Commentary

Since the revolutionary discovery of osseointegration, dental implants have been historically placed according to the Brånemark protocol.1 This method involved delicate surgical technique, whereby the implant was allowed to heal in situ without the burden of loading for a period of at least three to four months. This practice was aimed towards reducing the incidence of infection,2 reducing micro-movement and increasing primary stability of the implant;3 factors which would promote osseointegration and reduce the risk of fibrous encapsulation of the implant. As times have moved on, so have the wishes and expectations of our patients. Nowadays, patient-centred outcomes such as reduced treatment times, surgical interventions and costs are factors seen as desirable.4 Indeed, for patients suffering tooth loss and for those who are inexperienced with removable prostheses, it can be hard to tolerate a submerged healing period. As such, the advantages of an immediate loading protocol have fuelled much research in this area in recent years.

The meta-analysis by Chrcanovic et al.5 is well written and satisfies many of the criteria set out in the PRISMA appraisal tool.6 A review protocol is an important step to provide readers with a step-by-step methodology to enable transparency and reproducibility. Such a protocol enables replication of the analysis, which in turn imparts reliability of the analysis findings. However, the authors advise that a review protocol does not exist.

The authors utilise a detailed and repeatable search strategy and made efforts to make contact with authors in order to obtain missing data. Although Chrcanovic et al. did manually search a number of other journals, the authors note that only dental journals were included; and it is unclear why medical journals would be specifically excluded relating to dental implant placement. Ideally a search date should be given for manual journal searches in order to facilitate ease of replication of the analysis findings. The authors included a flow diagram detailing the numbers of studies screened, assessed for eligibility and included in the review, with reasons for exclusions given at each stage, however it may have been useful to have a reference list of all excluded studies. Equally it may have been useful to present confidence intervals for individual studies rather than the standard deviations (SD) that were given.

A quality assessment of the risk of bias across studies was performed using the Cochrane risk of bias tool; a detailed appraisal of bias risk across studies is also given. It would have been useful for added illustration to present this as a combined risk assessment figure, such as would be expected in a Cochrane review.

The systematic review highlights the difference in failure rate between implant placement techniques, with a good description of study limitations. The introduction provides a descriptive rationale of the advantages and disadvantages of single-stage implant surgery versus two-stage implant surgery and an explicit statement of the research question being addressed is given using the PICOS format. The problem with the placement of implants is that there are so many factors that influence success; from the detrimental relationship between smoking and peri-implant disease7 to the relationship between operator experience and implant success.8 Additionally, many studies in this field show considerable variability on what can be defined within the paradigm of ‘immediate loading’. Clinicians can load implants within 48 hours9 or delay the process for 20 days.10

A conventional meta-analysis is performed to combine the results of limited numbers of well-conducted RCTs with similar results, such that a summary effect can be synthesised. This is important because clinicians require concise, high quality guidance on whether or not an intervention can be effective. In order to increase sample sizes for the analysis, this meta-analysis included (non-randomised) controlled clinical trials and retrospective analysis, study designs which may be at high risk of bias. Many studies had small sample sizes and dissimilar results. Critics might argue that combining diverse studies in a meta-analysis may be an exercise in statistical alchemy, and may be akin to comparing apples with oranges.11

The very nature of many studies in dentistry preclude double-blinding in the study design. To include a patient in a study, a clinician may assess primary stability12 before making a decision on which loading technique they may choose. It is this clinical assessment and decision that may result in subsequent selection bias.

Multiple confounding factors were present in the reviewed studies including small cohort sizes, cigarette smoking, splinting of implants, operator's surgical experience, immediate implant placement in extracted sockets versus delayed placement and short term follow-up periods. Overall Chrcanovic et al. give a good description of limitations and confounding factors which may have influenced the long-term outcomes analysed and recognise that these factors limit the potential to draw robust conclusions about the loading of implants.

Practice point

  • It appears that delaying the loading of clinical implants increases their overall success rate.