney disease, or serious mental illness.

The true unit of efficiency is therefore not the processed case. It is the protected episode of care.

Key Takeaways

• Match evidence to consequence. Require stronger confirmation before terminating benefits than before correcting a minor administrative record.
• Separate uncertainty from ineligibility. An outdated address or missing form should trigger outreach and review, not serve as automatic proof that coverage should end.
• Measure continuity, not just throughput. Track reinstatement rates, treatment interruptions, successful contact, and avoidable coverage gaps alongside processing speed.
• Align contractor incentives with public purpose. If large insurers or technology vendors are involved, pay for accurate decisions and preserved access, not simply fewer open files.
• Design for correction. Make address updates, document submission, appeals, and rapid reinstatement available through multiple channels, including clinics, pharmacies, phone, text, and in person assistance.

The real price of administrative precision

Health policy debates often focus on visible numbers: the price of a drug, the size of a program, the cost of a contract, or the number of people enrolled. But the deeper issue is how institutions decide what counts as a fact.

The 340B program demonstrates that a public rule can use detailed formulas to constrain a price while pursuing a broader social purpose. Medicaid redeterminations demonstrate what happens when a social purpose is subordinated to a thin administrative signal. In one case, precision protects the program’s financial architecture. In the other, procedural simplification can quietly undermine the program’s human architecture.

The answer is not to eliminate rules, formulas, or technology. Large systems need all three. The answer is to apply precision where it matters most: not only to the calculation of what an institution pays, but also to the determination of what a person needs and what the evidence actually shows.

The most humane health system is not the one with no bureaucracy. It is the one that knows the difference between a missing record and a missing need.

That distinction should govern every redesign, whether the proposal involves a federal agency, a state office, a hospital, a pharmacy, or one of the country’s largest insurers. The question is never merely whether the system can process more information. It is whether the system can remain honest about uncertainty while protecting people from the consequences of being misclassified.

A drug price can be rounded from six decimal places to two because the rounding rule is known and the effect is bounded. A person’s health cannot be rounded that way. When the cost of error is a missed treatment, an untreated illness, or a preventable crisis, administrative convenience is not neutral. It is a decision to make the vulnerable absorb the system’s uncertainty.

The future of health care administration should therefore be judged by a different standard: not how cleanly it closes the file, but how reliably it keeps the right person connected to care.