Climate Science Glossary

Term Lookup

Enter a term in the search box to find its definition.


Use the controls in the far right panel to increase or decrease the number of terms automatically displayed (or to completely turn that feature off).

Term Lookup


All IPCC definitions taken from Climate Change 2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Annex I, Glossary, pp. 941-954. Cambridge University Press.

Home Arguments Software Resources Comments The Consensus Project Translations About Support

Bluesky Facebook LinkedIn Mastodon MeWe

Twitter YouTube RSS Posts RSS Comments Email Subscribe

Climate's changed before
It's the sun
It's not bad
There is no consensus
It's cooling
Models are unreliable
Temp record is unreliable
Animals and plants can adapt
It hasn't warmed since 1998
Antarctica is gaining ice
View All Arguments...

New? Register here
Forgot your password?

Latest Posts


Gas stoves pose health risks. Are gas furnaces and other appliances safe to use?

Posted on 20 March 2023 by Guest Author

This is a re-post from Yale Climate Connections by Sarah Wesseler. While predicated on accepted scientific findings, this article includes conclusions of the author and is presented to our readers as an informed perspective.

Poor air quality is a long-standing problem in Los Angeles, where the first major outbreak of smog during World War II was so intense that some residents thought the city had been attacked by chemical weapons. Cars were eventually discovered to be a leading cause of smog, but they weren’t the only ones. In 1978, the regional air quality authority created regulations aimed at reducing pollution from a surprising source: gas-powered water heaters found in homes throughout the city.

Gas stoves have become an unlikely front line in the culture wars thanks to growing awareness of their contribution to health problems like childhood asthma, not to mention their links to climate change. But the other gas-fueled appliances found in many American homes — water heaters, furnaces, and clothes dryers, to name a few — have received far less attention, although they also pose risks to public health and the environment.

“I’m not here to scare folks,” said Brittany Meyers, the national director of healthy indoor air policy at the American Lung Association. “But that said, we do know that there are impacts to both indoor and outdoor air quality that come from the burning of fuel inside the home that is vented outside, which is the appliances you’re talking about.”

Gas appliances can give off toxic carbon monoxide and other air pollutants

Approximately half of American households rely on gas appliances for heat and hot water. According to the Census Bureau, piped natural gas powered around 61 million water heaters, 58 million furnaces, and 20 million clothes dryers in 2021. Other common gas-powered appliances include fireplaces (approximately 7 million), air conditioners (around 2 million), and space heaters.

Health risks associated with gas appliances center on the chemical composition of the fuel they burn. Natural gas consists primarily of methane, a powerful heat-trapping gas that’s partly responsible for climate change, which the World Health Organization has called “the single biggest health threat facing humanity.” But it also contains other substances that, when released into the air through leaks or incomplete combustion, can more directly harm human health. The best known of these, carbon monoxide, causes at least 420 accidental poisoning deaths each year in the United States.

Drew Michanowicz, a scientist at research institute PSE Healthy Energy, has carried out several studies to understand exactly what’s in the natural gas that enters American homes, taking samples from more than 200 residences in California and Massachusetts. “They pretty much all contain a small suite of hazardous air pollutants that clearly we would not want to be exposed to,” he said. Among them: nitrogen dioxide, benzene, formaldehyde, particulate matter, and additional substances linked to asthma, cancer, heart disease, and other health problems.

Venting reduces indoor pollution from gas appliances — but it’s not foolproof

The growing concern over gas stoves stems largely from the fact that cooking appliances release these chemicals directly into areas of the home where residents spend much of their time. Stoves often aren’t vented to the outdoors when in use, either because the kitchen isn’t equipped with an exterior vent or because occupants don’t turn it on.

Outside of the kitchen, however, the most common gas-powered appliances — water heaters, furnaces, and clothes dryers — are required by building codes to be automatically vented outdoors.

This divergence in venting regulations reflects the fact that non-cooking appliances burn much more gas than stoves do, and therefore emit much more pollution.

