The recently published article by Al-Moraissi et al. has a number of technical errors, which we wish to draw to the attention of readers.
The authors clearly stated the intervention and comparator: piezoelectric device vs. conventional rotary bur for osteotomy performed for the surgical extraction of lower third molars. Yet, among the nine studies included by the authors in their analysis was a study by Chang et al. in which no osteotomy was performed; rather, a Piezotome tip was used for the elevation of roots (after odontotomy was performed in the test and control groups with a high speed hand-piece). (This article by Chang et al. is excluded from the discussion herein.)
The inclusion and exclusion criteria were also stated clearly (albeit very narrow) and none of the remaining eight studies included in the authors’ analysis were eligible for inclusion, for a number of reasons. These include participants not falling within the specified age range, participants not meeting the stated indications for extraction, and participants meeting the exclusion criteria (such as active pericoronitis at the time of operation).
For the outcomes facial swelling, trismus, and pain, the authors included in their meta-analysis for each outcome a range of follow-up days (days 1 to 30 for facial swelling and trismus, and days 1 to 15 for pain). We consider this inappropriate, as the results for each outcome would be expected to vary considerably between the early and late postoperative periods.
The measurements included in the pooled analyses were also often inappropriate. For example, for the outcome trismus (reported using the mean difference), postoperative measurements were pooled with a measurement that was a change from baseline. Notwithstanding the fact that the measurements must be the same when reporting a mean difference, the different measurements give opposite results: a greater postoperative maximum mouth opening is a better outcome, whereas a smaller change from baseline is a better outcome. For the outcome facial swelling, they pooled two-dimensional and three-dimensional measurements with a study (Sivolella et al. ) that reported the number and percentage of patients with ‘persistent oedema’ on certain follow-up days (days 7 and 30). In their analysis, for the mean facial swelling measurement (on both follow-up days), Al-Moraissi et al. incorrectly used the value for the number of patients with persistent oedema as if it was a measure of facial swelling in millimetres, and for the standard deviation they incorrectly used the value for the percentage of patients with ‘persistent oedema’. Further, the authors included a preoperative measurement (by Sortino et al. ) in this analysis, i.e., a measurement recorded before any intervention (either the test or control) had taken place.
There are other critical errors relating to selective analysis (in analyses that pool results for all follow-up days, omission of some of the follow-up days reported by included studies), weighting of studies in the analyses, and units reported (measurements in Figure 2 (facial swelling) were stated as being in millimetres, but some of the values included were measurements in centimetres).