Study design and sample

A prospective study design was used to address the research question. The study population comprised all patients presenting to the Department of Oral and Maxillofacial Surgery and the Department of Oral and Maxillofacial Radiology, Faculty of Dentistry, Karadeniz Technical University (Trabzon, Turkey) for the evaluation and management of mandibular third molars from January 2013 through May 2014. The criterion for inclusion in the study sample was that the patient had to have one or both impacted third molars with a close relationship between the third molar and the IAC on preoperative PAN images. These patients are considered at high risk for IAN injury. A close relationship was defined as the presence of at least one of six previously described radiographic markers : (1) interruption of the white line of the IAC; (2) darkening of the roots; (3) narrowing of the IAC or roots; (4) dark and bifid roots; (5) deflected roots; (6) diversion of the IAC.

The first sample of patients presenting to the unit were assessed only with PAN images during the preoperative evaluation. The second sample presenting to the unit underwent PAN and additional CBCT imaging. Patients were excluded from the study if they had any pathology (e.g., a tumour or cyst) around the third molar or IAC. Patients with systemic disorders were also excluded, because the neurological and surgical outcomes might be unfairly affected in such patients. Additionally, patients lost to follow-up were excluded. The study plan was approved by the relevant ethics committee. Each patient was informed about the possible risks and benefits of the procedure and the postoperative recovery time, and provided written consent to participate in the study.

Study variables

The primary predictor variable was the imaging method used for preoperative evaluation of the impacted third molar. The imaging method was a binary variable classified as digital PAN (PAN group) or digital PAN and additional CBCT (CBCT group). Digital PAN images were acquired with a panoramic unit (Kodak 9000C; Kodak Dental Systems, Carestream Health, Rochester, NY, USA) and Kodak Dental Imaging Software. Acquisition parameters were 70 kV, 6.3 mA, and a scan time of 14.3 s. Consecutive cross-sectional images were acquired using a CBCT device (Kodak 9500 Cone Beam 3D System; Kodak Dental Systems, Carestream Health, Rochester, NY, USA) and Kodak Dental Imaging Software. Scans were performed with a medium field of view (9 cm × 15 cm), a voxel size of 0.2 mm, and a scan time of 10.8 s. Volumetric data were reconstructed and sectioned on the coronal plane.

The imaging method was number-coded with slips of paper by a surgeon who was not associated with the study. The numbers were chosen by the surgeon. This allowed the subjects to be assigned to groups in a random fashion.

The primary outcome variable was IAN injury, i.e., the presence of sensory disturbances of the lower lip and/or chin. Evidence of IAN injury was assessed at 7 days after surgery. All patients were questioned about any sensory disturbance of the IAN and lingual nerve using a visual analogue scale (VAS). However, detailed objective assessment of lingual nerve injury was not done, as the primary focus of the study was to investigate IAN injury. The status of the sensory disturbance of the IAN was then confirmed by two-point discrimination test, pinprick test, and light touch test. The two-point discrimination test measures the minimum distance a patient can discriminate between two separate points. This test was performed using a calliper and the result recorded as a numeric value in millimetres. The pinprick test was performed by touching the skin with a sharp needle to test pain perception (the needle was applied to the skin through the loop using its own weight). For the light touch test, a cotton wisp was used to test tactile stimulation by gently touching the skin and the detection threshold of the patient was evaluated. In cases where there was reduced sensation, a neurological examination was performed to describe the type of IAN injury. The type of IAN injury was divided into the following subgroups: 1, no sensation (anaesthesia); 2, reduced sensation; 3, hypersensitive or spontaneous pain; this was monitored after 7 days and 1, 3, and 6 months until total recovery. If recovery was not complete after 6 months, the injury was considered permanent.

The secondary outcome variables were the pre-surgical plan and the surgical outcomes including IAN exposure and duration of surgery. The pre-surgical plan for each extraction was prepared based on preoperative radiographic findings and was determined by the surgical technique and the direction of tooth removal in both groups. The surgical technique was selected intuitively after consideration of the clinical and radiographic factors and was categorized using a difficulty rating on a two-point scale: 1, ostectomy; 2, ostectomy and crown/root sectioning (complex extraction). Similarly, the direction of the tooth removal was planned and divided into subgroups as follows: 1, buccal and/or lingual; 2, undeterminable. When the IAC was positioned buccal to the third molar, the surgeon planned to luxate the third molar in a buccal direction, thereby rotating the apex into a lingual direction. For the same reason, when the IAC was positioned lingual to the third molar, it was planned to luxate the tooth into a lingual direction, thereby rotating the roots in the opposite direction to the IAC.

IAN exposure was defined as direct visualization of the IAN after third molar removal. Each extraction site was inspected using direct vision after copiously irrigating the extraction site with normal saline, and IAN exposure was noted (yes/no). The surgical time was defined as the duration of surgery in minutes (from the first incision to insertion of the last suture) and was divided into two categories of <20 min or ≥20 min for one tooth in a particular patient.

Other variables were divided into two groups: demographic and anatomical/radiographic. Demographic variables included sex and age. Anatomical and radiographic variables included the anatomical position of the third molar (Pell and Gregory classification, Winter classification, type of mucosal or bony coverage), PAN findings of the third molar according to the criteria established by Rood and Shehab, side of the third molar extraction (left or right), and number of roots. These variables were recorded based on PAN images in both groups. The spatial relationship between the tooth and the IAN, distance from the IAN to the tooth (0 mm, 0–1 mm, and >1 mm), and cortication status of the IAC acquired from the CBCT scans were also recorded in the CBCT group. The spatial relationship was divided into the following subgroups: 1, buccal; 2, lingual; 3, in line (beneath the roots); 4, interradicular (between roots). True contact was thought to exist in the absence of cortical bone between the third molar and IAC.

Summary of operative methods

The surgical removal of the third molars was performed under local anaesthesia by the same surgeon who has at least 8 years of experience with the procedure. Along with the careful application of elevators and burs, the operational technique consisted of bone removal, crown and root sectioning (if necessary), and direction of tooth removal (if necessary) by preventing root movements, which put pressure on the IAC. Lingual flaps were not employed in any of the cases.

Data collection methods

The radiographic evaluations and measurements were performed independently by two maxillofacial surgeons and by a maxillofacial radiologist. The three observers had at least 8 years of clinical and radiological experience in maxillofacial structures. Cases in which there was disagreement with regard to the radiographic findings were discussed again and a final decision made to reach a statistical analysis. The neurosensory testing of all patients was performed by the same surgeon. The surgical time of each operation was measured using a digital chronometer by an impartial observer who was not associated with the study. IAN exposure was noted by the surgeon who performed the operations. The pre-surgical plan for each extraction was achieved by consensus of the two clinical observers (two maxillofacial surgeons). All of the data collected were recorded on a spreadsheet.

Due to the nature of an image-based study, it was impossible to blind the surgeons or the participants to their allocation to the CBCT or PAN group. However, the investigator who performed the neurosensory testing was blinded to the patient group allocation (CBCT or PAN).

Data analysis

The statistical analysis was performed using SPSS statistical software Graduate Package 17.0 (SPSS Inc., Chicago, IL, USA). When necessary, the normality of the distribution was tested using the Kolmogorov–Smirnov test. Descriptive statistics were calculated. In the event of IAN injury, comparisons between the groups were tested with Pearson’s test, Fisher’s test, or the χ ² test. Comparisons between the groups (primary predictor) for primary and secondary outcomes were analyzed with the Mann–Whitney U -test. Multivariate logistic regression analysis was performed for the primary outcomes. The confidence level was set at 95.0% for all statistical analyses.