“An average home in the Northeast uses like 60 MMBtu [metric million British thermal units, a standard unit for measuring the heat content of energy sources] per year for space heating and another 20 for water heating,” said Matt Rusteika, the director of market transformation at the Building Decarbonization Coalition, a nonprofit focused on building electrification. “By comparison, you only use about two MMBtus per year of gas if you have a gas stove.”

A robust set of government regulations and industrial standards has been developed to counter the threat of indoor air contamination from these appliances, said Iain Walker, a building scientist at Lawrence Berkeley National Laboratory who specializes in residential ventilation and home decarbonization. “Effectively, no matter where you go in the country, the same requirements are going to be met about the size of the vent and so on and how it’s all connected,” he said.

As a result, as long as these appliances are installed properly, the risk that they will release harmful pollution into homes is low, Walker said. “The vast majority of times, [for] a heating device, including water heaters, with a correctly installed flue, you are not going to have any problems.” Groups like the Lawrence Berkeley National Laboratory and weatherization programs throughout the country have amassed a substantial body of evidence confirming that this approach is effective, he said.

But though the system generally works, sometimes it doesn’t. Walker cautioned against undue alarmism, but he said that indoor air pollution from gas appliances is very difficult to eliminate completely. “Unfortunately, we know enough to know that it’s still an intractable problem.”

Joe Roy of energy efficiency company CMC Energy Services said that over the course of thousands of home energy assessments conducted in Connecticut, the company’s technicians have found malfunctioning gas appliances or gas lines in approximately 15% of all residences built after 1980. In homes dating from before the Reagan era, this figure climbs to 25%.

Leaking gas lines are the most common source of unwanted chemicals that enter the living space, Roy said, particularly in older homes.

But a variety of issues can lead to pollution from vented appliances themselves. A furnace component known as a heat exchanger may crack over time, which can lead to dangerous levels of carbon monoxide being distributed through a home’s duct system. Water heaters can experience a phenomenon known as backdrafting that causes combustion byproducts to enter the living space. Residents sometimes move clothes dryers without realizing that an easily damaged gas line lies behind them, with predictable results. And a practice known as sidewall venting, in which high-efficiency gas equipment is vented through a hole in the wall rather than through the chimney, as was standard practice in the past, occasionally leads to problems with debris, bird nests, snow, or other substances blocking the outlet air.

In new ruling by key engineering standards body, unvented gas appliances are deemed unsafe

A separate class of gas appliances that aren’t required to be vented to the exterior, including fireplaces and heaters, poses a different set of challenges. These machines can be cheaper and more flexible than their vented cousins due to the lack of need for a flue, but concerns have been raised about their safety.

The regulatory environment for these appliances is changing because of a recent update to the engineering standards that underlie most local building code ventilation requirements in the United States.

The Lawrence Berkeley National Laboratory’s Walker, who recently served as chair of the committee tasked with this update, said that one of the core debates during the process centered on how long users leave these appliances on during a given session — and therefore how much pollution might accumulate in the living space.

“Historically, these things have been allowed because the concept was that they would be low-capacity and not operate very much,” he said. “A lot of the wrangling [in the recent committee] has not been over, ‘Does combustion produce contaminants?’ [Instead] the discussions were about, ‘Well, what if it turns on and operates for several hours? Then what?’”

The committee’s final recommendation was to prohibit the use of unvented gas space heaters or fireplaces in homes seeking to meet the engineering standard, Walker said.

Gas furnaces and other appliances can cause outdoor air pollution and harm health

Although venting gas appliances is the best way to reduce risk within the home, emissions don’t simply disappear when they make it outdoors, but instead circulate within communities.

“There’s millions of boilers and furnaces and water heaters, and they’re all pumping combustion fumes up into the air, and it’s bound to affect people’s health,” said the Building Decarbonization Coalition’s Rusteika.

One key area of concern is nitrogen oxides, or NOx, a family of chemicals formed by fuel combustion in air that has been linked to respiratory problems like asthma, in addition to cardiovascular and other health issues. NOx also reacts with other chemicals in the air to form smog, which is itself associated with a variety of health problems ranging from reproductive harm to early death.

