Disability discrimination in the provision of health insurance: Article 25(e) of the UN convention on the rights of persons with disabilities

Abstract

Article 25(e) of the CRPD obliges State Parties to “prohibit discrimination against persons with disabilities in the provision of health insurance… which shall be provided in a fair and reasonable manner”. Equal access to health insurance is critical in ensuring that persons with disabilities can enjoy the highest attainable standard of health. However, the scope and substance of Article 25(e) have not been examined in existing scholarship on the CRPD. Standing at the intersection of international human rights law, disability studies and insurance law, this article fills the lacuna by offering a balanced interpretation of Article 25(e) that aims to strike a balance between actuarial fairness and the CRPD’s prohibition against all forms of disability discrimination. Greater clarity in the interpretation of Article 25(e) would therefore assist State Parties in properly complying with their obligations under the CRPD. Civil society, including disabled people and their representative organizations, would also be empowered to hold their governments to account where the laws and policies in the country have fallen short of what is expected under Article 25(e).

Introduction

At the height of the COVID-19 pandemic, disabled persons in India reported that they could not obtain health insurance; in one instance, a blind person’s application was rejected even without a medical screening (Bhuyan, 2020). In the same year, an autistic activist in Singapore called for government intervention after he and other autistic persons faced significant difficulties in obtaining private health insurance (Loh, 2020). The potential higher utilization rate of health insurance by those with disabilities also means that private insurers are often reluctant to offer health insurance to persons with disabilities on an equal basis as those without disabilities, if at all (Crossley, 2005: 91–95; Manning, 2000).

While both India and Singapore are party to the UN Convention on the Rights of Persons with Disabilities (CRPD), it is not entirely clear what their obligations are in respect of disability discrimination in the provision of health insurance under Article 25 of the CRPD. Specifically, Article 25(e) of the CRPD obliges State Parties to “prohibit discrimination against persons with disabilities in the provision of health insurance… which shall be provided in a fair and reasonable manner”.

In the insurance context, the disparate treatment of different groups based on their risk profile has been justified by the principle of actuarial fairness that underpins the underwriting process and the operation of the industry itself (Landes, 2015). Opponents to anti-discrimination regulation in the insurance industry argue that an overly interventionist approach may result in adverse selection – where lower-risk members exit the market – and a “death spiral” which undermines the entire market (Avraham et al., 2013). Yet, there remains scant scholarship on when and how actuarial fairness crosses into the realm of disability discrimination (Kuschke, 2018; Silvers and Francis, 2013).

In Singapore’s case, the Committee on the Rights of Persons with Disabilities (CRPD Committee) has recommended that legislation be enacted to ensure that all persons with disabilities have access to health insurance on an equal basis with others and are guaranteed equal conditions without discrimination (Committee on the Rights of Persons with Disabilities, 2022: 50(a)). Yet, it is uncertain how the prohibition of discrimination under Article 25(e) should be interpreted in light of the qualification that such provision shall be “provided in a fair and reasonable manner”. While the CRPD Committee has referred to this obligation in its concluding observations on a handful of other State Parties – namely Peru, Hong Kong, Thailand, Jordan and Israel – the scope and meaning of Article 25(e) remain unsettled. The interpretation of Article 25(e) similarly has not been clarified in existing international scholarship on the CRPD (Pavone, 2017; Stein et al., 2009; Weller, 2018).

Equal access to health insurance is important so that persons with disabilities can enjoy the highest attainable standard of health. As set out in the preamble of Article 25 of the CRPD, disabled persons have the right to the enjoyment of the highest attainable standard of health without discrimination on the basis of their disability. In this regard, disability discrimination in the health insurance context has serious effects on whether disabled persons are able to access and afford healthcare. This is especially when disabled persons are more likely to require more – and more expensive - healthcare compared to non-disabled persons. For example, according to a study on the Affordable Care Act (ACA) in the United States (US), persons with disabilities were more likely to have been admitted to the hospital in the last year (19.4%) than those without disabilities (4.7%) and (Kennedy et al., 2017). The study also found that the average annual cost for a disabled person ($13,492) was significantly higher than for those without disabilities ($2,835), including out-of-pocket costs which are more than twice as high ($1,053 v. $486).

Standing at the intersection of international human rights law, disability studies and insurance law, this article fills the lacuna by offering an interpretation of Article 25(e) in respect of the provision of health insurance that aims to strike a balance between actuarial fairness and the CRPD’s prohibition against all forms of disability discrimination. First, in relation to the meaning of the word “discrimination”, Article 25(e) should be understood to refer to what has been coined the “progressive definition of discrimination” in the context of the CRPD, where government regulation of the private health insurance market should absolutely prohibit direct discrimination and prohibit indirect discrimination insofar as it cannot be justified by the insurer (Gurbai, 2020). Second, the phrase “provision of health insurance” in Article 25(e) should be understood to refer not only to open enrolment and premium pricing but also benefit design. Greater clarity in the interpretation of Article 25(e) would therefore assist State Parties in properly complying with their obligations under the CRPD. Civil society, including disabled people and their representative organizations, would also be empowered to hold their governments to account where the laws and policies in the country have fallen short of what is expected under Article 25(e).

