Friday, January 29, 2016

Gifford-Jones

The Windsor Star keeps promoting Dr. Gifford-Jones and I have to keep scratching my head as to why? ( http://www.donotlink.com/foj6 ). He's just as nutty as Dr. Oz and promotes misunderstanding of science and likes to fear-monger. Here's a debunking of one of his past articles to show you the level of ignorance we are playing with: http://www.skepticnorth.com/2014/02/does-gifford-jones-understand-science/ .

In this latest article, Jones decides to take old news from 2012 and try to make it new again. He exaggerates the risks. In the actual study (http://www.cancer.org/cancer/news/study-examines-possible-link-between-dental-x-rays-and-meningioma-risk) you'll note that the researchers plainly state right in the second paragraph "The study does not prove that dental x-rays cause these tumors". I'm guessing Gifford missed that (note sarcasm). 

You'll also note this as well "The study has some drawbacks that make the link between dental x-rays and meningiomas far from certain. Perhaps most importantly, it relied on participants’ memories about their history of dental x-rays (rather than on dental records themselves)." This makes it very inaccurate and hard to trust the information due to recall bias and the fallibility of the human mind (consider just how poor people remember even minutes after an accident or crime when police get stories from eyewitnesses). 

Now let's figure out the actual risk. I'll be honest here and say I'm horrible at math, so if someone can check and make sure it's correct me if I'm wrong. The point though will still be that the risk is extremely low even with any slight errors. It is estimated according to the University of California that 6500 people in the USA are diagnosed with meningioma. The study was done in 2012, so the population of the USA was 314.1 million. So this gives us a risk of 0.000002% of the population which is extremely low. So if we input the 4.9 times as likely we get the original risk of 0.0000098% of the population. We are dealing with very low risks here, so doubling one's risk of a low number, is still low.

Now while it's true you never want to do more xrays then necessary, there are certain other factors that you should keep in mind that are problematic with the study. As noted by the ADA (http://www.ada.org/epubs/science/2012/april/page.shtml) the study did not appear to be dose related, which one would think would be important (dose make the poison) and that there have been advances in xray technology and imaging which reduces the risk even further (again consider the study was a recall from people aged 20-79 which included remember when they were younger then 10, so that can put the time frame between 1943-2002 for technology, which in itself is very expansive). 

So I urge the Windsor Star to get an actual science/health editor to help prevent nonsense like that Gifford-Jones promotes and to present some real journalism.

How does one know a good scientific study from a bad one?

How does a regular Joe know a good scientific study from a bad one?

This is a very good question. For laypersons (and even those knowledgeable in the area), trying to find out if a certain study is significant or not can be very confusing. It is not always a simple straight answer as there are many factors to consider. None of the following by themselves is necessarily a sure sign of a bad study, but they can definitely be used for the final consideration, especially if the study has multiple infractions.

The first thing I look at is where the study was published. This is usually a good first indication. If a study is published in a respected scientific journal, its odds of being a good one increases. Of course some bad studies still do get published (some are deliberate hoaxes testing the reliability of said journals and their process), but in time they are often retracted. Searching on Google, the top journals can give you decent resources such as:http://www.scimagojr.com/journalrank.php andhttps://en.wikipedia.org/wiki/List_of_scientific_journals.

So how does a study get into a respected journal? There is usually a process of evaluation (part of the scientific method) and review. They are often checking the methodology of the study. It's also good to know what type of study it is.

Is it just a review study looking at certain literature? I find these types are only ideal for proposing further studies and not necessarily for drawing any solid conclusion. Now this shouldn't be confused with a meta-analysis, which looks more in-depth at a larger selection of studies to try to determine a reasonable consensus and include more complex data algorithms and thus hold more weight.

Is the study relying on people's reporting/memory recall (which is often fallible, inaccurate, or prone to bias?) Studies that rely on people reporting from memory how they ate 5 years ago (or even 5 months ago) is problematic.

