3.1 Eligible studies and study characteristics
A total of 1178 records were found by keyword searching in the PubMed (n=312), Cochrane Library (n= 34), EMBASE (n= 618), Medline (n=9), Web of Science (n=200) and Sigle (n=5) databases. Subsequently, 558 duplicates were removed from the pooled database, and 620 unrelated studies were excluded by screening the titles and abstracts. Following full-text assessments, 25 articles were excluded: 5 articles had no control groups; 15 publications had low quality; 2 studies did not present metrics of interest; 2 studies were meta-analyses (although we excluded these articles, we included all the original studies); and one study’s subjects were older than the target age range. Finally, a total of 6 studies met the inclusion criteria for meta-analysis[23-28]. Among them, two studies contained subgroups. One article was grouped by sex, and in another study, a second cephalometric analysis was performed a year later in the same population without any intervention. For the latter, we include only the initial measurement data. The publication time of the included studies ranged from 2009 to 2018. The flow diagram of the literature search and review process based on the PRISMA statement is shown in Appendix B.
In this review, there were a total of 722 subjects;331 children with mouth breathing were included in the experimental group and 391 children with normal nasal breathing were included in the control group. The age range included in these studies was 3 to 16 years old. Of the 6 articles included, mouth breathing due to adenoid/tonsil hypertrophy was studied in 4 articles, obstructive sleep apnoea syndrome (OSAS) was studied in 2 articles. The cephalometric analysis indicators in all the included studies were statistically analysed, and the indicators that appeared 2 times or more were selected for consolidation. The included indicators were SNA, SNB, ANB, SN-OP, SNGoGn, MP-H, 1-NA, 1. NA, 1. NB, 1-NB, SPAS, PAS, and C3-H. The characteristics of the included studies are summarized in Table 2.
3.3Risk of bias assessment
All the included studies were from a specific population, so the representativeness of the included studies was not high. Meanwhile, the included studies were all retrospective studies, so the problem of non-response did not exist. The quality assessment of nonrandomized studies is shown in Table 3. About the bias assessment, three articles had low risk and three articles had medium risk (Table 4). Since there were less than 10 studies included in the meta-analysis, we did not conduct funnel plots or Begg's rank correlation tests.
3.4 Primary outcome measures
3.4.1 Sagittal direction
After the meta-analysis with Review Manager 5.3, 1. NB was not statistically significant (fixed: MD, random, 95% CI, P>0.050). As illustrated in Fig. 2, the indicators of sagittal direction are as follows. Two indicators in mouth-breathing children was lower than that in nasal-breathing children: SNA (4 studies, MB:215, NB:315, MD: -1.61, 95% CI: -2.31 to -0.91, P <0.0001, I²=0%), SNB (5 studies, MB:328, NB:428,MD: -1.92, 95% CI: -2.74 to -1.10, P <0.0001, I²=48%). However, four parameters showed higher values in children with mouth breathing than in children with nasal breathing: ANB (5 studies, MB:276, NB:336,MD: 0.79, 95% CI: 0.10 to 1.49, P = 0.0200, I²=72%), 1. NA (3 studies, MB:171, NB:286,MD: 1.98, 95% CI: 0.30 to 3.66, P = 0.020, I²=45%), 1-NA (3 studies, MB:171, NB:286,MD: 0.72, 95% CI: 0.23 to 1.20, P = 0.0040, I²=24%), and 1-NB (3 studies, MB:156, NB:274,MD: 1.06, 95% CI: 0.55 to 1.57, P < 0.0001, I²=0%).
3.4.2 Vertical direction
The vertical indicators are shown in Fig. 3. The following indexes were higher in mouth-breathing individuals than in nasal-breathing individuals: SN-OP (3 studies, MB:171, NB:286,MD: 3.20, 95% CI: 2.44 to 3.97, P < 0.0001, I²=0%), and SNGoGn (6 studies, MB:383, NB:483,MD: 4.46, 95% CI: 3.52 to 5.39, P < 0.0001, I²=40%).
After meta-analysis, MP-H were not statistically significant (fixed: MD, random, 95% CI, P>0.050). As shown in Fig.4, the airway data of children in the experimental group were lower than those in the control group: SPAS (3 studies, MB:156, NB:274,MD: -5.23, 95% CI: -5.95 to-4.51, P < 0.0001, I²=64%), PAS (3 studies, MB:171, NB:286,MD: -2.11, 95% CI: -2.90 to -1.32, P < 0.0001, I²=42%), and C3-H (3 studies, MB:171, NB:286,MD: -1.34, 95% CI: -1.96 to -0.72, P < 0.0001, I²=0%).
The heterogeneity of the other outcomes mentioned above was acceptable.
3.4.3 Subgroup analysis
Subgroup analysis was performed for all included studies based on the etiology of mouth breathing. In mouth breathing children with adenoid/tonsil hypertrophy, SNA was not statistically significant. As shown in Appendix C, SNB(5studies, MB:172, NB:154,MD: -1.80, 95% CI: -3.25to -0.15, P =0.0300, I²=54%) is higher than normal children. While, ANB(5studies, MB:224, NB:246,MD: 0.92, 95% CI:0.04to 1.81, P =0.0400, I²=77%) and SN-GoGn(6studies, MB:227, NB:209,MD: 4.11, 95% CI: 2.87to5.35, P<0.0001, I²=35%) is lower in mouth breathing children with adenoid/tonsil hypertrophy. As shown in Appendix D, five outcomes were higher in mouth breathing children with OSAS: SNA (4studies, MB:156, NB:274,MD: -1.63, 95% CI: -2.40 to-0.87, P<0.0001, I²=0%),SNB (5studies, MB:156, NB:274,MD: -2.07, 95% CI: -3.00 to-1.33, P<0.0001, I²=20%),SPAS (2studies, MB:156, NB:274,MD: -5.23, 95% CI: -5.95 to-4.51, P<0.0001, I²=64%), PAS (2 studies, MB:156, NB:274,MD: -2.06, 95% CI: -2.99 to-1.14, P<0.0001, I²=59%), C3-H (2 studies, MB:156, NB:274,MD: -1.30, 95% CI: -1.94 to-0.66, P<0.0001, I²=0%). The rest of the results were lower in mouth breathing children with OSAS:SN-OP(3studies, MB:156, NB:274,MD: 3.08, 95% CI: 2.30 to3.87, P<0.0001, I²=0%), SN-GoGn (5studies, MB:156, NB:274,MD: 4.87, 95% CI:3.37 to6.37, P<0.0001, I²=49%), 1-NA (3studies, MB:156, NB:274,MD: 0.85, 95% CI: 0.43 to1.27, P<0.0001, I²=0%), 1.NA (3studies, MB:156, NB:274,MD: 2.47, 95% CI:1.05 to3.89, P=0.0006, I²=24%), 1-NB (2studies, MB:156, NB:274,MD:1.06, 95% CI: 0.55 to1.57, P<0.0001, I²=0%). 1.NB and C3-H were not statistically significant.