The drafting of article 25(e)

In December 2006, then United Nations (UN) Secretary-General Kofi Annan declared the “dawn of a new era” as the UN General Assembly adopted the CRPD (UN News, 2006). According to Theresia Degener, former chairperson of the CRPD Committee, the CRPD “seeks to bring about a paradigm shift in disability policy that is based on a new understanding of disabled persons as right holders and human rights subjects” (Degener, 2016: 1). The prohibition against discrimination on the basis of disability is a central principle of the CRPD and is repeated numerous times in the document. For example, Article 3(2) recognizes non-discrimination as one of the eight core principles of the CRPD while Article 5(2) states that States Parties “shall prohibit all discrimination on the basis of disability.”

To determine the meaning of Article 25(e), this article begins by turning to the travaux préparatoires to trace how this provision came to be included in the text of the CRPD. These records offer valuable insight into the intention and concerns that were contemplated by the state delegates and disability organizations which participated in the drafting process of the CRPD.

In 2000, the Beijing Declaration of Rights of Persons with Disabilities in the New Century called for the creation of a new UN human rights convention that would legally bind countries to promote the full inclusion of disabled persons and the elimination of disability discrimination (Kanter, 2014: 39). Following the Beijing Declaration, the UN General Assembly adopted GA Resolution 56/168 formally establishing the Ad Hoc Committee on a Comprehensive and Integral International Convention to Promote and Protect the Rights and Dignity of Persons with Disabilities (Ad Hoc Committee). After the Ad Hoc Committee held its first session in July 2002, a Working Group comprising representatives from UN Member States and non-governmental organizations representing disabled people was established at its second session in June 2003 to prepare and present a draft text of the Convention.

At the Working Group’s meeting in January 2004, proposals relating to insurance discrimination that were submitted to the Ad Hoc Committee were considered. For instance, the World Blind Union proposed enshrining a right to insurance, which would ensure that disabled persons can obtain insurance cover for health, life and any other insurance protection at no additional cost (World Blind Union, 2003: 9). China also proposed the recognition of the right to social security, insurance and welfare, and a duty on states to “incorporate persons with disabilities into all kinds of insurance programs and eradicate all discriminatory provisions against persons with disabilities” (Ad Hoc Committee, 2003)^. Mexico similarly proposed that states should “establish norms whereby persons with disabilities are not discriminated against regarding the access to social security and public and private medical insurance” (Ad Hoc Committee, 2002).

In the draft text that was presented at the third session, the right to health was enshrined under a draft Article 21, which provided that:

States Parties recognize that all persons with disabilities have the right to the enjoyment of the highest attainable standard of health without discrimination on the basis of disability. States Parties shall strive to ensure no person with a disability is deprived of that right, and shall take all appropriate measures to ensure access [75] for persons with disabilities to health and rehabilitation services.

Though the issue of insurance discrimination was not included in the draft text itself, a footnote 75 noted that “some Working Group members suggested that affordability, and access to health insurance by persons with disabilities without discrimination on the basis of disability, should be addressed in the Convention” (Working Group to the Ad Hoc Committee, 2004: 26). Uganda also proposed adding the word “insurance” at the end of the article such that it would read “… to health insurance and rehabilitation services” (Working Group to the Ad Hoc Committee, 2004: 51).

Under Article 23 on social security and an adequate standard of living, sub-clause (f) stated that State Parties should “ensure that persons with disabilities are able to access life and health insurance without discrimination on the basis of disability” (Working Group to the Ad Hoc Committee, 2004: 65). A footnote appended to this sub-clause noted that “the Ad Hoc Committee may wish to consider the extent to which States parties can determine the provision of insurance, which in many countries is typically the domain of the private sector”. Multiple state parties proposed amendments (Working Group to the Ad Hoc Committee, 2004: 30). For example, Chile proposed adding the sentence “Whether the system is for individual or capitalization saving, the cost for the person with disability should be the same as that applied to all persons.” In contrast, the European Union proposed a complete rewording: “take appropriate measures to enable persons with disabilities to have access to life and health insurance based on objective criteria.”

Subsequently, at the sixth session, some delegates suggested that sub-clause (f) under the draft Article 23 should be moved to draft Article 21 on health instead (Ad Hoc Committee, 2005: 23). The International Disability Caucus also submitted a proposal for a sub-clause (l) under Article 21 which would oblige states to ensure that “public and private health insurance are available for persons with disabilities on an equal basis with others” (International Disability Caucus, 2005). Israel also proposed a new Article 23.bis on insurance specifically, which would require states to “ensure by way of legislation and other measures that insurance is neither denied on the basis of disability or made available to persons with disabilities on terms less favorable to persons with disabilities than to others, unless such treatment is based on sound actuarial and professionally recognized information and is fair and reasonable in the circumstances” (State of Israel, 2005). During the discussions, Israel explained that this clause would enshrine non-discrimination in insurance coverage, and require that insurance decisions such as denials, terms and conditions, be made on sound actuarial data (Rehabilitation International, 2005: 19–20).

