In the United States Court of Appeals For the Seventh Circuit ____________________

No. 23-2100 ANGELA MIDTHUN-HENSEN and TONY HENSEN, on behalf of their daughter K.H., Plaintiffs-Appellants,

v.

GROUP HEALTH COOPERATIVE OF SOUTH CENTRAL WISCONSIN, INC., Defendant-Appellee. ____________________

Appeal from the United States District Court for the Western District of Wisconsin. No. 21-cv-608-slc — Stephen L. Crocker, Magistrate Judge. ____________________

SUBMITTED FEBRUARY 15, 2024 — DECIDED AUGUST 5, 2024 ____________________

Before SYKES, Chief Judge, and EASTERBROOK and KIRSCH, Circuit Judges. EASTERBROOK, Circuit Judge. Angela Midthun-Hensen and her husband Tony Hensen asked their health insurer to cover certain therapies between 2017 and 2019 for their child K.H.’s autism. The insurer, Group Health Cooperative, refused. Based on its review of the medical literature, it determined 2 No. 23-2100

that evidence did not support speech therapy as a treatment for autism for a child K.H.’s age and did not support the use of sensory-integration therapy (a form of occupational ther- apy) as a treatment for autism at any age. Because the em- ployer-sponsored plan in which the family was enrolled co- vers only treatments that are “evidence-based”, Group Health Cooperative deemed these therapies ineligible for coverage. (Developments in the medical literature led the insurer to begin covering these treatments in October 2020. Plaintiffs do not contest the benefits K.H. received after this change.) After several lengthy medical-review and appeals pro- cesses confirmed Group Health Cooperative’s conclusion that then-available evidence did not support the requested thera- pies, Hensen and Midthun-Hensen sued, contending that the insurer violated provisions of the Employee Retirement In- come Security Act (ERISA), which regulates employer-spon- sored group health plans, as well as a state law regarding cov- erage for autism. The district court found nothing wrong with the insurer’s decisions. 672 F. Supp. 3d 662 (W.D. Wis. 2023). Plaintiffs no longer contest the district court’s conclusion that the insurer did not violate state law or deprive K.H. of bene- fits to which she was entitled under the plan. They focus in- stead on their argument that Group Health Cooperative’s lim- its violated the Mental Health Parity and Addiction Equity Act (MHPAEA), 29 U.S.C. §1185a (§712 of ERISA). The Parity Act requires, as a general matter, that health in- surers place coverage for mental conditions on an equal foot- ing with coverage for physical conditions. One way it does this is by requiring that “treatment limitations applicable to … mental health or substance use disorder benefits are no more restrictive than … treatment limitations applied to No. 23-2100 3

substantially all medical and surgical benefits covered by the plan”. 29 U.S.C. §1185a(a)(3)(A)(ii). As plaintiffs see it, Group Health Cooperative violated this prohibition by applying its requirement that treatments be “evidence-based” more strin- gently to mental-health benefits for autism than it did to one medical benefit, chiropractic care. (We use the statute’s dis- tinction between “mental health benefits” and “medical or surgical benefits”, though we recognize that mental condi- tions are themselves medical conditions.) Plaintiffs point out that, although their plan did not cover K.H.’s proposed therapies until 2020, it did cover (in certain situations) chiropractic care for musculoskeletal conditions in pediatric patients—a course of treatment that they contend lacks scientific support. They assert that, given this lack of support, Group Health Cooperative’s imposition of an age- based treatment limitation for certain autism treatments but not for chiropractic care violated the Parity Act. (Plaintiffs also contend that evidence available at the time had supported the treatments K.H. sought—but the district court found this un- substantiated when it determined that K.H. was not entitled to coverage for these treatments under the plan. 672 F. Supp. 3d at 675–76.) The district court did not see evidence that the difference between benefits for autism and benefits for musculoskeletal conditions could be attributed to any difference in the way the insurer treated mental and physical conditions. Id. at 678–80. The judge concluded that differences in coverage reflected differences in the medical literature on which the insurer re- lied. The surveys of autism research that Group Health Coop- erative consulted made treatment recommendations that de- pended in part on patients’ ages, whereas its sources 4 No. 23-2100