Several recent reports have called attention to gas appliances’ role in outdoor air pollution in California. One 2022 publication jointly produced by SPUR, a nonprofit focused on California cities, and environmental nonprofits Sierra Club and RMI explored the links between residential and commercial appliances and smog. It found that these machines generate approximately four times as much NOx as the state’s electric utilities and around two-thirds as much as its light-duty passenger vehicles.

Another 2022 report from the UCLA Fielding School of Public Health found that if all residential gas appliances were replaced by clean electric models, the reductions of NOx and particulate matter in the outdoor air would lead to 354 fewer deaths and 596 fewer cases of acute bronchitis each year in the state.

Detailed information from other parts of the nation is harder to come by. The American Lung Association’s Brittany Meyer said that there’s a need for more research on how gas appliance emissions affect outdoor air quality. One of the surprises of the association’s 2022 literature review on the health impacts of residential combustion was “the lack of studies on some of the health impacts of vented combustion-based appliances on people not directly using the appliance,” she said.

But Seth Hartley, an atmospheric scientist at consultancy ICF and a co-author of the 2022 report, said that existing data is clear on the broad outlines of the issue, if not the details. “We know that [these appliances] emit NOx. We know that NOx causes smog. We know that NOx is a lung irritant and causes other things like that. But to quantify that relationship — still, there’s a lot missing.”

A growing focus on the health risks of gas appliances

To date, building code–mandated ventilation has been the primary mechanism through which U.S. governments at any level have sought to limit health risks from gas-powered appliances.

“The U.S. EPA doesn’t regulate indoor air quality, so there hasn’t been that much focus, really, on controlling the indoor air quality aspects of these [appliances]. And generally, I don’t think there has been that much focus on the outdoor impacts, either,” Hartley said.

There are exceptions, however. Other regions of California have joined L.A. in setting limits on NOx emissions from appliances, and the state’s recent ban on sales of new gas furnaces and water heaters, which goes into effect in 2030, was motivated in part by air pollution concerns. Utah has NOx emission limits for water heaters, as does Texas for small water heaters and boilers. And in New York State, a recently passed law seems likely to lead to new emissions limits on appliances, according to the Building Decarbonization Coalition’s Rusteika.

In general, policy responses to health impacts from gas appliances are “starting to bubble up,” Rusteika said. “The overlap with climate is maybe mainly what’s driving that movement in some places. In other places, the focus is on NOx, which we’ve been regulating in other sectors for a long time.”

How to make sure gas appliances aren’t making you sick in your home

People who are concerned about the impacts of these appliances in their homes can take steps to reduce the risk of indoor air pollution. The Centers for Disease Control and Prevention recommends installing a battery-powered carbon monoxide detector on each level of the home and having all gas-powered appliances professionally serviced each year.

According to Joe Roy of CMC Energy Services, it’s critical to ensure that this regular maintenance includes an assessment of the home’s gas lines. So make sure that the HVAC contractor’s scope of work includes checking gas lines for leaks, Roy said.

Consumer-grade NOx monitors can also help provide peace of mind within the home, he said.

Another option for mitigating the risks posed by gas appliances is to take advantage of Inflation Reduction Act incentives for buying and installing their electric counterparts.

Rusteika said that rising sales of heat pumps show that ordinary Americans are increasingly convinced of the benefits of moving away from gas.

“People are used to burning fuel in their house, but people used to be used to sitting next to people smoking in restaurants as well,” he said. “We know that electrification is essential for the climate. We also know that electrification improves indoor air quality. We know that it’s more comfortable. We know that it can save you a lot of money in certain contexts. And so now is the time, right?”

0 0

Printable Version  |  Link to this page


Comments 1 to 19:

  1. The gas stove / asthma study uses a population attribution fraction (PAF) analysis. PAF analysis assumes causality based solely on correlation which is a very weak methodology. Below is a link to two articles describing the weakness of PAF (one of which is a PAF analysis with obesity, though the shortcomings in the PAF analysis are similar).