The suggestion to move sub-clause (f) to the article on health was taken in by the Ad Hoc Committee in its draft text presented at the seventh session. A new sub-clause (e) appeared under the draft Article 25 on health and required State Parties to “[p]rohibit discrimination against persons with disabilities in the provision of health insurance, and life insurance where permitted by national law, which shall be provided in a fair and reasonable manner” (Ad Hoc Committee, 2006b: 21). Israel’s proposed language also seems to have been partially incorporated with the inclusion of the final part of sub-clause (e) which states that such insurance “shall be provided in a fair and reasonable manner.”

This sub-clause (e) was the subject of some debate at the seventh session, particularly the phrase “where permitted by national law”. For example, Chile proposed deleting this phrase to avoid weakening the article by existing national laws that allow discrimination in health and life insurance while India expressed concern that this phrase could allow for discrimination in insurance premiums (Rehabilitation International, 2006: 24–25). In response, the Chair noted that this phrase was included to qualify life insurance where such insurance is not permitted by law in some countries and took up Kenya’s suggestion to add “where such insurance is permitted by national law” to clarify this issue.

Finally, at the eighth session, the language of Article 25(e) was finalized in the first revised text prepared by the Drafting Group on 13 September 2005 and did not go through any further changes in subsequent revision cycles (Ad Hoc Committee, 2006a). As passed by the UN General Assembly in December 2006, Article 25(e) reads as follows:

“State Parties shall… prohibit discrimination against persons with disabilities in the provision of health insurance… which shall be provided in a fair and reasonable manner”.

The drafting history of Article 25(e) thus suggests that it was not intended to adopt an absolutist definition of discrimination. First, the draft Article 23(f) did not initially include the qualifier that health insurance should be “provided in a fair and reasonable manner”. This language only appeared in the draft text presented at the seventh session and appears to have been informed by Israel’s proposal to allow private insurers to make insurance decisions that are less favorable to disabled persons based on sound actuarial data. Second, India and Chile’s concern that some form of discrimination in the provision of health insurance may be permitted under Article 25(e) because of the phrase “where permitted by national law” was not directly addressed by the Chair. This strengthens the non-absolutist reading of Article 25(e) that it was not intended to prohibit all forms of discrimination against persons with disabilities so long as it is “provided in a fair and reasonable manner”. At the same time, it is not obvious what it means for health insurance to be provided in a fair and reasonable manner and the specific forms of insurance practices that would fall foul of Article 25(e). To answer these questions, the next section considers how the words and phrases used in Article 25(e) ought to be interpreted.

The meaning of “discrimination”

As a starting point, Degener has observed that “equality and anti-discrimination is at the heart of” the CRPD, with every single article of the convention referring to these concepts (Degener, 2016: 15). Hence, Article 4(1)(e) obliges State Parties to “take all appropriate measures to eliminate discrimination on the basis of disability by any person, organization or private enterprise”. This obligation is reiterated in Article 5(2), which obliges State Parties to “prohibit all discrimination on the basis of disability and guarantee to persons with disabilities equal and effective legal protection against discrimination on all grounds”.

Article 2 defines “discrimination on the basis of disability” as “any distinction, exclusion or restriction on the basis of disability which has the purpose or effect of impairing or nullifying the recognition, enjoyment or exercise, on an equal basis with others, of all human rights and fundamental freedoms in the political, economic, social, cultural, civil or any other field”. By prohibiting any distinction, exclusion or restriction which has the “purpose or effect” of an unequal enjoyment of human rights by disabled people, this definition thus prohibits both direct discrimination (purpose) and indirect discrimination (effect).

The CRPD Committee clarified the obligations to prohibit discrimination in General Comment No. 6 (2018). According to the CRPD Committee, the CRPD develops a new model of equality which it coins “inclusive equality” that embraces a “substantive model of equality and extends and elaborates on the content of equality in [inter alia] a fair redistributive dimension to address socioeconomic disadvantages” (para. 11). In this regard, the CRPD’s human rights model of disability emphasizes the indivisibility and interdependence of both civil and political rights as well as social, economic and cultural rights (Degener, 2017).

Traditionally, the right to equality and non-discrimination is understood as a civil right enshrined in Article 26 of the International Convention on Civil and Political Rights while the right to enjoy the highest attainable standard of physical and mental health is enshrined in Article 12 of the International Covenant on Economic, Social and Cultural Rights. Yet, because existing human rights conventions provide “general protections to all persons”, they have had “limited success at ensuring persons with disabilities can exercise their rights” (Harpur, 2012: 4). Hence, the CRPD was drafted with the intent of expressly ensuring that people with disabilities are able to enjoy the fundamental rights set out in both categories of international human rights conventions (Degener, 2017: 94).

The CRPD Committee has clarified that the word “discrimination” under Article 5(2) includes four main forms of discrimination: direct discrimination, indirect discrimination, denial of reasonable accommodation and harassment.

