Author Topic: Benefits to supplementing collagen?  (Read 568 times)

0 Members and 1 Guest are viewing this topic.

Online stands2reason

  • Empiricist, Positivist, Militant Agnostic
  • Poster of Extraordinary Magnitude
  • **********
  • Posts: 10464
Re: Benefits to supplementing collagen?
« Reply #15 on: July 02, 2019, 02:29:07 PM »
or this:

https://www.ncbi.nlm.nih.gov/pubmed/18416885

Not that I have a "bone" in this fight either way, but...

Quote
When data from all subjects (n = 97) were evaluated, six parameters showed statistically significant changes with the dietary supplement collagen hydrolysate (CH) compared with placebo: joint pain at rest, assessed by the physician (CH vs. placebo (-1.37 +/- 1.78 vs. -0.90 +/- 1.74 (p = 0.025)) and five parameters assessed by study participants: joint pain when walking (-1.11 +/- 1.98 vs. -0.46 +/- 1.63, p = 0.007), joint pain when standing (-0.97 +/- 1.92 vs. -0.43 +/- 1.74, p = 0.011), joint pain at rest (-0.81 +/- 1.77 vs. -0.39 +/- 1.56, p = 0.039), joint pain when carrying objects (-1.45 +/- 2.11 vs. -0.83 +/- 1.71, p = 0.014) and joint pain when lifting (-1.79 +/- 2.11 vs. -1.26 +/- 2.09, p = 0.018). When a subgroup analysis of subjects with knee arthralgia (n = 63) was performed, the difference between the effect of collagen hydrolysate vs. placebo was more pronounced. The parameter joint pain at rest, assessed by the physician, had a statistical significance level of p = 0.001 (-1.67 +/- 1.89 vs. -0.86 +/- 1.77), while the other five parameters based on the participants' assessments were also statistically significant: joint pain when walking (p = 0.003 (-1.38 +/- 2.12 vs. -0.54 +/- 1.65)), joint pain when standing (p = 0.015 (-1.17 +/- 2.06 vs. -0.50 +/- 1.68)), joint pain at rest with (p = 0.021 (-1.01 +/-1.92 vs. -0.47 +/- 1.63)), joint pain when running a straight line (p = 0.027 (-1.50 +/- 1.97 vs. -0.80 +/- 1.66)) and joint pain when changing direction (p = 0.026 (-1.87 +/- 2.18 vs. -1.20 +/- 2.10)).

...if I understand this, they didn't state their hypothesis and sub-groups in advance. It looks like they did subgroup analysis; that's basically the same problem as p-hacking. What's that called, p-value cherry-picking, p-picking?

Online CarbShark

  • Poster of Extraordinary Magnitude
  • **********
  • Posts: 11337
Re: Benefits to supplementing collagen?
« Reply #16 on: July 02, 2019, 02:53:26 PM »
or this:

https://www.ncbi.nlm.nih.gov/pubmed/18416885

Not that I have a "bone" in this fight either way, but...
Ha ha! I see what you did there, nice...
Quote

