TL;DR: Taiwan’s NHI maps thirty major Catastrophic Illness categories to five certificate duration tiers: permanent, five-year, three-year, one-year, and case-by-case. In practice, the case-by-case tier also includes shorter durations, three months, six months, and others, which this article groups together for clarity. The design is not merely administrative. It reflects thirty years of actuarial judgment about prognosis and long-term cost. Each tier corresponds to a specific kind of status assessment: whether the condition is reversible, whether it requires lifelong monitoring, whether a cure is possible, and whether the patient is in a recovery window following an acute event. Viewed from the AI era, this structure looks remarkably like an early local prototype for tiered reimbursement.

▶ Listen to Summary
AI-synthesized voice · cloned from the author's own

As I mentioned in the previous piece, I first opened the NHI Administration’s Catastrophic Illness page while preparing materials for a Japanese partner organization. That page lists thirty major Catastrophic Illness categories, each with a column indicating the certificate’s validity period.

Look closely and a striking contrast appears.

Rare diseases: permanent. Most congenital metabolic disorders: permanent. Post-organ-transplant follow-up treatment (excluding bone marrow): permanent. Motor neuron disease: permanent. Aplastic anemia, liver cirrhosis with severe complications: five years. Thyroid cancer, Stage I breast cancer, Stage I cervical cancer, Stage I oral cancer: three years. Myasthenia gravis: three years. Burns covering twenty percent or more of total body surface area: one year. Acute cerebrovascular disease: per NHI regulations, physician certification must be completed within one month of acute onset.

Same system. Same regulatory framework. Yet the certificate durations differ enormously. Why?

The short answer: rare diseases are generally treated by the system as long-term, irreversible conditions requiring lifelong care, so they receive permanent status. Cancers, by contrast, vary widely in prognosis across types and stages, and with treatment advancing rapidly, some early-stage cancers can remain stable for many years after therapy. The system therefore uses three-year or five-year intervals as a rhythm for re-confirmation.

This article tries to answer the question more fully. The design logic behind these durations is deeper than it first appears, and it is worth examining through an AI-era lens. This tiered duration system, which has been operating for thirty years, already carries the early architecture of tiered reimbursement, and it resonates directly with the actuarial, individualized reimbursement trends now emerging in the AI era.


The Tiers Reflect Actuarial Judgment, Not Just Administrative Convenience

A first glance at this comparison table tempts most people to read it as an administrative convenience or a straightforward resource-allocation tradeoff. In practice, the duration design can be understood as one of the accumulated actuarial outcomes of thirty years of NHI institutional development. Actuarial here means more than financial calculation. It encompasses a composite judgment about disease prognosis, long-term treatment cost, and the appropriate frequency of re-assessment. Behind that single table are multiple simultaneous considerations: whether a condition is reversible, how quickly medical technology is advancing, and how the institution should weigh all of these together.

A tiered design is necessary precisely because the system must answer several questions at once. Will this diagnosis change over time? Does the patient require lifelong medication or monitoring? Is there any possibility of exiting Catastrophic Illness status? Might a condition today classified as severe become curable or manageable within five years, given medical progress?

Different diseases give different answers to those questions. The same disease may give different answers after time passes. The tiered duration structure is the institutional mechanism for making this “status assessment” concrete, matching each disease to a reassessment rhythm that fits its clinical reality.

Five Duration Tiers, Five Status Assessments

Setting aside the fine print of specific regulations, and resorting the thirty major categories by certificate duration from a prognosis perspective, a structure of roughly five status assessments emerges. The NHI Administration’s published tables also include shorter durations, three months, six months, which this article folds into the case-by-case tier for simplicity.

