Adjunctive effect of locally delivered antimicrobials in periodontitis therapy: A systematic review and meta-analysis
Abstract
Aim
To answer the following PICOS question: in adult patients with periodontitis, which is the efficacy of adjunctive locally delivered antimicrobials, in comparison with subgingival debridement alone or plus a placebo, in terms of probing pocket depth (PPD) reduction, in randomized clinical trials with at least 6 months of follow-up.
Material and methods
A systematic search was conducted: 59 papers, reporting 50 different studies, were included. Data on clinical outcome variables changes were pooled and analysed using weighted mean differences (WMDs) and 95% confidence intervals (CI), and prediction intervals (PI), in case of significant heterogeneity.
Results
Statistically significant differences were observed, in 6- to 9-month studies, for PPD (WMD = 0.365, 95% CI [0.262; 0.468], PI [−0.29; 1.01]) and clinical attachment level (CAL) (WMD = 0.263, 95% CI [0.123; 0.403], PI [−0.43; 0.96]). For long-term studies, significant differences were observed for PPD (WMD = 0.190, 95% CI [0.059; 0.321]), but not for CAL. For adverse events, no differences were observed. Results were affected by study design (split-mouth versus parallel studies) and assessment (full- or partial-mouth), as well as by the formulation tested.
Conclusions
The use adjunctive locally delivered antimicrobials in periodontitis therapy results in statistically significant benefits in clinical outcomes, without relevant side effects.
Clinical Relevance
Scientific rationale for the study: The efficacy of the adjunctive use of local antimicrobials to scaling and root planing (SRP), on different clinical outcome measures, is insufficiently clear to provide solid recommendations in clinical practice.
Principal findings: Local antimicrobials show significant benefits in probing pocket depth (PPD) and clinical attachment level (CAL) changes, in short-term studies, with no associated adverse effects. In long-term studies, significant benefits were observed for PPD, but not for CAL.
Practical implications: There is consistent evidence showing that the adjunctive use of local antimicrobials improves the outcomes of SRP, with no associated adverse effects.
1 INTRODUCTION
Periodontitis are infectious diseases associated with the dysbiosis of the subgingival microbiota. They are very relevant conditions due to their prevalence (Carasol et al., 2016) and global burden (Kassebaum et al., 2014), their impact on the quality of life (Ferreira, Dias-Pereira, Branco-de-Almeida, Martins, & Paiva, 2017) or on systemic health (Sanz et al., 2018). In the mouth, the main consequence of periodontitis is the destruction of the tooth-supporting tissues, eventually leading to tooth loss.
Basic or initial periodontal therapy, including subgingival debridement or scaling and root planing (SRP), has demonstrated an important clinical impact (Badersten, Nilvéus, & Egelberg, 1981). However, SRP is not free of limitations, and its impact in some patients (e.g. grade C periodontitis; Tonetti, Greenwell, & Kornman, 2018) or in specific sites may be not enough to achieve the desired results. Therefore, other forms of therapies, including different debridement approaches or adjunctive therapies (antimicrobials, probiotics, anti-inflammatory drugs, antioxidant micronutrients), have been proposed and tested (Graziani, Karapetsa, Alonso, & Herrera, 2017).
Antimicrobials can be used locally or systemically in the treatment of periodontitis. The main advantages of the local treatment are fewer side effects and improved compliance, in comparison with drugs used systemically, and reduced chances of developing bacterial tolerance to medications (Rams & Slots, 1996). Several studies and a few systematic reviews have assessed the effects of local antimicrobials delivered in fibres, gels, chips or microspheres, mainly in untreated patients but also in treated sites, with poor response or with recurrent disease (Bonito, Lux, & Lohr, 2005; Hanes & Purvis, 2003; Herrera, Matesanz, Bascones-Martínez, & Sanz, 2012; Matesanz-Pérez et al., 2013; Smiley et al., 2015). Although some of these studies have shown significant additional benefits with the use of certain agents/devices, the clinical value of these effects and the cost–benefit of these treatments have been controversial. Thus, the precise indications of locally delivered antimicrobials have not been clearly established.
Thus, the objective of the present systematic review was to answer the following PICOS question: in adult patients with periodontitis, which is the efficacy of adjunctive locally delivered antimicrobials, in comparison with subgingival debridement alone or plus a placebo, in terms of probing pocket depth (PPD) reduction, in randomized clinical trials (RCTs) with at least 6 months of follow-up?
2 MATERIAL AND METHODS
2.1 Protocol and registration
A protocol was prepared by the authors and presented to the Workshop Committee for the XVI European Workshop. Before starting the study, the protocol was approved and registered in the International Prospective Register of Systematic Reviews PROSPERO (CRD42019142370).
2.2 Eligibility: inclusion and exclusion criteria for studies
2.2.1 Population
Patients with periodontitis, older than 18 years, with any type of untreated periodontitis (aggressive or grade C periodontitis patients were analysed separately), including patients with already treated or “refractory”/recurrent periodontitis, also considered as a subgroup. Studies exclusively on patients with diabetes or smokers were excluded.
2.2.2 Interventions
For the test groups, SRP plus an adjunctive locally delivered antimicrobial, including antibiotics and antiseptics (e.g. chlorhexidine), at concentrations/dosages, recommended by the manufacturer, with sustained delivery (at least 24 hr, according to the product information), adjunctive to subgingival debridement, either full-mouth or localized or in single or repeated sessions.
2.2.3 Comparisons
Control groups received subgingival debridement plus a placebo or a negative control, including subgingival debridement alone.
2.2.4 Outcome
Primary outcome is change in PPD, either full-mouth and at selected treated areas. Secondary clinical outcomes were also registered: changes in clinical attachment level (CAL), “pocket closure” (from PPD ≥ 4 to PPD ≤ 3 mm), frequency distribution of pockets in different categories and bleeding on probing (BOP). Patient-based outcomes were also extracted: patient-reported outcome measures (PROMs), possible adverse effects and oral health-related quality of life (OHRQoL).
2.2.5 Study design and duration
Only RCTs were included, since they provide the best level of evidence in the evaluation of medications, of a minimum duration of 6 months, with parallel or split-mouth designs.
2.3 Information sources and search
The search strategy is presented in Appendix S1, with a combination of MeSH terms and free text words. Due to limitations in timing, no hand-search was done and only publications written in English were included. Three databases were searched: MEDLINE/PubMed, EMBASE and Cochrane.
2.4 Study selection
Study selection was based on a two-step approach: (a) screening of titles and abstracts analysis; and (b) full-text analysis, with reasons for exclusion in this section reported. Both steps were performed in duplicate and results of the agreement are reported.
Two reviewers (PM, DH) selected eligible studies by reviewing the list of titles and abstracts and considering the inclusion and exclusion criteria. The complete articles sourced via eligible titles and abstracts were obtained and examined independently to determine eligibility. Discrepancies between these reviewers pertaining to the selection and inclusion of any specific paper were discussed until either a consensus was reached, or a third reviewer (CM) determined inclusion or exclusion. All reports excluded at this stage were formally recorded, as well as the reason/s for their exclusion. Inter-observer agreement value for the screening of complete articles was assessed via κ score.
2.5 Data collection process and items
Data collection was done in specifically designed Excel sheet and included aspects related to methods and results from the selected studies (Appendix S2). Based on the Cochrane recommendations, a standardized, pre-piloted data extraction form was designed and used. Data were extracted from eligible studies and recorded by an initial reviewer. Second and third examiners cross-checked the accuracy and validity of all the data obtained from the studies.
