Background

Periodontitis is a widespread inflammatory disease of the tooth-supporting soft and hard tissues [1–3] with an intermittent destruction process. It progresses either chronically or aggressively [4], but in either case, bacterial involvement in biofilms is regarded as the primary etiologic factor for both disease initiation and progression [5, 6]. Accordingly, the pivotal aim of cause-related periodontal therapy is based on the removal of the pathogenic microbial challenge and the successful prevention of its re-establishment in the ecological niches [7]. Clinically, this is achieved by mechanical debridement using scalers, curettes and/or ultrasonic instruments along with proper oral hygiene instruction [8, 9]. In this context, however, a complete root surface cleaning has been shown to be an unrealistic aim: Especially in pockets exceeding a depth of 6 mm, a perfect debridement is impossible - even when performed by experienced operators [10, 11]. Despite these technical limitations, relevant outcome parameters like depth and number of pockets can be significantly reduced and maintained irrespective of the initial probing depth [12, 13]. However, in many situations periodontitis is not completely resolved by non-surgical mechanical means alone [14], especially in difficult to clean areas such as multi-rooted teeth and complex bone defect configurations [15].

Thus, the use of antimicrobials is a viable approach to improve the clinical outcomes. The adjunctive administration of systemic antibiotics for instance has been shown to offer special healing benefits to improve the mechanical debridement in critical sites [16]. In addition, periopathogenic bacteria are known to colonize not only subgingival tooth surfaces but also hide in oral niches like deep plications of the tongue, crypts of the palatopharyngeal tonsils or the inner buccal mucosa and its recesses, where they are mostly out of the reach of mechanical treatment [17, 18]. Noteworthy, some bacteria were even shown to invade periodontal soft tissue cells [19–21], where they remain inaccessible for conventional mechanical debridement as well. Therefore, antibiotic therapy has gained a long tradition in periodontitis therapy [22]. However, well-controlled studies are limited to specific agents [23], among which, amoxicillin, metronidazole and their combination being the most frequently investigated antibiotics [24]. To date, a considerable number of studies have consistently shown a superiority of the systemic administration of these agents together with scaling and root planning (SRP), mainly in terms of probing pocket depths (PPD), clinical attachment levels (CAL) and changes as compared to SRP alone [24]. However, the problem of adverse side-effects and especially a seemingly ever increasing risk of bacterial resistance [25] urge clinicians to balance risks and benefits well with each individual patient.

Among parameters for oral hygiene, marginal inflammation and gingival recession, periodontal pocket depth (PPD) and clinical attachment loss (CAL) are still the most important surrogate parameters for clinical changes. Whereas CAL indicates the amount of periodontal destruction that will not necessarily be recovered in most cases with successful periodontal treatments, PPD is the parameter that should improve significantly during therapy. As PPD values up to 3 mm are regarded as being compatible with periodontal health, pockets exceeding 5 or 6 mm might not align with immediate treatment success or long-term stability. As these pockets show a significantly enhanced risk for further bacterial regrowth and attachment loss [26], they constitute an indication for additional – in most cases surgical - treatments. This fact is well reflected in the cut-off values for pocket depths of the Community Periodontal Treatment Index of Treatment Needs (CPITN) and the Periodontal Screening Record (PTR) [27, 28]. hoo.de > ly dation section tge language t in its te added. the phrase into. “ text as follows: ults in enhanced heterogeneit.

Following the guidelines for the conduction of the respective studies, systematic reviews with meta-analyses present differences of various treatment modalities expressed as means and standard deviations of the above-mentioned outcome parameters (e.g. PPD and/or CAL) in millimeters [29–31]. Despite being statistically flawless, this mode of data presentation renders it difficult for clinicians and patients to estimate the clinical benefit in terms of an adjunctive treatment [32], as direct information on the degree of clinical success rate is not provided. Regarding a clinically applicable success estimation after periodontal treatment, the reduction of the periodontal pocket depth on a physiologic level of up to 3 mm, i.e. the clinical pocket closure, remains the most important end parameter. Beyond that, a further distinction between sites with moderately enhanced pocket depths that might remain stable over long time periods and those, which most probably need further invasive therapy, seems reasonable [26]. Thus, distinct cut-off values of ≤ 3 mm and ≤ 6 mm PPD might constitute important landmarks to clinicians and patients for every day decision-making.

