Part 4: Preventing 7,000 deaths from heart disease alone among nonwhites each year

Clark et al at the University of Minnesota produced a research paper titled:

National Patterns in Environmental Injustice and Inequality: Outdoor NO2 Air Pollution in the United States

 They tell us in the abstract:

"For example, we estimate that reducing nonwhites’ NO2 concentrations to levels experienced by whites would reduce Ischemic Heart Disease (IHD) mortality by ,7,000 deaths per year..."

I have been looking at their data, and their claim of a 7,000 reduction in IHD deaths, in the last three posts.  In Part 3 I wrote this:

To me, this calculation of 7,000, or 6,579, or 5,638 is not within the realm of reality. Could NO2 contribute to 5,000 plus IHD deaths per year? Yeah...that's possible if Jerrett's RR of 1,06 is a direct result of an increase in NO2. [Bowman]

Here is what I come up with when I looked at their data.

Let's assume some things first.

1. The relative risk for IHD with 4,1 ppb NO2 is 1.06.
2. The incidence of IHD mortality is 109 per 100,000
3. The population and concentrations for NO2 for 448 urban areas in the US as found in the Excel spreadsheet file "journal.pone.0094431.s001.XLSX" are valid.
4. The column in that Excel file called "Difference Between LIN and HIW (ppb)" is a valid number.

I am writing this in "real" time, so from this point on, I have no idea what my way of calculating this is going to show.

To start, I needed to know what percent of nonwhites fall below the poverty line. That line seems to be consistent with what the Clark et al authors use for their "Low-Income Nonwhite (LIN) Population-weighted Concentration (ppb)"

I asked Google to get me this information and I found a reputable site with this information.

Next, I went looking in the Excel spreadsheet for the urban areas where the authors calculated a difference between the Low-Income Nonwhite (LIN) Population-weighted Concentration (ppb) and the High-Income White (HIW) Population-weighted Concentration (ppb) of greater than 4.0 ppb.

I chose 4.0 ppb to be consistent with the author's claim that "average NO2 concentrations are 4.6 ppb (38%, p,0.01) higher for nonwhites than for whites" and to be close to the relative risk of 1.06 reported by Jerrett et al for 4.1 ppb NO2.

Okay...I think I am being reasonable and fair here. Let's start.

Assuming that a 4.1 ppb difference in NO2 is associated with a 6% increase in risk for IHD mortality, what was the population of the 448 urban areas where the difference between the low income folks and the high income folks was greater than 4 ppb?

Time for another assumption. I am going to assume that the IHD morality is 109 per 100,000 for the white population in the HIW area.

This group of 12 urban areas represents a total population of 45.2 million people.

Let's do another assumption. Let's assume that within these 12 urban areas the percentage of nonwhites is 40%.

That would mean that the nonwhite population in these 12 urban areas where there is a difference of  at least 4.0 ppb NO2 between low income nonwhites and high income whites, is 18.1 million nonwhites.

Looking at the low income percentages provided by the Kaiser Family Foundation,

1. 26% of Blacks are low income (26% of 18.1 = 4.7 million)
2. 24% of Hispanics are low income (24% of 18.1 = 4.3 million)
3. 15% of Other nonwhites are low income (15% of 18.1 = 2.7 million)

This equals a total of 11.7 million nonwhites living in areas where the difference between NO2 population-weighted concentrations for high-income whites and low-income nonwhites is greater than 4.0 ppb.

Assuming that there is a 6% increase in IHD mortality risk for 4.1 ppb NO2, a risk of 109 per 100,000 for high-income whites would rise to 116 per 100.000 for low-income nonwhites. There would be an additional 7 deaths per 100,000 for a total added deaths from IHD for these 12 urban areas of 823.

Using the same language as in the abstract:

Reducing low-income nonwhites’ NO2 concentrations to levels experienced by high-income whites in 12 urban areas would reduce Ischemic Heart Disease (IHD) mortality by 823 deaths per year. [Bowman]

For that number - 823 - to be correct, the relative risk for NO2 must be 1.06 for 4.1 ppb NO2 and the concentration of NO2 for 11.7 million nonwhites must be 4 ppb higher than for their white neighbors living in the same urban area.

