.png)
Brazilian health insurers operate under historically high loss ratios that pressure margins and force premium increases the insurers themselves recognize as unsustainable in the long term.¹ Much of the public debate concentrates on two axes: judicialization and aging of the insured population. There is a third axis, less visible and operationally addressable, that rarely enters the agenda with the depth it deserves: the systematic payment of duplicate laboratory tests that the insurer cannot identify before settlement.
Laboratory duplication is not fraud. It is a direct consequence of the data architecture connecting the insurer and its credentialed providers.
The operational scenario where duplication is born
A mid-sized insurer typically credentials hundreds of laboratories and clinics in its network. Each provider operates its own Laboratory Information System, with local nomenclatures, proprietary codifications, and nomenclature variations between departments. When a beneficiary undergoes a complete blood count at facility A on Monday and another at facility B on Wednesday, two TISS files arrive at the insurer at different moments, from different providers, with identifiers that frequently do not allow automatic recognition that this is the same procedure on the same patient within a clinically close window.
The audit rule should be straightforward: a CBC performed within an interval shorter than clinically justifiable is eligible for denial by the insurer, except where medical justification is documented. In practice, that rule can only be executed automatically if the insurer can compare in real time the procedure being billed by facility B with the history of procedures already paid for the same beneficiary. This comparison depends on two simultaneous requirements that most insurers do not meet: semantic identity between procedures recorded by both facilities, and availability of the beneficiary's longitudinal history in a comparable structure.
When the insurer lacks this infrastructure, duplication auditing becomes a manual sampling process. Technical audit teams review a fraction of claims, identify some obvious cases, deny what they can document, and the rest goes through to payment.
The scale of the problem
The literature on health interoperability documents that fragmentation of data between providers is one of the main causes of unnecessarily repeated tests. Variations of just 10% in reference values between laboratories can generate unnecessary follow-up tests in up to 30% of cases.² That percentage measures repetitions driven by physician interpretive uncertainty, not pure duplicates, but the vector is the same: lack of comparability between data produced by different providers turns already-performed tests into clinically unusable information, which both leads the physician to order the test again and leads the insurer to pay for the second one.
Brazil recorded more than 2.5 billion laboratory tests in 2024, of which 46.4% were performed in the private network.³ For an insurer with one million beneficiaries, this represents a monthly volume of laboratory procedures in the hundreds of thousands. Even a low percentage of undetected duplication, applied to this base, produces relevant financial impact. The question is not whether duplication exists. It is how much of it passes through auditing without being identified, and why.
What's missing for automated auditing to work
Automated duplication auditing systems exist in mature insurers, but operate with structural limitations when input data is not harmonized. The detection algorithm needs to answer an apparently simple question: is procedure X billed by facility B today the same as procedure Y already paid to facility A last week? If both facilities sent identical TISS codes and equivalent nomenclatures, the system recognizes the match and triggers the duplication rule. If one facility sent "Complete Blood Count" under the official TISS code and the other sent "Hemogram with platelets" under a proprietary table code mapped inconsistently, the system treats them as two distinct procedures.
Lack of clinical data interoperability is frequently identified as one of the largest sources of operational inefficiency in mid-sized health organizations.⁴ In duplication auditing, this inefficiency materializes on three fronts: payments that should have been denied before settlement, audit team rework after payment (trying to recover funds already transferred to providers), and the inability to identify systemic duplication patterns at specific providers.
The regulatory consequence is also relevant. The TISS Technical Standard, established by ANS, requires insurers to process provider information in structured and auditable form.⁵ The insurer that cannot audit duplication with precision is not only paying more. It is operating with reduced margin of compliance with technical audit requirements that ANS itself expects to see implemented.
The premise shift: the insurer as data integrator
Most insurers operate under an implicit premise: data quality responsibility belongs to the provider. Each laboratory sends its data in whatever format it operates, the insurer receives, processes, and pays. When there is inconsistency, it's the provider's problem.
This premise worked while data volume was manually processable and while duplication was the exception. In a contemporary operation, with hundreds of credentialed providers and millions of monthly procedures, it has stopped making economic sense. The insurer that continues treating data received from the provider as pre-validated truth is paying, in undetected duplications and audit rework, more than it would cost to build an internal layer that harmonizes data on intake.
The premise inversion is direct: the insurer becomes the data integrator of its credentialed network, semantically harmonizing procedures recorded by different providers before processing for payment. This does not require each provider to change its system. It requires the insurer to operate a harmonization layer that recognizes that "Complete Blood Count" in provider A's LIS and "Hemogram with platelets" in provider B's LIS refer to the same procedure, with verifiable semantic identity mapped to the corresponding TISS code.
What changes when the harmonization layer works
Insurers that implement semantic harmonization layers before settlement report measurable gains on three fronts. Duplication auditing starts operating on 100% of claims, not on sampling. Duplication patterns at specific providers become visible on management dashboards, enabling contractual negotiation based on data rather than suspicion. And the beneficiary's longitudinal database, harmonized across time, starts feeding population health management programs with quality that was previously inaccessible.
This last point deserves attention. The insurer that harmonizes laboratory data of its beneficiaries doesn't just reduce undue payments. It builds an informational asset that sustains actuarial risk management, chronic disease prevention programs, and population stratification. The same investment that solves the duplication problem solves, in parallel, the data foundation the insurer needs for every health management initiative that depends on longitudinal clinical data.
This is the context in which OpenHealth Technologies operates. The platform automatically correlates multiple data streams with rigorously validated logical layers of laboratory tests, receiving TISS files and clinical data from any credentialed provider and delivering to the insurer a harmonized, semantically equivalent representation of those procedures. For duplication auditing, this means the comparison between what is being billed and what has already been paid operates on truly comparable data, in real time, before settlement. For population management, it means each beneficiary's longitudinal history exists in structured and actionable form, regardless of how many different providers in the network contributed data over time.
Learn how your insurer can transform fragmented credentialed-network data into automated auditing and population management based on reliable data.