    The second obvious weakness in the study are simple math computational errors. For example gas stove use in Pennsylvania is 79% while florida, the gas stove use is only 9%. The difference in the rate of asthma is approx 1% (8.5% vs 7.3%). If the 12% causation was a reasonable estimate, then the difference in asthma rates would be much larger. 6-10%.

    What is disappointing is that this study has not only received tremendous fanfare, but has been embraced by groups that should have recognized the weaknesses and shortcomings in the conclusions.


    0 0
  2. David-acct:

    PAF is widely used in epidemiology, where controlled experiments are difficult (and run into ethics complications - you can't really select children from across the country and force them to live in houses with/without gas stoves and no other differences in their lives).

    The paper you link to does not give a blanket condemnation of PAF studies - it just points out that there are issues that need to be considered. In the Introduction, it states (emphasis added):

    " certain settings, the value of a population attributable fraction estimate may be questionable..."

    ...and in their conclusion, they state:

    The assumptions underlying valid population attributable fraction estimation include the following...

    ...Such assumptions are often not justified. Those who present population attributable fractions have a duty to ensure that estimates are correctly computed and that their limited meaning is correctly communicated...

    Do you dismiss PAF in its entirety? The reference you give does not, as far as I see.

    As for the specific study mentioned in this OP - I presume that this is the link you are referring to - you make your own calculation comparison using a 12% value. When I read the paper, I see that it gives 12.7% as a national value. In the abstract, it states:

    "We found that 12.7% (95% CI = 6.3–19.3%) of current childhood asthma in the US is attributable to gas stove use.

    ...and the same sentence appears in the results section, referring to figure 1. We also see in the conclusion:

    In conclusion, 12.7% of current childhood asthma nationwide is attributed to gas stove use, which is similar to the childhood asthma burden attributed to secondhand smoke exposure.

    But figure 1 of their paper also gives state-specific estimates of PAF. Pennsylvania's PAF is 13.5%, while Florida's PAF is only 3%. In their Materials and Methods section, they state (emphasis added):

    ...we used the summary statistics of the proportion of children exposed to gas cooking to generate distributions of proportions nationally and for each state for which we have data.

    It looks to me like they have accounted for gas stove use, and calculated individual state PAFs, and Florida shows a low PAF for asthma attributable to gas stove use.

    In your calculation, you seem to be assuming that gas stove use will be the only factor affecting asthma rates, which is clearly not what PAFs look at. The PAF of 3% for Florida suggests to me that this study is entirely consistent with most Florida asthma being due to other causes.

    Why did you pick the 12.7% national rate, and exactly what do you think the "computational error" in the study is?

    0 0
  3. David-Acct,

    In post 1 You say:

    "The second obvious weakness in the study are simple math computational errors. For example gas stove use in Pennsylvania is 79% while florida, the gas stove use is only 9%. The difference in the rate of asthma is approx 1% (8.5% vs 7.3%). If the 12% causation was a reasonable estimate, then the difference in asthma rates would be much larger. 6-10%."

    Bob Loblaw shows that the paper actually says:

    "We found that 12.7% (95% CI = 6.3–19.3%) of current childhood asthma in the US is attributable to gas stove use."

    That would mean that gas stoves increase the asthma rate in Pennsylvania by about 8.5 x .127 x .79= 0.85% and Florida by about 7.3 x .127 x .09 = .079%.  Your claim that " the difference in asthma rates would be much larger. 6-10%." is simply a gross computational error on your part.  I note that Florida in general has cleaner air than Pennsylvania which probably accounts for the slightly higher asthma rate in Pennsylvania.

    In general it is best to figure peer reviewed papers do not have computational errors unless someone who is qualified to make the claim has produced evidence first.  The problem here is you have made a gross calculation error, not the peer reviewed paper.

    0 0
  4. My calculation is slightly off.  The actual asthma rates due to gas stoves would be a little higher.

    I never accept calculations by amateurs on line, even my own.

    0 0
  5. I was going to wait and see if David-acct wanted to defend his calculations and claims of errors in the paper, but Michael has advanced the issue.

