Treating periodontitis-A systematic review and meta-analysis comparing ultrasonic and subgingival hand scaling at different pocket probing depths.

Background: Mechanical plaque removal has been commonly accepted to be the basis for periodontitis treatment. The study aims to compare the effectiveness of ultrasonic subgingival scaling and subgingival hand scaling at different initial pocket probing depths in periodontitis treatment. Methods: Public databases were searched. Weighted mean pocket probing depths and clinical attachment loss reduction differences estimated by random effects model. Results: Ten randomized controlled trials were included out of 1,434 identied. Selected outcomes were pocket probing depth and clinical attachment loss. Initial pocket probing depth and follow-up periods formed subgroups. For 3-month follow-up: (1) too few shallow initial pocket studies available; (2) medium depth studies were unmergeable; (3) deep studies were adequate. No statistical differences between pocket probing depth nor clinical attachment loss reduction between ultrasound and hand groups. For 6-month follow-up: (1) too few shallow initial pocket probing depth studies for analysis; (2) medium initial pocket probing depth studies favored hand scaling. No statistical differences observed in clinical attachment loss reduction between the two approaches; (3) deep initial pocket probing depth studies showed hand scaling superior by both measures. Conclusion: When initial pocket probing depths were ≥ 4mm, pocket probing depth results, clinical attachment loss reduction, and other outcomes indicated subgingival hand scaling was superior. When operation duration and comfort were considered, ultrasonic debridement was. only 3 months, CAL caused great heterogeneity. Heterogeneity was also large in the following groups: 1) deep pocket, at 3-months follow-up; 2) medium and deep pockets, at 6-months follow-up [16]. After excluding this study, heterogeneity decreased from: 87%; 52%; and 71%, to 24%; 0%; and, 0%. This study’s heterogeneity was quite big and should be excluded. After exclusion, at a 6-month follow-up, after UD, CAL reduction was more than SRP and were statistically signicant different.

Only hand-or UD-treated periodontitis patients in a completely randomized controlled trial (RCT) were included. Subjects were classi ed clinically by groupin initial pocket probing depth (PPD) into: (1) shallow: PPD ≤4mm; (2) medium: 4mm< PPD ≤6mm; or, (3) deep: PPD >6mm. Studies had to be in-vivo and compare UD with SRP for inclusion. A group could not be simultaneously included in different PPD categories. Root bifurcations were not considered in the selection process.
Studies meeting the following conditions were excluded: (1) follow-up in less than 3 months; (2) treatment during follow-up. Cluster trials not included. Studies with primary or secondary outcomes were included.

Participants:
Adults (age ≥18) diagnosed with periodontitis unaccompanied by any other oral or systemic disease and not taking antibiotics. Intervention: Meta-analysis sought to eliminate bias caused by different initial PPD to compare UD and SRP. There are several ultrasonic device makers. A maker may produce more than one model. All such devices share the same mechanism. Manual instruments operate on a different theory. No distinctions were made between UD makes or models. A Gracey scraper was chosen as the SRP instrument. Outcomes: A study may have primary and secondary indicators. Each indicator was processed differently. Not all outcomes were consistent with the criteria. The outcomes included appear in tables. (Table 1 and 2).
(1) Primary outcomes: PPD and clinical attachment loss (CAL) were the primary outcomes to compare different subgroup outcomes.
(2) Secondary outcomes: Bleeding on probing (BOP), GR, and post-scaling residual dental calculus were used as measures. These indicators are of interest. Data about them could not be extracted for meta-analysis due to the limited number of studies, their different measurements, and de nitions. They were analyzed by comparing each study's results.

Search strategy
PubMed, Cochrane Central Register of Controlled Trials, EMBASE, Medline, and ClinicalTrials.gov were searched until January, 2019 for relevant studies. The search was performed using a combination of controlled vocabulary and key words (Table S2). No language or time restrictions were imposed Potentially eligible studies were considered for review irrespective of primary outcomes or language. A manual search using reference lists of related articles was performed.

