What Makes a Benefit Design Discriminatory

Answer first: the design itself is regulated, not only the decision to sell

45 CFR 156.125(a) states the core rule in one sentence: "An issuer does not provide EHB if its benefit design, or the implementation of its benefit design, discriminates based on an individual's age, expected length of life, present or predicted disability, degree of medical dependency, quality of life, or other health conditions."

Three things in that sentence do the work. It reaches the design and "the implementation of" the design, so a rule that is neutral on paper but applied selectively is inside the prohibition. The listed characteristics are health-related rather than demographic — degree of medical dependency and quality of life are not categories that appear in most civil rights statutes. And the consequence is stated in benefits terms rather than penalty terms: an issuer whose design discriminates "does not provide EHB," meaning it fails the essential health benefits requirement rather than merely committing a separate violation.

The same section adds a positive requirement with a start date. "Beginning on the earlier of January 1, 2023 (the start of the 2023 plan year) or upon renewal of any plan subject to this rule, a non-discriminatory benefit design that provides EHB is one that is clinically-based." That converts the test from a purely negative one into a standard the design must meet.

And the section preserves an exception, in its own words: "Nothing in this section shall be construed to prevent an issuer from appropriately utilizing reasonable medical management techniques." The two adverbs are load-bearing. Reasonable medical management is protected; management that is neither reasonable nor appropriately utilized is not carved out by this sentence.

A second, demographic prohibition sits alongside it

Section 156.125(b) requires an issuer providing essential health benefits to comply with § 156.200(e). That provision states that a qualified health plan issuer "must not, with respect to its QHP, discriminate on the basis of race, color, national origin, disability, age, or sex (which includes discrimination on the basis of sex characteristics, including intersex traits; pregnancy or related conditions; sexual orientation; gender identity; and sex stereotypes)."

So two lists operate together. The § 156.125 list is about health status and its proxies — age, expected length of life, present or predicted disability, degree of medical dependency, quality of life, other health conditions. The § 156.200(e) list is the familiar civil rights set, with the parenthetical defining what sex includes for this purpose.

A benefit design can run into either. A rule keyed to how much care someone needs engages the first list without mentioning any demographic category at all; a rule keyed to who someone is engages the second. The regulation does not require a design to fail both.

Marketing is regulated on the same theory

The rules do not stop at what a plan covers. Section 147.104(e), inside the guaranteed availability section, addresses how coverage is sold: a health insurance issuer and its officials, employees, agents and representatives "must comply with any applicable State laws and regulations regarding marketing by health insurance issuers and cannot employ marketing practices or benefit designs that will have the effect of discouraging the enrollment of individuals with significant health needs in health insurance coverage."

The theory behind that clause is worth stating plainly. A plan that cannot refuse to sell to someone can still shape what it sells, and how it advertises, so that the people it does not want do not come. The rule is aimed at that substitution, and its test is effect rather than intent — "will have the effect of discouraging the enrollment."

The same paragraph carries a discrimination list of its own, prohibiting marketing practices or benefit designs that discriminate on the basis of race, color, national origin, present or predicted disability, age, sex as defined in the same parenthetical used in § 156.200(e), expected length of life, degree of medical dependency, quality of life, or other health conditions. It combines both lists in a single sentence, which is a useful reminder that the health-status and demographic theories are not kept in separate boxes.

Section 147.104 also states its own boundaries. Paragraph (h) provides that the section does not apply to grandfathered health plans in accordance with § 147.140, and paragraph (j) provides that nothing in the section should be construed to require an issuer to offer coverage otherwise prohibited under applicable federal law.

The section 1557 rule adds a coverage-specific list

45 CFR 92.207 governs a covered entity providing or administering health insurance coverage or other health-related coverage. Paragraph (a) sets the general rule: such an entity "must not, in providing or administering health insurance coverage or other health-related coverage, discriminate on the basis of race, color, national origin, sex, age, disability, or any combination thereof."

Paragraph (b) then enumerates specific prohibited practices. The first two are the ones most relevant to design. A covered entity must not "deny, cancel, limit, or refuse to issue or renew health insurance coverage or other health-related coverage, or deny or limit coverage of a claim, or impose additional cost sharing or other limitations or restrictions on coverage" on the basis of the listed characteristics. And it must not "have or implement marketing practices or benefit designs that discriminate" on those bases.

A further item addresses disability specifically: a covered entity must not have or implement benefit designs that fail to provide or administer coverage "in the most integrated setting appropriate to the needs of qualified individuals with disabilities, including practices that result in the serious risk of institutionalization or segregation." The remaining enumerated items in paragraph (b) concern coverage decisions based on sex assigned at birth, gender identity or gender otherwise recorded, and categorical or specific exclusions for services related to gender transition or other gender-affirming care.

Paragraph (d) prevents the list from being read as a ceiling: "The enumeration of specific forms of discrimination in paragraph (b) of this section does not limit the general applicability of the prohibition in paragraph (a)." An unlisted practice is not thereby permitted.

What the rules preserve, and the words that limit the preservation

Every one of these provisions contains a safe harbor, and every safe harbor is qualified.

Section 156.125(c) preserves the appropriate use of reasonable medical management techniques. Section 92.207(c) is more explicit about the same idea and is the clearest statement of the boundary anywhere in this set: "Nothing in this section requires coverage of any health service where the covered entity has a legitimate, nondiscriminatory reason for denying or limiting coverage of the health service or determining that such health service fails to meet applicable coverage requirements, including reasonable medical management techniques such as medical necessity requirements."

The next sentence is the limit on that limit: "Such coverage denial or limitation must not be based on unlawful animus or bias, or constitute a pretext for discrimination." So a medical necessity requirement is a legitimate tool, and the same requirement deployed as a pretext is not made legitimate by being called one. The regulation also preserves whatever protections a covered entity may have under §§ 92.3 and 92.302, which it names by cross-reference.

Put together, the rules do not say that a plan must cover everything, or that any particular exclusion is unlawful. They say that a benefit design is itself a regulated object, that a design keyed to health status or its proxies is reachable, that effect can be enough without proof of intent, and that the availability of a medical management justification depends on whether the justification is the actual reason.

Why the distinction between a rule and a verdict matters here

It is easy to read these provisions and conclude that a given exclusion in a given plan must be unlawful. Nothing in the text supports that step, and this article does not take it.

Whether a specific benefit design discriminates is a fact-bound determination. It depends on what the design does, whom it affects, what the issuer's stated reason is, whether that reason is clinically based, whether the technique used is a reasonable medical management technique appropriately applied, and whether the stated reason is the real one. Those questions are resolved by regulators and courts on a record, not by reading a plan document against a list.

What the rules do give a reader is a set of questions with regulatory weight behind them. Is the design clinically based, as § 156.125 requires from the 2023 plan year or renewal onward. Does a limit key on degree of medical dependency, present or predicted disability, or other health conditions. Would it have the effect of discouraging enrollment by people with significant health needs, in the language of § 147.104(e). And is the reason given for a denial a legitimate, nondiscriminatory one, or is it, in the words of § 92.207(c), a pretext.

One last scope note, because these four sections do not all reach the same plans. Sections 156.125 and 156.200 govern issuers of essential health benefits and qualified health plans. Section 147.104 applies to issuers in the guaranteed availability context and, by its own paragraph (h), not to grandfathered health plans. Section 92.207 applies to covered entities as part 92 defines them. A plan's own documents, and the part of the Code that governs it, determine which of these is in play.