3 ms·

I wonder if some of this is due to a subtle intervening variable like conscious or subconscious (i.e. fidgeting) decisions to exercise more accumulated over tim

by cvsh 9y ago

I wonder if some of this is due to a subtle intervening variable like conscious or subconscious (i.e. fidgeting) decisions to exercise more accumulated over time because of increased energy, rather than caffeine per se.

MichaelBurge 9y ago
If that were true, decaf wouldn't show the same effect because it wouldn't give you increased energy. (I don't drink decaf)
- devotedtoneu 9y ago
  Well decaf does contain caffeine, just less of it.
  - viridian 9y ago
    Typically less than a twentieth of it with most modern caffeine methods.
epmaybe 9y ago
In epidemiological studies we refer to these variables as confoundimg variables. There's another term as well that I can't seem to remember. When you try and correlate one specific variable, like coffee, to trends like mortality, you must control for every other variable. Socioeconomic factors like income generally cause confounding of your results. For example, there was a recent paper that showed that women with gum disease were at higher risk for certain cancers. Is it a wide stretch to believe that people with gum disease may be of lower socioeconomic status, and also probably not take care of their diets and health anyways? Can someone link the PDF? My login for the acp journal article the parent linked to describing the health correlations won't work for some reason. I want to see how they controlled for some of these factors. Edit: I found the full PDF through my institution. They note that their study may contain confounds, and note possible ones that they tried to control for. The paper isn't perfect, but I applaud the authors for being as thorough as they could. That being said, this sort of article is bound to go viral as soon as possible, with many misleading headlines.
- tertius 9y ago
  > epidemiological studies Tough field... How do you survive with the overwhelming stigma of doubt connected to it?
  - epmaybe 9y ago
    I'm not involved in any epidemiological studies, simply a medical student :). But that being said, I think a lot of the stigma comes from researchers using biased data, or acting unethically with their data. Good epidemiological studies exist that have improved the health of the general public. As a guide for new readers of public health research, here are my suggestions: 1. Read the abstract. Think about the claims that are being made. For example, if someone tells me that drinking 2 cups of coffee a day leads to a lower risk of cancer, my first thought is: what kinds of things do people drinking 2 cups of coffee a day do that would lower their risk of cancer? Could they have a different diet? Do they make more money, leading to them being able to afford preventative health? Do coffee drinkers just have more awareness about their health? Do their regular schedules and sleep cycles give them a health advantage? Keep these questions in mind when reading the rest of the paper. See if the researchers tried to control for these easy alternative explanations. 2. Get a fundamental understanding of some of the statistical terms and calculations used, and make sure you are not misinterpreting them. Thinks like odds ratio, attributable risk, relative risk, etc. Each of them means something specific, and should not be used for extrapolation. Google is your friend! 3. Read the paper as completely as you can! 4. Especially look at the discussion section for maybe reasons that their data isn't great, or also discussing alternative explanations. Also look at their appendix for their exclusion criteria. This can be a clue towards whether or not a researcher is being genuine about their dataset or data collection. In the ACP article about coffee drinkers, there were quite a few exclusions made, and the population they pulled data from seemed biased (women who went to mammogram screenings might care a bit more about their health than the average person) That's all I have for now. I'd welcome any actual researcher to provide feedback and suggestions!

by cvsh 9y ago

MichaelBurge 9y ago

If that were true, decaf wouldn't show the same effect because it wouldn't give you increased energy. (I don't drink decaf)

devotedtoneu 9y ago
Well decaf does contain caffeine, just less of it.
- viridian 9y ago
  Typically less than a twentieth of it with most modern caffeine methods.

epmaybe 9y ago

In epidemiological studies we refer to these variables as confoundimg variables. There's another term as well that I can't seem to remember. When you try and correlate one specific variable, like coffee, to trends like mortality, you must control for every other variable. Socioeconomic factors like income generally cause confounding of your results. For example, there was a recent paper that showed that women with gum disease were at higher risk for certain cancers. Is it a wide stretch to believe that people with gum disease may be of lower socioeconomic status, and also probably not take care of their diets and health anyways? Can someone link the PDF? My login for the acp journal article the parent linked to describing the health correlations won't work for some reason. I want to see how they controlled for some of these factors. Edit: I found the full PDF through my institution. They note that their study may contain confounds, and note possible ones that they tried to control for. The paper isn't perfect, but I applaud the authors for being as thorough as they could. That being said, this sort of article is bound to go viral as soon as possible, with many misleading headlines.

tertius 9y ago
> epidemiological studies Tough field... How do you survive with the overwhelming stigma of doubt connected to it?
- epmaybe 9y ago
  I'm not involved in any epidemiological studies, simply a medical student :). But that being said, I think a lot of the stigma comes from researchers using biased data, or acting unethically with their data. Good epidemiological studies exist that have improved the health of the general public. As a guide for new readers of public health research, here are my suggestions: 1. Read the abstract. Think about the claims that are being made. For example, if someone tells me that drinking 2 cups of coffee a day leads to a lower risk of cancer, my first thought is: what kinds of things do people drinking 2 cups of coffee a day do that would lower their risk of cancer? Could they have a different diet? Do they make more money, leading to them being able to afford preventative health? Do coffee drinkers just have more awareness about their health? Do their regular schedules and sleep cycles give them a health advantage? Keep these questions in mind when reading the rest of the paper. See if the researchers tried to control for these easy alternative explanations. 2. Get a fundamental understanding of some of the statistical terms and calculations used, and make sure you are not misinterpreting them. Thinks like odds ratio, attributable risk, relative risk, etc. Each of them means something specific, and should not be used for extrapolation. Google is your friend! 3. Read the paper as completely as you can! 4. Especially look at the discussion section for maybe reasons that their data isn't great, or also discussing alternative explanations. Also look at their appendix for their exclusion criteria. This can be a clue towards whether or not a researcher is being genuine about their dataset or data collection. In the ACP article about coffee drinkers, there were quite a few exclusions made, and the population they pulled data from seemed biased (women who went to mammogram screenings might care a bit more about their health than the average person) That's all I have for now. I'd welcome any actual researcher to provide feedback and suggestions!