In case of missing data, an attempt to contact primary authors was done in the systemic antimicrobials section. Studies without enough data for meta-analyses were kept in the systematic review, but excluded from the meta-analyses.
2.6 Risk of bias in individual studies
Quality assessment was conducted by two reviewers, following the recommendations by Cochrane for randomized clinical trials (Higgins et al., 2011), using the original Risk of Bias (RoB) tool. When the papers adequately showed a random sequence allocation and allocation concealment (selection bias), blinding of participants and personnel (performance bias), no bias of outcome assessment (detection bias), complete outcome data (attrition bias), no selective reporting (reporting bias) and no other bias (conflict of interest bias), the studies were defined as low risk of bias. When one of these criteria was not fulfilled, the study was classified as moderate potential risk of bias and, when missing two or more criteria, as a high potential risk of bias (Ten Heggeler, Slot, & Weijden, 2011).
2.7 Summary measures
The main outcome variables have already been described.
2.8 Data analyses and synthesis of the results
Meta-analyses on the mean treatment effects were performed (changes from v1 to vn) with v1 being baseline, and vn being 6-month, 12-month and final visits.
To compare the selected studies, data on the primary and secondary outcomes were pooled and analysed using weighted mean differences (WMDs) and 95% confidence intervals (CI). When the differences between (∆) baseline-end were not reported, they were calculated using baseline and final values. The variance of ∆Var was estimated with the formula: SVar2 = SVar12 + SVar22 − (2*r*SVar1*SVar2), where SVar2 is the variance of the difference, SVar12 is the variance of the mean baseline value, and Svar22 is the variance of the mean end value. A correlation r of .5 was assumed.
The statistical heterogeneity among studies was assessed using the Q test according to Dersimonian and Laird (DerSimonian & Laird, 1986). As a complement to the Q test, the I2 index was calculated in order to know the percentage of variation in the global estimate that was attributable to heterogeneity (I2 = 25%: low; I2 = 50%: moderate; I2 = 75%: high heterogeneity).
The study-specific estimates were pooled using both the fixed effect model (Mantel-Haenszel-Peto test) and the random effect model (Dersimonian-Laird test). If a significant heterogeneity was found, the random effect model results were presented. For subgroup analysis, meta-regression was applied, in case of enough available data.
Forest plots were created to illustrate the effects in the meta-analysis of the global estimation and the different sub-analysis. STATA® 14 (StataCorp LP) intercooled software was used to perform all analyses. Statistical significance was set at p ≤ .05.
In case of heterogeneity in the primary outcome, in addition to the summary estimate (WMD) and CI, prediction intervals (PI) were reported to allow more informative inferences and illustrate which range of true effects can be expected in future settings, presenting the heterogeneity in the same metric as the original effect size measure (IntHout, Ioannidis, Rovers, & Goeman, 2016).
2.9 Risk of bias across studies: publication bias
Egger's test and funnel plots, in case of a sufficient number of included studies (at least 10 studies), were used to assess publication bias. Sensitivity analysis was performed assessing the contribution made to the totality of the evidence by each study after omitting each of them.
3 RESULTS
3.1 Study selection
A total 2,184 references were identified by the electronic search. After the screening of titles and abstracts (κ = 0.658, 95% CI [0.587; 0.728]), 97 papers were selected, plus two additional references, which were associated with one of selected references. After full-paper assessment (κ = 0.873, 95% CI [0.775; 0.972]), 59 articles were included in the review, while the remaining 40 were finally excluded (see Appendix S3).
In total, data from 50 independent investigations were extracted since, in six cases, data came from two papers [(Eickholz et al., 2002; Ratka-Krüger et al., 2005); (Goodson et al., 2012; Socransky et al., 2013); (Wong, Lu, Liu, & Hou, 1999; Wong, Lu, Liu, Hou, & Chang, 1998); (Jeffcoat et al., 1998, 2000); (Palmer, Matthews, & Wilson, 1998, 1999); (Tomasi, Koutouzis, & Wennström, 2008; Tomasi & Wennstrom, 2011)], and for one study, data came from three papers (Colombo et al., 2003; Gonzales et al., 2011; Rodrigues et al., 2004). When reporting these studies, only the primary clinical papers were quoted (Eickholz et al., 2002; Gonçalves, Rodrigues, Feres-Filho, & Colombo, 2004; Goodson et al., 2012; Jeffcoat et al., 1998; Palmer, Matthews, & Wilson, 1998; Tomasi et al., 2008; Wong et al., 1998). In addition, one paper reported a longer-term follow-up (5 years) in a subset of patients (Wilson, McGuire, Greenstein, & Nunn, 1997), from a study sample already described in another manuscript (Newman, Kornman, & Doherty, 1994).
3.2 Study characteristics
3.2.1 Study design
Among the 50 selected studies, there were differences in study design (split-mouth, parallel or both), the number of centres involved, the type of clinical setting (private, university or both) and study duration (Table 1). Although most of the studies were carried out in USA and Europe, selected studies were performed in 16 different countries from four continents.
| Study reference | Blinding | Design | Centres | Setting | Country | Follow-up |
|---|---|---|---|---|---|---|
| Agan et al. (2006) | Single | Split-mouth | Single | University | Turkey | 6 |
| Ahamed et al. (2013) | NR | Parallel | Single | University | India | 6 |
| Aimetti et al. (2004) | Single | Split-mouth | Single | University | Italy | 12 |
| Akncbay, Senel, and Ay (2007) | NR | Parallel | Single | University | Turkey | 6 |
| Azmak, Atilla, Luoto, and Sorsa (2002) | Single | Split-mouth | Single | University | Turkey | 6 |
| Bogren et al. (2008) | Single | Parallel | Multi (3) | Private practice and university | USA and Sweden | 36 |
| Buduneli, Tünger, Evrenosoglu, and Bilgiç (2001) | Single | Split-mouth | NR | NR | Turkey | 12 |
| Carvalho, Novak, and Mota (2007) | Single | Split-mouth | Single | University | USA | 9 |
| Cortelli, Querido, Aquino, Ricardo, and Pallos (2006) | Double | Parallel | Single | University | Brazil | 12 |
| D'Aiuto et al. (2006) | Single | Parallel | Single | University | UK | 6 |
| Dannewitz et al. (2009) | Single | Parallel | Single | University | Germany | 12 |
| Eickholz et al. (2002) and Ratka-Krüger et al. (2005) | Double | Split-mouth | Multi (3) | University | Germany and the Netherlands | 6 |
| Flemmig et al. (1996) | Single | Split-mouth | Single | University | Germany | 6 |
| Friesen, Williams, Krause, and Killoy (2002) | Single | Split-mouth | Single | University | USA | 6 |
| Gonçalves et al. (2004), Colombo et al. (2003) and Rodrigues et al. (2004) | Single | Parallel | Single | University | Brazil | 12 |
| Gonzales et al. (2011) | Double | Parallel | Single | University | Germany | 6 |
| Goodson et al. (2012) and Socransky et al. (2013) | Single | Parallel | Multi (NR) | NR | USA and Sweden | 24 |
| Goodson et al. (1985) | NR | Split-mouth | Single | University | USA | 12 |
| Griffiths et al. (2000) | Single | Split-mouth | Dual | University | UK | 9 |
| Grisi et al. (2002) | Single | Parallel | Single | University | Brazil | 9 |