    Looking at the original paper, they state in the abstract:

    Effect sizes previously reported by meta-analyses for current asthma (Odds Ratio = 1.34, 95% Confidence Interval (CI) = 1.12–1.57) were utilized in the PAF estimations.

    An Odds Ratio of 1.34 means that there is a 34% increase in the likelihood of getting asthma if a child is exposed to a gas stove (versus not). The Odds Ratio is the number that you need for RR, in their equation that calculates PAF:

    PAF = {p × (RR − 1 )} / {p × ( RR − 1) + 1}

    where p is the proportion of households with children exposed to gas stoves and RR is the relative risk of developing asthma given exposure to gas stoves.

    When you stick in  RR= 1.34 and appropriate values for p for Illinois and Florida (0.791 and 0.091), you get the PAF values in the paper (0.211 and 0.03 respectively).

    So, I am not seeing any "simple math computational errors" in the paper.

    Over to you, David-acct.

    0 0
  6. Also, note that the paper itself says:

    In Illinois, for example, approximately 79.1% of households with children cook with gas, whereas in Florida, the figure is only 9.1%.

    I think that when David-acct said Pennsylvania, he might have meant Illinois. The paper does not appear to give a figure for the number of households with children that have gas stoves. David-acct may be getting numbers from somewhere else?

    0 0
  7. Bob -You are correct , I inadvertantly misstated Pennsylvania instead of Illinois

    I took a second look at the asthma rates for both adults and children by state and the gas stove use by state from the

    state - gas stove % - adult asthma rates - child asthma rates

    maine - 7% - 10.6 - 8.0

    new york 62% 9.5 - 8.2

    Michigan - 45 - 11 - 8.4

    Missouri - 27% - 9.3 - 8.9

    Rhode Island 36% - 12.1- 9.5

    mississippi - 17% - 8.9 - 9.6

    georgia 35% - 9 - 9.8

    pennsylvania 37% - 10.2- 11.5

    DC 62% - 10.4 -11.7

    Conn 27% - 10.6 - 11.8

    California 70% - 9.3

    nevada 60% 9.5

    Illinois - 67% - 8.5 - 9.5

    Alabama 18% - 9 - 12.3

    As the data above shows, there is near zero correlation with asthma rates and gas stove use which should have raised red flags. NY has 9x the gas stove usage of Maine yet has near identical asthma rates which is a strong indication that gas stoves have little if any effect on asthma rates.    

    0 0
  8. David-acct:

    You just made a classic mistake again - assuming that gas stoves are the only cause of asthma.

    Many other factors can cause asthma - for example, there are may other factors that influence air quality, and many other factors that influence health conditions that can correlate with asthma.

    You need to properly account for all those other factors before you can start to say that you have data that reflects only the effect of gas stoves.

    I think it is you that is making simple errors - errors in statistical modelling.

    0 0
  9. Bob  - that is not the claim that I made.

    It was the study that made the claim that  approx 12% of asthma cases where caused by gas stoves.  

    I only pointed out the complete lack of correlation of asthma rates and gas stove usage which renders the conclusion of that study very suspect.  

    Bob - its true that there are many factors that contribute to asthma - though keep in mind the PAF analysis used in this study  only takes in the single factor -gas stove usage.  As I detailed above, the correlation is non - existant.   Further, the lack of correlation raises the question as to whether paf analysis is even appropriate for searching for causality of asthma and gas stove usage.    

    0 0
  10. David -acct: I figured you would double-down on your misunderstanding.

    The lack of correlation in the data you provided does not mean what you think it means. Let me show an example, using a made-up set of values and a fictitious cause-effect relationship. The advantage of this is that I can generate values from known mathematical relationships, then apply your statistical test, and see what happens.

    We will start with your per-state values of gas stove use. We will then assume that the world works as follows:

    • For some strange reason, every tenth gas stove purchase came with a free ABBA CD, so that 10% of the households that own gas stoves also own an ABBA CD that came from the stove purchase.
    • But gas stove owners also might have bought ABBA CDs on their own. It turns out that a survey asked people if they liked ABBA music in each of those states, and if they had bought an ABBA CD to fill that desire. It turns out that in every state, 50% of the people that liked ABBA music had bought a CD. (The other 50% either listen to it on the radio, stream it, have MP3s, etc., or like it but don't listen to it on their own.)
    • So, when you take a count of the number of ABBA CDs in houses with gas stoves, the total number of ABBA CDs comes from two sources - the gas stove purchase, and people that bought them on their own.

