Hacker News Reader

Comment by cwkoss | Hacker News Reader

The Cochrane review is weird. It's a meta analysis that excluded all studies but one - which did find an effect:

"This trial randomly assigned 146 participants to either an allicin‐containing garlic capsule (dose unspecified) or a placebo (once daily) for 12 weeks. The trial reported 24 occurrences of the common cold in the garlic intervention group compared with 65 in the placebo group (P < 0.001), resulting in fewer days of illness in the garlic group compared with the placebo group (111 versus 366). The number of days to recovery from an occurrence of the common cold was similar in both groups (4.63 versus 5.63)."

Why do you interpret this as "otherwise"?

View on HN

> Authors' conclusions

> There is insufficient clinical trial evidence regarding the effects of garlic in preventing or treating the common cold. A single trial suggested that garlic may prevent occurrences of the common cold but more studies are needed to validate this finding. Claims of effectiveness appear to rely largely on poor‐quality evidence.

If garlic is so obviously effective as the other commenter claims then there should be plenty of supporting evidence, it's not that hard or expensive to test for it, yet, this meta-analysis could only find a single good quality study that showed some effect. This is called chance.

XKCD on significance (p-value): Green jelly beans linked to acne! https://xkcd.com/882/

Absence of evidence is not evidence of absence. But there is only an absence if you choose to throw out the hundreds of "not good enough" studies indicating effects in the same direction.

There is a weird contrarianism a lot of people seem to relish in when it comes to natural medicine having an effect.

The reason pharmaceuticals have studies that prove effects is because there is a giant profit engine powering them and a regulatory system requiring it. It's hard and expensive to do a huge well controlled study of garlic because most modern people eat garlic as a regular part of their diet. It would be hard to find statistically significant effects for prozac if the control group used it as a seasoning.

sorry, do it not pass the test? usually any drug should pass test.

> It's hard and expensive to do a huge well controlled study of garlic because most modern people eat garlic as a regular part of their diet. It would be hard to find statistically significant effects for prozac if the control group used it as a seasoning.

If everybody already takes this "medicine" then what's the problem? We are saved! Or... Wait a minute! How come common cold is still common?!

Also, if this were true it'd be like saying air saves us from suffocating. Well, duh.

Lots of correlations should exist with your chemical, and I don't see them, so if it floats it's a witch.

> If garlic is so obviously effective as the other commenter claims then there should be plenty of supporting evidence, it's not that hard or expensive to test for it, yet, this meta-analysis could only find a single good quality study that showed some effect.

This line of reasoning is fraught with pitfalls. An inferred absence of True Scotsmen to use in an argument by authority doesn't prove a negative. Add that to the fact that there's a selection bias in favor of studies that appear to disprove commonly held beliefs (and generally almost no interest in studies that confirm them) and you really have to use more judgement if you're trying to get at the truth and not just win the argument.

https://www.bmj.com/content/363/bmj.k5094

Reply on news.ycombinator.com