Therefore, it was the aim of the present study to conduct a meta-analysis based on data of the existing literature on combined administration of amoxicillin and metronidazole as an adjunct to SRP, calculating the probability of clinical success by using these relevant cut-off values of 3 an 5 mm PPD to provide estimated for pocket closure and avoidance of surgery after scaling and root planing with systemic antibiotics.

Methods

This study was planned and conducted in accordance to the PRISMA guidelines for systematic reviews [29]. Modifications were made with regard to the study specific presentation of the outcomes expressed as means and standard deviations instead of estimated probability for the cut-off values.

The focused question according to the PICO criteria was:

“What is the outcome after non-surgical subgingival debridement with or without systemic administration using a combination of amoxicillin and metronidazole in healthy humans with chronic or aggressive periodontitis in terms of the estimated odds ratio for pocket closure (i.e. PPD ≤ 3 mm) or avoidance of surgery (i.e. PPD ≤ 5 mm)?”

A meta-analysis was conducted for data at 3 and 6 month after intervention.

Search strategy

A literature search up to June 2013 was conducted in the US National Library of Medicine (PubMed), the Exerpta Medical Database (Embase) and the Cochrane Central Register of Controlled trials (CENTRAL) using the search terms and combinations presented in Figure 1. After title and abstract screening, an additional hand search was performed in the reference lists of all full texts of interest and the index of contents of Journal of Clinical Periodontology, Journal of Periodontal Research and Journal of Periodontology. The search was conducted without language restriction. The literature search was performed by two independent reviewers (Kolakovic and Sahrmann). In case of discrepancies, study exclusion was determined after discussion. The search strategy is depicted in Figure 2.

Figure 1
figure 1

Search items for the electronic literature search. MeSH – Mdical Subject Headings, TIAB – Title and Abstract.

Full size image
Figure 2
figure 2

Screening strategy performed by two independent reviewers. vWC – vanWinkelhoff Cocktail.

Full size image

Eligibility criteria

In order to include data from studies of highest quality, only randomized controlled clinical trials were considered. Studies comparing the clinical outcomes of non-surgical periodontal treatment with and without adjunctive systemic antibiotic therapy focussing on the combination of amoxicillin and metronidazole, in otherwise healthy patients were included. Studies had to report data for periodontal probing depths after a time interval of at least 3 months after treatment, presented as means and standard deviations, which displayed normal data distribution. Studies on patients with known diseases or drug intake that potentially affects progression and therapy of periodontitis (diabetes, immunosuppressive medication etc.) were excluded. In order not to exclude an entity that is often specifically treated with a concomitant antibiotic medication smokers were not excluded.

Assessment of heterogeneity

To assess the comparability of the selected studies, data on diagnosis, patient populations, exclusion criteria, treatment protocols including pre-treatment, interventions and maintenance protocols of each study were extracted.

Quality assessment

To estimate the potential bias of the different studies included, the described method of randomization, the concealment strategy of the allocation and the blinding of the operator performing the clinical examination were assessed.

Statistical analysis

From each study, we extracted the number of participants and the mean pocket depth and standard deviation at the follow-up examination(s). If not exactly described in the respective statistical methodology section in the original paper, we assumed that the pocket depths were normally distributed if they were presented as mean and standard deviation. Based on available data sets, the probability of clinical success, expressed as the proportion of pockets < 3 mm, and the proportion of persisting pockets > 3 mm and > 5 mm, respectively, using the method proposed by Hauri and co-workers [32] was determined. For this purpose, the odds ratios (OR) and their 95% confidence intervals from the derived event rates in experimental and control group for each of the studies were calculated. For pooling of these ORs a fixed effects meta-analysis model was used. All analyses were performed with R, a free software environment for statistical computing and graphics [33].

Results

Study selection

The electronic literature search provided 1603 potentially includable studies. Based on titles and abstracts, 1537 of these were excluded (93% agreement between reviewers prior to discussion). Based on the full text assessment further 40 studies were excluded due to administration of antibiotics other than the combination of interest, data presentation without means and standard deviations or inadequate intervention in either test or control group (see Figure 2 and Table 1[34–58]). In case of missing clinical data or unsuitable data presentation the corresponding authors were contacted via electronic mail requesting further information, [59, 60]. If no reply was received within 12 weeks, the respective study had to be excluded. The remaining studies could not be included into the meta-analysis due to their individual time points of data evaluation.

Table 1 Excluded studies
Full size table

Study heterogeneity and study characteristics

In some studies smokers were excluded [61–64], one study included only smokers [65] while others [66–75] included both or even did not report on the smoking status of their study population.