    So, let's look at what the data table from this would look like:

    Stategas stove %Like ABBA musicTotal ABBA CD ownership
    maine 7 19.8 10.6
    new york 62 6.6 9.5
    Michigan 45 13 11
    Missouri 27 13.2 9.3
    Rhode Island 36 17 12.1
    mississippi 17 14.4 8.9
    georgia 35 11 9
    pennsylvania 37 13 10.2
    DC 62 8.4 10.4
    Conn 27 15.8 10.6
    California 70 4.6 9.3
    nevada 60 7 9.5
    Illinois 67 3.6 8.5
    Alabama 18 14.4 9

    Now remember, we made this data up - at least, I made up the numbers for the "like ABBA music part" - so we know the exact relationships: gas stove ownership causes 10% of those owners to own an ABBA CD, and people that like ABBA music have a 50% chance of owning an ABBA CD.

    The final column is an exact match for your adult asthma rates. When we apply your correlation test, the correlation between gas stove use and ABBA CD ownership will be the same as your asthma rates - in spite of the fact that the data were generated using a model where there is a perfect relationship between owning a gas stove and an increased likelihood of owning an ABBA CD.

    In your analysis, you have no idea how "other factors" vary from state to state, and you have no idea what the relationships are between those factors and asthma rates, so you cannot determine from your data whether gas stoves have an effect or not. You are missing the "Like ABBA music" column and how it affects total ABBA CD ownership, and you are drawing a very, very wrong conclusion about the effect of gas stove use.

    You never bothered to read enough of the original paper to understand where they got their Odds Ratio of 1.34, did you?

    0 0
  11. As a further examination of the ABBA music scenario I presented in comment #10, let's see if there is a statistical process that can discover the underlying relationships amongst the variables in the table I created.

    Remember: David-acct has claimed in comment #9 that the data he presented in comment #7 indicates a "...complete lack of correlation of asthma rates and gas stove usage which renders the conclusion of that study very suspect. "

    So, in comment #10 I created a similar data set, called it "ABBA Music" instead of asthma rates, and created an extra artificially-generated column ("Like ABBA music") that follows an exact model that explains all the variation in ABBA CD ownership. Basically, what I have done is created a model of the form:

    Y = aX1 + bX2

    where a and b are constants (with values of 0.1 and 0.5 respectively), X1 is gas stove use, and X2 is "Like Abba Music".

    David-acct has said "Y is not correlated to X1", which is correct when you completely ignore that X2 is a factor. If we do a linear regression between Y and X1, we get a slope of -0.007, an intercept of 10.2, and an r2 of 0.03. But that analysis assumes that b = 0, which we know is incorrect. The regression result does not get the correct value of a = 0.1 with an intercept of 0.

    Likewise, if we do a linear regression of Y vs. X2, we get a slope of 0.11, an intercept of 8.6, and an r2 of 0.30.  A better fit, but still wrong. We should get a slope of 0.5 and an intercept of 0.

    So, finally, we do multiple linear regression where we fit Y to both X1 and X2, and we find that a=0.1, b=0.5, the intercept is 0, and r2 = 1. This regression correctly identifies the effects of both X1 and X2.

    Why does David-acct's analysis go wrong? In my synthetic ABBA music example,  the values I used for "Like Abba music" are strongly (negatively) correlated with gas stove use: slope -0.21, intercept 20.3, and r2 = 0.84. David-acct's "no correlation" ignores confounding variables - other factors that vary from state to state and can affect asthma rates, and affect the validity of statistical testing.

    And in my "synthetic" ABBA music example, the numbers I use for gas stove % are exactly the same as David'acct's values, and the numbers I use for ABBA CD ownserhip are the same as David-acct's adult asthma rates. If his analysis does not work for ABBA music, it will not work for asthma rates, either.

