Evaluation of Recently Proposed Scales as Predictors of Mandibular Third Molar Extraction Difficulty

Background: The Prediction of extraction difficulty of impacted mandibular third molar (M3M) is extremely important for both patients and clinicians. Recently, many new difficulty-estimating indices had been proposed, among them are, Zhang et al., Kim et al., Pernambuco, Lainez et al., and Roy et al. indices. This study aimed to evaluate the validity of these new scales as preoperative predictors of the difficulty of surgical removal of impacted M3M. Material and Methods: The five scales under study predicted extraction difficulty of a series of 50-impacted M3M preoperatively, and postoperative difficulty was assessed with Parant scale (PS) and by the time required for surgery (TS). Results: The proposed indices had low to moderate sensitivity (21%-45%, 41%-67%) and variable in their specificity (21%-86%, 36%-85%). Only three out of five evaluated indices have shown a statistically significant correlation with both, the operation time and the surgical technique; namely, Zhang et al., Pernambuco, Lainez et al. indices. Conclusions: Zhang et al., Pernambuco, Lainez et al. indices can be used as


Introduction
Prediction of the degree of impacted mandibular third molar (M3M) extraction difficulty is essential to plan treatment options, and to limit the risk of complications.
Classic difficulty scoring models were based on radiographic variables [7][8][9], while the recent ones had associated additional clinical, non-radiographic variables [10,11]. Pederson scale, among these scales, is widely used as a prediction tool of extraction difficulty of M3M [12]. However, many researchers have questioned its performance [5,12].
Other indices have been proposed for preoperative estimation of difficulty, but they found invalid [4,5,11,12] or of limited clinical use [5,11,[13][14][15]. Due to these drawbacks, there is a continuous need for developing an index that can precisely determine the extraction difficulty of M3M. In the last years, many new difficultyestimating indices had been proposed; they are Zhang [16][17][18][19][20]. Some of them based on radiographical variables only [17,19], while others involve additional clinical and demographic variables [16,18,20]. Authors of these indices claimed that they are valid and reliable prediction tools. The aim of this study was to evaluate the prediction accuracy of these new scales.

Material and methods
Surgical extraction of fifty M3M were evaluated for patients who presented to the private clinic of authors located in Mosul city, Iraq from June to December 2021. All patients signed informed consent and the study approved by the local ethics committee. All operations executed according to standard protocols under local anesthesia by two surgeons (A.A., G.M.) who had eleven and fifteen years of experience in oral surgery. Preoperatively, the authors of this study predicted the difficulty of extraction from panoramic radiographs according to five indices; Zhang et al., Kim [16][17][18][19][20] (Figure 1-5).
Any disagreement among authors solved by consensus. Two outcome variables were considered to assess extraction difficulty: the surgical technique using Parant scale (PS), and the time required for surgery (TS) (from start of incision to final suture) ( Table  1).
Statistical analysis using descriptive statistics of IBM SPSS Statistics 23, sensitivity, specificity and likelihood ratios were calculated considering the PS and ST as a reference. In addition, the correlation between the operative time and the difficulty of operation as proposed by all scales were also assessed by analysis of variance test. A probability value (P) of less than 0.05 was considered significant.

Results
Fifty patients (26 female and 24 male) between 17 and 42 years of age (mean age of 26.9 ± 6.35 years) were analyzed. Right mandibular (n=23) and left (n=27) wisdom teeth were extracted. Table 2 illustrate the difficulty of fifty extraction as classified by preoperative prediction scales and postoperative PS and TS.

Classification of difficulty
Actions required for extraction      Table 3). Comparing to other indices, Pernambuco index is the most sensitive one (66.7%) with a significant correlation with TS (P=0.001) but with limited specificity (46.9%) ( Table  4). The likelihood ratios were not significant (Table 3 and 4). A significant correlation (P=0.12, 0.000) exist between index prediction with both; the PS and TS (Table  5).

