I think it is unfortunate that this happens. In fact, research is not about being a competent researcher. It's about knowing how to sell yourself. The professors that are best in selling their research get most funding. Everything is also driven by the papers that are published. Department/faculty with largest amount of publications published in journals with high impact factors get most money from university for funding their research. So professors put their PhD students and Postdoc students under heavy pressure to publish papers.
I doubt it is fraud. Scientists are too well aware that if their discovery is significant their colleagues will try to repeat it and in that way science is self correcting. Animal studies very often lead on to wildly expensive clinical trials, so no group is going to base those on fraudulent reports if it can possibly help it. The professor who did the analysis in this report points out that many studies are not statistically relevant........the scientist has only used 5 mice when he should have used a hundred (why, do you think?), the scientist has only noted the positive effects and not the negative (competition as MisterX says or perhaps a prejudicial desire to find a cure). The author suggests that sharing all positive and negative results with other scientists studying the same intervention would make assessments more reliable and lead to more successful clinical trials.