Well that was a fun way to spend my vacation on Friday and this rainy Saturday.

To conclude...there will not be a savings of 7000 nonwhite lives if we lower the concentration of NO2 to that which whites experience. At best I calculate 823 and that number is based on a lot of assumptions all being correct.

Thanks for reading

Jeff

Part 3: Preventing 7,000 deaths from heart disease alone among nonwhites each year

Let's get into the press release statement once more:

Gap results in an estimated 7,000 deaths each year among people of color from heart disease alone

...is that number calculated correctly?

According to their paper, the relative risks in Ischemic Heart Disease mortality from increasing NO2 concentrations by 4.1 ppb is 1.066. This came from the 2003 paper by Jerrett et al.

Reading the Jerrett paper we are told:

All RR estimates are given over the interquartile range of each pollutant.

This value of 1.066 was calculated for the NO2 concentration values between Q1 and Q3

If you look at the Table in the Jarrett paper we see this:

Q1 therefore is the value for NO2 at 25% and Q3 is the value for NO2 at 75%.

Q1 = 10.21

Q3 = 14.33

The difference between Q1 and Q3 is 4.1 ppb.

The RR they calculated for 4.1 ppb is 1.066.

This would lead us to conclude that those who had NO2 exposure between 10.21 and 14.33  had a 6% increase in risk of ischemic heart disease IHD mortality.

Now for me, when I look at stuff like this, I try to look at it collectively. The mean - average - from the Excel file is at the Q1 value in the Jarrett paper.

What I see when I look at this data, is if the IHD is "109 deaths per 100,000 people," that rate takes into account this exposure range. And that value includes the deaths of whites and nonwhites.

The claim of Clark et al, is that "gap results in an estimated 7,000 deaths each year among people of color from heart disease alone."

That value of 7,000 deaths is based on the author's assumption that every nonwhite in the US lives in areas where the N02 concentration is at 14.5 ppb. This assumption - using the Jarrett data for California, places every nonwhite in the US in the forth quartile for distribution.

In other words, based on the Jarrett Table 2 data, 25% of the distribution of NO2 by population contains 100% of the nonwhite population.

That...that cannot be what the Clark et al folks are saying...or is it?

Follow me on my analysis here, just to make sure I am seeing this correctly...

Source

Now they (Clark et al) calculated the Relative Risk (RR) in a weird way (in my opinion):

Relative risks (RR) for NO2 concentrations experienced by nonwhites and whites calculated using: RR = exp(βc), where c is the NO2 concentration (units: ppb), and β = ln(1.066)/(4.1 ppb) = 0.0156 ppb-1.

They then show these calculations:

I think there is an error in this calculation. The second calculation is for the amount of IHD deaths anticipated for whites which should be divided by 1.167 and not 1.254. If I understand their thinking on this, they are comparing one population at an RR of 1.254 (nonwhites) to one at 1.167 (whites).

If I am correct, then the white IHD deaths would be 13,570 for a difference of 5,638 IHD deaths per year.

That amount, however, assumes that the entire US population of nonwhites is exposed to 14.5 ppb of NO2 while the entire population of whites is exposed to 9.9 ppb NO2.

To me, this calculation of 7,000, or 6,579, or 5,638 is not within the realm of reality. Could NO2 contribute to 5,000 plus IHD deaths per year? Yeah...that's possible if Jerrett's RR of 1,06 is a direct result of an increase in NO2.

That's not what Clark et al are wanting to convey:

Gap results in an estimated 7,000 deaths each year among people of color from heart disease alone

They assert that compared to whites, 7,000 more deaths from IHD happen because of NO2 concentrations for this group being 4.6 ppb higher than what whites are exposed to.

That high of a "gap result" is not supported by their data.

So what does the data they present actually tell us?


Next post: Part 4: Preventing 7,000 deaths from heart disease alone among nonwhites each year

Part 2: Preventing 7,000 deaths from heart disease alone among nonwhites each year

The amount of peer reviewed research papers that contain exaggerated claims of association and statistical - but meaningless - results that I come across is disheartening.

I don't want to be skeptical of every peer reviewed research paper I read, but I am.

The authors of this paper "National Patterns in Environmental Injustice and Inequality: Outdoor NO2 Air Pollution in the United States," are professors in the Civil engineering department at the University of Minnesota.