| Heasman et al. (2001) | Single | Split-mouth | Single | University | UK | 6 |
| Henderson et al. (2002) | Single | Split-mouth | Single | University | New Zealand | 6 |
| Jeffcoat et al. (1998) and Jeffcoat et al. (2000) | Double | Parallel & split-mouth | Multi (10) | University | USA | 9 |
| Jones et al. (1994) | Single | Parallel | Single | University | USA | 6 |
| Kasaj et al. (2007) | Single | Split-mouth | NR | NR | Germany | 6 |
| Killeen et al. (2016) | Single | Parallel | Single | University | USA | 12 |
| Kinane and Radvar (1999) | Single | Parallel | Single | University | UK | 6 |
| Lauenstein, Kaufmann, and Persson (2013) | Single | Parallel | NR | NR | Switzerland | 6 |
| Leiknes et al. (2007) | Single | Split-mouth | Single | University | Norway | 6 |
| Lie et al. (1998) | Single | Split-mouth | Single | University | Norway | 6 |
| Matesanz et al. (2013) | Triple | Parallel | Single | University | Spain | 6 |
| Mizrak et al. (2006) | Single | Parallel | Single | University | Turkey | 6 |
| Newman et al. (1994) and Wilson et al. (1997) | Single | Split-mouth | Multi (7) | Private practice | USA | 6 |
| Palmer et al. (1998) and Palmer, Matthews, and Wilson (1999) | Single | Parallel | Single | University | UK | 6 |
| Paolantonio, Dolci, et al. (2008b) | Single | Split-mouth | NR | University | Italy | 6 |
| Paolantonio, D'Angelo, et al. (2008a) | Single | Split-mouth | Multi (4) | University | Italy | 6 |
| Paolantonio et al. (2009) | Single | Split-mouth | Multi (4) | University | Italy | 6 |
| Romano, Torta, Debernardi, and Aimetti (2005) | NR | Split-mouth | Single | University | Italy | 12 |
| Sakellari et al. (2010) | Single | Parallel | Single | University | Greece | 6 |
| Soeroso et al. (2017) | Single | Parallel | Single | University | Indonesia | 6 |
| Stelzel and Florès-de-Jacoby (2000) | Single | Split-mouth | NR | University | Germany | 9 |
| Tabenski et al. (2017) | Single | Parallel | Single | University | Germany | 12 |
| Timmerman et al. (1996) | Double | Parallel | NR | NR | Netherlands | 18 |
| Tomasi et al. (2008), Tomasi and Wennstrom (2011) | Single | Parallel | Single | University | Sweden | 9 |
| Tonetti et al. (2012) | Single | Parallel | Multi (5) | Private practice | Italy, Germany, Greece, the Netherlands, Switzerland | 12 |
| Tonetti et al. (1998) | Single | Parallel | Multi (6) | Private practice | Italy | 6 |
| Van Dyke, Offenbacher, Braswell, and Lessem (2002) | Double | Parallel | Single | University | USA | 6 |
| Williams et al. (2001) | Single | Parallel | Multi (18) | University | USA | 9 |
| Wong et al. (1998) and Wong et al. (1999) | NR | Split-mouth | Single | University | China | 6 |
| Zingale, Harpenau, Bruce, Chambers, and Lundergan (2012) | Single | Split-mouth | Single | University | USA | 6 |
Note
- Multi (n), multicentre study (number of centres).
- Abbreviation: NR, not reported.
3.2.2 Disease definition
In 22 studies, periodontitis was defined as chronic or adult; in eleven studies, the terminology used was recurrent/“refractory”/relapsing in already treated patients or patients in supportive periodontal therapy (SPT), while in five studies the only definition of disease was “periodontitis” and it was not reported in 9 studies (Appendix S4, Table A4.1). In two studies, two groups were included: both aggressive and chronic periodontitis (Agan, Sönmez, & Serdar, 2006), and untreated and recurrent (Eickholz et al., 2002). In just one study, additional microbiological criteria were used (Jones, Kornman, Newbold, & Manwell, 1994).
With regard to the extension of the disease, in two studies it was considered as localized and in four as generalized. For severity, in 20 studies it was “moderate-severe” or “advanced,” while in two were “severe” or “advanced” and in other two “mild” or “initial to moderate.”
3.2.3 Selected samples
The characteristics of the population samples recruited in the different studies are shown in Table A4.2. Among the selected papers, sample sizes strongly differed, ranging from large samples, with more than 100 patients (Bogren et al., 2008; Eickholz et al., 2002; Goodson et al., 2012; Jeffcoat et al., 1998; Newman et al., 1994; Paolantonio, D'Angelo, et al., 2008a; Tonetti et al., 1998, 2012; Williams et al., 2001) to just 10 patients (Agan et al., 2006; Goodson, Hogan, & Dunham, 1985).
The age of participants ranged from 20 (Kasaj, Chiriachide, & Willershausen, 2007) to 85 (Killeen, Harn, Erickson, Yu, & Reinhardt, 2016) while percentage of females ranged from 31.4% to 77.8%. With regard to smoking profiles, twelve studies included only non-smoker patients, while in studies including smoker patients, the proportion of smokers ranged 4%–47%. Minor differences were detected in the distribution of patients, age, gender and smoking profiles among the study groups (Table A4.3).
3.2.4 Outcome assessment
Seventeen studies used a full-mouth approach to assess clinical outcome variables, either by evaluating all sites, or a group of sites according to a part of the mouth (e.g. a quadrant) or according to a clinical criterion (e.g. PPD > 4 mm); in contrast, 36 studies selected some specific sites/teeth for evaluation, based on clinical, radiological or biomarkers criteria, including furcation lesion sites (Dannewitz, Lippert, Lang, Tonetti, & Eickholz, 2009; Tomasi et al., 2008; Tomasi & Wennstrom, 2011; Tonetti et al., 1998) (Tables A4.1 and A4.4). In three studies, both full-mouth and partial-mouth evaluations were reported (Gonçalves et al., 2004; Timmerman et al., 1996).
The number of sites measured per tooth was 4–6 in 25 studies, and just one in 14 (in two of them, two sites were measured but just the maximum (Leiknes, Leknes, Böe, Skavland, & Lie, 2007), or the mean (Lie, Bruun, & Boe, 1998) were reported). In 21 studies, only one examiner was used, with 13 studies not reporting on this. A higher quality of probing was considered when describing calibration exercises (22 studies), using force-controlled (7 studies) or force-controlled and computer-assisted probes (9 studies), using stents (6 studies), duplicate measurements (5 studies) or both (3 studies).
Most studies used a patient-based approach for data analysis (34). Additional non-clinical outcomes were evaluated in 36 studies, including microbiological evaluation in 23 studies, biomarkers in gingival crevicular fluid (GCF) in 11, and also systemic or radiological outcomes.
In 26 studies, adverse events and/or PROMs were evaluated, most frequently with no clear specification of the findings (“unspecified by authors”). Some studies clearly identified these effects, including gingival/periodontal abscess/infection, gingival tenderness/gingivitis, pain and root/tooth sensitivity.
3.3 Type of interventions
3.3.1 Interventions before the study
For studies dealing with untreated patients, a minimum period with no periodontal therapy was normally suggested as inclusion criterion, ranging from 3 to 12 months; for refractory/recurrent cases, patients were either enrolled in a SPT program or had been periodontally treated within the previous 3–6 months; no criteria were described in five studies (Table A4.5).
In most cases, the studies described periodontal therapies which were rendered before the main intervention, for all study groups, including oral hygiene instructions (OHI) alone (n = 15) or in combination with supragingival professional mechanical plaque removal (PMPR) (n = 12) or with SRP (n = 4); in some studies, the intervention was PMPR alone (n = 3), and in 16 studies, no intervention was mentioned.