Was it done with experiments on actual people? Many studies are done just in petri dishes or maybe in just certain animals. While these give some good starting points for pursing further studies, they can't always be relied on to discern the reactions within the human body, which is quite different then a petri dish and rats, for example.

How many people were used? A small number of people used in a study can contain a lot of “noise” in regards to more closely representing what can happen in the general public. A study done on 10,000 people is definitely more robust than one done on 10. For example, you could get 3 out of 10 with a reaction and thus conclude a 30% effect. But in the 10,000 study one could find only 100, which then is only 1%.

Was the study blinded and have controls? Although not always possible, being randomized, blinded and having controls to compare to can increase the robustness of the research. Having it where participants do not know if they are getting a placebo or not is ideal, and even more ideal if the researchers are also blind to that fact as well (usually relying on an impartial 3rd party to keep track of that info for later review).

Did the study list conflicts of interest? A good study will list any conflicts of interest that could bias their results. It's about open honesty. One that hides any conflict raises serious red flags. To find a conflict of interest, one may have to do a bit of searching to discover the conflict of interest.

Has it been peer reviewed? With science part of the process is having your research critiqued. This can be others checking your math or even better seeing if when they repeat your experiment, can they replicate the results. This type of review of a study done by competent peers is the reason some studies have been retracted from journals. They found serious flaws in the data, conclusions and reproducibility of the studies. This is why good science can sometimes take a while, but it eventually corrects itself.

I find that media reports on science studies can sometimes misrepresent or exaggerate what is said in an actual study. Headlines are usually done to grab your attention by asserting a certain point like “chocolate can help with_______ study says.” What is often the case is a certain ingredient was found to have an interesting result (quite often in that petri dish we talked about earlier) and that more study is needed. So don't take headlines at face value. It's always a good idea to look at the actual study to see its conclusions (if any), or you could just ASK BILL wink emoticon (or any other friend who is proficient in finding out that info).

There are many more factors to consider as well, but just knowing some of the above, one can start to get an understanding of what makes a good study and will hopefully be able to disregard some “bad science.” I hope that answers your question.

Carrot Ingredients


Was this year the hottest July?

“Was this year the hottest July (world average) in our recorded history? Somebody showed me some results that said otherwise. I didn't get to check the sources because I was out and it was on his phone. What do the stats and experts say regarding it and where are these false stats coming from?”

ANSWER:

According to reports from NOAA (http://www.ncdc.noaa.gov/sotc/global/201507) , NASA (http://data.giss.nasa.gov/gistemp/tabledata_v3/GLB.Ts+dSST.txt) and JAPAN METEROLOGICAL AGENCY(http://ds.data.jma.go.jp/tcc/tcc/products/gwp/temp/jul_wld.html), this past July (2015) was the hottest recorded.

Some climate change deniers like Joanne Nova (Codling) and Roy Spencer are making claims to the contrary. They are correct in stating that some of the headlines get it wrong, such as those claiming “hottest in 4000 years” because we haven't been recording temperature for that long. The two naysayers though seem to rely on data strictly from UAH satellite data. The problem with this is that satellites do not measure temperature directly, but are inferred from radiance. The UAH data has had problems with the inferred temperatures not matching the actual site surface temperature (http://www.nature.com/nature/journal/v394/n6694/full/394661a0.html). Thus their claims are suspect and need to be investigated further. Even if it wasn't the hottest July on record, it would not indicate a lack of climate change. Variations go up and down in the short term and that is expected. It's the long term trend that is most important and that clearly shows a warming trend.

It is interesting to note that Joanne Nova has a degree in microbiology and molecular biology and not climatology. Roy Spencer is a meteorologist and not a climatologist. The major difference between the two is time frames. Meteorologists produce forecasts in a window no larger than 10 days and usually localized (weather). Climatologists are concerned with long term climate conditions. When considering sources, one would want it from the highest quality and someone actually in the field of studying climate. 97% of publishing climate scientists agree that humans are causing global warming.

To learn more, I suggest taking this free, self-paced course: https://www.edx.org/course/making-sense-climate-science-denial-uqx-denial101x-0

I hope this answers your question.