Protocol and Registration

This SR was conducted according to a specific protocol developed and piloted with respect to the guidelines outlined in the Preferred Reporting Items for Systematic review and Meta-Analysis Protocols (PRISMA-P).¹⁹ The study was registered at PROSPERO (CRD42023331648).

Eligibility Criteria

Eligibility criteria were based on the PICOS (Participants, Intervention, Comparison, Outcomes, Study design) strategy (Table 1). The population was patients of any age and sex, in permanent dentition, with or without previous orthodontic treatment, presented with impacted or semi-impacted or erupted mandibular third molars. These patients were compared with patients with absence of mandibular third molars (agenesis or extracted).

Table 1. Eligibility Criteria Based on the PICOS (Participants, Intervention, Comparison, Outcomes, Study Design) Strategy

View Fullscreen

Finite element analyses, follow-ups, case reports, reviews (systematic and literature), author responses, letters to the editor, editorial summary, books and/or book chapters, abstracts, congress abstracts, personal opinions, author debates, summary articles, non-English articles, or no author response to inquiry email for data clarification were excluded.

Information Sources and Search Strategy

The search databases included PubMed, Scopus, ProQuest, and Google Scholar. The databases were searched from inception until April 30, 2022. Research was specified and performed to identify any relevant study based upon various combinations of key words (Table 2).

Table 2. Electronic Search Strategy

View Fullscreen

Developed detailed search strategies for each database were conducted by one of the authors (GP) and assessed by the other two authors (AM and IS) independently. The search strategies developed for MEDLINE were adopted and were revised appropriately for each database to include differences in controlled vocabulary and syntax rules (Table 2). No restrictions were applied on the date and status of publication except that only English language papers were considered for inclusion. The reference lists of all the eligible studies, as well as these of relevant reviews, were searched. The authors of retrieved studies were contacted for further clarification if needed.

Study Selection

Two authors (GP and AM) assessed the titles and the abstracts of the retrieved records for inclusion independently and in duplicate. They were not blinded to the identity of the authors, institution of origin, or the results of the research. Any article title that met the objectives of the study but did not have an available abstract was fully analyzed in the final evaluation. Full texts of the selected studies were obtained and evaluated to verify whether they fulfilled the eligibility criteria. Disagreements were resolved by discussion or consultation with a third author (IS) before the final decision. The reference lists of all retrieved full text articles were searched for relevant articles that might have been missed through the electronic search. Finally, the articles that did not answer the clinical research questions were excluded.

Risk of Bias in Individual Studies

Two authors (GP and AM) assessed the risk of bias in individual studies independently and in duplicate using the “ROBINS-I” (RoB) assessment tool. Any disagreements were resolved by discussion or consultation with the last author (IS).²⁰

Data Collection and Data Items

The same two authors (GP and AM) performed data extraction independently and in duplicate. Any disagreements were resolved through discussion with the last author (IS). Predetermined and prepiloted data collection forms were used to record the necessary information: first author, age, type of study, previous orthodontic treatment, sample size, presence of third molars, analysis tools, crowding measurement, angulation of lower teeth, and outcome of the study.

Summary Measures and Synthesis of Results

The summary characteristics of the included studies were reviewed. The mean values and the corresponding SDs were extracted independently from each article that used quantitative measures for outcome assessment. The random effects model was used. Forest plots were also used to visualize the pooled estimates across studies. Statistical analysis of the data was conducted with the statistical package R (www.r-project.org, v3.6.2) and the level of statistical significance was set at 0.05.

Additional Analyses and Risk of Bias Across Studies

Egger’s test was used to assess potential publication bias in the meta-analysis via funnel plot asymmetry. If there were at least two studies with the same outcome and having sufficient homogeneity regarding the compared groups and the time frame, meta-analysis was planned. Otherwise, a qualitative synthesis of the results was reported. Statistical heterogeneity was inspected with the DerSimonian and Laird’s Q statistic and I² statistic. If split-mouth studies were included in meta-analysis, the correlation coefficient was planned to be computed to calculate the adjusted standard error (SE), as described in chapter 23 of the Cochrane handbook. The difference in means of both arms and the adjusted standard error were used in the meta-analysis.²¹

Study Selection

The results of literature searching identification, inclusion, and exclusion of the articles are presented in the flow diagram according to the PRISMA statement (Figure 1). Initially, 847 relevant records were identified after the electronic and manual search, while only 798 remained after manual duplicate removal. After title and abstract screening, 27 articles were selected for full-text review according to the inclusion and exclusion criteria. Finally, nine papers were excluded because four were reviews, five did not meet the criteria for outcome and comparator, two were case series, one pilot study, one questionnaire, and one unclassified. The remaining five did not meet the inclusion criteria for outcome and comparator. Finally, 13 articles were included in the qualitative synthesis.^1,22–33 However, only seven articles were included in the quantitative synthesis since the remaining studies were heterogeneous.^{1,22–24,26,28,30}