3.3.2 Subgingival debridement
The study intervention was local SRP in 19 studies, full-mouth SRP in 22 studies, while supragingival PMPR was the main mechanical therapy in two studies (Gonzales et al., 2011; Heasman, Heasman, Stacey, & McCracken, 2001) (Tables A4.5 and A4.6). Information on the performance of subgingival debridement is described in the tables. It should be highlighted that 48 out of the 50 studies clearly explained that the local antimicrobial was placed/delivered immediately after debridement, with two exceptions: in one study, it was placed before debridement (Tonetti et al., 1998), and SRP was rendered at fibre removal; and in other study, it was placed one week after debridement (Flemmig et al., 1996).
3.3.3 Local antimicrobials and control groups
In 35 studies, there were just two study groups, while 8 had three groups and 6 had four groups (Table 2). One study had eight groups (Goodson et al., 2012). Forty-three studies had SRP alone as main control group, while eight had a vehicle control (placebo), with three of the studies presenting both control groups. Four studies presented an additional untreated control, while one study presented two SRP alone control, one in adjacent sites and another in remote sites (Henderson, Boyens, Holborow, & Pack, 2002).
| Study reference | n groups | Control group | Additional control group | Test group | Additional test group/s |
|---|---|---|---|---|---|
| Agan et al. (2006) | 2 | SRP alone | Atridox | ||
| Ahamed et al. (2013) | 2 | SRP alone | Atridox | ||
| Aimetti et al. (2004) | 2 | SRP alone | Actisite | ||
| Akncbay et al. (2007) | 3 | SRP alone | Chitosan and Metronidazole | Chitosan | |
| Azmak et al. (2002) | 2 | SRP alone | PerioChip | ||
| Bogren et al. (2008) | 2 | SRP alone | Atridox | ||
| Buduneli et al. (2001) | 2 | SRP alone | Elyzol | ||
| Carvalho et al. (2007) | 2 | SRP alone and cyanoacrylate | PerioChip | ||
| Cortelli et al. (2006) | 2 | SRP and vehicle | Arestin | ||
| D'Aiuto et al. (2006) | 2 | SRP alone | Arestin | ||
| Dannewitz et al. (2009) | 2 | SRP alone | Ligosan | ||
| Eickholz et al. (2002) and Ratka-Krüger et al. (2005) | 3 | SRP alone | SRP and vehicle | Ligosan | Vehicle control |
| Flemmig et al. (1996) | 2 | SRP alone | Actisite | ||
| Friesen et al. (2002) | 4 | SRP alone | untreated | tetracycline strips (one) | Tetracycline strips (multiple) |
| Gonçalves et al. (2004), Colombo et al. (2003) and Rodrigues et al. (2004) | 3 | SRP alone | Actisite | Systemic tetracycline | |
| Gonzales et al. (2011) | 2 | SRP and vehicle | PerioChip | ||
| Goodson et al. (2012; Socransky et al. (2013) | 8 | SRP alone | Actisite | Six additional groups | |
| Goodson et al. (1985) | 4 | SRP alone | untreated | Actisite | Actisite alone |
| Griffiths et al. (2000) | 2 | SRP alone | Elyzol | ||
| Grisi et al. (2002) | 2 | SRP alone | PerioChip | ||
| Heasman et al. (2001) | 2 | SRP alone | PerioChip | ||
| Henderson et al. (2002) | 3 | SRP alone (remote) | SRP alone (adjacent) | Arestin | |
| Jeffcoat et al. (1998; Jeffcoat et al. (2000) | 3 | SRP alone | SRP and vehicle | PerioChip | |
| Jones et al. (1994) | 4 | SRP and vehicle | untreated | minocycline powder | Minocycline powder alone |
| Kasaj et al. (2007) | 2 | SRP alone | PerioChip | ||
| Killeen et al. (2016) | 2 | SRP alone | Arestin | ||
| Kinane and Radvar (1999) | 4 | SRP alone | Dentomycina | Actisite, Elyzol | |
| Lauenstein et al. (2013) | 2 | SRP alone | Periofilm | ||
| Leiknes et al. (2007) | 2 | SRP alone | Elyzol | ||
| Lie et al. (1998) | 3 | SRP alone | Elyzol | Aureomycin | |
| Matesanz et al. (2013) | 2 | SRP and vehicle | Chlosite | ||
| Mizrak et al. (2006) | 2 | SRP alone | PerioChip | ||
| Newman et al. (1994) and Wilson et al. (1997) | 2 | SRP alone | Actisite | ||
| Palmer et al. (1998) and Palmer et al. (1999) | 2 | SRP alone | Elyzol | Systemic metronidazole | |
| Paolantonio, Dolci, et al. (2008b) | 2 | SRP alone | PerioChip | ||
| Paolantonio, D'Angelo, et al. (2008a) | 2 | SRP alone | PerioChip | ||
| Paolantonio et al. (2009) | 2 | SRP alone | Chlosite | ||
| Romano et al. (2005) | 2 | SRP alone | Actisite | ||
| Sakellari et al. (2010) | 2 | SRP alone | PerioChip | ||
| Soeroso et al. (2017) | 2 | SRP alone | Perioclinea | ||
| Stelzel and Florès-de-Jacoby (2000) | 2 | SRP alone | Elyzol | ||
| Tabenski et al. (2017) | 3 | SRP alone | Arestin | SRP & PDT | |
| Timmerman et al. (1996) | 2 | SRP and vehicle | Perioclinea | ||
| Tomasi et al. (2008), Tomasi and Wennstrom (2011) | 2 | SRP alone | Atridox | ||
| Tonetti et al. (2012) | 2 | SRP alone | Ligosan | ||
| Tonetti et al. (1998) | 2 | SRP alone | Actisite | ||
| Van Dyke et al. (2002) | 4 | SRP alone | no SRP | Arestin | Arestin alone |
| Williams et al. (2001) | 3 | SRP alone | SRP and vehicle | Arestin | |
| Wong et al. (1998) and Wong et al. (1999) | 2 | SRP alone | Actisite | ||
| Zingale et al. (2012) | 4 | SRP alone | Arestin |
- Abbreviations: PTD, photodynamic therapy; SRP, scaling and root planing.
- a Same formulation with different brand names.
The test groups with commercialized local antimicrobials aimed to assess tetracyclines, including tetracycline [Actisite (10), Aureomycin (1)], minocycline [Arestin (8); Dentomycin (1) and Periocline (2), same formulations with different brand names] and doxycycline [Atridox (4), Ligosan (3)]; chlorhexidine [Chlosite (2), PerioChip (11)]; metronidazole [Elyzol (7)]; and piperacillin and sodium tazobactam [Periofilm (1)]. Among those not commercially available, chitosan (1), chitosan with metronidazole (1), minocycline powder (1) and tetracycline strips [just using one (1), or multiple (1)] were tested. In the text, brand names are used for consistency, but a detailed information on composition is presented in Table A4.7.
Additional test groups, not relevant for this review, included local antimicrobials alone, systemic antimicrobials or photodynamic therapy.
The number of applications (Table A4.8) varied among products and study protocols, being the most frequent just one application, in 34 study groups; two applications were performed in 10 study groups and more than two in five. In six study groups, one initial application was performed, while a second (three studies) or a third one (three studies) was decided based on the dislodging on the first application or on the presence of pockets. When more than one application was scheduled, the protocols were highly heterogeneous. In some cases (16 study groups), a dressing was used after the local antimicrobial application, with cyanoacrylate or a periodontal dressing, that was kept on place for 3–13 days; dislodging of the antimicrobial or dressing was recorded in 12 study groups.
