In the biomedical research field, everybody fabricates results. Or selects them. Or fails to do the research properly and contaminates the experiment. That's why so few experiments are easily reproduced, and a good chunk of published literature eventually gets refuted, or at the very least, refined. The only thing is that scientists don't attempt to reproduce most experiments. So nobody really knows for sure what's real and what's not.
There's pressure to publish, but there's also pressure to selectively publish positive results. No journal will publish experimental dead-ends, so nobody writes papers saying their experiment failed. And so people who spend years of their life working on an experiment will force their paper through by forcing a positive result from the data if it doesn't go the way they intend.
Things are only really called out when there's a lot of money involved in the results of an experiment, and there's pressure from private industry to monetize, stem cells in this case, clones in the other. Otherwise, things languish for 30 or more years before somebody takes a good, long hard look at the data. Sometimes, it's because the original researcher has to die first before the revocation is even allowed to happen.
Look at that fish oil B.S. recently. It was known fact for 30+ years that fish oil, specifically omega 3 fatty acid was beneficial for the cardiovascular system. Turns out somebody fabricated it, and nobody caught on for the next 30 years.
Or Dolly, the cloned sheep who turned out not to be a real clone. Or any early literature on vitamins supplements, whose effects are largely found to have been literally pissed away. Or the ever-changing food pyramid and other nutritional recommendations.
And I don't even have to mention all of the monetary incentive-based skewed research.
Pure research, that which deals in either the microcosm or the macrocosm exclusively, is not subjected to this fault. The former tends to not have such ambiguities, and the former is taken as ambiguous by nature. But beware reseach that tries to tie the two worlds together (for example, asserting that a protein deficiency is correlated to a certain disease, or tying cell phones to medical disorders) and presents some solid, concrete result. Both socially and academically, we're still unable to support that kind of research the way it should be supported. And so it's really a crapshoot which paper ends up withstanding the test of time and which one doesn't.