O R I G I N A L A R T I C L E
Evaluation of bifid mandibular canals with cone-beam computed
tomography in a Turkish adult population: a retrospective study
Kaan Orhan•Sec¸il Aksoy•Burak Bilecenoglu•Bayram Ufuk Sakul•Candan Semra Paksoy
Received: 19 October 2010 / Accepted: 3 December 2010 / Published online: 16 December 2010 Ó Springer-Verlag 2010
Abstract
Purpose Knowing the anatomic location and also varia-tions of the mandibular canal is especially important for surgical procedures on mandible such as dental implant surgery, impacted molar extraction and sagittal split ramus osteotomy. The purpose of this study was to clarify the incidence and location of bifid mandibular canals in an adult Turkish population to avoid complications during surgical procedures.
Methods A retrospective study using cone beam CT images was performed to evaluate bifid mandibular canal in mandible of 242 patients. Both right and left sides were studied (n = 484). Axial, sagittal, cross-sectional and panoramic images were evaluated, and three-dimensional (3D) images were also reconstructed and evaluated, as necessary. The course and length of bifid mandibular canals and the superior and inferior angles between canals were measured.
Results Bifid mandibular canals were observed in 225 (46.5%) of 484 sides examined. The most frequently encountered type of bifid canal was the forward canal (29.80%), followed by the retromolar (28.10%) the buc-colingual (14.50%) and the dental canal type (8.30%). Mean lengths of bifid canals were 13.6 mm in the right side and 14.1 mm in the left side. Mean superior angles were 139° on the right and 141° on the left side, whereas mean inferior angles were 38° on the right side and 32° on the left side. No statistically significant differences were found in the lengths or angles between the right and left sides and also for gender (p \ 0.05).
Conclusions This study, which utilized CBCT images, uncovered a higher prevalence of bifid mandibular canals than what has been reported in previous studies using conventional radiography techniques.
Keywords Cone-beam CT Anatomy Bifid mandibular canal Implant surgery Third molar extraction
Introduction
The mandibular canal is a conduit for the inferior alveolar artery, the inferior alveolar vein and the inferior alveolar nerve, a branch of the third division of the trigeminal nerve that runs through the inferior mandible from the mandib-ular foramen to the mental foramen. Dental and incisive branches of the inferior alveolar nerve exit the canal to supply teeth and adjacent structures, while a terminal branch exits the canal at the mental foramen to become the mental nerve [21]. Radiographically, the mandibular canal appears as a dark, linear shadow with thin, radiopaque superior and inferior borders cast by the lamella of bone that bounds the canal [22].
K. Orhan (&) C. S. Paksoy
Department of Oral, Teeth and Jaw Radiology, Faculty of Dentistry, Ankara University, 06500 Besevler, Ankara, Turkey e-mail: call53@yahoo.com K. Orhan S. Aksoy
Department of Oral, Teeth and Jaw Radiology, Faculty of Dentistry, Near East University, Mersin 10, Turkey
B. Bilecenoglu
Department of Anatomy, Faculty of Medicine, Ufuk University, Ankara, Turkey
B. U. Sakul
Department of Anatomy, Faculty of Dentistry, Ankara University, 06500 Besevler, Ankara, Turkey DOI 10.1007/s00276-010-0761-y
Knowledge of the location as well as the configuration of the mandibular canal is important for surgical proce-dures that involve the mandible [14]. Although the canal has generally been known to be a single, structure, since 1973, variations such as bifid and trifid canals have been reported [8,11,17,21].
The term bifid is derived from the Latin word meaning cleft into two parts or branches [20]. Chavez et al. [3] has suggested that three distinct inferior dental nerves inner-vating three groups of mandibular teeth—incisors, primary and permanent molars—are fused together during embry-onic development to form a single nerve, with bifid and trifid mandibular canals occuring as a result of incomplete fusion of these three nerves.