Data Collection and Analyses
(1) Study selection and quality assessment.
Three review authors (Zhang, Hu, and Zhu) independently searched and included eligible studies. The quality of each study was reviewed and evaluated and relevant data extracted. When there was a disagreement whether to include or not, a discussion including consulting corresponding author (Chen and Li) was held and an agreement reached on inclusion or exclusion.
A study's methodological quality was assessed using the original publication. Trial quality was evaluated using Cochrane review bias assessment risk criteria [15].
This included random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel blinding (performance bias), outcome assessment blinding (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other biases. Possible ratings were ranked by risk: low (L); high (H); uncertain (U).
(2) Statistical analyses When appropriate, data extracted was combined for meta-analysis using Review Manager 5.3. Effect size was estimated and reported as the mean difference (MD) for continuous variables with a 95% con dence interval (CI). Weight was calculated in individual studies based on the inverse of variance. This study used a random-effects model for analyses due to expected heterogeneity of the studies selected. Study statistical homogeneity was assessed using a Cochran test and by examining the observed variances in effect sizes and residual variance. I 2 was calculated to quantify heterogeneity. I 2 >50% was considered signi cant [15]. No statistical corrections were used to adjust for multiple analyses.
According to Cochrane reviews [15], in meta-analysis, studies with baseline changes as outcomes could be combined with those with nal measurements as outcomes. In randomized trials, differences in mean values obtained from baseline changes were usually analyzed on the basis of nal measurements and obtained the same effects. In this meta-analysis, baseline changes and nal measurements from different studies were combined.
A reason for disagreement among previous studies was that PPD in uence had not been eliminated. The instant study established three subgroups based upon initial pocket depth: shallow; medium; and deep, to control for initial PPD effects.
If a study had two initial PPD groups that could be included in one subgroup, data was combined to conform to the depth classi cation, using the formulas in Table S3 [15].

Study selection
There were 1,434 studies searched. References in selected papers were searched and no additional studies were located. After reading full the texts, ten studies were selected for qualitative synthesis. Process selection appears in Figure 1.
There were 495 duplicates in the 1,434 studies. There were 875 studies ruled out after reading titles and abstracts. Another 43 studies were ruled out after reading the full text. There were 11 unavailable or awaiting classi cation. Finally, 10 studies were included.

Study characteristics
All research included were RCTs. Follow up periods were 3 and 6 months. Characteristics of articles selected for primary outcomes appear in Table  1. Characteristics of articles selected for secondary outcomes appear in Table 2. Reasons for study exclusion appear in Table S4.
Limited information of 11 studies could be obtained from the publication. Inclusion, or exclusion could not be decided because of unobtainable full texts and unknown speci c conditions. These appear in Table S5.
Quality and risk of bias assessment Bias analysis results for the studies appear in Figures S1 and S2. Most studies do not have high bias risks. Funnel plots could not be done due to limited number of studies (<10).

Meta-analysis Results
The follow-up period lengths of the studies varied. Most were 3 and 6 months. This allowed for grouping into 3-and 6-months and reduced heterogeneity.
(2) 6 months: a) Initial PPD ≤4mm Two articles [16,17] met the criteria as they used baseline changes as outcomes and one baseline changes value of 0. A meta-analysis could not be done using the two studies. They reported no statistical differences between UD and SRP. b) Initial PPD >4mm When initial PPD was medium, differences between SRP and UD were not statistically signi cant (MD 0.19, 95%CI [0.11, 0.27], P = 0.22).

Secondary outcome measures
(1) GR Sculean et al. [18] indicated no statistical differences studying single or multiply-root teeth between SRP and UD at 6-months when initial PPD was deep. Kargas et al. [19] noted that, at medium depths, there were no statistical differences between ultrasonic and SRP at either 3 or 6months.
(2) BOP: Christgau et al. [17] found that at 6-months, SRP showed greater BOP reduction for initial deep pocket compared to ultrasound.
(3) Residual dental calculus: Schwarz et al. [20] indicate that for single-root teeth at deep initial depth, UD was superior to SRP in removing subgingival dental calculus. Yukna et al. [21] found no statistical differences in residual dental calculus rates between UD and SRP with initial PPD in 5-6mm/7-8mm/ > 9mm. Gellin et al. [22] found no statistical differences in dental calculus clearance rates between the two when initial PPD was 0-3mm; 4-5mm; or, 6-12mm. When SRP was combined with UD, the effect was superior to either UD or SRP individually.