Traditionally, differential treatment does not constitute discrimination if the purpose or effect of the differential treatment is to achieve a legitimate aim and if the differential treatment can be justified objectively and reasonably. This is also commonly referred to as the “pragmatic definition” of discrimination (Vandenhole, 2005: 76). Though the CRPD Committee initially recognized this pragmatic definition in its draft general comment in relation to indirect discrimination, the final version of the document made no mention of this pragmatic definition at all (Gurbai, 2020: 284–285). According to Gurbai’s survey of the CRPD’s jurisprudence, while the CRPD Committee did not rely on the pragmatic definition of discrimination in all three cases relating to direct discrimination, it did adopt this pragmatic definition in four cases out of seven cases relating to indirect discrimination (Gurbai, 2020: 288). Gurbai thus proposed what she coins the “progressive definition of discrimination”, which permits the justification defense in relation to indirect discrimination but not direct discrimination (Gurbai, 2020: 289).

Direct discrimination

First, direct discrimination refers to a situation where a disabled person is treated less favorably than other persons because of their disability. Hence, a person whose health insurance is denied by an insurer because they have a disability would have suffered direct discrimination. The refusal to offer insurance to a person on the basis of a protected characteristic has been described as the “harshest type of discrimination (Avraham, 2018: 342). They have also been outlawed in various countries; for example, in the US, section 1557 of the ACA (42 U.S. Code § 300gg–1), also known as the guaranteed issue provision, prohibits insurers from refusing to offer health insurance coverage to a person on the basis of a pre-existing condition. However, though an insurer may be compelled to offer health insurance to high-risk individuals despite the policyholder’s higher cost of risk in the US, they are not prohibited from charging higher premiums to account for this higher cost of risk.

The CRPD Committee has made clear that private insurers should not be permitted to reject insurance coverage solely on the basis of a person’s disability. For example, in its Concluding Observations on Thailand, the Committee expressed concern that private health insurance companies had discriminated against persons with disabilities by refusing to sell them health insurance policies because of their disability (Committee on the Rights of Persons with Disabilities, 2016: 49–50). Similarly, in its Concluding Observations on Hong Kong, it expressed concern that “many insurance companies reject the applications of persons with disabilities, thus leaving them unable to pay the medical fees”, and recommended that the government “arrange the cooperation of the insurance companies” (Committee on the Rights of Persons with Disabilities, 2012: 76).

Indirect discrimination

Second, indirect discrimination may arise in situations where policies or practices that appear neutral at face value have a disproportionate negative impact on a person with disability. Examples of indirect discrimination in health insurance include charging a disabled person higher premiums for health insurance on the basis that they are classified under a higher risk group according to the insurer’s risk classification policy or when an insurance policy imposes a limit on certain benefits that disproportionately affects a disabled person’s ability to receive the healthcare services they need. In these situations, while the risk classification policy or the insurer’s benefit design do not necessarily single out disabled persons, it may nevertheless still have the effect of disproportionately affecting their ability to access or afford health insurance compared to non-disabled persons.

The meaning of “fair and reasonable manner”

Informed by Gurbai’s analysis of CRPD jurisprudence, this article takes as a starting point that direct discrimination cannot be justified but indirect discrimination may be. Accordingly, the requirement that insurers provide health insurance in a “fair and reasonable manner” can be understood to apply only to permit indirect discrimination where it is “fair” and “reasonable” but not direct discrimination, which is absolutely prohibited. How then are these references to fairness and reasonableness supposed to interact with the progressive definition of discrimination to determine when indirect discrimination in the provision of health insurance may be justified?

Fairness

As a starting point, the concept of fairness is nebulous and amorphous. As Yves Thiery and Caroline Van Schoubroeck (2006: 191) have observed, “lawyers and legislators on the one hand, and insurers on the other, have profoundly different views on the concept of fairness in insurance classification”. The former group conceive of fairness from an “individualist” approach while insurers approach the issue from a “group” approach. From the “individualist” perspective, it is wrongful to treat a person differently on the basis that they are a member of a particular group. However, insurers do not conceive of themselves as treating any specific individual differently; instead, it is fair if individuals belonging to the same group because of their risk characteristics are treated similarly.

This “group” approach to fairness is also known as actuarial fairness, which requires that individuals assume, through adjusted premiums, the costs of the risk that they ‘import’ to the insurance pool (Landes, 2015: 521). As Kenneth Abraham and Pierre-André Chiappori (2015: 292) put it:

“[C]ontracts sold to individuals with different risks are different commodities, even if the losses covered are the same (for instance, an auto- mobile insurance contract sold to a young driver or to an experienced one are viewed as two different goods). As a consequence, charging different prices to parties posing different risks is price differentiation (differences in prices reflect differences in commodities sold), whereas charging different prices to parties posing identical risks but different price elasticities is price discrimination.”

The principle of actuarial fairness is said to be necessary to avoid the mechanism of adverse selection, which occurs when both lower-risk and higher-risk individuals are charged the same premiums. This results in the subsidization of solidarity, where lower-risk individuals are paying for the loss suffered by those at higher risk (Thiery and Van Schoubroeck, 2006: 196). Rationally, because the lower-risk individuals would be averse to such subsidization, they would exit the risk pool either by purchasing insurance from another insurer that charges lower premiums based on their risk level or not purchasing insurance at all. This may eventually lead to a “death spiral” when adverse selection completely undermines the insurance market (Avraham et al., 2013: 204). Hence, adverse selection may arise when insurers are prohibited by law from charging different premiums based on risk level.

