At 3-year follow-up, bone loss was noted in one patient (reclassified as peri-implantitis at the 4-year follow-up) and an important bone loss (due to poor oral hygiene and bruxism; two implants) in a patient with psychosocial issues who could not be treated during the study. Such a patient would not have been included in an RCT.

Consequently, three implants were lost based on the bone loss criterion. Being able to measure bone level changes is also dependent on the availability of evaluable radiographs. In our study, these were taken as per standard clinical protocol using the available equipment, which may differ to that available in a university clinic, a setting commonly found in controlled clinical studies. Thus, some radiographs were not digitized and were difficult to read. Also, if the protocol does not stipulate radiography, then the natural behaviors of patients in private practice are revealed. Some patients refused radiographs, other patients were followed up by referring dentists, and radiographs were not exchanged. Additionally, if radiographs are routinely acquired, the clinician is still reliant on follow-up attendance. Accordingly, the success rates measured in the present study should be assessed collectively. Other studies not assessing bone level changes may report higher success rates than those achieved if bone level changes were evaluated [4, 5].

Other factors need to be considered when reporting success [32]. Papaspyridakos et al. reported a relationship between the number of success criteria and the success rate: the higher the number of success criteria, the lower the reported success rate [32]. Also, the common criterion of bone loss being < 2.0 mm during the first year of function, followed by < 0.2 mm annually thereafter, may no longer be suitable, particularly with new implant systems, such as platform-switching implants, which lead to minimal crestal bone remodeling (Prosper et al. and Trammell et al. cited in [32]).