Introduction

The removal of the maxillary third molar is commonly performed in oral and maxillofacial surgery [1, 2]. Berge et al. reported a high rate of postoperative complications associated with third molar extractions [3, 4], a finding replicated by investigators focusing specifically on postoperative complications following maxillary third molar extraction [2, 5–7]. In general, postoperative complications after third molar removal are less common in the maxilla than in the mandible. Surgical removal of the maxillary third molar is often easier because there are no large blood vessels or nerves near the extraction site and the maxillary bone is less dense and more flexible than mandibular bone [2]. However, oroantral perforation is a common operative complication during maxillary third molar removal [2, 8], as are rupture of the maxillary tuberosity, root fracture, and the displacement of the tooth/roots into the maxillary sinus. The risk of experiencing an oroantral perforation is associated with the spatial proximity of the maxillary third molar to the maxillary sinus floor [2, 9]. Panoramic radiography is an imaging modality commonly used by oral surgeons to visualize the third molar and to determine the risk of oroantral perforation. Archer et al. suggested a classification for the anatomic location of the impacted maxillary third molar [10], but this was based mainly on its spatial relationship to the second molar. Therefore, the application of this radiographic classification to determine the risk of oroantral perforation remains controversial [2, 9, 11]. Furthermore, few studies have quantitatively analyzed the factors contributing to oroantral perforation [2, 9], and none have evaluated the multivariate relationship(s) between these risk factors. We hypothesized that the identification of one or more risk factors with strong association to postoperative oroantral perforation could enhance treatment outcomes for patients. In this study, we investigated the multivariate relationships between the risk factors for oroantral perforation during surgical extraction of the maxillary third molar.

Patients and methods

Surgical removal of maxillary third molars was performed at the Department of Oral and Maxillofacial Surgery, Kobe University Graduate School of Medicine and Kakogawa East City Hospital between April 2009 and March 2013. All patients received written information regarding possible surgical complications, including oroantral perforation, and informed consent to proceed with extraction was acquired for each patient. For patients aged <18 years, informed consent was sought from their parent/guardian. This retrospective study has been conducted in full accordance with the World Medical Association Declaration of Helsinki and was approved by the institutional review board of Kakogawa East City Hospital (Authorization number: 27006).

Surgical methods

Mucoperiosteal flaps were raised to expose impacted teeth, followed by bone removal as necessary to establish application points for elevation and a free path of removal for the tooth/roots. Incisions were closed using 3–0 silk sutures and removed after 7 days. Oroantral perforation was comprehensively identified by the creation of oral bubbles following a pressure test (nose blowing), careful probing of the extraction socket using a blunt sinus probe, and audible changes to the nature of the noise produced by vacuum suction across the socket. Antibiotics and anti-inflammatory drugs were prescribed as necessary.

Study design

Panoramic radiography was performed for all patients using an Orthoceph OC100CT (Yoshida, Tokyo, Japan). The predictive variables of preoperative panoramic findings were decided by the first and second authors. The spatial relationship between the root apices and the floor of the maxillary sinus (root-sinus [RS] classification) was diagnosed based on panoramic radiography and classified as types 1–5 (Fig. 1). In type 1, the inferior border of the maxillary sinus curved gently and was clearly distinct from the root tip. In type 2, the inferior border of the maxillary sinus curved sharply and was partially superimposed across the root tip. In type 3, the inferior border of the maxillary sinus dropped extensively between the premolar and the maxillary third molar and was superimposed across most or all roots. In type 4, the inferior border of the maxillary sinus impinged on the root tip but dropped only between roots. In type 5, the relationship between the inferior border of the maxillary sinus and the root structure was indistinct. In this study, this classification was adopted to identify the association to postoperative oroantral perforation from an existing Japanese method for categorizing aberrant teeth [12].

Fig. 1
figure 1

The relationship of the roots to the maxillary sinus (root-sinus [RS] classification). Diagrams show typified examples of different spatial relationships between the root apices and the sinus floor (inferior border of the maxillary sinus). Type 1 shows clear distinction between the two features, whereas types 2 and 3 have different degrees of radiographic superimposition of the sinus floor across the roots. In type 4, there is a close proximity of the sinus and roots but clear demarcation between the sinus floor and the roots. Type 5 is an indistinct relationship between the roots and the sinus floor

Full size image

The relationship between the maxillary third molar and the second molar was also classified, according to a modified version of the Archer classification [10]. The relative depth of the maxillary third molar in the bone was classified according to the position of the lowest portion of the crown of the impacted maxillary third molar relative to the occlusal plane and cervical line of the second molar (class A—between the occlusal plane and the cervical line, class B—between the cervical line of the second molar and the middle third of its root, class C—at or above the apical third of the root of the second molar) (Fig. 2). The angulation of the maxillary third molar to the long axis of the second molar (vertical, mesioangular, distoangular, or other) was also classified. Procedure-related factors, including the performance of an incision/bone removal, demographic factors (patient age/gender), and other factors (preoperative complaints, eruption status, and surgeon experience levels), were analyzed and compared. The surgical treatments performed for correction of an oroantral perforation were also recorded.

