Commentary

Smoking in spite of everything is one of the main public health concerns worldwide regardless of the efforts to encourage smoking cessation due to the well known impact of smoking on both general and oral health. As oral health providers we still have to face issues that patients who still smoke may have due to the negative impact in varying degrees on oral health and on the success of some dental procedures. The aim of the review was to evaluate the marginal bone loss and implant failure rates between smokers and non-smokers, which have been investigated in other previous reviews. 1, 2, 3, 4, 5

The review appropriately followed the methodology suggested and adapted the PRISMA statement to conduct the systematic review and searched several databases (three) to look for articles that met their inclusion criteria. Prospective, retrospective and even randomised clinical trials to assess smoking and non-smoking groups were included in the search strategy. As expected the search strategy produced observational studies that were accepted for the review, 15 in total of which 10 were retrospective cohort studies and five were prospective cohort studies.The quality assessment was carried out using a tool for appraisal of non-randomised clinical trials, the Newcastle–Ottawa scale (NOS) and the authors mentioned that with the maximum score of nine on the scale, only two studies scored less than six stars.

The overall results were presented in the narrative section of the published article and were represented in a forest plot figure (meta-analysis). The results analysing marginal bone loss included seven studies in the meta-analysis. The heterogeneity among the studies was very high (the authors calculated I2: 98%, (usually less than 40% is acceptable: the ideal is be close to zero).

The results arrived at a statistically significant result with standard mean difference of 0.49 (95% CI of 0.07-0.90), with a confidence interval that translates to a modest clinical relevance, for a range of bone loss from 0.07 mm to 0.9 mm between smokers and non-smokers, favouring the non-smoking group. Another result examines marginal bone loss and the difference between the mandible and the maxilla. Four studies were analysed for that and the results had a heterogeneity of 0% and statistically significant results with a standard mean difference of 0.40 (CI 0.24- 0.50) favouring the mandible. The maxilla may be the area for more bone loss, but again with a precise and narrow confidence interval, translating to a bone loss between 0.24 mm and 0.5 mm.

For the outcome of failure, the authors divided the results into less than a year, two years, three years, less than five years and more than five years. The overall result of the meta-analysis with a heterogeneity of 23% presented an OR of 1.96 (CI 1.68-2.30) in favour of the non-smoking group. However, the authors reported on a subgroup analysis for follow-up time which revealed no significant increase in implant failure proportional to the increase in follow-up time.

Previous reviews from 2006 and 2007 1,2 reported similar results with respect to implant failure. The success rate of implants is still considered high and the failure rate is low. The overall implant failure rate reported in studies is more or less 5% 3, 4, 5 and is dependent upon multiple variables: in other words, the failure rate in smokers may be increased to another 5% or less (considering the results of the odds ratio of almost two). A fairly small increase in the failure rate should not preclude clinicians using implants in patients who are smokers, and probably a greater discussion should be considered for heavy smokers. The multiple cofounders are difficult to control in observational studies.

Several reviews1, 2, 3, 4, 5 have examined this same topic with similar or identical outcomes.

We can say that the results show that the implant failure risk is higher in smokers than in non-smokers. However, we should note that while the risk is low for implant failure and marginal bone loss we should interpret the result with caution as the evidence is extracted from observational studies where cofounders are difficult to control and associations can be incorrectly assumed.