3.4 Risk of bias within studies
In most studies (Table A4.9), quality parameters were considered unclear or not fulfilled, and all the selected studies, except three (Eickholz et al., 2002; Killeen et al., 2016; Tabenski et al., 2017), were qualified with a high risk of bias.
3.5 Results of individual studies
These results are shown in the different forest plots associated with the meta-analyses. It was not possible to report information on pocket closure or on the frequency distribution of PPD categories, since these outcomes were not reported in the individual studies.
3.6 Synthesis of the results
3.6.1 Probing pocket depth, short-term (6–9 months)
Data from 38 comparisons between test and control groups were combined in a meta-analysis, showing a statistically significant benefit (p < .001) for test groups (WMD = 0.364, 95% CI [0.236; 0.491]), with significant heterogeneity (I2 = 96.8%; PI [−0.29; 1.01; Table 3a and Appendix S5, Figure A5.1).
| (a) | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Number of | Weighted mean difference (WMD) | Heterogeneity | |||||||
| Studies | Patients | WMD | 95% CI | p-value | I2 (%) | p-value | |||
| Control | Test | Lower | Upper | ||||||
| All | 38 | 2,137 | 2,111 | 0.364 | 0.236 | 0.491 | <.001 | 96.8 | <.001 |
| Parallel | 21 | 1,259 | 1,233 | 0.231 | 0.136 | 0.326 | <.001 | 91.3 | <.001 |
| Split-mouth | 17 | 878 | 878 | 0.521 | 0.322 | 0.72 | <.001 | 95.1 | <.001 |
| Partial-Mouth | 31 | 1,468 | 1,467 | 0.427 | 0.307 | 0.547 | <.001 | 97.4 | <.001 |
| Full-Mouth | 7 | 669 | 644 | 0.249 | 0.179 | 0.32 | <.001 | 36.7 | .148 |
| Vehicle | 5 | 621 | 595 | 0.308 | 0.29 | 0.327 | <.001 | 0.0 | .858 |
| SRP | 33 | 1,516 | 1,516 | 0.384 | 0.238 | 0.531 | <.001 | 97.2 | <.001 |
| Untreated | 25 | 1,565 | 1,545 | 0.291 | 0.208 | 0.375 | <.001 | 91.0 | <.001 |
| Refractory | 11 | 356 | 350 | 0.443 | 0.25 | 0.636 | <.001 | 62.9 | .003 |
| Both | 2 | 216 | 216 | 0.562 | 0.306 | 0.818 | <.001 | 23.9 | .252 |
| Test product | |||||||||
| Atridox | 2 | 19 | 19 | 0.8 | 0.084 | 1.516 | .028 | 0.0 | .502 |
| Actisite | 7 | 255 | 257 | 0.729 | 0.696 | 0.761 | <.001 | 0.0 | .834 |
| Periochip | 9 | 718 | 719 | 0.23 | 0.12 | 0.341 | <.001 | 96.4 | <.001 |
| Elyzol | 5 | 136 | 135 | 0.14 | −0.041 | 0.322 | .130 | 0.0 | .783 |
| Ligosan | 3 | 236 | 232 | 0.525 | 0.283 | 0.767 | <.001 | 3.9 | .353 |
| Arestin | 6 | 567 | 564 | 0.279 | 0.203 | 0.356 | <.001 | 0.0 | .91 |
| Dentomycin | 2 | 65 | 41 | 0.377 | −0.036 | 0.79 | .073 | 0.0 | .915 |
| Periofilm | 1 | 14 | 18 | −0.1 | −1.053 | 0.853 | .837 | – | – |
| Aureomycin | 1 | 18 | 18 | 0.6 | −0.339 | 1.539 | .219 | – | – |
| Chlosite | 2 | 109 | 108 | 0.486 | −0.238 | 1.211 | .188 | 89.3 | 0.002 |
| Meta-regression | Coefficient | 95% CI | p-value | |
|---|---|---|---|---|
| Lower | Upper | |||
| Parallel/split-mouth | 0.282 | 0.098 | 0.466 | .004 |
| Partial/full-mouth | −0.248 | −0.479 | −0.016 | .037 |
| Vehicle/SRP | 0.085 | −0.194 | 0.363 | .54 |
| Untreated/refractory | 0.138 | −0.091 | 0.367 | .229 |
| (b) | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Number of | Weighted mean difference (WMD) | Heterogeneity | |||||||
| Studies | Patients | WMD | 95% CI | p-value | I2 (%) | p-value | |||
| Control | Test | Lower | Upper | ||||||
| All | 10 | 291 | 230 | 0.19 | 0.059 | 0.321 | .004 | 28.1 | .185 |
| Parallel | 7 | 228 | 167 | 0.079 | −0.098 | 0.257 | .381 | 0.0 | .912 |
| Split-mouth | 3 | 63 | 63 | 0.313 | −0.164 | 0.789 | .198 | 72.1 | .028 |
| Partial-Mouth | 7 | 137 | 130 | 0.243 | 0.07 | 0.415 | .006 | 48.1 | .072 |
| Full-Mouth | 3 | 154 | 100 | 0.118 | −0.082 | 0.319 | .247 | 0.0 | .948 |
| Vehicle | 2 | 90 | 40 | 0.144 | −0.165 | 0.452 | .361 | 0.0 | .803 |
| SRP | 8 | 201 | 190 | 0.2 | 0.056 | 0.344 | .007 | 43.3 | .089 |
| Untreated | 5 | 136 | 86 | 0.061 | −0.193 | 0.314 | .640 | 0.0 | .781 |
| Refractory | 5 | 155 | 144 | 0.236 | 0.084 | 0.389 | .002 | 57.5 | .052 |
| 12 months | 6 | 111 | 104 | 0.082 | −0.191 | 0.354 | .526 | 46.4 | .097 |
| 18 months | 2 | 90 | 40 | 0.144 | −0.165 | 0.452 | .458 | 0.0 | .803 |
| 36 months | 1 | 64 | 60 | 0.1 | −0.164 | 0.364 | .361 | – | – |
| 60 months | 1 | 26 | 26 | 0.35 | 0.128 | 0.572 | .002 | – | – |
| Test product | |||||||||
| Atridox | 2 | 77 | 73 | 0.112 | −0.147 | 0.372 | .396 | 0.0 | .614 |
| Actisite | 2 | 45 | 45 | 0.41 | 0.201 | 0.618 | .028 | 56.9 | .128 |
| Elyzol | 1 | 18 | 18 | −0.21 | −0.72 | 0.3 | .420 | – | – |
| Ligosan | 1 | 19 | 15 | 0.18 | −0.633 | 0.993 | .664 | – | – |
| Arestin | 2 | 42 | 39 | −0.191 | −0.643 | 0.261 | .408 | 0.0 | .747 |
| Dentomycin | 2 | 90 | 40 | 0.144 | −0.165 | 0.452 | .361 | 0.0 | .803 |
| Meta-regression | Coefficient | 95% CI | p-value | |
|---|---|---|---|---|
| Lower | Upper | |||
| Parallel/Split-mouth | 0.240 | −0.125 | 0.604 | .168 |
| Partial/Full-Mouth | −0.077 | −0.567 | 0.413 | .728 |
| Vehicle/SRP | 0.039 | −0.556 | 0.633 | .884 |
| Untreated/Refractory | 0.182 | −0.291 | 0.656 | .4 |
| Follow-up months | 0.005 | −0.006 | 0.015 | .315 |
- Abbreviations: CI, confidence interval: SRP, scaling and root planing.