Quote
When data from all subjects (n = 97) were evaluated, six parameters showed statistically significant changes with the dietary supplement collagen hydrolysate (CH) compared with placebo: joint pain at rest, assessed by the physician (CH vs. placebo (-1.37 +/- 1.78 vs. -0.90 +/- 1.74 (p = 0.025)) and five parameters assessed by study participants: joint pain when walking (-1.11 +/- 1.98 vs. -0.46 +/- 1.63, p = 0.007), joint pain when standing (-0.97 +/- 1.92 vs. -0.43 +/- 1.74, p = 0.011), joint pain at rest (-0.81 +/- 1.77 vs. -0.39 +/- 1.56, p = 0.039), joint pain when carrying objects (-1.45 +/- 2.11 vs. -0.83 +/- 1.71, p = 0.014) and joint pain when lifting (-1.79 +/- 2.11 vs. -1.26 +/- 2.09, p = 0.018). When a subgroup analysis of subjects with knee arthralgia (n = 63) was performed, the difference between the effect of collagen hydrolysate vs. placebo was more pronounced. The parameter joint pain at rest, assessed by the physician, had a statistical significance level of p = 0.001 (-1.67 +/- 1.89 vs. -0.86 +/- 1.77), while the other five parameters based on the participants' assessments were also statistically significant: joint pain when walking (p = 0.003 (-1.38 +/- 2.12 vs. -0.54 +/- 1.65)), joint pain when standing (p = 0.015 (-1.17 +/- 2.06 vs. -0.50 +/- 1.68)), joint pain at rest with (p = 0.021 (-1.01 +/-1.92 vs. -0.47 +/- 1.63)), joint pain when running a straight line (p = 0.027 (-1.50 +/- 1.97 vs. -0.80 +/- 1.66)) and joint pain when changing direction (p = 0.026 (-1.87 +/- 2.18 vs. -1.20 +/- 2.10)).

...if I understand this, they didn't state their hypothesis and sub-groups in advance. It looks like they did subgroup analysis; that's basically the same problem as p-hacking. What's that called, p-value cherry-picking, p-picking?

Where are you getting that?
and Donald Trump is President of the United States.

I'm not a doctor, I'm just someone who has done a ton of research into diet and nutrition.

Online stands2reason

  • Empiricist, Positivist, Militant Agnostic
  • Poster of Extraordinary Magnitude
  • **********
  • Posts: 10464
Re: Benefits to supplementing collagen?
« Reply #17 on: July 02, 2019, 03:49:31 PM »
Oh, actually, it was just the one sub-group mentioned "subjects with knee arthralgia (n = 63)".

Offline jt512

  • Frequent Poster
  • ******
  • Posts: 2447
    • jt512
Re: Benefits to supplementing collagen?
« Reply #18 on: July 03, 2019, 06:09:22 AM »
or this:

https://www.ncbi.nlm.nih.gov/pubmed/18416885

Not that I have a "bone" in this fight either way, but...
Ha ha! I see what you did there, nice...
Quote

Quote
When data from all subjects (n = 97) were evaluated, six parameters showed statistically significant changes with the dietary supplement collagen hydrolysate (CH) compared with placebo: joint pain at rest, assessed by the physician (CH vs. placebo (-1.37 +/- 1.78 vs. -0.90 +/- 1.74 (p = 0.025)) and five parameters assessed by study participants: joint pain when walking (-1.11 +/- 1.98 vs. -0.46 +/- 1.63, p = 0.007), joint pain when standing (-0.97 +/- 1.92 vs. -0.43 +/- 1.74, p = 0.011), joint pain at rest (-0.81 +/- 1.77 vs. -0.39 +/- 1.56, p = 0.039), joint pain when carrying objects (-1.45 +/- 2.11 vs. -0.83 +/- 1.71, p = 0.014) and joint pain when lifting (-1.79 +/- 2.11 vs. -1.26 +/- 2.09, p = 0.018). When a subgroup analysis of subjects with knee arthralgia (n = 63) was performed, the difference between the effect of collagen hydrolysate vs. placebo was more pronounced. The parameter joint pain at rest, assessed by the physician, had a statistical significance level of p = 0.001 (-1.67 +/- 1.89 vs. -0.86 +/- 1.77), while the other five parameters based on the participants' assessments were also statistically significant: joint pain when walking (p = 0.003 (-1.38 +/- 2.12 vs. -0.54 +/- 1.65)), joint pain when standing (p = 0.015 (-1.17 +/- 2.06 vs. -0.50 +/- 1.68)), joint pain at rest with (p = 0.021 (-1.01 +/-1.92 vs. -0.47 +/- 1.63)), joint pain when running a straight line (p = 0.027 (-1.50 +/- 1.97 vs. -0.80 +/- 1.66)) and joint pain when changing direction (p = 0.026 (-1.87 +/- 2.18 vs. -1.20 +/- 2.10)).