At the same time, numerous countries have introduced regulations over the private health insurance industry that mandate open enrolment (where insurers accept all applicants for insurance cover) or community rating (i.e., where insurers cannot vary premiums payable by individuals on the basis of their risk profile). An example is Ireland’s Health Insurance Act 1994, which involves both open enrolment and community rating. However, researchers have found that adverse selection has occurred in Ireland’s private health insurance market because each insurer has a different risk profile from the others (Turner and Shinnick, 2013). To overcome these challenges, the Irish government operates a risk equalization scheme which offers credits to insurers for covering older members though it has not fully eliminated risk selection in the market (Keegan et al., 2017).

While it is beyond the scope of this article to delve into the nuances of actuarial fairness (Stone, 1993), it suffices to say that what is fair is ultimately context-specific. For example, Ronen Avraham has observed that “it is possible that fairness requires insurance companies to admit people with various diseases and disabilities to their pool, and yet fairness might still allow insurers to charge these people a higher premium” (Avraham, 2018: 342). In his two-stage framework that considers whether an insurance regime is fair and just, the first stage determines the issue of fairness in respect of three distinct factors, namely the discriminated characteristic, the type of insurance and the type of discriminatory treatment (Avraham, 2018: 341–342). Specifically, in relation to disability discrimination in health insurance, what is considered to be fair in one jurisdiction may not be in another. Relevant factors would include the broader healthcare and disability policy landscape, the prevailing political ideology in that jurisdiction as to the role of government vis-à-vis the market, and the socio-economic circumstances in that country in relation to the government’s ability to subsidize or correct the market failures which may arise from policies mandating open enrolment or community pricing in the private health insurance market.

Reasonableness

In relation to the concept of reasonableness, this article suggests that it refers simply to the pragmatic definition’s reasonable justification defense. In the context of health insurance, any form of discrimination must be capable of justification by actuarial or statistical data. This interpretation is supported by the fact that the CRPD Committee has not expressed any concern with existing anti-discrimination statutes that enshrine this exemption for such rational discrimination.

For example, in both of its Concluding Observations on Australia in 2013 and 2019, the CRPD Committee did not take issue with Australia’s exemption for the insurance industry under the Section 46 of the Disability Discrimination Act (ADDA), which permits the differential treatment of disabled persons in two situations. First, the differential treatment must be based upon actuarial or statistical data on which it is reasonable to rely, and the discrimination is reasonable having regard to the matter of the data and other relevant factors. Second, in a case where no such actuarial or statistical data is available and cannot reasonably be obtained, then the discrimination must be reasonable having regard to any other relevant factors.

The CRPD’s interpretation of Article 25(e) in this regard is consistent with the European Court of Justice’s (ECJ) decision in Association belge des Consommateurs Test-Achats ASBL v Conseil des ministers (2011) C-236/09. In this case, the issue is whether a Belgian statute - Article 10 of the Law of 10 May 2007 – which allows differential treatment between men and women “on the basis of relevant and accurate actuarial and statistical data” violates the Article 21 and 23 of the Charter of the Fundamental Rights of the European Union (European Charter), which prohibit any discrimination on grounds of sex and require equality between men and women to be ensured in all areas. This Belgian statute is in turn enacted to implement Directive 2004/113, which sought to lay down a framework for combating discrimination based on sex in access to and supply of goods and services. Specifically, Article 10 of the Law of 10 May 2007 is mutatis mutandis as Article 5(2) of Directive 2004/113.

On the basis that the principle of equal treatment “requires that comparable situations must not be treated differently, and different situations must not be treated in the same way, unless such treatment is objectively justified” (at [28]), the ECJ held that Directive 2004/113 is based on the premise that the respective situations of men and women with regard to insurance premiums and benefits contracted by them are comparable. Specifically, Recital 18 to Directive 2004/113 provides as follows: “in order to guarantee equal treatment between men and women, the use of sex as an actuarial factor must not result in differences in premiums and benefits for insured individuals”. Accordingly, Article 5(2) of Directive 2004/113 and Article 10 of the Law of 10 May 2007 were found to be in violation of Articles 21 and 23 of the European Charter.

This case highlights that the CRPD does not impose the same exacting demand on Member States. Unlike Directive 2004/113, there is nothing in the CRPD which requires that there must be no difference between disabled and non-disabled persons in the provision of health insurance. Instead, Article 25(e) makes clear that what is prohibited is only “discrimination”. Actuarial and statistical data may therefore justify differences in premium pricing and – as the next section will discuss – benefit design. The question then arises as to how such differences are to be justified.