Generally, periodontitis cases were classified as generalized chronic or aggressive periodontitis or were not further classified. Not every study reported explicitly to what extent oral hygiene instructions were given before treatment. Frequency and method of supragingival cleaning remains unclear in some studies [67, 69, 70, 73–75]. Test- and control interventions were performed either as full mouth or quadrant-wise treatments with either hand instruments, ultrasonic devices or both. All studies used local anesthesia during subgingival cleaning. Prescribed antibiotics varied in concentration (375–500 mg for amoxicillin and 250/400/500 mg for metronidazole, three times a day each) and the period of intake (7, 10 or 14 days). Different modes of controlling the drug adherence were described. The post-interventional care varied in terms of the use of antiseptic solutions like chlorhexidine of different concentrations and pharmacological forms (gel, mouth washes) and concentrations (0.1/0.12/0.2/1.0%). The periods of investigation varied from 3 to 24 months (Table 2).

Table 2 Study description
Full size table

Quality assessment

The quality assessment is presented in Table 3. If the method of randomization, concealment or the blinding of the examiner was clearly described, the quality was rated as “+” if it was claimed that randomization, concealment or the blinding was performed but no information about the mode of performance was provided the rating was “(+)” and if no concealment or blinding was stated, the rating was “–”. Based on this rating, study quality was assessed as moderate to high.

Table 3 Quality assessment
Full size table

Study outcomes

For the re-evaluation time points we performed two meta-analyses, one at 3 months after treatment and one after 6 month, which included 10 and 7 studies with a total of 521 and 448 patients, respectively.The meta-analyses revealed that the use of the combination of amoxicillin and metronidazole together with SRP increased the chance of pocket closure by a factor of 3.55 three month after the therapy (Figure 3) and a pronounced 4.43 fold chance six month after the treatment (Figure 4).

Figure 3
figure 3

Meta-analysis of the chance for pocket closure after 3 months. OR – odds ratio, 95-CI – 95% confidence intervall, w – weight, p – level of significance.

Full size image
Figure 4
figure 4

Meta-analysis of the chance for pocket closure after 6 months. OR – odds ratio, 95-CI – 95% confidence intervall, w – weight, p – level of significance.

Full size image

We found that it was not possible to calculate the risk estimation for residual pockets exceeding 5 mm, as there was an estimated risk for residual pockets over 5 mm of 0 for both treatment types, which rendered the comparative calculation impossible.

The estimated percentage for pertinent pockets exceeding 3 and 5 mm for the additional use of antibiotics and for SRP alone is presented in Table 4a (at time point 3 months) and b (at time point 6 months).

Table 4 Percent of persisting pockets deeper than 3 mm and 5 mm
Full size table

For two additional studies [62, 71] with data given for both the means and standard deviations and the exact proportion of residual pockets, we performed the same estimation like for the included studies in order to re-validate the statistical model (Table 5). The comparison of published and calculated ratios show a qualitative accordance. However, some subgroups showed considerable differences in size.

Table 5 Comparison of published and calculated OR for the use of antibiotics for studies providing both means and standard deviations and percentages of residual pockets
Full size table

Discussion

This study aimed to estimate the chance of pocket closure or avoidance of surgical therapy after non-surgical periodontal treatment comparing the treatment with and without the additional use of the combination of the amoxicillin and metronidazole. Other than in conventional systematic reviews and different to the data presentation recommended by the PRISMA statement [76], this review did not present the differences by means and standard deviations, but estimated the likelihood for the attainment of clinical relevant surrogate parameters. We believe that this kind of data presentation provides easier and clinically more relevant interpretations of the clinical effectiveness, as in periodontal treatment the main target is the reduction of pockets below a cut-off pocket depth of less than 3 mm or not exceeding 5 mm [26]: The first benchmark indicates that the pockets are “closed” with no further treatment needs, whereas the second benchmark indicates the avoidance of specific needs for a surgical intervention, which is classically still indicated if pockets of 6 mm depth and deeper persist after treatment due to their significantly enhanced risk for disease recurrence [26].

This study clearly elucidated an enhanced chance for pocket closure when antibiotics were used in combination with mechanical root surface debridement. On the other hand, the calculations could not be performed for the case of a cut-off value > 5 mm. This implied that the statistical model indicated a 100% elimination of pockets > 5 mm for both the treatment with and without the use of antibiotics. This fact depicts a shortcoming of the performed statistical estimation, as the single studies in fact reported isolated residual pockets.