...if I understand this, they didn't state their hypothesis and sub-groups in advance. It looks like they did subgroup analysis; that's basically the same problem as p-hacking. What's that called, p-value cherry-picking, p-picking?

Where are you getting that?

They actually state in the paper that the knee-pain subgroup analysis was not planned in advance.  Therefore, those results should be considered exploratory. 

But I don't have much confidence in the main analysis.  The authors write, "The primary end points of the study were defined as the comparison of the numerical differences of the scores of the VAS between both groups (collagen hydrolysate vs. placebo) after 24 weeks of treatment. These scores were calculated by subtracting the score of a particular parameter of visit 1 from the score recorded during visit 5."  But then they report that they dropped 11 subjects because they "presented fewer than three times for the scheduled visits, thus rendering a meaningful evaluation of their data impossible."  Huh?  Since the primary endpoints were calculated using only the visit 1 and visit 5 scores, then why does a meaningful evaluation of a subect's data require three visits?  Nowhere in the paper is data from intermediate visits used or even mentioned.  Dropping these 11 subjects is unjustified and suspicious. 

After dropping 11 subjects for not having data that wouldn't have been analyzed anyway, the authors find that out of 15 primary endpoints 6 are statisitcally significant.  Of those, 5 have p-values between .01 and .05, a range which Bayes analyses have shown to provide little to no evidence, and which we typically see when the results have been p-hacked.

In summary, I see little to no evidence of effectiveness in this paper.
« Last Edit: July 03, 2019, 06:14:50 AM by jt512 »
Ich verstehe nur Bahnhof.

Offline jt512

  • Frequent Poster
  • ******
  • Posts: 2447
    • jt512
Re: Benefits to supplementing collagen?
« Reply #19 on: July 03, 2019, 12:26:19 PM »
or this:

https://www.ncbi.nlm.nih.gov/pubmed/18416885

Not that I have a "bone" in this fight either way, but...
Ha ha! I see what you did there, nice...
Quote

Quote
When data from all subjects (n = 97) were evaluated, six parameters showed statistically significant changes with the dietary supplement collagen hydrolysate (CH) compared with placebo: joint pain at rest, assessed by the physician (CH vs. placebo (-1.37 +/- 1.78 vs. -0.90 +/- 1.74 (p = 0.025)) and five parameters assessed by study participants: joint pain when walking (-1.11 +/- 1.98 vs. -0.46 +/- 1.63, p = 0.007), joint pain when standing (-0.97 +/- 1.92 vs. -0.43 +/- 1.74, p = 0.011), joint pain at rest (-0.81 +/- 1.77 vs. -0.39 +/- 1.56, p = 0.039), joint pain when carrying objects (-1.45 +/- 2.11 vs. -0.83 +/- 1.71, p = 0.014) and joint pain when lifting (-1.79 +/- 2.11 vs. -1.26 +/- 2.09, p = 0.018). When a subgroup analysis of subjects with knee arthralgia (n = 63) was performed, the difference between the effect of collagen hydrolysate vs. placebo was more pronounced. The parameter joint pain at rest, assessed by the physician, had a statistical significance level of p = 0.001 (-1.67 +/- 1.89 vs. -0.86 +/- 1.77), while the other five parameters based on the participants' assessments were also statistically significant: joint pain when walking (p = 0.003 (-1.38 +/- 2.12 vs. -0.54 +/- 1.65)), joint pain when standing (p = 0.015 (-1.17 +/- 2.06 vs. -0.50 +/- 1.68)), joint pain at rest with (p = 0.021 (-1.01 +/-1.92 vs. -0.47 +/- 1.63)), joint pain when running a straight line (p = 0.027 (-1.50 +/- 1.97 vs. -0.80 +/- 1.66)) and joint pain when changing direction (p = 0.026 (-1.87 +/- 2.18 vs. -1.20 +/- 2.10)).