Their press release for their paper is titled:

Groundbreaking nationwide study finds that people of color live in neighborhoods with more air pollution than whites  Gap results in an estimated 7,000 deaths each year among people of color from heart disease alone

That claim of 7,000 deaths is front and center.

When I read the press release I was skeptical of that claim. I go where the numbers take me. I am just having a difficult time with their findings:

Assuming a 6.6% change in IHD mortality rate per 4.1 ppb NO2 [39] and US-average IHD annual mortality rates (109 deaths per 100,000 people [40]), reducing NO2 concentrations to levels experienced by whites (a 4.6 ppb [38%] reduction) for all nonwhites (87 million people) would be associated with a decrease of ∼7,000 IHD deaths per year.

I got hung up on the "87 million people." Where did that number come from? I don't mean where did they get it from, I mean why did they use it?

Once that brain nugget planted in my thinking as I read, I started trying to figure out what they were going after. I can only speculate as to their motives. What I write about are my questions and research into the validity of their claim.

The best place to start, me thinks, is with this table tucked away in their supporting data.

Looks like they used 2000 census data for the nonwhite population. What I cannot figure out is how they calculated the 14.5 ppb NO2 concentration for nonwhites. That number is critical in their claim of 7000 additional deaths from IHD.

According to their paper, that concentration is "population weighted."

If I understand this correctly, each "Block Group" or "BG" - defined as "the smallest Census geography with demographic data (race-ethnicity, household income, poverty status, education status, and age) reported in the 2000 Census" with a mean BG sizes of "1.1 km2 (urban), 185 km2 (rural),
and 45 km2 (mixed)" and a  mean (standard deviation) BG population of "1,350 (890) people."

The "population-weighted NO2 concentration" was calculated by summing up the concentration of NO2 in the BG multiplied by the population of the group. This sum was then divided by the sum of all of that group's population.

Okay...so I cannot offer an opinion on this method's validity or accuracy. I can, however, ask a couple of questions.

First, based on the data in the Excel spreadsheet they link to, this is what I calculated for the data identified as "urban":

Those averages are calculated from 448 urban areas within the US containing 119,643 BGs and a population of  160.8 million folks.

Call me skeptical, but if you are going to tell me the mean NO2 concentration for whites is 9.9 ppb - which is close to the mean concentration for 448 urban areas - then how do you support nonwhites in these 448 urban areas living in an average concentration of 14,5 ppb?

This means that within an urban area, whites live within the mean concentration of NO2 and nonwhites live in areas two standard deviations outside of this - as an average??

Call me intrigued.

What looks strange to me is this; if the population-weighted concentration is calculated and used, then wouldn't we see a very large difference in the concentration of NO2 for low-income nonwhites when compared to high-income whites?

If environmental justice is in play, pollution is more prevalent in low-income areas. High-income areas would be further away from highways and would have less NO2 spewing industries in their backyards.

If my logic is correct, low-income nonwhites would be exposed to more NO2 than high-income whites. Wouldn't we see a difference in the population-weighted NO2 concentration greater than 4.6 ppb (the difference between the population-weighted mean NO2 concentration for whites and nonwhites)?

Call me confused, but the data in their Excel sheet shows an average Low-Income Nonwhite (LIN) Population-weighted Concentration  of 11.1 ppb for 448 urban areas and an average High-Income White (HIW) Population-weighted Concentration of 9.8 ppb.

Call me perplexed, but that's a difference of 1.3 ppb for these two VERY different groups.

Are we sure about that number of 14.5 ppb NO2 represents the exposure concentration for nonwhites?

And are we sure that 87.000.000 people - the total nonwhite population in the US in the year 2000 - lives in this concentration of 14.5 ppb while all whites live in areas with 9.9 ppb?

Call me skeptical, but that's what I see is the problem with their conclusion. They claim:

Gap results in an estimated 7,000 deaths each year among people of color from heart disease alone.

That's based on an NO2 concentration of 14.5 ppb for 87,000,000 people - which is the total population of nonwhites in the US.