In all the included studies, we assumed a normal distribution of the data [77]. With relatively small case numbers, this expectation might have distorted the calculated results to some extent. However, the effect of possible statistical misclassification was likely to be similar in test and control group due to the randomization of the treatment allocation of the studies, thus limiting the disturbing bias again.

We tried to verify the adaptability of the statistical model using the calculation in studies that provided both, means and standard deviations and the exact distribution of residual pockets of either > 3 mm or > 5 mm depth. In the data of one study group [62, 78] we found a good correlation of true and calculated results. However, in another the true and estimated values varied to a greater extent [71], despite the fact that important factors such as sample size were comparable.

The calculation model has been previously published and more studies using this analysis methodology have been demanded [32]. With the present data, its applicability can be better understood and its limitation to studies with higher numbers of participants appears recommendable. In conclusion, the data presentation of the exact distribution of the pocket depths over 3 and 5 mm – as already presented in the actual literature – should be provided in future studies as it was done in the classical studies as well. However, such a request needs time to push through and as long as this claim is not generally implemented, the proposed statistical model offers a useful alternative method to combine and compare study results in such a way.

The pronounced effect of the antibiotics during healing after the first three months as compared to healing after SRP alone is well reflected well in our analysis: For pocket closure, there is an enhanced chance after 6 month if antibiotics had been used. This finding is in accordance with the existing literature [30, 79, 80].

A large heterogeneity existed for the included data: Smoking status, diagnosis of aggressive or chronic periodontitis and the detailed treatment scheme used in the studies showed substantial variations. With the cumulative analysis there was a certain risk of comparing apples with oranges. However, this approach offered the possibility to generate a universal conclusion on the antibiotic treatment of periodontitis, regardless of which patients were treated with which protocol. Furthermore, and another limitation of our approach, we could not include important studies assessing the issue of interest because of the way on which their data was presented: Several authors presented their data well and even with the similar aim to refer to distributions of specific benchmark values, but unfortunately other cut-off values than ours were chosen, which rendered a comparison impossible.

With 10 and 7 included studies for the time points 3 and 6 months after treatment, respectively, only a relatively small number of studies dealing with the issue of antibiotics in periodontitis treatment could be included. Zandbergen et al. assessed a body of 24 studies in a classic review [30]. However, aiming to perform a meta-analysis they could only calculate the overall change of PD and CAL for SRP in combination with antibiotics. Neither a direct comparison to the treatment without antibiotics, nor an estimate of the treatment success in terms of pocket closure or avoidance of surgical therapy was possible in their review. These aspects, however, are important to both the practitioner and the patient.

The benefits of antibiotic treatment always have always to be balanced against their possible adverse reactions. For amoxicillin allergic skin reactions, joint swelling and – in few cases – anaphylactic reactions are well documented [81]. Metronidazole has frequently been reported to cause – among other discomforts and indispositions - nausea, diarrhea and headache [82]. The dimension of the potential risk of causing resistant strains against these antibiotics must also be kept in mind, even if the discussion about this issue is controversial in the current literature [25, 82, 83]. For the clinician a clear prediction of the benefits of adjunctive antibiotic therapy in therms of residual treatment needs after non-surgical treatment might be an easier and better tool for the consideration of a possible antibiotic prescription than rather abstract means and standard deviations as predominantly presented in conventional reviews.

Conclusion

Using a distribution based statistical approach, it was shown that there is a clear benefit in terms of an enhanced chance for pocket closure by co-administration of the combination of amoxicillin and metronidazole as adjunct to non-surgical mechanical periodontal therapy. However, based on the currently available data a potential benefit in terms of the possible avoidance of surgical interventions could not be delineated.

Appendix

The following formula for the calculation of the probabilities P(X”x) for clinical success, ie, for pockets >3 mm was used:

F(X)(x) describes the cumulative distribution function, the mean, the standard deviation and x the cutoff value, eg 3 mm or 5 mm.

For the expression:

One can determine the probabilities by consulting normal distribution tables [58]. stands for the distribution function of the standard normal distribution. The probability values in the Tables published by Stahel are only valid for standard normal distribution. For example, for a group of nine patients, the mean pocket depth at followup is 3.53 mm with a standard deviation of 0.62. Therefore, for:

We get ((3.00 - 3.53)/0.62) = -0.85. The probability for pockets ≤ 3 mm is 0.20, as derived from the Table in Stahel [58]. Hence, the number of patients with pockets ″Accordingly, the number of patients with pockets ≤ 3 mm is 9 × 0.20 and the number of patients with pockets >3 mm is 0.80×9 [32].