...if I understand this, they didn't state their hypothesis and sub-groups in advance. It looks like they did subgroup analysis; that's basically the same problem as p-hacking. What's that called, p-value cherry-picking, p-picking?

Where are you getting that?

They actually state in the paper that the knee-pain subgroup analysis was not planned in advance.  Therefore, those results should be considered exploratory. 

But I don't have much confidence in the main analysis.  The authors write, "The primary end points of the study were defined as the comparison of the numerical differences of the scores of the VAS between both groups (collagen hydrolysate vs. placebo) after 24 weeks of treatment. These scores were calculated by subtracting the score of a particular parameter of visit 1 from the score recorded during visit 5."  But then they report that they dropped 11 subjects because they "presented fewer than three times for the scheduled visits, thus rendering a meaningful evaluation of their data impossible."  Huh?  Since the primary endpoints were calculated using only the visit 1 and visit 5 scores, then why does a meaningful evaluation of a subect's data require three visits?  Nowhere in the paper is data from intermediate visits used or even mentioned.  Dropping these 11 subjects is unjustified and suspicious. 

After dropping 11 subjects for not having data that wouldn't have been analyzed anyway, the authors find that out of 15 primary endpoints 6 are statisitcally significant.  Of those, 5 have p-values between .01 and .05, a range which Bayes analyses have shown to provide little to no evidence, and which we typically see when the results have been p-hacked.

In summary, I see little to no evidence of effectiveness in this paper.


In addition, although the paper states that the main analysis was adjusted for multiple comparisons, it looks to me like this was never done.  When I apply the multiple comparisons procedure the authors say they used to the reported p-values, I get that none of them are statistically significant.
Ich verstehe nur Bahnhof.

Offline daniel1948

  • Isn’t a
  • Reef Tank Owner
  • *********
  • Posts: 8414
  • I'd rather be paddling
Re: Benefits to supplementing collagen?
« Reply #20 on: July 03, 2019, 04:26:53 PM »
It's a real red flag for me when they claim significance based on the results of one sub-group that was not specified before the study, and other similar-sounding sub-groups didn't show improvement.
Daniel
----------------
"Anyone who has ever looked into the glazed eyes of a soldier dying on the battlefield will think long and hard before starting a war."
-- Otto von Bismarck

Offline God Bomb

  • Seasoned Contributor
  • ****
  • Posts: 504
Re: Benefits to supplementing collagen?
« Reply #21 on: July 21, 2019, 06:17:33 PM »
OK i've seen references to new data now that notes that hydrolyzed collagen contains certain peptides which can survive digestion.  And these have shown to improve elasticity of skin in older test subjects.  i think they were around 50ish.  Could be promising.
Fell deeds awake. Now for wrath, now for ruin, and the red dawn.

Offline daniel1948

  • Isn’t a
  • Reef Tank Owner
  • *********
  • Posts: 8414
  • I'd rather be paddling
Re: Benefits to supplementing collagen?
« Reply #22 on: July 21, 2019, 08:35:46 PM »
OK i've seen references to new data now that notes that hydrolyzed collagen contains certain peptides which can survive digestion.  And these have shown to improve elasticity of skin in older test subjects.  i think they were around 50ish.  Could be promising.

In the study of the effect of those peptides on skin, were they applied to the skin directly, or were they taken internally? If the former, it says nothing about how collagen taken internally would affect the skin. And no effect on the skin would tell us anything about any effect on the joints.

Skin cream makes your skin softer. That doesn't mean it has any effect on your joints.

Is your point that collagen might be good for your joints, or that, applied externally, it might be helpful for your skin?
Daniel
----------------
"Anyone who has ever looked into the glazed eyes of a soldier dying on the battlefield will think long and hard before starting a war."
-- Otto von Bismarck

 

personate-rain