Sensitivity analysis
Outcome: PPD At 3-months, at medium depths, heterogeneity was great (I 2 = 90%). After sensitivity analysis, four studies were found highly heterogeneous to one another and was unsuitable for meta-analysis. After a bias analysis, the heterogeneity source was thought to be: (1) small number of studies; (2) the fact that tissue healing takes time and early probes disrupted attachment gains. At 3-months, PPD and CAL reductions were unstable, causing large heterogeneity.

Outcome CAL
When was initial PPD was medium, at 3-months follow-up, results were similar to PPD reduction with high heterogeneity (Tau² = 0.01; Chi² = 10.92, df = 3 (P = 0.01); I² = 73%). According to a sensitivity analysis, four papers were highly heterogeneous with each other, making them unsuitable for meta-analysis (Fig. 3B). The same argument applies to a 3-month follow-up when initial PPD was medium: (1) too few studies; (2) tissue healing took time and early intervening probing may damage attachment gain. When the follow-up period was only 3 months, CAL were unstable which caused great heterogeneity.
Heterogeneity was also large in the following groups: 1) deep pocket, at 3-months follow-up; 2) medium and deep pockets, at 6-months follow-up [16]. After excluding this study, heterogeneity decreased from: 87%; 52%; and 71%, to 24%; 0%; and, 0%. This study's heterogeneity was quite big and should be excluded. After exclusion, at a 6-month follow-up, after UD, CAL reduction was more than SRP and were statistically signi cant different.

Discussion
In the same paper, PPD and CAL heterogeneity in the studies was much greater at 3-months than at 6-months. The effect was quite apparent at medium PPD, 3-months. Deep PPD heterogeneity at either 3-months or 6-months was acceptable. This suggests that deep pocket heterogeneity has no relationship with time. The reason may be that the tissue took time to heal. In deep pockets, tissue contacted and attached to the bone better, resulting in shorter healing time and more stablility within 3 months. At medium pocket depths, tissue does not contact bone as readily as at deeper pockets, so healing time is longer. Probing too early in healing process may damage tissue and in uence attachment gain and lead to unstable results.
Sensitivity analysis suggests that two few articles (1-2) met the requirement of shallow initial PPD. This meant that no reliable conclusions could be reached and more studies are required. PPD results were not reliable for medium pocket depths at the 3-month follow-up whose heterogeneity of four included articles was quite large and requires more studies to reach reliable conclusions. CAL medium pocket depths at a 3-month follow-up was similar. Post analysis: 1) medium depth [19] and deep depth [23] at 6-month follow-up for PPD, 2) deep pocket at 3-month follow-up, medium, and deep pocket at 6-months follow-up for CAL [16] were excluded.
After excluding a major source of heterogeneity, the deep initial PPD at 3-month follow-up and all data from the 6-month follow-up were analyzed and following results were returned: A. deep pocket at 3-month follow-up: PPD and CAL reductions showed no differences between ultrasound and hand groups. B. 6-month follow-up: a) for shallow initial PPD, there were no differences between hand and UD; b) at medium initial PPD, PPD reduction showed SRP was better. CAL reduction showed no differences between ultrasound and hand groups; c) at deep initial PPD, PPD and CAL indicated SRP was superior.
The number of roots, either single or multiple, had little effect on either hand or ultrasound at medium or deep PPD in terms of GR [18,19].
BOP results from a study [17] showed superior BOP reduction at deep pocket depths with SRP than with ultrasound. Residual calculus provided different results. Two studies [21,22] indicated that, regardless of depth, there were no statistical differences in calculus clearance rates between ultrasound and hand treatment. Schwarz [20] indicated when PPD was deep, for a single-root tooth, ultrasonic dental calculus removal was more e cient than SRP.
There were others interesting results in various indicators.
Studies of bacteria populations returned complex results. In terms of bacterial reductions, some people indicated no differences between hand and ultrasound [17,19], while other study [24] found SRP to be better. As for Treponema denticola, Ioannou [16] found SRP to be better when PPD ≥4mm, while Derdilopoulou [25] found ultrasonic treatments is. But Petelin [26] indicated that both have advantage in bacteria reduction and Breininger [27] found SRP somewhat more prone to leave apical, or calculus-associated, plaque.
When measuring cementum thickness, periodontal membrane cells and broblasts, Bozbay [28] found that UD retained more cementum. Schwarz [29] found periodontal membrane cells /mm² treated by ultrasound was signi cantly higher. SPR was similar to UD concerning broblast morphology in Foroutan [30]. When measuring in ammation, Malali [24] indicated white blood cell count decreased more after SRP but Torfason [31] found gingival crevicular uid change was similar after SRP and UD.
According to the newly-released 2018 periodontitis classi cations, there are some disputed understandings: 1. the differing economic and health care developments between developed and developing countries different in uences on periodontitis [42,43].Primary CAL in developing countries was three times of developed countries [43]. Only one study [24] involved a developing country (Turkey). Whether the conclusions reached in this paper apply to developing countries is unknown. (1) It may be premature to probe within 3 months after therapy. It may lessen attachment gain, slow or interrupt recovery and lead to inaccurate measurements. Only 6-month follow-up results are relied upon by the instant study to reach following conclusions: a) When initial PPD was shallow, no conclusions were drawn due to the limited number of studies. b) When initial PPD was medium, PPD reductions proved SRP as superior. CAL and GR results showed no statistical differences. More studies are needed before any conclusion can be drawn. c) When initial PPD was deep, SRP was superior in terms of PPD reduction, CAL and BOP. This conclusion also needs more study because of the limited number of studies.
(2) Considering operating time, patient preferences, adverse reactions, and operator sensitivity, UD was superior to SRP.
(3) In terms of residual dental calculus, bacteria changes, cementum thickness, periodontal adhesion, and root surface roughness, no conclusions could be drawn.
(4) The studies used are mainly from developed countries and conclusions may not be applicable to developing countries.
(5) UD were recognized to be superior when initial PPD was shallow; SRP was found superior when initial PPD >4mm.
Although UD is becoming increasingly popular, SRP is still irreplaceable.

