Adenovir's Posterous

The O’Neill’s “Health Status, Health Care and Inequality: Canada vs. the U.S.“

 http://www.pnhp.org/facts/single-payer-faq#rationing

A recent paper by June and Dave O’Neill contests previous research findings that health outcomes are better in Canada than in the U.S. The O’Neills also claim that income-based health disparities are larger in Canada than in the U.S., that access to care is better in theU.S. and that cancer screening and survival are worse in Canada.

The O’Neills collected no new data. Their analysis rests on idiosyncratic, highly selective and overtly biased reinterpretations of previously published data — mostly from the Joint Canada/U.S. Survey of Health (JCUSH), a population-based survey conducted jointly by theU.S. and Canadian government statistical agencies. While they extensively cite the few pieces of published data that supports their grim view of Canada’s health system, they ignore a large body of research and statistics that conflicts with their portrayal.

We will briefly discuss the main assertions in the O’Neill paper.

1-Canada’s lower mortality rates are not a result of better health care

The O’Neills assert that Canada’s longer life expectancy (2 years longer) and lower infant mortality rate (5.3 deaths/1000 live births vs. 6.8 in the U.S.) have nothing to do with health care. Rather, they claim that non-medical factors such as substance abuse, obesity, low education and “cultural factors” explain the U.S.’ poor performance.

They argue that high infant mortality in the U.S. simply reflects the high frequency of preterm births and low birth weight, and especially the very poor outcomes among African-Americans. American’s short life expectancy, they say, comes from high rates of obesity, as well as homicides and accidents.

But the U.S.’ high rate of prematurity and low birth weight is, largely, a result of poor care — inadequately treated infections and chronic illnesses among pregnant women, and the shockingly frequent failure to deliver adequate prenatal care. 16% of pregnant women in theU.S. receive no care at all in the first trimester of pregnancy, far higher than in Canada.

While the O’Neill’s dwell on the high obesity rates in the U.S. as an important non-medical cause of mortality differences, they ignore Canada’s significantly higher smoking rate — a graver threat to health than obesity. Moreover, they cite data from the OECD that exaggerates the obesity differences, ignoring the more reliable JCUSH data that they use for most of their other comparisons (presumably because the JCUSH found more modest differences in obesity rates).

They emphasize that accidents and homicides account for a large fraction of theU.S./Canada mortality difference among young adults, age 20-24. But deaths are rare in this age group, and accident/homicides account for virtually none of the difference in older age groups, where almost all of the deaths occur. In fact, differences in heart disease cause most of the Canadian advantage — a difference that almost certainly reflects, at least in part, better access to care in Canada.

The O’Neills also ignore the fact that the U.S. had a lower infant mortality rate than Canada’s until the passage of national health insurance (NHI) in Canada, after which Canada’s rate fell sharply. Similarly, they never mention that most of Canada’s advantage in life expectancy emerged shortly after NHI was implemented.

In many of their data tables on mortality and other health measures, the O’Neills’ separate out white from minority Americans, and indicate that Canadians’ health outcomes are similar to those of white Americans. Of course, excluding minorities in the U.S. means excluding one third of the entire population, and more than three quarters of the poor. In essence, they want to exclude the 100 million poorest and sickest Americans, and compare the remainder to a cross section of Canadians, including the sick and poor.

2-Other measures besides mortality rates are better indicators of the quality of health care in the two nations, and the U.S. comes out well on these.

The O’Neill’s simply assert that self-reported health status (the answer to the question “How would you rate your health? Excellent, good, fair or poor?”) is a better measure of the impact of the health care system than are mortality rates. Yet this measure has never been validated for cross national comparisons of the type they make, and it seems very likely to be greatly affected by cultural norms. And, as with death rate comparisons, only by eliminating minority Americans from the comparison can they conclude that the U.S. looks slightly better than Canada.

They then compare the two nations using a “health utility index” and the percent of people with pain that limits their activities. For both of these, Canadians do better than Americans, until minorities are subtracted from the population.

Finally, they compare the prevalence of chronic condition like diabetes, emphysema and arthritis in the two nations, and the proportion of people with each condition who are getting treatment. They conclude that while slightly more Americans are chronically ill, more of them are getting treatment. But millions of uninsured Americans with chronic illnesses like diabetes or high blood pressure are unaware of their diagnoses because they can’t afford the doctors visit or lab test needed to make the diagnosis. Surveys will not identify undiagnosed persons as having chronic disease. Hence, the proportion getting treatment is falsely inflated in the U.S. Moreover, even the differences they cite to favor the U.S. are not statistically significant. Hence, a more accurate depiction of the data would state that among people who know of their diagnoses, rates of care are similar in the two nations.

3-More Americans get cancer screening and the U.S. has more high tech health resources than Canada

The O’Neills cite higher screening rates in the U.S. for cervical cancer (PAP smears), breast cancer (mammography), colon cancer (colonoscopy or sigmoidoscopy) and prostate cancer (PSA testing). Only the small PAP smear difference is real.

For mammography, they include all women 40-69 in their calculation of screening rates. But neither the American College of Physicians nor the Canadian Task Force on Preventive Health Care recommend mammograms for all women 40-50. Mammograms for women in this age group leads to more breast surgery and other cancer treatments, but has not been shown to lower overall mortality. It is likely that most of the difference in breast cancer screening is due to higher screening rates among young women in the U.S., who may even be harmed by excessive mammograms.

There’s a similar problem with their analysis of colon cancer screening. They include people age 40-69. Yet standard guidelines do not recommend colon cancer screening in normal risk individuals before age 50. They’ve fudged the data to get a result they want.