Fig. 2
figure 2

The relationship between the maxillary third molar and the second molar was also classified, according the Archer classification. Class A: occlusal plane of third molar is between occlusal plane and cervical line of second molar, class B: occlusal plane of third molar is between cervical line and middle third of second molar roots, and class C: occlusal plane of third molar is within the apical third of the second molar roots

Full size image

Data were introduced into a multiple logistic regression model, where the patients were divided according to their RS classification (type 3 vs. other types) and the angulation of the maxillary third molar to the long axis of the second molar (mesioangular vs. other angulation).

Statistical analysis

SPSS 15.0 (SPSS, Chicago, IL) and Stat-View-J-4.5 software (Abacus Concepts, Berkeley, CA, USA) were used for the statistical analyses. The association of each variable with the oroantral perforation was analyzed by the nonparametric Mann-Whitney U test for ordinal variables, and either Fisher’s exact test or chi-squared test for categorical variables. Probabilities of less than 0.05 were accepted as significant. All of the variables associated with oroantral perforation were introduced into a multiple logistic regression model. Forward stepwise algorithms were used, with the rejection of those variables that did not fit the model significantly. Odds ratio (OR) and 95 % confidence intervals (CIs) were also calculated.

Results

In total, 741 teeth from 315 male patients and 426 female patients were included in this study. Patient ages ranged from 16 to 80 years (mean age 32.7 ± 11.9 years). Of these teeth, 132 (17.8 %) were extracted under general anesthesia and 609 (82.2 %) were extracted under local anesthesia. None were extracted under conscious sedation, which is not available at our institution.

Demographic factors

The overall incidence of oroantral perforation was 3.1 % (23 of 741 teeth). There was no significant difference in the mean age of patients experiencing oroantral perforation (34.0 ± 7.7 years, range 21–54 years) relative to those that did not (32.7 ± 12.0 years, range 16–80 years), or in their gender (Table 1). Most oroantral perforations occurred in patients with no preoperative complaints, but the relative prevalence of oroantral perforation was significantly higher for patients with pericoronitis or sinusitis than it was for those with caries. Univariate analysis showed that full impaction of the tooth was the factor most significantly associated with oroantral perforation (p < 0.001) (Table 1). However, these demographic factors were not indicated to have a statistically significant effect according to multivariate analysis.

Table. 1 Characteristics and incidence rates of radiographic signs and their relationship to oroantral perforation after maxillary third molar extraction
Full size table

Radiographic factors (panoramic radiographs)

The most common depth and angulation was class A (582 teeth) and vertical (563 teeth), respectively. Univariate analysis showed the rate of oroantral perforation to be significantly higher for class B and/or mesioangular teeth (p < 0.001) (Table 1). However, class B depth classification was not a significant factor according to a multivariate analysis.

The most and least common RS classifications were type 3 (315 teeth, 42.5 %) and type 5 (two teeth, 0.3 %), respectively. The ratio of oroantral perforation was significantly higher in type 3 patients than in those with other classifications (p < 0.001) (Table 1).

Procedure-related factors

Table 1 summarizes the data relating to the incidence of oroantral perforation relative to various procedural factors. There were no significant differences in the method of anesthesia received by patients with and without oroantral perforation. The incidence of oroantral perforation was higher for surgeons with 1–4 years of experience, but not significantly different. Univariate analysis suggested that the incidence of oroantral perforation was significantly higher in patients that underwent incision or bone removal during extraction than in those that did not (p < 0.001), and in those suffering a fracture of the maxillary tuberosity during extraction than those whose tuberosity was intact (p = 0.009). However, neither bone removal nor fracture of the maxillary tuberosity was found to be significantly associated with oroantral perforation in the multivariate analysis. Other complications during extraction (e.g., root fracture, remarkable hemorrhage) were not significant risk factors for oroantral perforation.

In contrast, multivariate analysis showed various signs, both radiographic (type 3 RS classification, mesioangular angulation) and non-radiographic (performance of an incision), to be significant variables in oroantral perforation (Table 2). The discriminant analysis hit ratio (96.9 %) for this study was considered to be excellent.

Table 2 Results of multivariate logistic regression analysis of the risk factors for oroantral perforation
Full size table

Treatments for oroantral perforation

The corrective treatments provided in cases of oroantral perforation are listed in Table 3. The mean period until closure of the oroantral perforation (as assessed by pressure test (nose blowing) and careful probing of the fistula) was 21.1 ± 30.2 days (range 1–120 days). The most common treatment provided was simply to place a suture (seven teeth, 30.4 %). More complex treatments were applied in some cases, including the creation of a buccal advancement flap, the use of oxidized regenerated cellulose, splinting and/or a periodontal pack.

Table 3 Relative frequency of use and efficacy of various corrective treatments for oroantral perforation
Full size table

Discussion

Oroantral perforation sometimes occurs after the extraction of the maxillary first and second molars [13–16], but is also a common complication during the extraction of the maxillary third molar [2]. Rothamel et al. reported that the incidence of oroantral perforation was 13 % after maxillary third molar removal [2]. Bodner et al. reported that it was 5 % after maxillary premolar and molar removal [17]. In this study, the incidence of oroantral perforation was 3.1 %, which is lower than in other studies [2, 17]. This may be for demographic reasons: tooth root length in most Japanese patients is shorter than that of Caucasian subjects [18].