In this regard, the Supreme Court of Canada’s (SCC) decision in Zurich Insurance Co. v. Ontario (Human Rights Commission) [1992] 2 S.C.R. 321 (Zurich Insurance) is instructive. The case involved the insurer, Zurich, charging young, single, male drivers higher car insurance rates than young, single, female drivers and young, married, male drivers or any drivers older than 25 years of age. The insurer alleged that this practice was justified on reasonable and bona fide grounds based on available statistics which show that single males under the age of 25 had the highest claim frequency, the highest loss per car insured and the highest average claim cost of any of the categories for which statistics were kept (at 347).

The relevant statutory provision in this case is section 21 of the Ontario Human Rights Code 1981, which permits insurers to make distinctions in insurance policies on “reasonable and bona fide grounds” on account of various protected characteristics. Specifically, the court had to determine what this phrase means. According to the majority (at 342), a discriminatory practice is “reasonable” within the meaning of section 21 if two requirements are satisfied. First, the discriminatory practice must be one which is desirable to adopt for the purpose of achieving the legitimate business objective of charging premiums that are commensurate with risk. Second, there must be no practical alternative.

In relation to the first requirement, both the majority and minority agreed that statistics, in and of itself, is insufficient to justify discriminatory treatment on the basis of a protected characteristic, in this case, age, gender and marital status. As the majority put it (at 349), “To allow discrimination simply on the basis of statistical averages would only serve to perpetuate traditional stereotypes with all of their invidious prejudices”. The dissenting judgment of L'Heureux-Dubé J clarifies that statistical correlation is insufficient and discriminatory treatment is permissible only if the insurer can prove that there is statistical causation between the protected characteristic and the insured risk (at 371):

“Discrimination based on statistical correlation is simply discrimination in a more invidious form. Strong statistical proof is required to demonstrate a rational connection between the discriminatory classification and high risk. This proof, however, must not simply be one of correlation, but one of causal connection.”

Put simply, the statistical data relied on by the insurer must not merely demonstrate, as Zurich did, that young, single, male drivers had the highest average claim frequency or claim cost compared to other groups. Instead, the statistics must show that this group had the highest claim frequency or claim cost because of their gender, age and/or marital status by demonstrating that the relevant protected characteristic has a statistically significant impact on those actuarial measures. From a statistical perspective, this would involve further data analysis such as hypothesis tests, correlation coefficients and multiple regression to remove confounding factors (Scheines, 2007).

In this case, the majority allowed the insurer’s appeal on the exceptional basis that there was no available alternative criteria that it could have utilized to set premiums due to the impracticability and impossibility to obtain actuarially reliable figures at the time in 1983 (at 352). However, the majority also cautioned that this did not mean that Zurich – or any other insurer – could continue to price its premiums on the same basis if more data becomes available. Accordingly, the majority’s decision was reached on the basis that Zurich’s premium pricing was based on the “only statistics available… at the time in question”.

Both dissenting judges however pointed to the fact that Zurich ultimately bore the burden of proving that no reasonable alternatives exist but failed to discharge this burden by its own failure to collect the required data (at 383). As McLachlin J. put it, the majority had confused the absence of a reasonable alternative with absence of proof of such a reasonable alternative. In doing so, all the insurer had to do is “show that it cannot produce evidence as to the existence or non-existence of a reasonable alternative” (at 384). The burden of proof is then shifted to the victims of such discriminatory treatment who have to prove that a reasonable alternative exists.

The dissenting judges’ analysis is more consistent with the CRPD in that the burden of proof must ultimately lie with insurers to justify that their insurance practice in respect of premium pricing is fair and reasonable, not with the disabled person to prove the opposite. Crucially, a lack of statistical data that demonstrates causality between the protected characteristic and the insured risk is no excuse for discrimination in premium pricing since it essentially means that the insurer is relying not on statistics but, in the words of L'Heureux-Dubé J, “stereotype and entrenched practice” (at 366).

Informed by the justification test developed in European human rights jurisprudence in relation to the vertical relationship between the state and its citizens (Thiery and Van Schoubroeck, 2006: 204) and the insights gleaned from Zurich Insurance, this article proposes a more flexible four-stage test which should be enshrined in domestic law to determine whether indirect discrimination can be justified in the context of a horizontal relationship between a non-state actor like a private insurer and an individual insured disabled person.

At the first stage of this flexible four-stage test, there must be a legitimate aim for the differential treatment (the legitimacy test). This can usually be satisfied by pointing to the insurer’s profit motive and the mechanism of adverse selection. Second, there must be a pertinent relationship between the target characteristic and the risk insured (the pertinency test). This will require the insurer to adduce empirical evidence to show that there is either an actual or predicted causal relationship between this target characteristic and the risk insured.

Third, the less restrictive alternative test requires the insurer to show that the discriminatory policy or practice in question is a less restrictive alternative. For example, to address adverse selection, the insurer can charge higher premiums rather than deny insurance coverage at all. Alternatively, the insurers can achieve the same legitimate aim through changes in the benefit design such as the introduction of higher deductibles or benefit limits may be less restrictive.

Finally, the proportionality test aims to balance the importance of the insurance type to the individual consumer and the public welfare with the insurer’s profit interest. An example is the use of pregnancy as a risk variable, which would impose higher premiums on women even though society benefits generally from procreation (Thiery and Van Schoubroeck, 2006: 208).

