Clinical trials of new drugs, or drugs on the market, are expensive, and have to have a clear purpose. For a drug that's in development, and hasn't yet been approved for sale, the purpose is usually to generate data on efficacy and safety, and eventually get approval. For a drug that's about to get approved, it may be to build hype for the forthcoming product (on the assumption it will indeed get approved). For drugs that have safety problems, the manufacturer may need to promise to do safety studies after the drug is approved, to gain more information about the safety problem.
And sometimes, drugs have trials done for no apparent purpose except to allow the company to brag about their product. So, if "Drug B" has already been approved, and the company wants to create buzz about how good it is, they might set up a trial comparing Drug B against a competitor, such as "Drug J". The study would be short in duration, include only a few patients, would not include standard endpoints (such as comparison of improvements of A1C levels), and would be announced at a meeting and via press release, rather than be published in peer-reviewed scientific journals.
The Reuters news agency describes just such a study: "Data presented at the annual meeting of the European Association for the Study of Diabetes in Rome, showed that in a four-week study, patients taking Byetta showed more efficient use of their body's insulin than those taking ... Januvia." More efficient use of insulin? Well, the 61-patient, 4 week study compared the effect of Byetta and Januvia on glucose levels two hours after eating. It was reported that in patients treated with Byetta, glucose levels fell an average of 112 mg/dl compared to a decline of 37 mg/dl for those taking Januvia.
But that's not the point of diabetes therapy. The point is to assure practicing physicians and patients that the drug can decrease the likelihood of complications relating to blood glucose level, and the gold standard is whether it lowers A1C. So, if the company was serious, it would have run the trial in a large number of patients, over at least 3 months, would have measured A1C levels, and would have reported whether Drug B or Drug J had better A1C-lowering effects, or whether Drug B or J had a worse adverse events (side effect) profile.
Or, as an analyst is quoted by Forbes as saying, "This is clearly a marketing study to give their reps more stuff to talk about... You wouldn't do this study unless you knew your drug was better." And a representative from the maker of Drug J (oops, Januvia) was blunt: the study failed to compare the side effects of the two drugs, and "The study is really looking at the peak of when Byetta is in the body after it is injected; on the other hand, Januvia works for a 24-hour period and has a steady effect". "It's a mismatch in comparing the two in the way that the study is done."
Sounds to me like it was indeed a clinical trial for marketing purposes.