It's slow, tedious, and peppered with errors and biases. If done correctly, it rechecks itself and corrects any errors. Sometimes it can take decades to draw any strong conclusions about the effectiveness or efficacy of an intervention (snail out of the well).
1. Is a pilot study
2. Has a small sample population
3. Has a hypothesis with a low plausibility
4. Is older than 7-10 years after it is first published
(There are more factors but I'm not going to list them all.)
We know today that smoking increases the risks of heart and respiratory diseases. However, back then when “I Love Lucy” was a very popular show in the U.S., there is little confirmation or scientific evidence that indicates so. Even ads portray doctors recommending certain brands of cigarettes.
Many studies since the 1950s have published a strong correlation between smoking and longevity and risk of cardiovascular and respiratory problems. Strong evidence includes a British study that observed and followed over 34,400 male physicians who smoked over a 50-year period (1951-2001) and a systematic review of over 8,000 abstracts between 1966 to 2010 that found no significant differences of health risks between female and male smokers.
Combined with public awareness and education about smoking and smoking cessation programs, such evidence have contributed to a decrease of smoking among teens and adults, which have reached an all-time low.
Compared to other healthcare professions, like nursing and cardiology, massage research and applying scientific thinking and reasoning in practice and education are still in their budding state. The aforementioned meta-analysis is one example of how the position of the evidence can change over time, and we remain cautious and should not justify anything with absolute certainty because we could be wrong.
Progress and knowledge advancement may sometimes seem to slip down a few inches when we read about a review based on low-quality studies and methodology, but at least we know — with optimism — that we can correct these errors and produce better evidence that can help us climb out of the well.