Papers provide evidence for or against various hypotheses, but no one paper is ever a definitive statement on an issue. It is expected that different trials will have different outcomes, both due to differences in the studies themselves (different inclusion/exclusion criteria, different study population, different dose/procedure, etc) and random statistical variation.
It's also very important to recognize that most studies are designed to test whether there is sufficient evidence to reject a null hypothesis of no effect. Failure to reject that null hypothesis does not mean that there is no effect, it means that the results aren't different enough from what you'd expect to see if there was no effect. It may be that there is a smaller effect than the study was powered to detect. Also, consider that a properly powered study generally has a true-positive detection rate of somewhere between 80 and 90%; that means, 10-20% of the time when there's a real effect of the expected size, the study won't be able to conclude there is a significant difference.
So, when there are multiple studies with different results, how do we decide what is "right"? Well, scientific consensus is built over many studies. In clinical research, the best evidence is when you have not only clinical evidence of an effect, but also an understood biological mechanism. At just the clinical level, though, studies can be combined through techniques of meta-analysis. Meta-analysis combines results from multiple studies, weights them by the size/quality of the study, and makes a comprehensive quantitative estimate from the covered literature. If you want to look for scientific consensus on something, you often want to look for recent meta analyses of that topic. In the medical field, there are also often consensus papers written on particular topics. Here's an example:
Priori, S. G., Wilde, A. A., Horie, M., Cho, Y., Behr, E. R., Berul, C., ... & Tracy, C. (2013). HRS/EHRA/APHRS expert consensus statement on the diagnosis and management of patients with inherited primary arrhythmia syndromes: document endorsed by HRS, EHRA, and APHRS in May 2013 and by ACCF, AHA, PACES, and AEPC in June 2013. Heart rhythm, 10(12), 1932-1963.
These statements are written by a bunch of scientists/physicians gathering together and comprehensively reviewing literature on some topic. They typically will weigh not only evidence for effect like in meta analysis, but also consider the implications of treatment options including costs, side effects, and the consequences of getting it wrong (for example, if a treatment following a wrong diagnosis masks some other condition that then goes untreated).