The meaning of “provision of health insurance”

The foregoing discussion has established that Article 25(e) prohibits private insurers from outrightly refusing to insure disabled persons on the basis of their disability and any indirect discrimination in the context of premium pricing is permitted only if it can be justified. However, insurance discrimination extends beyond the refusal to offer insurance coverage; it also includes issues of benefit design that may prevent disabled persons from accessing the healthcare that they need on an equal basis as others. The phrase “provision of health insurance” in Article 25(e) should thus be broadly interpreted to not only refer to the availability of health insurance to disabled people but also its affordability and accessibility. Article 25(e) should therefore prohibit disability discrimination not only in coverage denial and unjustified premium pricing but also benefit design. This is for two reasons.

First, in its Concluding Observations on Jordan’s initial report, the CRPD Committee called on Jordan to enact legislation to ensure that disabled persons can “enjoy all private health insurance services covered by private insurance companies on an equal basis with others” (Committee on the Rights of Persons with Disabilities, 2017: 48(a)). Though the CRPD Committee did not clarify what it meant by “all private health insurance services”, this suggests that Article 25(e) does not merely apply to whether insurers will offer or deny insurance coverage. Similarly, in its Concluding Observations on Singapore’s initial report, the CRPD Committee urged Singapore to ensure not only that all persons with disabilities “have access to health and life insurance on an equal basis with others” but also “that they are guaranteed equal conditions, without discrimination” (Committee on the Rights of Persons with Disabilities, 2022: 50(a)). This reference to “equal conditions” buttresses the notion that the terms on which health insurance is provided to persons with disabilities cannot be discriminatory on the basis of disability as well.

Second, this broad interpretation is consistent with the model of “inclusive equality” that the CRPD promulgates, which includes “a fair redistributive dimension to address socioeconomic disadvantages” (Committee on the Rights of Persons with Disabilities, 2018: 11). A narrow meaning of “provision” to only refer to the availability of health insurance to disabled people would run contrary to this redistributive role of the CRPD and practically undermine the operation of Article 25(e) since insurers may charge unreasonably high premiums to deter disabled people from purchasing health insurance or design the terms of the insurance policy such that disabled people cannot receive the healthcare that they need.

While it is more intuitive why insurance denial and higher premium prices can result in the exclusion of disabled people from health insurance, it may be less obvious to non-disabled persons how discrimination in benefit design can hinder disabled people from meaningful access to healthcare. Drawing on comparative equality jurisprudence on discrimination in the provision of various insurance products on the basis of disability and other protected characteristics, the remainder of this section briefly addresses disability discrimination that arises in the context of insurance denial decisions and higher premium prices before offering a deeper discussion on how discriminatory benefit design can hinder the ability of people with disabilities to receive the healthcare they require.

As a starting point, benefit design is a term of art that refers to the set of rules that describe how a health insurance policyholder can access medical services. This can include policy terms that determine which medical or related services will be covered by or excluded from the plan, the providers from which a member of the plan can receive a covered service, and the cost-sharing amounts a member of the plan will be responsible to pay when receiving a service (Delbanco et al., 2016: 11). How an insurance policy is structured can seriously affect a person with disability’s ability to access and afford the healthcare services that they require.

First, insofar as direct discrimination goes, any discriminatory policy term that is expressly based on a person’s disability would violate the CRPD’s prohibition against discrimination. For example, in the Australian case of Ingram v. QBE Insurance (Australia) Ltd (Human Rights) [2015] VCAT 1936, the insurer’s policy exclusion in a travel insurance policy for claims arising directly or indirectly due to mental illness was held to constitute direct discrimination under section 44 of Victoria’s Equal Opportunity Act. Though the insurer had sought to rely on the statutory exemption to justify this policy exclusion on actuarial or statistical data, this was rejected by the court on the evidence. (Specifically, the insurer acknowledged that there was an absence of evidence which demonstrated that it made the decision to incorporate the mental illness exclusion on the basis of contemporaneous actuarial data.) In any case, as discussed in the section above, Article 25(e) would not permit insurers to rely on the justification defense in relation to direct discrimination claims.

Indirect discrimination in benefit design has mostly arisen in the context of US law, where there remains some controversy over whether disparate impact discrimination in benefit design is prohibited as a matter of US disability law (Francis and Silvers, 2008, 2017). This refers to a situation where a policy term is facially neutral and does not expressly target a person’s disability but disproportionately affects the ability of persons with disabilities to enjoy “meaningful benefit” of health insurance.

This issue was first addressed by the US Supreme Court in Alexander v. Choate, 469 U.S. 287 (1985), where the state of Tennessee’s proposed reduction in the number of annual inpatient hospital days that state Medicaid would pay hospitals on behalf of a Medicaid recipient from 20 days to 14 days was challenged for having a disproportionate effect on disabled Medicaid recipients. Under Section 504 of the Rehabilitation Act (29 U.S.C.A. § 701), recipients of federal financial assistance – in this case, the state Medicaid program – cannot exclude from participation in, deny the benefits of, or subject to discrimination a person with disability on the basis of their disability. The proposed change to the Medicaid program will disproportionately affect disabled individuals, who are more likely to be admitted for hospitalization for medical care. Though the court acknowledged that disabled persons must be provided with “meaningful access to the benefit” of the Medicaid program, it held on the facts that the reduction in the number of annual inpatient hospital days still provided meaningful access to disabled Medicaid recipients since they can still “benefit meaningfully from the coverage they will receive under the 14-day rule” (Francis and Silvers, 2008: 452).

