The article "Clinical Performance of Bonded Amalgam Restorations at 42 Months" by Dr. William D. Browning and colleagues (May JADA) reported equivalence between amalgam restorations placed with copal varnish or with an adhesive liner.
However, the investigators did not use a statistical equivalency test. Instead, their conclusion was based on an inability to show a statistically significant difference between the two groups.
While this was an attractive conclusion, it was inappropriate based on the statistical tests used by the authors. Indeed, similar misuses of statistical tests designed to deter differences have been discussed repeatedly in the biomedical literature by many, including me (
Rethman M, Nunn M. Clinical versus statistical significance; guest editorial. J Periodontol 1999;70[6]:700–2[Medline]
.
Rethman M. Power—the other important probability for statistical comparisons [Letter]. J Clin Periodontol 1991;18[10]:788–9[Medline]
).
Unfortunately, the readers of this manuscript were left with no appreciation of the probability that the authors’ conclusion of equivalency might be untrue—a key qualifier whenever probability tests are utilized to underpin scientific claims.
Admirably, the ADA Seal of Acceptance Program has specific rules for equivalency testing that these authors might have adapted to support their conclusions. However, I suspect that in light of the typical longevity of amalgam restorations, the study’s size and observation periods were inadequate to demonstrate any difference, even if one really exists. But we simply don’t know, do we? And that’s what’s wrong with this study.