Next post: Part 3: Preventing 7,000 deaths from heart disease alone among nonwhites each year

Part 1: Preventing 7,000 deaths from heart disease alone among nonwhites each year

The press release from the University of Minnesota's web site reads:

A first-of-its-kind study by researchers at the University of Minnesota found that on average nationally, people of color are exposed to 38 percent higher levels of nitrogen dioxide (NO2) outdoor air pollution compared to white people.

The press release goes on to tell us:

...researchers estimate that if nonwhites breathed the lower NO2 levels experienced by whites, it would prevent 7,000 deaths from heart disease alone among nonwhites each year.

7,000 deaths - each year - prevented by breathing less NO2? That number seems....high...really high.

So I went and read their research paper.

An important issue is whether the NO2 disparities described above are relevant to public health. To investigate that question, we consider here one illustrative example: ischemic heart disease (IHD) annual deaths associated with NO2 concentration disparities between nonwhites and whites.

Here is what the author's write:

Assuming a 6.6% change in IHD mortality rate per 4.1 ppb NO2 [39] and US-average IHD annual mortality rates (109 deaths per 100,000 people [40]), reducing NO2 concentrations to levels experienced by whites (a 4.6 ppb [38%] reduction) for all nonwhites (87 million people) would be associated with a decrease of ,7,000 IHD deaths per year.

Since there are numbers quoted and cited, I went looking for [39] to see what those numbers mean.

The "6.6% change in IHD mortality rate per 4.1 ppb NO2" comes from this paper. Table 4 shows the "Relative risks ...for the interquartile range of exposure in each pollutant (i.e., 5.3037 mg/m3 for PM2.5, 4.1167 ppb NO2, and 24.1782 ppb for O3). Values in parentheses are 95% confidence intervals."

Based on the results in this table, exposure to 4.1 ppb NO2 showed a 6.6% increase [(1.066 - 1) * 100] in risk for IHD. Another way to look at this result is:

• The risk of IHD in those exposed to 4.1 ppb of NO2 was 1.06 times as high as the risk of IHD compared to subjects who were exposed to less than 4.1 ppb NO2.

I am not going to question if, or if not, that number of 6.6% is valid. I am going to assume that it is.

Going back to the Clark et al paper, we are told that the annual death rate for IHD is "109 deaths per 100,000 people." Again...I am going to assume that this number is correct.

So we have these numbers. Assuming they are correct, how do they get us to preventing "7,000 deaths from heart disease alone among nonwhites each year?"

...reducing NO2 concentrations to levels experienced by whites (a 4.6 ppb [38%] reduction) for all nonwhites (87 million people) would be associated with a decrease of 7,000 IHD deaths per year.

Oh...I see what they did...

From Wikipedia: The United States Census Bureau defines White people as those "having origins in any of the original peoples of Europe, the Middle East, or North Africa. It includes people who reported "White" or wrote in entries such as Irish, German, Italian, Lebanese, Near Easterner, Arab, or Polish." Whites constitute the majority of the U.S. population, with a total of about 245,532,000 or 77.7% of the population as of 2013.

So...with a total population of 323,730,000 in the US, 22.3% would be "nonwhites" for a total of 72 million. Since the authors don't cite where they got their "87 million people" from, I can only assume it was based on this type of thinking. [Note: I found it in the PDF attachment]

Let's use their number of 87 million (nonwhite) people.

If the incidence of IHD is 109 deaths per 100.000, then we would expect 94,830 of those 87 million to die from IHD each year.

If 4.1 ppb NO2 is associated with a 6.6% increase in risk and the risk of IHD is 109 per 100,000, we should see 109 x 1.066 = 116 per 100,000 or 7 more per 100,000 IHD deaths per year.

Based on my calculation, I get 6,258 more deaths. I think my estimate is correct based on this:

For example, say the absolute risk of a work injury is two per 100 workers. Due to an intervention, it drops to one injury per 100 workers. This yields a relative risk reduction of 50 per cent. Overall, in absolute terms, this means one less injured worker per 100.

All this number crunching is really not the issue here. What is bugging me is how the statement...:

 "it would prevent 7,000 deaths from heart disease alone among nonwhites each year."

...becomes the takeaway message from their research.

That number isn't even close to the reality at hand.

Next post: Part 2: Preventing 7,000 deaths from heart disease alone among nonwhites each year