Signi cance to research
(1) Inclusion and exclusion criteria could refer to the new classi cation to reduce bias; (2) Studies should consider other indicators such as BOP, PI, and GI, bacterial changes, when comparing in different PPDs; (3) More studies are needed in developing countries; (4) Age should be considered; (5) 12-month follow-ups are suggested to determine reliable results; (6) Single and multiple root teeth should be measured separately; (7) Follow-up treatment shouldn't be carried out after initial therapy; (8) Further studies should enlarge sample sizes to improve credibility; (9) Inclusion or exclusion criteria for smokers should be standardized.

Conclusions
Results varied between SRP and UD on different initial PPD: (1) the limited number of studies of shallow initial PPDs allow for no conclusion; (2) when initial PPD was medium, PPD reduction proved SRP superior, but CAL and GR reduction showed no statistical differences; (3) when initial PPD was deep, PPD and CAL reductions suggest that SRP was superior to UD. Other factors such as operating time, patient comfort, adverse reactions, and operator operating sensitivity proved UD superior.
Suggestions: (1) when initial PPD is shallow, use UD; (2) when initial PPD is medium or deep, use ultrasonic and SRP together. UD has become increasingly popular but cannot replace SRP. Declaration 1) Ethics approval and consent to participate ( not applicable for that section).
2) Consent for publication( not applicable for that section).  (h) Presence of at least ten teeth for each dental arch. Pregnant or nursing females were excluded from the study. SRP, scaling and root planing; UD, ultrasonic debridement; PPD, probing pocket depth; CAL, clinical attachment level.
All contents of the table are from the reference paper directly. Except for the serial number, there is no modification. *: The data is from two study centers: Italy and Sweden.   PPD>6mm Figure 2B. Forest plot comparing PPD at 6-months with hand versus UD of initial PPD>4mm in terms of the following: 1.2.1 PPD4-6mm; 1.2.2 PPD>6mm Figure 2C. Forest plot comparing CAL at 3-months with hand versus UD of initial PPD>4mm in terms of the following: 2.1.1 PPD4-6mm; 2.1.2 PPD>6mm Figure 2D. Forest plot comparing CAL at 6-months with hand versus UD of initial PPD were shallow in terms of the following: 2.2.1 PPD≤4mm. Figure 2E. Forest plot comparing CAL at 6-months with hand versus UD of initial PPD>4mm in terms of the following: 2.2.1 PPD4-6mm; 2.2.2 PPD>6mm.