kcat
Lv 4
kcat asked in Science & MathematicsMedicine · 1 decade ago

Statistics question about quartiles: Need help understanding a journal article.?

At http://jama.ama-assn.org/cgi/content/abstract/286/... under results, it says:

------

Baseline levels of IL-6 (P<.001) and CRP (P<.001) were significantly higher among cases than among controls. The relative risks of future Diabetes for women in the highest vs lowest quartile of these inflammatory markers were 7.5 for IL-6 (95% confidence interval [CI], 3.7-15.4) and 15.7 for CRP (95% CI, 6.5-37.9).

-----

Ok so women with more IL-6 in their blood were more likely to develop diabetes. First of all, what does P<.001 mean?

Secondly, "The relative risks of future DM for women in the highest vs lowest quartile of these inflammatory markers were 7.5 for IL-6 (95% confidence interval [CI], 3.7-15.4)"

Does this mean they measured the IL-6 content of women's blood, subtracted the lowest quartile from the highest quartile and got 7.5 mg/L? Also, does it mean that the first quartile number was 3.7 mg/L and the third quartile was 15.4 mg/L? Or are 3.7 and 15.4 the min and max values or....?

Or if not, what does this mean?

2 Answers

Relevance
• 1 decade ago
Favorite Answer

"P<0.001" is talking about the chance that their conclusions are wrong. In statistics, there is a thing called a "p-value" (which is what "P" stands for). A p-value is the probability that a person's conclusion is wrong, and it is calculated using statistics.

In this article, the researchers found out that their p-value was less than 0.001. This means the chance that their results are wrong is less than 0.001 or 0.1%, which is a good thing (you want your results to have little error). It's saying that there is a 0.1% chance their results just happen to be wrong due to random chance - kind of like saying "s--- happens."

In the context of this article, they're saying the chance that the case studies (diabetic women) and controls (women who don't have diabetes) have the same baseline levels of IL-6 and CRP is less than 0.001 or 0.1%. This means they can safely say that is the wrong conclusion, and instead say that IL-6 and CRP levels are statistically higher for diabetic women than they are for non-diabetic women.

Now you might be wondering, why use 0.001? 0.001 is just a number they chose to compare their p-value to. It's a number that in statistics is called a "significance level". If your p-value is less than the significance level, then you can say with fairly good certainty that your results are correct (and any other conclusions are wrong).

For your second question, I think you're right, when you say the lower quartile is the first quartile and the higher quartile is the third quartile. But I don't think the first quartile is exactly 3.7 and the third quartile 15.4. 3.7 and 15.4 refer to the min and max values of a confidence interval. That's something totally different in statistics, and I won't explain it if you don't want me to.

Hope this helps. I know that this may be too much information, but I wanted to explain the best way possible. Any questions?

Source(s): UW pharmacy student. I have taken biostatistics in pharm school.
• Login to reply the answers
• 1 decade ago

First a few definitions for your further understanding of how they get these numbers:

The P-value is a statistical value that I could not fully explain because I am a biology major and not a mathematics major but I do know that an acceptable P-value for a study is <.05. Your study has P<.001. This strengthens the data by saying that it is statistically significant that the women with more IL-6 in their blood were more likely to develope DM. In other words, it is not random.

The most important thing if you are seeking this information for yourself is to see how much you have in common with the test group. If you are African decent and grew up in the USA and the test group is asians who live in siberia than this study a) is not relevant to you and b) needs to be repeated with different or more diverse populations.

Relative Risk (RR), in a nutshell, is the RATIO of risk. In this context with the quartile comparison, it means that if you take all the participants and divided the results into quartiles (quarters or 1/4) then the RR of the Highest 1/4 compared to the lowest 1/4 is 7.5. Translated to english...people in the higher 1/4 were 7.5 times (also read 7500%) more likely to develop DM than those in the lower 1/4. for example and perspective if RR=1 than they would be equally likely and if RR=0.5 then the higher 1/4 would be at 50% less risk.

Confidence Interval (CI) is another statistical number that says: I am 95% confident that all values fall between 3.7 and 15.4. This is also acceptable.

I know this is wordy, but you can be assured that if an article is published in a reputable journal the data is sound and the study put together well.

The fact that this is a nested case control study adds strength to the results. In this case the original study is a cohort (pick people and see if they develop the disease) and from that they chose the group of people who developed the disease and did a case-control study (pick people with the disease and see what is different)

Source(s): I am an undergraduate student in Microbiology and Molecular Biology and I read these journals regularly...and as you can see I know what they mean
• Login to reply the answers
Still have questions? Get your answers by asking now.