Significant differences (p < .001), favouring the test groups, were found independently of study design, type of assessment, use or not of placebo or periodontitis type (untreated versus “refractory/recurrent”). Meta-regression showed a significant influence of study design and type of assessment.
Ten different local antimicrobials were included in the analysis, with five of them demonstrating statistically significant benefits for test groups, four of them with no heterogeneity (Atridox: n = 2,WMD = 0.800; Actisite: n = 7, WMD = 0.729; Ligosan: n = 3, WMD = 0.525; Arestin: n = 6, WMD = 0.279) and another with significant heterogeneity (Periochip: n = 9, WMD = 0.23). Among the other products, four demonstrated not significant benefits in test groups (Elyzol: n = 5, WMD = 0.140; Dentomycin: n = 2, WMD = 0.377; Aureomycin: n = 1, WMD = 0.600; Chlosite: n = 2, WMD = 0.486), while Periofilm (n = 1) showed non-significant differences favouring the control group.
3.6.2 Probing pocket depth, long-term (12–60 months)
Data from 10 comparisons between test and control groups were combined in a meta-analysis, showing a statistically significant benefit (p < .05) for test groups (WMD = 0.190, 95% CI [0.059; 0.321]), with no heterogeneity (I2 = 28.1%) (Table 3b and Figure A5.2).
Significant differences (p < .05) were also found in studies with partial-mouth assessment, studies with “refractory/recurrent” patients and those with a 60-month follow-up (p = .002), but not in other groups of studies. Meta-regression did not show a significant impact for any of the factors considered.
Six different local antimicrobials were included in the long-term analysis, with four of them demonstrating benefits for test groups (Atridox: n = 2, WMD = 0.112; Actisite: n = 2, =0.410; Ligosan: n = 1, WMD = 0.180; Dentomycin: n = 2, WMD = 0.144), and the other two, in favour of control groups (Elyzol: n = 1, WMD = −0.210; Arestin: n = 2, WMD = −0.191). However, the only comparison showing statistically significant differences was for Actisite (p = .028).
3.6.3 Clinical attachment level, short-term (6–9 months)
Data from 37 comparisons between test and control groups were combined in a meta-analysis, showing a statistically significant benefit (p < .001) for test groups (WMD = 0.263, 95% CI [0.123; 0.403]), with significant heterogeneity (I2 = 85.0%; PI [−0.43; 0.96]) (Table 4a and Figure A5.3).
| (a) | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Number of | Weighted mean difference (WMD) | Heterogeneity | |||||||
| Studies | Patients | WMD | 95% CI | p-value | I2 (%) | p-value | |||
| control | test | Lower | Upper | ||||||
| All | 37 | 1,717 | 1,667 | 0.263 | 0.123 | 0.403 | <.001 | 85 | <.001 |
| Parallel | 17 | 782 | 732 | 0.05 | −0.134 | 0.234 | .596 | 69.8 | <.001 |
| Split-mouth | 20 | 935 | 935 | 0.44 | 0.404 | 0.475 | <.001 | 82.6 | <.001 |
| Partial-Mouth | 31 | 1,527 | 1,503 | 0.357 | 0.211 | 0.503 | <.001 | 82.5 | <.001 |
| Full-Mouth | 6 | 190 | 164 | −0.155 | −0.494 | 0.184 | .370 | 78.3 | <.001 |
| Vehicle | 4 | 409 | 361 | 0.116 | −0.021 | 0.252 | .666 | 55.8 | .079 |
| SRP | 33 | 1,308 | 1,306 | 0.293 | 0.139 | 0.447 | <.001 | 85.3 | <.001 |
| Untreated | 25 | 1,156 | 1,111 | 0.264 | 0.076 | 0.451 | .006 | 82.3 | <.001 |
| Refractory | 10 | 345 | 340 | 0.266 | 0.056 | 0.476 | .013 | 62.8 | .004 |
| Both | 2 | 216 | 216 | 0.4 | 0.046 | 0.754 | .027 | 0.0 | 1.000 |
| Test product | |||||||||
| Atridox | 1 | 6 | 6 | 0.64 | −0.005 | 1.285 | .052 | - | – |
| Actisite | 8 | 265 | 267 | 0.276 | −0.172 | 0.723 | .227 | 89 | <.001 |
| Periochip | 10 | 745 | 749 | 0.183 | −0.055 | 0.421 | .132 | 86.1 | <.001 |
| Elyzol | 4 | 124 | 122 | 0.035 | −0.175 | 0.245 | .746 | 0.0 | .538 |
| Ligosan | 3 | 236 | 232 | 0.408 | 0.063 | 0.753 | .020 | 0.0 | .982 |
| One_TetraStrip | 1 | 24 | 24 | 0.44 | −0.025 | 0.905 | .064 | – | – |
| Multiple_TetraStrip | 1 | 24 | 24 | 0.48 | 0.087 | 0.873 | .017 | – | – |
| Arestin | 4 | 67 | 66 | 0.517 | 0.151 | 0.883 | .019 | 38.5 | .181 |
| Dentomycin | 3 | 110 | 61 | −0.116 | −0.371 | 0.138 | .426 | 14.6 | .31 |
| Aureomycin | 1 | 18 | 18 | 1,000 | −0.204 | 2.204 | .103 | – | – |
| Chlosite | 1 | 98 | 98 | 0.84 | 0.452 | 1.228 | <.001 | – | – |
| Meta-regression | Coefficient | 95% CI | p-value | |
|---|---|---|---|---|
| Lower | Upper | |||
| Parallel/Split-mouth | 0.374 | 0.040 | 0.708 | .029 |
| Partial/Full-Mouth | −0.535 | −0.946 | −0.124 | .012 |
| Vehicle/SRP | 0.237 | −0.322 | 0.795 | .396 |
| Untreated/Refractory | −0.009 | −0.43 | 0.412 | .965 |
| (b) | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Number of | Weighted mean difference (WMD) | Heterogeneity | |||||||
| Studies | Patients | WMD | 95% CI | p-value | I2 (%) | p-value | |||
| control | test | Lower | Upper | ||||||
| All | 10 | 288 | 224 | 0.09 | −0.253 | 0.433 | .607 | 75.3 | <.001 |
| Parallel | 6 | 215 | 151 | 0.025 | −0.208 | 0.259 | .862 | 48.6 | .083 |
| Split-mouth | 4 | 73 | 73 | 0.305 | −0.339 | 0.95 | .353 | 88.7 | <.001 |
| Partial-Mouth | 7 | 169 | 134 | −0.001 | −0.518 | 0.518 | .999 | 78.4 | <.001 |
| Full-Mouth | 3 | 119 | 90 | 0.27 | 0.036 | 0.504 | .024 | 0.0 | .897 |
| Vehicle | 2 | 90 | 40 | 0.264 | −0.127 | 0.655 | .186 | 0.0 | .759 |
| SRP | 8 | 198 | 184 | 0.062 | −0.351 | 0.475 | .768 | 79.6 | <.001 |
| Untreated | 5 | 133 | 83 | 0.126 | −0.094 | 0.347 | .430 | 29.4 | .225 |
| Refractory | 5 | 155 | 141 | 0.192 | −0.464 | 0.848 | .567 | 86 | <.001 |
| 12 months | 5 | 98 | 88 | 0.119 | −0.682 | 0.919 | .772 | 83.4 | <.001 |
| 18 months | 2 | 90 | 40 | 0.264 | −0.127 | 0.655 | .186 | 0.0 | .759 |
| 24 months | 1 | 10 | 10 | 0.33 | −0.079 | 0.739 | .114 | – | – |
| 36 months | 1 | 64 | 60 | 0.2 | −0.185 | 0.585 | .309 | – | – |
| 60 months | 1 | 26 | 26 | −0.43 | −0.752 | −0.108 | .009 | – | – |
| Test product | |||||||||
| Atridox | 1 | 64 | 60 | 0.2 | −0.185 | 0.585 | .309 | – | – |
| Actisite | 3 | 55 | 55 | 0.493 | −0.5 | 1.487 | .330 | 92.5 | <.001 |
| Elyzol | 1 | 18 | 18 | −0.06 | −0.426 | 0.306 | .748 | – | – |
| Ligosan | 1 | 19 | 12 | 0.38 | −1.25 | 2.01 | .648 | – | – |
| Arestin | 2 | 42 | 39 | −0.586 | −1.048 | −0.125 | .013 | 0.0 | .482 |
| Dentomycin | 2 | 90 | 40 | 0.264 | −0.127 | 0.655 | .186 | 0.0 | .759 |
| Meta-regression | Coefficient | 95% CI | p-value | |
|---|---|---|---|---|
| Lower | Upper | |||
| Parallel/Split-mouth | 0.372 | −0.666 | 1.411 | .432 |
| Partial/Full-Mouth | 0.271 | −0.824 | 1.365 | .584 |
| Vehicle/SRP | −0.151 | −1.499 | 1.198 | .803 |
| Untreated/Refractory | 0.163 | −0.911 | 1.238 | .735 |
| Follow-up months | −0.010 | −0.043 | 0.023 | .514 |
- Abbreviations: CI, confidence interval; SRP, scaling and root planing; TetraStrip, tetracycline strip.