Neither the U.S. Preventive Services Task Force nor its Canadian counterpart recommend routine PSA testing because its not at all clear that such testing does more good than harm — it turns up lots of false positives, including many small tumors that would never cause serious problems if left untreated. Routine screening may well lead to many unnecessary operations that leave men incontinent and impotent. Yet the O’Neill’s interpret Canada’s lower PSA screening rate as an indicator of poor quality care.

We may ultimately find that PSA screening or early mammography saves lives — or causes more harm than good. But at present, we just don’t know whether the lower use of these technologies in Canada is a good thing or a bad one.

Their analysis also trumpets the greater number of CT scanners and MRI machines in theU.S. as an indicator of better quality. Yet recent estimates suggest that in the coming years radiation from CT scans may cause as many of 2% of all cancer deaths in the U.S. — about 30,000 excess deaths annually. It is far from clear that the greater use of CT scanners in theU.S. (relative to Canada) causes more good than harm.

4-Waits for care compromise access in Canada, and these access problems are worse than those in the U.S.

The O’Neill analysis admits that fewer Canadians than Americans report an unmet health need (11.3% vs. 14.4%). In the U.S., cost is the big problem, while waits for care are more prominent in Canada. They try to obfuscate the Canadian advantage on access measures by presenting a complex sub-group analysis of pain suffered by those unable to get care. But when you cut through their obfuscation, even this measure favors Canada; about 12% more Americans who say they’re unable to get care report being in pain.

5-Cancer mortality rates are higher in Canada, indicating worse cancer care

The O’Neills claim that mortality rates for lung, breast, colon and prostate cancers are lower in the U.S. than in Canada. They calculate mortality rates by dividing cancer deaths by the number of cases of cancer.

When cancer death rates are calculated in a more standard fashion, i.e. the number of deaths per thousand people in the population, age adjusted cancer mortality is actually lower in Canada than in the U.S. for all of these cancers except colorectal cancer. But there are more cancers diagnosed in the U.S. Hence, the death rate among those who are diagnosed — the figure the O’Neills choose as the most important - is lower in the U.S.

This higher incidence of cancers diagnosed in the U.S. probably reflects more intensive screening programs, which diagnose more people with cancer. But, as stated above, its far from clear that diagnosing small prostate cancers based on PSA screening causes more good than harm. Most elderly men with prostate cancer do not die of that disease, but of heart disease or some other illness. (It is likely that some breast and lung cancers that are diagnosed through screening would also never come to light without screening.) In the U.S.,these men with small, non-lethal cancers appear in the denominator of the O’Neill’s calculation of cancer mortality rates, but not the numerator. In Canada, they appear in neither the numerator nor denominator. Thus, their estimate of cancer mortality rates is biased against Canada because of the higher screening rate in the U.S.

6-Income-based health disparities are, if anything, steeper in Canada than in the U.S.

The O’Neills admit that health differences between those above and below the median income are sharper in the U.S. than in Canada. In fact, the differences between the top and bottom 10% are also bigger in the U.S., as are those between the top and bottom 25% etc.

But the O’Neills want to measure health inequities on a new scale. They observe that the rich in the U.S. are much richer than the rich in Canada, and the poor are much poorer. Rather than comparing high and low income persons, they decide to analyze how much worse health gets for each dollar decrease in income. Since the U.S. income gradient is much steeper, this analysis automatically makes the health per dollar gradient less steep. Notice that this method would find that a nation with almost no income inequality would automatically have very steep health inequalities.

Here’s an example. In Country A, the top 1% has an average income of $60,000 and a mortality rate of 100 per 1000. The bottom 1% has an average income of $30,000 and a mortality rate of 130 per 1000. Then, according to the O’Neills’ method for each $1000 increase in income, the mortality rate rises 1 per 1000.

In Country B, the top 1% has an average income of $603,000 and a mortality rate of 100 per 1000. The bottom 1% has an average income of $3,000 and a mortality rate of 300 per 1000. Then, according to the O’Neills’ method for each $1000 increase in income, the mortality rate rises only 1 per 3000.

So the O’Neills’ calculus would judge the income/health gradient less steep in Country B(where the poor have a death rate 300% higher than the wealthy) than in Country A (where the poor have a death rate 10% higher than the wealthy).

7-What’s left out?

The O’Neill’s paper cites dozens of references. But they fail to mention any of the numerous previous studies that directly address the questions they seek to answer. These include:

1-Previous analyses of the JCUSH data by the National Center for Health Statistics (http://www.cdc.gov/nchs/pressroom/04news/firstjointsurvey.htm) and by our group at Harvard (http://www.ajph.org/cgi/content/abstract/96/7/1300), which found a far different result.

2-The many published studies directly comparing the quality of medical care in the two nations for cancer patients, renal dialysis patients etc. 38 of these studies were included in a systematic review which concluded that, on average, mortality rates are 5% lower in Canada (http://www.openmedicine.ca/article/view/8/1).

3-The large body of literature showing that Canada’s health care system is far more efficient, with administrative overhead that is a small fraction of the U.S. level (http://content.nejm.org/cgi/content/short/349/8/768)

4-A recent analysis of deaths that could be prevented by good medical care ranked the U.S.worst among the 19 nations studied, well behind Canada which ranked 6th. Moreover, while Canada’s ranking improved between 1997 and 2003, the U.S. fell further behind. (http://content.healthaffairs.org/cgi/content/abstract/27/1/58)

There is much to criticize in Canada’s health care system. But the O’Neill’s analysis strays far from legitimate scientific discourse, mixing selective citation and creative accounting that is intellectually dishonest.