When two active treatments improve from baseline, the key question is how their changes compare under the study's planned statistical framework. Primary study
What was reported
The phase 2 trial randomized 232 people with diabetic peripheral neuropathy to acetyl-L-carnitine or methylcobalamin. It used a blinded double-dummy, active-control design intended to assess noninferiority. Both groups improved on reported symptom and disability scores over 24 weeks, without significant between-group differences on those measures. There was no placebo arm. Primary report
Our appraisal: “not significantly different” is not the whole noninferiority argument
Noninferiority requires an acceptable difference to be defined in advance and an interval estimate to be judged against that margin. A conventional superiority test that fails to find a difference does not, on its own, establish noninferiority or equivalence.
Analysis populations also matter. Intention-to-treat and per-protocol analyses can behave differently in an active-control design, especially when adherence or follow-up is incomplete. A careful review should examine whether conclusions remain consistent under the planned analyses.
The lack of a placebo arm sets another boundary. Improvement in both groups cannot directly show how much either improvement exceeds the change that would have occurred without the active intervention. Natural variation, background care, and repeated assessment remain possible contributors.
Double-dummy blinding is useful because it can help conceal differences between treatment preparations. It does not replace a justified margin, appropriate analysis, or a clinically meaningful endpoint.
This trial offers a comparison between two specified interventions in diabetic neuropathy. It should not be summarized as proof that the compounds are identical, that each is superior to placebo, or that every formulation of carnitine and vitamin B12 produces the same result.
Frequently asked questions
What was the comparator?
Methylcobalamin.
Was there a placebo arm?
No.
Was a double-dummy design used?
Yes.
Did both groups improve from baseline?
Reported scores improved in both.
Were the highlighted between-group differences significant?
No.
Does nonsignificance alone establish noninferiority?
No.
Are all vitamin B12 and carnitine formulations interchangeable?
No.