Significant differences (p < .001), favouring the test groups, were also observed in split-mouth studies, those with partial-mouth assessment, or with SRP alone as control, and in all types of periodontitis (p < .05). Meta-regression detected a statistically significant impact of study design and type of assessment.
Eleven different local antimicrobials were included in the analysis, with four of them demonstrating statistically significant benefits for test groups, with no heterogeneity (Ligosan: n = 3, WMD = 0.408; multiple tetracycline strips: n = 1, WMD = 0.480; Arestin: n = 4, WMD = 0.517; Chlosite: n = 1, WMD = 0.840). Among the other products, six demonstrated benefits for test groups, but not significant (Atridox: n = 1, WMD = 0.640; Actisite: n = 8, WMD = 0.276; Periochip: n = 10, WMD = 0.183; Elyzol: n = 4, WMD = 0.035; one tetracycline strip: n = 1, WMD = 0.440; Aureomycin: n = 1, WMD = 1.000), while Dentomycin (n = 3, WMD=−0.116) showed not significant differences favouring the control group.
3.6.4 Clinical attachment level, long-term (12–60 months)
Data from 10 comparisons between test and control groups were combined in a meta-analysis, showing no significant benefits (p = .607) for test groups (WMD = 0.090, 95% CI [−0.253; 0.433]), with significant heterogeneity (I2 = 75.3%; PI [−1.01; 1.19]) (Table 4b and Figure A5.4).
Significant differences were also observed, in favour of the test groups, in studies with full-mouth assessment (p = .024) and 60-month follow-up (p = .009). Meta-regression did not show a significant impact for any of the factors considered.
Six different local antimicrobials were included in the long-term analysis, with four of them demonstrating benefits for test groups (Atridox: n = 1, WMD = 0.200; Actisite: n = 3, WMD = 0.493; Ligosan: n = 1, WMD = 0.380; Dentomycin: n = 2, WMD = 0.264), and the other two, in favour of control groups (Elyzol: n = 1, WMD = −0.060; Arestin: n = 2, WMD = −0.586). The only comparison showing statistically significant differences was for Arestin (p = .013).
3.6.5 Bleeding on probing
Data from nine comparisons (only short-term) between test and control groups were combined in a meta-analysis, showing a non-statistically significant overall benefit for test groups (WMD = 2.495, 95% CI [−1.996; 6.986], p = .276), with significant heterogeneity (I2 = 86.9%; PI [−10.63; 15.62]) (Table 5 and Figure A5.5).
| Number of | Weighted mean difference (WMD) | Heterogeneity | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Studies | Patients | WMD | 95% CI | p-value | I2 (%) | p-value | |||
| control | test | Lower | Upper | ||||||
| All | 9 | 199 | 195 | 2.495 | −1.996 | 6.986 | .276 | 86.9 | <.001 |
| Partial-Mouth | 6 | 88 | 85 | 4.71 | −2.198 | 11.618 | .181 | 64.8 | .014 |
| Full-Mouth | 3 | 111 | 110 | 1.255 | −9.38 | 11.889 | .817 | 88.0 | <.001 |
| Untreated | 5 | 127 | 126 | 1.496 | −3.652 | 6.644 | .569 | 92.2 | <.001 |
| Refractory | 4 | 72 | 69 | 4.547 | −2.433 | 11.526 | .334 | 55.8 | .079 |
| Test product | |||||||||
| Atridox | 1 | 6 | 6 | 12.5 | 1.746 | 23.254 | .023 | – | – |
| Actisite | 2 | 30 | 29 | −0.458 | −1.543 | 0.628 | .408 | 0.0 | .351 |
| Dentomycin | 1 | 20 | 21 | 0.08 | −13.343 | 13.503 | .991 | – | – |
| Elyzol | 3 | 106 | 104 | 4.315 | −1.239 | 9.87 | .169 | 11.1 | .325 |
| Chlosite | 1 | 12 | 10 | 22.0 | 5.957 | 38.043 | .007 | – | – |
| Periochip | 1 | 25 | 25 | −7 | −8.793 | −5.207 | <.001 | – | – |
| Meta-regression | Coefficient | 95% CI | p-value | |
|---|---|---|---|---|
| Lower | Upper | |||
| Partial/full-mouth | −3.917 | −17.988 | 10.154 | .531 |
| Untreated/refractory | 2.742 | −12.171 | 17.655 | .677 |
- Abbreviation: CI, confidence interval.
No significant differences were detected in studies with different types of assessment or of patients. Meta-regression did not show a significant impact for any of the factors considered.
When the meta-analysis was performed by local antimicrobial, three products achieved statistically significant differences, but with just one study each: Atridox (WMD = 12.500), Chlosite (WMD = 22.000) and PerioChip (WMD = −7.000, favouring the control group). Only two products showed results for two or more studies: Actisite (n = 2, WMD = −0.458) and Elyzol (n = 3, WMD = 4.315), and none of them showed statistically significant differences. Finally, Dentomycin, with one study, reported a WMD = 0.080, and no significant differences.
3.6.6 Patient-reported outcome measures and adverse event
In 26 studies, PROMs were evaluated, focusing on adverse events, and normally with no clear description of the events (Tables A5.1 and A5.2). In the cases in which they were identified, their description included gingival/periodontal abscess/infection, gingival tenderness/gingivitis, pain and root/tooth sensitivity.
Twenty-seven comparisons were available for adverse events not specified by the authors, being the most frequent result “no adverse event” (for 19 test and 21 control groups), with 19 comparisons with the frequency of not specified adverse event being 0% for both groups. The remaining eight comparisons, coming from five different studies, were meta-analysed showing an odds ratio (OR) = 0.983 (95% CI [0.775; 1.246]), with no significant differences and with no heterogeneity (I2 = 20.4%); p = .268). Similar results were observed when the analysis was repeated with the different subgroups. Five local antimicrobials were included in the analysis: Ligosan, Actisite, tetracycline strips (either one or multiple) and Arestin.