View Figure

• Download as Powerpoint
• Download Figure

Risk of Bias in Individual Studies

Risk of bias for the included studies^1,22–33 is depicted in Supplemental Table 1 and Table 3 (in summary). Eight studies were assigned an overall risk of bias of moderate risk,^{1,23,25,29–31,33} while the rest (five studies) were of high risk.^{22,26–28,32}

Table 3. Risk of Bias Summary

View Fullscreen

General Characteristics of Included Studies

The general characteristics of the studies included in the present SR, as well as the sample characteristics, are depicted in Tables 4 and 5. A total of 1342 participants was calculated from the included studies. All included studies were observational by design (cross-sectional and prospective nonrandomized controlled trials)^1,22–33 and used two-dimensional radiographs (orthopantomographs, lateral cephalometric radiographs) and casts as analysis tools, except two studies that used only casts²⁹ or cone beam computed tomography (CBCT).³⁰ Crowding was quantified with Little’s irregularity index in eight studies,^{22,23,26–30,33} three used Lundstrom analysis,^1,20,31 and two studies used Ganss ratio.^31,33 Five studies measured arch length^{22,26–28,32} and four studies measured arch width.^1,25,26,28 In four studies, the participants had undergone orthodontic treatment.^22,26–28 Four out of the 13 eligible studies supported a cause-and-effect relationship among third molars and anterior crowding.^29–32

Table 4. Summary Characteristics of Included Studies

View Fullscreen

Table 5. General Characteristics of the Studies Included in the Quantitative Synthesis

View Fullscreen

Quantitative Synthesis of Included Studies: Synthesis of Results

Seven studies were included in the quantitative analysis and three different meta-analyses were conducted. The first was conducted in patients without previous orthodontic treatment,^1,23,24,30 and the second in patients with previous orthodontic treatment.^22,26,28 The third analysis was implemented in combination for patients with and without previous orthodontic treatment.^{1,22–24,26,28,30}

The first meta-analysis of four studies was performed on patients without previous orthodontic treatment and compared mean differences in crowding between cases with or without third molars (agenesis or extracted). Crowding was statistically significant with random-effects model (**P = .01). According to the pooled results of meta-analysis, random effects model yielded a significant benefit for those without third molars compared to those with third molars (Figure 2). Specifically, those with third molars had greater mean Little’s irregularity index scores compared to those without third molars. Egger’s test P value was .09, which implies borderline significant publication bias.

View Figure

• Download as Powerpoint
• Download Figure

The second meta-analysis of three studies on patients with previous orthodontic treatment compared mean differences in cases with and without third molars (agenesis or extracted) regarding crowding, arch width, and arch length. The pooled estimate of crowding and arch width was not statistically significant (P = .23, P = .69). According to the pooled results of meta-analysis, random effects model showed that those with third molars had lower mean arch length scores than those without third molars (P = .07) (Figure 3). Additionally, Egger’s test P value was .73, which implies no publication bias.

View Figure

• Download as Powerpoint
• Download Figure

The third meta-analysis of seven studies on patients with and without previous orthodontic treatment compared mean differences in crowding between patients with and without third molars (agenesis or extracted). Crowding was statistically significant with random effects model (**P = .005). According to the pooled results of meta-analysis, random effects model yielded a significant benefit for those without third molars compared to those with third molars. Specifically, those with third molars had greater mean Little’s irregularity index scores compared to those without third molars (Figure 4). Egger’s test P value was .10, which implies no publication bias.

View Figure

• Download as Powerpoint
• Download Figure

Strengths and Limitations

A protocol was developed a priori and the present review followed clear-cut guidelines. The search strategies employed were exhaustive, covering electronic, manual, and gray literature material up to April 2022. Their character was comprehensive, irrespective of date and status of publication. Every effort to decrease bias in the methodology employed was made. Screening, verification of eligibility, abstraction of information, assessment of risk of bias, and of the quality of evidence were all performed in duplicate, and any disagreement was resolved by discussion or consultation until a final consensus was achieved.

However, a potential source of bias in the present review could be the exclusion of articles not written in English. Further limitations in this study arose mainly from the nature and the characteristics of the retrieved data during the review process. Factors such as extractions, interproximal enamel reduction, or compliance with fixed/removable retention protocols were not evaluated in the included studies. However, the main limitation was related to the heterogeneity of the studies and the inadequate control of confounding in the included studies. These were nonrandomized trials, with risk of bias and differences in the methods of assessing the crowding. Differences in the study groups (impacted or erupted or present third molars vs agenesis or extracted) may have confounded the present findings and the impact of the third molars on the mandibular incisors may have occurred before their extraction. Additionally, each of the first and second meta-analyses included three studies. Egger’s test may lack the statistical power to detect bias when the number of studies is small (<10). Additionally, there are several other factors that may affect anterior crowding, such as growth-related changes, muscular factors, periodontal ligament traction, bone adaptation, masticatory force, etc.^17,18 Further research is necessary to clarify the etiology of mandibular anterior crowding and to quantify the relapse tendency in each individual case.