Discussion
The classic Pell and Gregory, and Winter classifications of impacted M3M based on their relative occlusal depth, the relation to the mandibular ramus and the tooth angulation in respect to the long axis of the adjacent second molar. Over decades, many modifications of these scales have been proposed to improve the prediction of extraction difficulty [19].
Many radiographical and clinical parameters should be considered before surgery for correct evaluation and prediction of M3M extraction difficulty. They help in drawing of correct treatment plan to improves patient's outcomes [12,21].
Different scales were proposed as predictors of M3M extraction difficulty; however, some of these scales have drawbacks. Bali et al.  [22] in their meta-analysis study concluded that Pederson scale is not valid index in M3M. MRACBS scale [15] need to cone beam computed tomography in classification of wisdom teeth, giving a limited practical implication. WHARFE index [23] and Sammartino Index [24] is rarely used in practice owing to its complexity [12]. Koerner index [9] is similar to Pederson index in that it measures the same radiographical parameters. However, these indices have not been validated [18].
Yuasa index [5] and Kharma scale [14] consider not only the relative depth and relation with the mandibular ramus as Pederson index, but also the root width and form. Gbotolorun et al. [11] proposed an index depends on four variables: two clinical and two radiographic. It is differed from Pederson index in that it does not consider neither the tooth relation to mandibular ramus nor the tooth angulation.
In the present study, we consider PS and ST to determine extraction difficulty like many previous studies [Zhang et al., Pernambuco, Roy et al.], they considered as a standard protocol to accurately assess surgical difficulty [16,18,20]. The proposed indices had low to moderate sensitivity (21%-45%, 41%-67%) and variable in their specificity (21%-86%, 36%-85%), and may be related to some limitations.
For instance, In Roy index, pericoronal or periradicular radiolucency, the number roots, root proximity to adjacent second molar or inferior dental canal (IDC) were not considered during difficulty assessment. In addition to absence of important clinical variables such as body mass index (BMI) and age that could influence the level of difficulty in M3M surgery as reported by other researchers [11].
Both, Kim et al. [17] and Lainez et al. [19] index depended only on radiographical parameters and did not consider any clinical factor like BMI, tongue size, cheek flexibility, and mouth opening. These factors also not addressed in Pernambuco index along with pericoronal or periradicular radiolucency or root relation to IDC.
Again, these factors also not considered in Zhang et al. index [16] in addition to lack of detailed description of relative tooth angulation, depth and relation to mandibular ramus.
Three of the five difficulty indices evaluated in the present study have shown a statistically significant correlation with the operation time and the surgical technique; namely, Zhang et al., Pernambuco, Lainez et al. indices [16,18,19]. Although came from different populations, studies that reported these three indices were almost similar to our study in regard to patient's average age, surgical technique, and patterns of the impacted teeth. This coincidence may explain the high predictability of these indices for the level of surgical difficulty assessed in the present study. A common radiographic factor that was considered in calculating each of the three indices was the number and morphology of roots of the impacted M3M. The difficulty of extraction of these teeth is directly proportional to the number and complexity of their roots. Adding this factor to Pederson's scale would expectedly increase its reliability in assessing the surgical difficulty of these procedures. The coronal width of M3M is another local anatomic parameter which was measured by Lainez et al. study and might have improved the predictability of their scale [19]. The wider the crown of the tooth, the longer time would be required to Evidence has shown that both clinical and demographic factors should be considered in assessing the difficulty of impacted M3M surgery [11]. Of these factors, patient's age was considered in both Pernambuco and Zhang indices. It is obvious that surgical difficulty is increased in older patients due to changes in the dental and tooth investing tissues. Incomplete root formation, more elastic bone, and pericoronal follicle space seen in patients younger than 25 years of age are usually associated with less difficult surgery.
Two indices, Kim's et al. [17] and Roy's et al. [20] scales, correlated weakly with the time of operation and Parant's scale in the present study. In both of these proposed indices, patient's age and number of roots of impacted M3M were neglected in the final scoring. Kim et al. focused only on local radiographic parameters and depended on a modified Pederson scale with 4 instead of 3 categories of difficulty. Unlike our study, the most common pattern of impaction treated by these authors was the horizontal rather than mesioangular impaction, indicating the involvement of more difficult cases in their study [17]. The authors to their working in a tertiary medical institution linked this finding and that relatively simple cases were presumably referred to be treated in private clinics elsewhere.
In their index, Roy et al. also did not take into account demographic features of the patients. Roy's index consisted of many parameters and a maximum score of 33 with 3-step difficulty scale [20]. Such a detailed index with exhaustive graduation may not be straightforward for many surgeons to calculate. In addition, gathering of all potential factors in one index without calibration of the significance of each factor might have decreased rather than increased the sensitivity of the created index. The authors did not report the patterns of impaction in their study, and they performed bone removal and tooth division in all cases, which makes Roy's index more suitably applicable for difficult than for easy procedures. According to our best knowledge, the validity of above indices as prediction tools were not tested before except for Pernambuco index which proven to be a reliable index with high sensitivity (87.9%), specificity (93.1%). However, these results were not corresponded to ours.