Permanent covers conditions assessed as irreversible and requiring lifelong monitoring. This includes rare diseases (designated under the Rare Disease and Orphan Drug Act), most congenital metabolic disorders, motor neuron disease, chromosomal abnormalities, post-organ-transplant follow-up (excluding bone marrow), and systemic autoimmune syndromes requiring lifelong treatment. What these share is that once the diagnosis is established, the patient’s condition typically does not return to “normal.” Current medical technology generally does not frame them in terms of straightforward cure. Granting permanent status is the institution’s acknowledgment of their chronicity and irreversibility.

Five years covers conditions where treatment is effective but long-term monitoring remains necessary: aplastic anemia, liver cirrhosis with severe complications, post-bone-marrow-transplant follow-up, and congenital immunodeficiency, among others. These conditions may be relatively stable but still leave room for change, which is why periodic reassessment is warranted. The five-year interval is not purely administrative. It shares a comparable time horizon with the five-year survival rate, a clinical benchmark commonly used to assess long-term prognosis, though this does not mean the two are institutionally equivalent.

Three years applies mainly to early-stage cancers with high cure rates: thyroid cancer, Stage I breast cancer, Stage I cervical cancer, Stage I oral/oropharyngeal/hypopharyngeal cancer. If no recurrence appears within three years of treatment, the system can provisionally judge the treatment outcome as stable. This tier reflects how the institution uses time to observe whether treatment has durably succeeded.

One year covers the recovery window following acute events: burns covering twenty percent or more of total body surface area, and initial major trauma (Injury Severity Score ≥ 16). These situations have a clear starting point, and a reasonable initial prognosis assessment is possible within roughly one year of recovery.

Case-by-case applies to highly dynamic medical situations where validity is adjusted progressively as the patient’s condition evolves. For chronic renal failure patients requiring dialysis, the initial duration is only a few months while long-term dialysis has not yet been confirmed, requiring frequent reassessment; once long-term dialysis is established, the certificate converts to permanent. Patients requiring long-term ventilator use follow a similar logic: the initial certification lasts 42 days, the second is three months, and from the third certification onward the duration extends to one year, progressively relaxed as the duration of use lengthens. Acute cerebrovascular disease falls into a distinct acute-onset certification window: under current NHI regulations, a physician must certify Catastrophic Illness status within one month of acute onset, allowing the patient to receive co-payment exemption without first obtaining a certificate card. Actual durations and certification procedures should always be verified against the NHI Administration’s latest announcements and physician judgment. What all these situations share is rapid change in patient condition. A single fixed duration is institutionally inappropriate, so the design preserves space for physician judgment at the bedside, while allowing short initial durations to transition into longer or permanent status over time.

Five tiers. Five status assessments. Taken together, they constitute more than a resource-allocation rule. They form a disease classification system with built-in dynamic evaluation and actuarial logic.

A Taiwanese Local Design Built on Top of ICD-10

As I wrote in Part 1, my first encounter with ICD-10 was not through Wikipedia. I studied life sciences in college, and earlier had trained as a medical laboratory scientist. In clinical systems back then, this disease classification code was already running in the background.

The same coding system later entered daily life at home. A family member once went through the Catastrophic Illness system because of cancer, renewing the certificate every three years, waiting for follow-up results at each visit. That rhythm of waiting, I watched it, and I lived alongside it. So when I opened that NHI page, what I saw was not only a thirty-item comparison table. I also saw the time that accumulates between renewals, the time that never gets written into the table itself.

But ICD-10 on its own says nothing about how long a certificate should be valid under Taiwan’s Catastrophic Illness system. ICD-10 is the World Health Organization’s international disease classification: globally standardized, it describes how diseases are categorized in clinical medicine. Taiwan’s NHI duration tiers take that international coding framework and bind it to a local reimbursement system, then distribute different diagnoses, based on whether they are reversible, whether monitoring is required, whether cure or stable control is possible, across the permanent, five-year, three-year, one-year, and case-by-case tiers. NHI Administration announcements show that whenever ICD-10 is updated (as with the 2023 ICD-10-CM/PCS revision), the diagnostic code ranges in the Catastrophic Illness comparison tables are also adjusted accordingly.