Oroantral perforation is a well-known complication after the removal of teeth that are in close proximity to the sinus floor and can be a cause of significant morbidity to the patient. Therefore, the prediction of oroantral perforation before surgical treatment is advantageous for oral surgeons and patients alike. A preoperative radiological assessment should therefore be performed to identify the proximity of the maxillary third molar to the sinus floor. However, the relationship between the radiographic findings and oroantral perforation is controversial [2, 9, 11] and clinicians are therefore unable to precisely inform their patients of the risks. In this study, we investigated the multivariate relationships among risk factors for oroantral perforation during the extraction of the maxillary third molar. The aim of the study was to identify parameters that could help to predict the likelihood of this complication occurring.

The incidence of oroantral perforation in patients with preoperative complaints such as pericoronitis or sinusitis was found to be significantly increased compared with those with caries. A fully impacted tooth was found by a univariate analysis to pose the most significant risk of oroantral perforation (Table 1), which is logical considering that most caries occurs in erupted or partially impacted teeth, whereas fully impacted teeth are much closer in proximity to the sinus floor. However, multivariate analysis indicated that these demographic factors did not have a statistically significant impact on the incidence of oroantral perforation.

The results of the univariate analysis determined that bone removal and the fracture of the maxillary tuberosity were associated with significantly higher risks of oroantral perforation. The reason may be that these procedures cause the rupture of the sinus membrane. However, multivariate analysis indicated that neither of these factors had a statistically significant impact on the risk of oroantral perforation. Conversely, the performance of an incision was associated with a significantly higher risk of oroantral perforation by both univariate and multivariate analyses. In some cases of oroantral perforation, there were mesioangular or vertical and partially erupted teeth that required incision but not bone removal. Therefore, the multivariate analysis might indicate that it was the incision rather than the bone removal that had the statistically significant impact on the risk of oroantral perforation, even though the procedural aspects of the surgery (i.e., incision) do not directly cause the rupture of sinus membrane. However, with regard to predicting the factors involved with oroantral perforation, the radiographic factors may be more important than these procedural factors. In this study, mesioangular angulation and a type 3 RS classification were found by multivariate analysis to be highly significant risk factors for oroantral perforation, and thus represent the best indicators of the likelihood of this complication occurring (Fig. 3).

Fig. 3
figure 3

Panoramic radiograph demonstrating mesioangular angulation and type 3 RS classification. These aspects serve as the best indicators of a high risk of oroantral perforation

Full size image

Lim et al. also reported the relationship between oroantral perforation and the long-axis pattern of impaction. The relative depth of the maxillary third molar to the second molar in the bone according to the Archer classification was associated with significantly higher risk of oroantral perforation in univariate, but not multivariate analysis. The reason may be that the Archer classification unintentionally demonstrated the relationship of the maxillary third molar to the inferior border of the maxillary sinus, rather than to the second molar as intended. In this study, we first demonstrated the relationship between the position of the roots relative to the maxillary sinus (RS classification: types 1–5) and oroantral perforation. Therefore, this RS classification may offer a new predictive parameter for oroantral perforation. However, a panoramic radiograph is a two-dimensional image and may produce misleading data regarding the actual proximity of the maxillary third molar to the sinus floor. A three-dimensional image would be more suitable, so we conclude that radiographic findings alone are not a reliable method for predicting the possibility of an oroantral perforation and should be supported by additional computed tomography (CT) to clarify the three-dimensional relationship between the two structures. The validity of this approach will be investigated in future studies. However, the high cost and radiation exposure of CT scans may justify the use of panoramic radiography in the initial planning of surgery for maxillary third molar removal.

In 16 cases (69.6 %) of oroantral perforation, the defect was closed using a simple suture without extending a flap and with/without additional materials such as hemostatic agents, a periodontal pack or a covered splint. Most of these cases involved fully impacted teeth and the extraction socket could be covered without extending the flap. Von Wowern et al. showed that a primary simple suture resulted in a relatively high number of treatment failures [13]. Therefore, Visscher et al. suggested that a buccal advancement flap or a palatal rotational flap were preferable, especially for larger perforations [19]. Some investigators recommend the buccal fat pad as the first-line option for the closure of larger oroantral perforations [20, 21]. In the present study, we found that the use of a splint requires a longer period to produce effective results compared with other treatment modalities. We aim in future work to compare these different treatment options/outcomes for oroantral perforation in a larger cohort.

In conclusion, we successfully demonstrated, using multivariate analysis, that an increased incidence of oroantral perforation was significantly associated with various risk factors, notably the performance of an incision, mesioangular tooth angulation, and a type 3 RS classification. The preoperative identification of risk factors associated with postoperative oroantral perforation would enhance the treatment outcome in patients. Clinicians should give additional consideration to these risk factors prior to performing surgical operations to remove the maxillary third molar.