While this case establishes the requirement of “meaningful access” in the health insurance context, it did not fully flesh out what would satisfy as “meaningful access”. Scholars have also acknowledged that this is “not an easy task and it does not admit of a simple answer” (Francis and Silvers, 2017: 7). Two appellate courts have recently revisited this issue in two cases in relation to Section 504 of the Rehabilitation Act read with Section 1557 of the ACA. At issue in both cases was restrictions imposed by insurers on where individuals may obtain specialty medicine – specifically, HIV/AIDS medication. In Doe v. BlueCross BlueShield of Tenn., Inc., 926 F.3d 235 (6th Cir. 2019), the Sixth Circuit held that Section 504 does not prohibit disparate impact discrimination, which is synonymous with indirect discrimination in international human rights law. Since the plaintiff in that case could not prove that the insurer’s policy intentionally discriminated against him on the basis of his disability, his claim failed. Since this article has established above that Article 25(e) prohibits indirect discrimination, this court’s holding is not relevant.

In contrast, the US Ninth Circuit’s decision in CVS v. Doe, 982 F.3d 1204 (9th Cir. 2020) Ninth Circuit found in favor of the plaintiffs who argued that the insurer’s benefit design, which requires the plaintiffs to obtain specialty medication only through specific pharmacies (the Program), discriminated against them by eliminating various aspects of pharmaceutical care critical to their health. Specifically, the court agreed with the plaintiffs that because of their unique pharmaceutical needs, “changes in medication to treat the continual mutation of the virus requires pharmacists to review all of an HIV/AIDS patient’s medications for side effects and adverse drug interactions”. However, because they can only obtain medication at particular pharmacies, they have been denied a benefit they can no longer receive under the Program. As the court put it:

“…the structure and implementation of the Program discriminates against [the plaintiffs] on the basis of their disability by preventing HIV/AIDS patients from obtaining the same quality of pharmaceutical care that non-HIV/AIDS patients may obtain in filling non-specialty prescriptions, thereby denying them meaningful access to their prescription drug benefit.”

Though the insurer had initially filed an appeal to the US Supreme Court to challenge the Ninth Circuit’s decision, this was subsequently withdrawn.

The key takeaway from the foregoing discussion on US jurisprudence is to demonstrate how facially neutral benefit design can have a disproportionate impact on persons with disabilities. To ensure that they are able to receive the healthcare they need and realize their right to the highest attainable standard of health, Article 25(e) must necessarily prohibit disability discrimination in these circumstances. Though this will undoubtedly upend existing insurance practices of which many will become susceptible to scrutiny under Article 25(e), it is exactly this rethinking and reorganization of our society and laws that the CRPD demands. In accordance with Article 25(e)’s prohibition against discrimination in the provision of health insurance, private insurers must find alternative ways to balance their search for profit and economic considerations like adverse selection without infringing on the rights of persons with disabilities to equality and non-discrimination.

Conclusion

This article makes two contributions to the literature on disability discrimination under the CRPD. First, it engages with and builds on Gurbai’s progressive definition of discrimination by applying it in the health insurance context to examine what forms of discrimination are proscribed under the CRPD to ensure that disabled persons are able to access health insurance on an equal basis as others. Notably, while reasonableness may be operationalized into a legal test, the concept of fairness remains more slippery and must involve a more fact-specific and context-dependent analysis of the relevant factors in each jurisdiction or society. Second, it draws on comparative equality jurisprudence on insurance discrimination to elaborate on how the justification defense should operate in the context of indirect discrimination in the areas of premium pricing and benefit design.

In terms of future research directions, there remains a gap in relation to disability discrimination in the life insurance context, the other area of insurance where disability discrimination is prohibited under Article 25(e). It may also be worth examining how other human rights principles that underpin the CRPD can come into play in the interpretation of Article 25(e) to realize transformative equality in the health insurance context (Goldschmidt, 2017). For example, the participation and involvement of disabled persons and their representative organizations is a central concept in the CRPD, and it may be fruitful to examine how they should be involved in the actuarial and legislative process respectively when it comes to addressing disability discrimination in the insurance context.

It is hoped that this article offers a starting point for governments and civil society organizations in considering what ought to be done to combat disability discrimination in health insurance. The discussion may also be helpful for lawmakers and advocates concerned with introducing or reforming anti-discrimination legislation to ensure that such laws are appropriately drafted to ensure that disabled people are able to afford and enjoy meaningful access to health insurance and healthcare more generally. In addition, this article calls on the CRPD Committee to conduct a more searching scrutiny of the laws and policies of State Parties in addressing disability discrimination in the context of the provision of health insurance by private insurers.