To make this mapping quickly searchable, I compiled the thirty major categories, together with their ICD-10 codes, certificate durations, and sub-item breakdowns, into a structured tool page available in four languages. The page is based on the “Catastrophic Illness Categories and Certificate Validity Periods” published by the NHI Administration on September 16, 2024 (Republic of China Year 113), effective January 1, 2025, and notes the NHI Administration’s subsequent update to the rare disease diagnostic code table on April 28, 2025 (Year 114). For actual application, the NHI Administration’s latest announcements remain authoritative. That tool page makes visible how a single ICD-10 code, once it enters Taiwan’s NHI system, is further divided into major categories, sub-items, and multiple durations, at a level of granularity that far exceeds what the international coding standard itself expresses.

In other words, the thirty categories are not a straightforward ICD-10 classification. They are ICD-10 as filtered, processed, and stratified by Taiwan’s NHI institution. This institutional judgment has been continuously updated across successive revisions. The NHI Administration website shows amendment records for the “Catastrophic Illness Categories and Certificate Validity Periods” from at least 2015 (Year 104), 2019 (Year 108), 2022 (Year 111), and 2024 (Year 113). As medicine advances, some conditions previously assigned five-year durations may shift to permanent; others that were permanent may shift to periodic renewal as new treatments emerge. Cancer provides a clear example: the NHI has reclassified some previously permanent cancer categories to three-year or five-year renewal cycles, reflecting the system’s reassessment of its own rhythm as cure rates and five-year survival rates improve.

When I look at this system, I think of it as a classifier that has been continuously updated and fine-tuned over time. Its inputs are diagnosis, disease course, and individual patient status; its outputs are certificate duration, reimbursement eligibility, and certification scope. The system was not built with AI, but it already possesses several core characteristics that data systems in the AI era require: well-defined inputs, traceable classification rules, versioned updates, and outputs that connect to reimbursement outcomes. Over thirty years of institutional adjustment, this classifier has accumulated remarkably detailed judgment rules.

Why This System Is an Early Prototype for AI-Era Tiered Reimbursement

Viewing the system as a classifier reveals something else.

Health insurance systems worldwide are grappling with a shared challenge: when medical resources are finite and patient needs are increasingly individualized, how should reimbursement be allocated? The traditional approach favors one diagnosis, one reimbursement level. This served reasonably well in the twentieth century, but as precision medicine, gene therapy, cell therapy, and AI-assisted imaging diagnosis proliferate, the limits of uniform reimbursement grow harder to ignore.

The U.S. Medicare Value-Based Payment programs, the UK NHS’s NICE evaluation framework, and Japan’s Designated Intractable Disease Medical Expense Subsidy system are each exploring different forms of case-level, outcome-based, risk-stratified reimbursement. All of these trends share a common technical precondition: the system itself needs a disease classification and status assessment infrastructure that is granular enough, machine-readable, and continuously updatable.

Looking back now, Taiwan’s tiered duration system already possesses many of the elements that a tiered reimbursement regime requires: clear categorization, duration design, dynamic certification, versioned updates, and rules linked to reimbursement eligibility. It has operated for thirty years, is anchored to ICD-10 coding, and has accumulated a dynamic case-by-case evaluation mechanism. It was not designed for AI, but it is well-suited to be reread through an AI lens. Taiwan’s distance from AI-era tiered reimbursement may not lie in the institutional logic itself. What is missing is the work of re-interfacing and re-structuring that existing logic so it can become a data foundation that algorithms can draw on in real time.

This circles back to the point from Part 1: the data is already there. The interface layer is the next evolution.

An Administrative Table That Is, in Fact, a Policy Asset

This article does not set out to render a verdict on any particular system design or compare their merits. Contemporary health insurance systems are all still evolving. Taiwan’s thirty categories and Japan’s designated intractable disease framework are neither complete nor final. Both are institutional designs still under active adjustment.