    0 0
    Moderator Response:

    [BL] Self-edit. The original comment text had "adjusted r2" values, not raw r2 values. Values corrected to raw r2 values, which are slightly higher (more positive).

  12. Another note on David-acct's comment #9. He states:

    "...though keep in mind the PAF analysis used in this study only takes in the single factor -gas stove usage.

    This shows a misunderstanding of what PAF is and how it is used. Yes, it is looking at what would change if gas stoves were eliminated - a single factor - but the analysis of the effect of gas stoves on asthma is not represented by the PAF values. In comment #5, I showed the equation used in the study in question:

    PAF = {p × (RR − 1 )} / {p × ( RR − 1) + 1}

    and pointed out that RR is referred to in the paper as the Odds Ratio, with a value of 1.34. That is the number that tells us how gas stove use is related to other asthma causal factors. And the paper cited in the OP does not derive that Odds Ratio from their own data analysis, they do a meta-analysis of other papers that have specifically looked at asthma in more detail. If David-acct questions whether gas stoves affect asthma, he needs to look at those other papers cited in the one we are talking about here.

    The Odds Ratio tells that when all other asthma factors are taken into account, adding gas stoves to the mix increases the likelihood of asthma by 34% (1.34 times the background levels without gas stoves.)

    The PAF (equation above) just takes that gas stove effect and adjusts each state (or the nation) for actual gas stove usage. Obviously, eliminating gas stoves will have no effect on households that did not have a gas stove to begin with. If all houses had gas stoves, and we eliminated all, then the Odds Ratio of 1.34 would mean a reduction of 25% in asthma rates. (1/1.34 = 0.75). Fewer gas stoves? Less reduction through elimination of gas stoves.

    0 0
  13. Bob

    This argument boils down to whether the PAF metholodogly is valid in this case.  If the study was robust, then there would be some positive level of correlation.  As I have shown above, there is near zero correlation.


    As I previously noted , and as I linked above, PAF doesnt work when there are multiple cofounding variables.  With Asthma there are far too many variables which prevents an analysis using PAF to be robust.



    Here is another article on the short comings of PAF with multiple variables.


    The money quote in that article is the first sentence of the conclusion

    "As discussed previously in this article and as stated by Kempthorne (14), attempts to partition causality when multiple forces act together to produce the outcome are meaningless."




    0 0
    Moderator Response:

    [BL] Second link linkified.

  14. I will also note that the gas stove study was a meta study which limited the studies used to only those in North america and some europe.


    The selection omitted a much newer. more comprehensive and robust study. (by design?)

    0 0
  15. David-Acct,

    The study you link is much older than the stuy Bob Loblaw linked.  In fact, the study that Bob linked only reviewed studies that were pubished after the study you linked.  Your statement "a much newer. more comprehensive and robust study" is incorrect.  Bob's study also only used data from the USA and Europe while your link used worldwide data, mostly from open fires.

    The study that Bob linked used data from 9 states.  They are not the same states that you produced data for in post 7.  You have not linked where you obtained the data you posted.  Where did you get the data in post 7?  I note that the study Bob linked used all the states where the most comprehensive data was available for.  How were the states you selected chosen?  Was a list of all the states reviewed and these picked out for some reason? 

    0 0
  16. David-acct:

    Now you are tripling down on your erroneous analysis.

    As I have pointed out, your lack of correlation is meaningless. Your methodology gets no correlation even when applied to a set of data that is completely explained by two variables, including the one that you claim has "no correlation". A bogus search for correlation - by ignoring confounding variables - is not evidence for anything.

    And, as I have pointed out, PAF is not the measure that used to determine whether or not a single variable affects an outcome - it is an application of other results (the Odds Ratio, in this case) where gas stoves have been previously assessed against other possible causes of asthma.

    Your continued focus on PAF is unfortunate.

    Your 'by design?" quip in comment 14 is an accusation of dishonesty on the part of the authors of the study in question. That is a serious accusation.