For gingival/periodontal abscess/infection, three comparisons were available, with no significant differences (OR = 1.277, 95% CI [0.845; 1.931]) and no heterogeneity (I2 = 0%). Similar results were found for gingival tenderness/gingivitis (3 comparisons, OR = 1.194, 95% CI [0.774; 1.842]), pain (3 comparisons, OR = 1.154, 95% CI [0.754; 1.765]) and root/tooth sensitivity (3 comparisons, OR = 1.225, 95% CI [0.864; 1.735]).
3.7 Risk of bias across studies (publication bias) and sensitivity analyses
No publication bias was detected in the main outcome variable (p = .118; Egger´s test for changes in PPD) (Figure A5.6). The sensitivity analyses detected the influence of particular studies in the overall heterogeneity, but as the global estimator did not change significantly after omitting each of the contributing studies, it was decided to keep all selected studies (Figure A5.7).
4 DISCUSSION
The present systematic review was able to identify 50 RCTs (reported in 59 publications), assessing the use of locally delivered antimicrobials as adjuncts to subgingival debridement for, at least, 6 months. The overall meta-analysis, combining all test groups, demonstrated statistically significant PPD reductions and CAL gains (WMDs for 6- to 9-month studies, 0.365 mm and 0.263 mm, respectively), when compared with control groups. In addition, minor adverse effects were observed, with no differences between test and control groups. However, the significant heterogeneity observed in most of the analyses, together with the estimated prediction intervals calculated, highlighted the need for further analysis to deeply understand the results.
These results are similar to those reported in previous systematic reviews (Bonito et al., 2005; Hanes & Purvis, 2003; Matesanz-Pérez et al., 2013; Smiley et al., 2015), with additional PPD reductions ranging between 0.3 and 0.6 mm, and demonstrate that locally delivered antimicrobials, as adjuncts to SRP, can improve the clinical outcomes of mechanical treatment alone or with a placebo. As in the present work, benefits were statistically significant, although clinical relevance is more controversial, due to the small magnitude of some of the benefits. In addition, many factors have been identified as possibly influencing the outcomes and explaining the detected heterogeneity.
4.1 Heterogeneity explained by study design
In order to identify sources of heterogeneity, subgrouping and meta-regressions were performed, including factors such as study design, and types of assessment, of control or of periodontitis. Meta-regression (for PPD and CAL changes in 6- to 9-month studies) identified a significant influence of study design (with larger benefits for split-mouth studies) and type of assessment (with larger benefits for partial-mouth assessments). Although not significant, studies on untreated patients tended to achieve larger PPD reductions, and studies with placebo tended to achieve smaller benefits in both PPD and CAL changes.
Split-mouth studies offer a clear advantage of isolating treatment comparisons from intersubject variability and consequently have the potential to require less number of subjects than in a parallel design with the same power (Shoukri, Colak, & Donner, 2011). However, there is a risk of potential leakage of the treatment effect from one site to another, which is called the carry-across effect (Lesaffre, Philstrom, Needleman, & Worthington, 2009). This limitation is seldomly acknowledged by the authors, with one specific exception (Jeffcoat et al., 1998), who recorded but did not analyse data from contra-lateral sites in the same patient. In addition, not having a placebo/vehicle control may have consequences, on top of the lack of blinding for patients, affecting different types of bias; placebo-controlled studies minimize subject and investigator bias and increase the ability to detect adverse effects (Food & Drug Administration, 2001). The influence of partial/limited sites evaluation has been shown to clearly affect misclassification bias in periodontitis (Heaton, Garcia, & Dietrich, 2018; Romano, Perotto, Castiglione, & Aimetti, 2019), and it has also been reported that partial- and full-mouth assessments should not be combined in clinical trials (Chilton, Fertig, & Talbott, 1978). Finally, a poorer response can be expected in non-responding/refractory patients, when compared with untreated periodontitis, due to a larger potential for healing in untreated sites (Harrel & Nunn, 2001), or to specific microbiological profiles or immunological conditions in non-responding/refractory cases (Haffajee et al., 2004); however, non-responding sites after therapy, or recurrent disease during SPT, may represent a reasonable indication for local antimicrobials (since only localized sites/teeth may be affected).
4.2 Heterogeneity explained by different products/formulations
It may not be correct to discuss the overall use of sustained-released local antimicrobials, since each particular product poses unique properties that may prevent from assessing them globally. For example, the pharmacokinetic and pharmacodynamic characteristics of the different products show large variations. These have also consequences in the instructions of application/usage: strictly one application or repeated applications, or additional applications depending on the clinical outcomes; in addition, some products have to be removed after 7–10 days, and others may use some dressing or cyanoacrylate to protect the treated area.
The number of applications was not considered among those factors of study design discussed before, since in most cases the evaluated protocol was the one suggested by the manufacturer (see Table A5.3) and sometimes differed among the products evaluated.
The largest observed benefits were observed for doxycycline- or tetracycline-based products: Atridox (WMD = 0.8 mm for PPD), Actisite (WMD = 0.729 mm) and Ligosan, (WMD = 0.525 mm). Minocycline-based products (Dentomycin, Periocline) demonstrate similar benefits to the overall WMD for all products.
In the last years, an effort has been made to develop subgingival slow-released antiseptics, aiming at providing similar benefits to local antibiotics but with less adverse effects, especially the development of bacterial resistance. Two chlorhexidine-based products have been included in the present review, either delivered as chips (Periochip) or as gels (Chlosite). However, they seem to provide smaller benefits, when compared to the previously mentioned local antibiotics.
Other products have demonstrated limited improvements: Elyzol rendered no additional benefit over SRP alone, what is in agreement with previous systematic reviews (Bonito et al., 2005; Hanes & Purvis, 2003; Matesanz-Pérez et al., 2013; Smiley et al., 2015).
4.3 Limitations
The main limitation of the analysed data is the limited quality of the papers reviewed and the lack of appropriately conducted research. Although some methodological aspects, such as blinding and randomization, were acceptable in most cases, the global risk of bias was considered as high in most of the included publications, and only three of them were classified as having a moderate risk of bias (Eickholz et al., 2002; Killeen et al., 2016; Tabenski et al., 2017). In addition, when combining data (meta-analysis), statistically significant heterogeneity was observed for most of the analyses, what limits the results of this systematic review as well. Furthermore, the risk of bias of the selected studies may have increased by the participation of the manufacturing companies in most of the studies, either by sponsoring them or even by including their personnel in the research teams. Finally, relevant outcomes are seldomly reported (pocket closure, frequency distribution of PPD categories) and could not be evaluated in the present review.
Due to high risk of bias and the heterogeneity of the studies and of the reported results, it is difficult to define an evidence-based protocol of usage of locally delivered antimicrobials in the clinical practice. The available information is scarce regarding the clinical scenarios to use them, which product/s to choose and the profile of the patients who would eventually benefit the most from this treatment option. Baseline characteristics of the patient, availability of products in the national markets or added economic costs and cost/benefit ratio (Henke et al., 2001) are also factors that need to be carefully assessed.
5 CONCLUSIONS
- The adjunctive use of locally delivered subgingival antimicrobials results in statistically significant benefits in terms of PPD reduction and (only short-term) CAL gain.
- Significant heterogeneity was observed in most of the analyses, since they combined different products with different active agents, and also influenced by the study design (larger benefits for split-mouth studies and for partial-mouth assessments). Prediction intervals did not demonstrate significant benefits, and the magnitude of the benefits may not be clinically relevant.
- No increase in adverse effects or differences in PROMs were observed.
ACKNOWLEDGMENTS
The authors express their gratitude to all the authors answering their requests for additional information.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