What I want to do is restore to a frequently dismissed “administrative table” the policy context it deserves. This is an actuarial classification system that has been operating for thirty years. From an institutional design perspective, it can also be read as one of the earliest local prototypes for tiered reimbursement in the AI era. If global medical reimbursement moves over the next decade toward precision, individualization, and AI-assisted decision-making, then local designs like Taiwan’s Catastrophic Illness duration tiers are very much worth re-examining, re-interfacing, and placing on the international table.

That is part of what this Regenerative Medicine & Technology series is working toward, piece by piece.

Part 1 turned a single ODT page into a searchable tool. This piece has translated a comparison table into policy logic that is somewhat easier to follow. The institutional layer, the industry layer, and the technical layer each still need their own treatment, and they are coming.


Glossary

  • ICD-10-CM/PCS: The disease diagnosis and procedure coding foundation currently used by Taiwan’s NHI. ICD-10 is the International Classification of Diseases published by the World Health Organization; ICD-10-CM is a clinical modification of ICD-10 maintained primarily by the U.S. NCHS/CMS; ICD-10-PCS classifies inpatient procedures and is published by U.S. CMS. Taiwan’s NHI adopted and localized the 2023 version of ICD-10-CM/PCS. For a detailed introduction, see the Part 1 Glossary.
  • Catastrophic Illness: The thirty major illness categories designated in Appendix 1, Article 2 of the Regulations Governing Exemption from Co-Payment for NHI Beneficiaries. Patients who obtain a Catastrophic Illness certificate are exempt from co-payments for related treatment.
  • Rare Disease: A disease designated and publicly announced under Article 3, Paragraph 1 of the Rare Disease and Orphan Drug Act. Rare diseases are automatically listed as Category 30 of the Catastrophic Illness categories, with a permanent certificate duration.
  • ISS (Injury Severity Score): An internationally standardized trauma assessment tool. The body is divided into six regions (head and neck, face, chest, abdomen, extremities, and external soft tissue); the three highest AIS (Abbreviated Injury Scale) scores across those regions are each squared and summed. Higher scores indicate greater trauma severity; a score of 16 or above defines “major trauma.” Taiwan’s NHI uses ISS ≥ 16 as the qualifying threshold for Catastrophic Illness Category 12.
  • Value-Based Payment (VBP): A reimbursement model contrasting with traditional fee-for-service, in which payment levels are determined by treatment outcomes, patient prognosis, and degree of health improvement, among other indicators. U.S. Medicare has been progressively implementing VBP since the 2010s.
  • NICE (National Institute for Health and Care Excellence): The UK body responsible for evaluating the cost-effectiveness of new drugs, treatments, and medical technologies. Its assessments serve as a key input for NHS coverage decisions.
  • Designated Intractable Disease Medical Expense Subsidy Certificate (指定難病患者への医療費助成制度): Japan’s Ministry of Health, Labour and Welfare issues these certificates for designated intractable diseases (a concept similar to Taiwan’s rare disease designation), with tiered subsidies based on disease severity and income bracket. Renewal is generally required annually. Prior to the enforcement of the Act on Medical Care for Patients with Intractable Diseases in January 2015, these were known as Specified Disease Medical Expense Subsidy Certificates (特定疾患医療受給者証).
  • Tiered Reimbursement: An institutional design that sets differentiated reimbursement levels according to disease severity, treatment outcomes, individual case differences, and risk assessment, an advanced model relative to traditional uniform reimbursement.

Note: This article compiles publicly available information and presents institutional observations. It does not constitute medical, legal, or NHI application advice. Catastrophic Illness eligibility, certificate validity, and applicable scope should always be confirmed against the latest announcements from the National Health Insurance Administration, Ministry of Health and Welfare, physician diagnosis, and competent authority determinations.


Sources: