MAID and marginalization

A look at the Ontario data

Recently, there has been a lot of news coverage about three reports published by Ontario’s Office of the Chief Coroner (OCC). These reports summarize the 2023 data on medically assisted deaths in the province, and were released more broadly following the data being leaked. They haven’t been released in the sense of being published for the public, which is a shame, because it has led to a bunch of misunderstandings about what the reports actually say.

Although Ontario already had a MAID review team that reviewed each medically assisted death in Ontario, in January of this year, the OCC created the MAID Death Review Committee (MDRC), which consists of sixteen people from different disciplines as well as one member of the public. The committee represents a diverse range of backgrounds and viewpoints. Some of the members are well-known opponents of MAID, while others are more supportive.

The existence of both these review bodies, as well as the publication of their findings, is great. All provinces should have such a committee. The committee represents another way that MAID has much more oversight than any other area of end-of-life care. A couple weeks ago, the Associated Press journalist Maria Cheng published an article which, among other things, raised issues about the lack of oversight of MAID in Ontario. Somehow, her reporting failed to turn up the existence of this committee and its imminent report, which is too bad, because her article mischaracterized how MAID deaths are evaluated.

Anyway, the OCC released three documents. In this post, I’m going to cover one of them, which is called “Medical Assistance in Dying (MAiD): Marginalization Data Perspectives”. There’s a great deal of interest in the demographic characteristics of people who access MAID. People have raised concerns about track two MAID in particular, which occurs when a person’s natural death isn’t reasonably foreseeable. Track two became available in 2021, when parliament passed Bill C-7 in response to the 2019 Truchon decision. That decision, which I’ve written about here and here, occurred because two people with disabilities—Jean Truchon and Nicole Gladu—argued that they should be able to have MAID even though their deaths weren’t reasonably foreseeable. The Superior Court of Québec agreed, thereby making the law more in line with Carter, the 2015 Supreme Court of Canada decision that struck down the general prohibition on assisted dying.

That was then. Here’s the new data. In Ontario, there were 4,644 MAID deaths last year, which is an eighteen percent increase from 2022. The number of track two deaths went down slightly from 121 in 2022 to 116 in 2023. In 2022, track two deaths made up 3.17 percent of MAID deaths. In 2023, track two deaths made up 2.74 percent of MAID deaths. The overall increase in MAID deaths is unsurprising, but it’s worth emphasizing that the jump in track two deaths that many MAID opponents predicted hasn’t materialized, at least in Ontario.

The OCC report makes use of the Ontario Marginalization Index, which Public Health Ontario calls ‘ON-Marg’. Using census data, the index splits Ontario into ‘dissemination areas’: thousands of population cohorts from a few hundred up to just under thirty thousand. These dissemination areas are the basis for marginalization quintiles, which put each area into one of five groups based on that area’s level of marginalization. To better capture the nuance of the factors, there are four marginalization dimensions: Households and Dwellings, Material Resources, Age and Labour Force, and Racialized and Newcomer Populations. These dimensions have to be treated separately; there’s no way to aggregate them to get a single marginalization score, but you can view the entire map here.

Each dimension uses various indicators of marginalization. For example, the Household and Dwellings dimension gives a higher marginalization score when there’s a higher proportion of people living alone; more people who aren’t five to fifteen years old; more apartment buildings; a higher proportion of people who are single, divorced, or widowed; more renters; and a higher proportion of people who have moved in the last five years.

Already, this shows the difficulty with quantifying marginalization. Young people living in downtown Toronto renting single-bedroom apartments will make that area more marginalized, even though they might be doing just fine. An assisted living facility will score high for the same reason.

That means the value of the data is limited. The report tells us the marginalization distribution of people who had MAID, but this is only based on their dissemination area. Someone might live in a high-marginalization area and be well off or live in a low-marginalization area and be badly off. We can’t assume that the area characteristics match the individual’s situation. Doing that would be the ecological fallacy. Further, even if we knew someone’s specific income or housing status, that wouldn’t indicate that something was amiss with their use of MAID. Renters should have access to MAID, and being a renter and having MAID doesn’t mean the person had MAID because she’s a renter.  

As I mentioned earlier, the absolute number of track two deaths went down from 121 to 116 and the percentage of track two deaths went down to 2.74 percent. Since there are so few track two deaths, each one will have more of an effect on each marginalization score. One-fifth of 116 is 23.2 people, whereas one-fifth of 4,644 is 928.8 people. 

With all that in mind, let’s look at the charts.

Housing, Age, and Working

First, here’s the distribution of MAID recipients based on Households and Dwellings marginalization:

As you can see, both tracks skew more marginalized. With 48.3 percent of track two recipients living in the most marginalized areas, it’s particularly skewed.

Age and Labour Force is even more dramatic:

This has led some people, such as Sonu Gaind, to claim that this is evidence that Canada has fallen “over the cliff” with track two MAID.

However, the report itself cautions against this interpretation: 

[T]he metrics for Households and Dwellings and Age and Labour Force are, by definition, highly dependent on age and ability to engage in the work force, which are too closely correlated to provide new and meaningful information outside of confirming the older age and increased levels of disability as reported within the MAiD population.

Another way of putting this is that, if anything, the rate of marginalization among people having track two MAID is probably higher than ON-Marg indicates, but, in this context, ‘marginalization’ means “is older and doesn’t work”, which is neither surprising nor cause for concern. We already know from other data that MAID is used most by older people—the average age for track two MAID deaths in 2022 across Canada was 73.1—so to point to this rate of marginalization as evidence of a serious problem with MAID misunderstands the data. (In the chart above, the numbers in the column for track one MAID recipients only adds up to 71.2 percent, so there’s some kind of error there.)

Material Resources

The report says that the previous metrics aren’t meaningful, which is bad news for Gaind’s argument. So let’s look at Material Resources, where marginalization goes up if an area has a higher proportion of people without a high-school diploma, who receive government transfers, are unemployed, low-income, or live in dwellings needing major repair. As with the previous dimensions, some of these factors are neither surprising nor necessarily problematic, such as a retiree receiving pension payments.

For this dimension, the report helpfully compares the quintiles of various groups that had MAID to a broader category they belong to. This is important to avoid the base rate fallacy. For example, here’s the distribution of people who received MAID and were getting palliative care versus people receiving palliative care who didn’t get MAID:

There’s basically no difference in marginalization between all MAID recipients and MAID recipients receiving palliative care, but the chart shows that MAID recipients receiving palliative care tended to live in areas less marginalized than palliative care patients generally. This is consistent with data from Canada and elsewhere showing that MAID recipients tend to be more educated and have higher incomes than the general public.

Still, that’s only palliative care. While eighty percent of MAID recipients are getting palliative care, that’s going to be skewed by track one. Gaind and others are mainly worried about track two, so here’s neurological conditions, which accounted for 37.9 percent of track two cases in Ontario last year (and fifty percent of Canadian track two cases in 2022).

Here, the data is similar to palliative care. Among people with neurodegenerative diseases or Parkinson’s, they are less likely to be marginalized if they receive MAID. And this is not just compared to the quintile distribution of all people with the disease, but also compared to Ontarians generally.

MAID is Working Great

There’s nothing in this report that indicates that marginalized people are more likely to receive MAID or that marginalization is causing a disproportionate number of marginalized people to choose MAID.

Gaind, for his part, says that defenders of MAID are engaged in denialism. He claims that a “higher proportion of marginalized people [are] dying from Track two” and that Canada is committing ‘social murder’ that will require a future prime minister to issue a national apology. He compares MAID to tobacco companies for “failing to warn Canadians adequately of the risks of premature death posed by Track 2 MAiD to those suffering from social marginalization.”

But the report itself warns him against the analysis he takes, and a closer look at the other data in the report, which he leaves out of his article, shows that people who get MAID, including track two, are actually less likely to be marginalized. This is consistent with previous data, not a new finding.

This doesn’t mean there have been no difficult cases. I’ll discuss those in another post. But this report isn’t the smoking gun MAID opponents claim it is.