Data sources: PubMed/Medline, Scopus and Cochrane databases supplemented by hand searches in the journals; Clinical Oral Implants Research, Journal of Oral Rehabilitation, The Journal of Prosthetic Dentistry, Journal of Prosthodontics and Journal of Dentistry.

Study Selection: Randomised controlled clinical trials (RCTs), prospective studies with ≥ ten participants and follow-up periods ≥ six months, published in English, and comparing splinted and unsplinted attachment systems within the same study.

Data extraction and synthesis:Two investigators independently performed the electronic search; of which one collected the data while the other investigator verified it. A third investigator was involved in cases of disagreement. The Kappa test was also used to determine the inter-examiner agreement. The risk of bias was analysed using the Cochrane Risk of Bias Tool and the Newcastle-Ottawa scale.

Results: Nine studies were included; six RCTs, two prospective studies and one crossover study, involving the placement of 984 implants in 380 patients and a mean follow-up period of five years. All implants were placed in the mandibular arch. The included studies demonstrated a low or unclear risk of bias. Both splinted and unsplinted attachment systems performed similarly, with no statistically significant differences present in marginal bone loss, complications and implant survival.

Conclusions: The choice of attachment system does not seem to influence marginal bone loss, the incidence of complications or implant survival in mandibular overdentures.