    Have you provided the wrong link for your "much newer, more comprehensive" study mentioned in comment  14? That link leads to a paper from 2013 - the same year as the metastudy that is the first reference in the paper mentioned in the original blog post. In fact, the paper that you linked to is listed in the references of the metastudy. How was this study "omitted" when it is part of the chain of references?

    The authors also did a search for newer material, and assessed a number of papers, but none had any newer material. They state, in their Materials amd Methods section:

    Full manuscripts (n = 27) were independently reviewed by co-authors; none reported new associations between gas stove use and childhood asthma specifically in North America or Europe.

    As for the use of data from North America and some of Europe: the authors of the study referenced in the OP are explicit regarding this in their Materials and Methods section (emphasis added):

    We combined effect sizes for North America and Europe given the similarities in housing characteristics and gas-stove usage patterns across these geographies.

    What this means is that they are looking for results that can reasonably apply to the U.S. with minimum difficulty. Selecting data that comes from similar environments reduces the likelihood that there are confounding variables that are not controlled for. This is good scientific practise: search for relevant results that are directly applicable to your use case. Do not include results that have factors that are not present in the area that you are analyzing.

    Frankly, you are clutching at straws to try to dismiss the results of this study. It is clear that you do not understand the methodology, you do not understand the statistics, and you do not understand the chain of supporting evidence referenced in the sequence of papers cited in the relevant studies. You keep missing essential details in the papers that cover the "criticisms" that you are making. The only parts you seem to be reading are the parts that you erroneously seem to think invalidate the study.

    0 0
  17. To follow up on Michael Sweet's comment 15, pointing out that the study that David-acct has linked to is related to open fires, here is the relevant quote from the Interpretation section:

    The use of open fires for cooking is associated with an increased risk of symptoms of asthma and of asthma diagnosis in children. Because a large percentage of the world population uses open fires for cooking, this method of cooking might be an important modifiable risk factor if the association is proven to be causal.

    (The full paper is behind a paywall. Onlne availability at The Lancet is limited to this.)

    Questions for David-acct:

    • How may households in the U.S. use open fires for cooking?
    • Are open fires used for cooking expected to be a significant contributor to asthma in the U.S.?
    0 0
  18. And finally, in comment #13, David-acct refers to another article critical of PAF, and provides what he calls a "money quote".

    What David-acct does not do is include other quotes from the same paper's Conclusions section.

    The first quote I will provide talks about assumptions related to the use of PAFs:

    With respect to interpretation of an AF as the proportion of disease risk that could be eliminated if the excess risk associated with exposure were to be eliminated, there may be valuable meaning under a specific set of assumptions. In addition to the assumptions commonly listed in textbooks, there is one more critical assumption: that we can envision a specific intervention that will cause the estimated reduction in risk in the exposed while changing no other risk factor distributions.

    The second quote I will provide is the final paragraph of those conclusions (emphasis added):

    This paper is not an argument for never computing a population AF. It is an argument for more clarity, justification, and complex thinking when using this measure. AFs are only a beginning of the discussion of the public health consequences of intervening to reduce the prevalence of risk exposures.

    Once again, the paper does not support the argument that David-acct is trying to make. From his first post, he has been trying to dismiss the results of a study simply because it used PAF. None of the references he has provided represent a blanket condemnation of PAF that he wants to think they do. They recommend caution, which is an argument applicable to any statistical method - for example, over-stating the significance of a correlation (or lack thereof) that we have seen in certain comments here.

    0 0
  19. A quick reference to a news item I came across this week. Earlier this month, Desmog had an article about this gas stove issue, with the under-the-headline statement "The American Gas Association is trying to discredit research on the health impacts of gas stoves today. But newly revealed documents show it was discussing indoor air pollution concerns five decades ago."

    As described in the Desmog article, the pattern of behaviour by an industry trade group looks awfully familiar. It is a lot like the tactics that we have seen from the fossil fuel industry over climate change issues - and by the tobacco industry over smoking and cancer issues.

    0 0

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.

The Consensus Project Website


(free to republish)

© Copyright 2024 John Cook
Home | Translations | About Us | Privacy | Contact Us