regarding chiropractic care did not (though they recognized that there was a paucity of evidence specifically demonstrat- ing efficacy of chiropractic care in children). An insurer is en- titled to identify and rely on such literature so long as its pro- cess for doing so applies to mental-health benefits and medi- cal benefits alike. Limiting coverage to evidence-based treat- ments has the support of a regulation that plaintiffs do not contest. 29 C.F.R. §2590.712(c)(4)(I). That’s why the insurer prevailed. The district court’s conclusion comports with the medical evidence of record, even taking that evidence in the light most favorable to plaintiffs. Pre-2020 restrictions on coverage for autism therapies did not result from how the insurer assessed the literature regarding each condition. Rather, they reflect how the underlying literature assessed and accounted for age. The Parity Act permits health insurers, when determining what treatments to cover, to rely on the available medical lit- erature. They must make sense of this literature as they find it, no matter how thin or developing it may be. Cf. Smith v. Office of Civilian Health & Medical Program of Uniformed Ser- vices, 97 F.3d 950, 956–57 (7th Cir. 1996). The way in which the medical literature considers the efficacy of and makes recom- mendations regarding various treatments will vary for any number of reasons—from the availability of study partici- pants across demographics, to funding considerations, to judgments regarding study design, to which patient charac- teristics researchers expect to bear on treatments’ efficacy. Such variance affects the results—and treatment recommen- dations—of medical study. It’s unsurprising that literature on autism focuses more on efficacy by age than does literature on chiropractic care. No. 23-2100 5

Musculoskeletal conditions tend to develop with injury and age, which may lead researchers to focus on adult popula- tions. Meanwhile studies on autism, which is commonly di- agnosed and first treated in childhood, most often focus on children. That Group Health Cooperative’s policies reflect this differing focus does not pose a problem under the Parity Act. This is not all. Plaintiffs’ argument fails for a more funda- mental reason. Plaintiffs make their case by identifying a sin- gle medical benefit that was handled differently from the men- tal-health benefits K.H. sought. But the relevant statutory pro- vision requires that treatment limitations applicable to men- tal-health benefits be no more restrictive than treatment limi- tations “applied to substantially all medical and surgical bene- fits covered by the plan”. 29 U.S.C. §1185a(a)(3)(A)(ii) (em- phasis added). To evaluate whether a limitation applies to “substantially all medical and surgical benefits”, the plaintiff must focus on treatments as a whole rather than a single kind of treatment. “Substantially all” is less than all—but not much less. See Con- tinental Can Co. v. Chicago Truck Drivers Pension Fund, 916 F.2d 1154 (7th Cir. 1990). Regulations implementing the Parity Act define “substantially all” to mean “at least two-thirds” as con- cerns “financial requirements” or “quantitative treatment limitations” but are silent (for no reason we can discern) on what “substantially all” means for “nonquantitative treat- ment limitations” such as the one at issue here. See 29 C.F.R. §2590.712(c)(3)(i)(A). Plaintiffs do not contend that these reg- ulations are invalid. (ERISA authorizes rulemaking, see 29 U.S.C. §1135, and we need not address how Loper Bright 6 No. 23-2100

Enterprises v. Raimondo, 144 S. Ct. 2244 (2024), applies to regu- lations adopted under an express delegation.) We do not have to determine exactly what “substantially all” means, because “substantially all” does not mean “one.” Plaintiffs proceed as if they can prevail by showing that their insurer approached coverage for one mental-health benefit more restrictively than coverage for one medical benefit. They are mistaken. No matter how much space “substantially” leaves, a showing that an insurer limits a mental-health bene- fit more than it does one medical benefit cannot show that it so limits substantially all such benefits. Plaintiffs have not seri- ously tried to show that Group Health Cooperative, as a gen- eral matter, imposed age-based treatment limitations less stringently on medical (and surgical) benefits than on mental- health benefits. Plaintiffs maintain that they lacked adequate opportunity to make their best case because discovery was stayed by the district court early in the suit. They suggest that, given the lack of discovery, we should evaluate the district court’s judg- ment as we would a motion to dismiss. This is wrong for many reasons—for one, discovery is not required before sum- mary judgment. See Fed. R. Civ. P. 56. Litigants may ask the district court to hold off on deciding a summary-judgment motion until they can conduct further discovery, see Fed. R. Civ. P. 56(d)—and plaintiffs did file a motion asking the court to defer acting on the summary-judgment motion. But the dis- trict court said no, 2022 U.S. Dist. LEXIS 174594 (W.D. Wis. Sept. 27, 2022), and plaintiffs do not contend on appeal that the judge abused his discretion. (Litigants do not need discov- ery to find out the contents of medical literature.) To receive relief on appeal, a party must do more than express No. 23-2100 7

dissatisfaction with how things went in the district court—it must explain how an adverse order was reversible. That de- pends on showing both error and prejudice. See F.C. Bloxom Co. v. Tom Lange Co., No. 22-3268 (7th Cir. July 25, 2024), slip op. 16–18 (discussing a litigant’s need to show why a grant of relief under Rule 56(d) would have been likely to turn up im- portant evidence). We cannot consider—and plaintiffs cannot receive relief from—a determination they do not appeal. AFFIRMED