This article originally appeared in The Globe and Mail.
By Ramona Coelho, June 29, 2026
Thomas Dillon’s MAID assessment took place in 2023 outside a Tim Hortons in St. Thomas, Ont., after a psychiatrist had raised the option with him. He died at age 45, with his mental illness and addictions largely untreated. The MAID law was not precise enough to prevent cases like his – and that is a problem.
Canada’s parliamentary committee has recommended indefinitely pausing the 2027 planned expansion of medical assistance in dying (MAID) where mental illness is the sole underlying medical condition. I testified at the committee. The recommendation is necessary, but Parliament must go further. The mental illness debate has exposed a problem that runs through the whole MAID framework.
The debate over MAID for mental illness has focused on several concerns, including the impossibility of reliably determining whether mental illness is irremediable, and the significant role that community life and supports play in recovery. We also cannot distinguish a “reasoned MAID request” from suicidality, leaving assessments without an objective standard, and therefore allowing bias to shape who accesses MAID versus suicide prevention and care.
Proponents of expansion respond that uncertainty also exists in physical illness, and that MAID providers can be trusted to make these judgments.
But these arguments point to a deeper problem: how uncertainty is addressed across MAID decision-making.
In my work as a family physician serving marginalized patients, and as a former member of Ontario’s MAID Death Review Committee (MDRC), I have seen similar uncertainty extend beyond MAID for solely psychiatric illness. In one MDRC case, a socially isolated woman with severe obesity and depression was deemed eligible for MAID after refusing any diagnostic workup or treatments that might have improved or reversed her condition. Eligibility was not shaped by a clear disease trajectory but by treatment refusal and disengagement from care, with neglect interpreted as irremediability.
In another, a man in his 70s with essential tremor, which is not typically considered to cause a serious decline in capability, was approved for MAID, despite his request being mainly driven by spousal bereavement.
In both these cases, a chronic, manageable condition became grounds for death once isolation, poverty or lack of care entered the clinical picture. That is structural ableism – the institutional assumption that living well with disability is impossible, leading clinicians to view death rather than support as the more appropriate response.
Ontario data from Track 2 MAID – the option for those whose natural death is not reasonably foreseeable – confirms these concerns extend beyond individual cases. Nearly 30 per cent of Track 2 recipients were living in poverty, were less likely to have family members listed as their next of kin, and most were not offered mental-health, disability, housing or income supports. These patterns point directly to problems with the architecture of the law itself.
Eligibility is divided into Track 1 and Track 2 based on whether natural death is considered “reasonably foreseeable,” a distinction intended to separate those nearing death from those living with disability. In practice, this prognostic judgment is shaped by assessor interpretation. Research published in the Canadian Medical Association Journal found that assessors interpret the timeline of “reasonably foreseeable” as anywhere from six months to five years. That range is unacceptable. MAID is an irreversible procedure; there should be virtually no margin for error. Yet the documented variability confirms that a significant one exists.
If uncertainty is inherent in psychiatric prognosis, disability assessments, treatment refusal and end-of-life trajectories, then the concern extends far beyond MAID for mental illness.
As Parliament considers the next phase of MAID legislation, it must confront whether the law’s central concepts are precise. Avoiding unnecessary deaths requires clear statutory definitions of “reasonably foreseeable natural death,” “grievous and irremediable,” and “intolerable suffering,” along with oversight to stop interpretive drift and doctor-shopping. It requires restoring minimum waiting periods under Track 1 and introducing a real-time mechanism to pause MAID assessments when concerns are raised by families or clinicians.
The United Nations Committee on the Rights of Persons with Disabilities has called for changes, including repealing Track 2 MAID due to risks related to discrimination and social vulnerability. When a system cannot reliably protect those most likely to be harmed, it should not be allowed. Track 2 is exactly that.
MAID is a legal exemption to homicide. The least we owe Canadians is a law that makes wrongful deaths harder, not easier.
Ramona Coelho, MDCM, CCFP, is a senior fellow at the Macdonald-Laurier Institute, an adjunct research professor of family medicine at the University of Western Ontario’s Schulich School of Medicine and Dentistry, and the co-editor of Unravelling MAiD in Canada: Euthanasia and Assisted Suicide as Medical Care.