Dental panoramic radiography, computed tomography (CT) and cone-beam computed tomography (CBCT) have all been used to identify the prevalence of bifid mandibular canals. Studies [5, 11,17,20] in which dental panoramic radiographs have been used to identify bifid mandibular canals have reported incidences rates under 1.0%. The 2D nature of panoramic radiography may result in the appear-ance of thin, cortical outlines in images of the mandible that may simulate the presence of a bifid canal. These outlines may occur as a result of the imprint of the mylohyoid nerve on the medial mandibular surface at the point where it splits from the inferior alveolar nerve and travels to the floor of the mouth [1,20]. False images may also be produced by radiological osteocondensation caused by the insertion of the mylohyoid muscle into the internal mandibular surface [1]. Moreover, ghost shadows produced by the opposing side of the mandible, the pharyngeal airway, the soft palate and the uvula may hamper the localization of the mandib-ular canal using panoramic radiographs [22].
Some authors [1,20] found characteristic radiographic features that can aid in identifying suspected bifid or trifid mandibular canals from dental panoramic radiographs. Auluck et al. [1] suggested that clinicians remains to be aware of the possible existence of a bifid canal in cases where a panoramic radiograph shows the cortical outlines of different canals joining together to form a triangular island of bone, with the vertex of the triangle representing the root of separation of the canals. However, buccally and lingually bifurcated canals cannot be identified using pan-oramic radiographs [10]. According to Naitoh et al. [14], dental panoramic radiographs failed to identify three out of five bifid canals that were identified using multislice CT images. Moreover, Rouas et al. [18] also noted that dental panoramic radiography has limitations with regard to the diagnosis of bifid mandibular canals and recommended CBCT as an excellent, low-cost tool for the evaluation of these anatomical structures with only slightly more radiation than panoramic radiography and far less than a CT-scan.
A review of literature found no studies examining the presence of bifid mandibular canals in a Turkish popula-tion. Therefore, this study aimed to identify the incidence and locations of bifid mandibular canals in a population of Turkish adults using CBCT.
Materials and methods
The study population comprises 242 subjects [133 (55%) female, 109 (45%) male] who had undergone CBCT imaging for dental implant surgery, Le Fort I osteotomy, impacted third molar surgery or orthodontic treatment in Near East University, Faculty of Dentistry from Januray 2008 to 2010. The mean age of subjects was 36.7 years (age range 17–83 years). The study’s protocol was carried out according to the principles described in the Declaration of Helsinki, including all amendments and revisions. Only the investigators had access to the collected data. The institu-tional review board of the faculty reviewed and approved informed consent forms. There was no preference about gender regarding sample choice, however, only Turkish patients were included in the study. Only high quality scans were included. Images of low quality, such as scattering or insufficient accuracy of bony borders were excluded. Imaging using CBCT
CBCT scans were obtained using a Newtom 3G (Quantita-tive Radiology s.r.l., Verona, Italy). Despite recent studies indicating that small variations in head position do not influence accuracy of measurements from 3D CBCT [6], every CBCT scan which were obtained in our clinics, stan-dardize according to a strict, scanning protocol. Patients were placed in a horizontal position and stabilized with custom made head bands and chin support and monitored to ensure that they remained motionless throughout the dura-tion of the scan (36 s). All images were recorded at 120 kVP and 3–5 mA using a 9-inch field of view, an axial slice thickness of 0.3 mm and isotropic voxels. X-ray parameters of kV and mA are automatically determined from scout views by the NewTom 3G. Depending on the size of the patient and the extent of beam attenuation a variation in exposure of up to 40% was possible. All constructions and measurements were performed on a 21.3 inch flat panel color active matrix TFT medical display (Nio Color 3MP, Barco, Belgium) with a resolution of 2048 9 1536 at 76 Hz and 0.2115 mm dot pitch operated at 10 bit.
Image evaluation
All CBCT images were evaluated retrospectively by a sin-gle oral and maxillofacial radiologist (KO). Axial, sagittal,
cross-sectional and panoramic images were reconstructed for all semi-mandibles, and 3D reconstructions were used as necessary (Fig.1). The course and length of bifid mandib-ular canals were measured either in sagittal or panoramic reconstructed CBCT images using the CBCT system’s own software which allows the observer to measure straight or curved structures (like bifid mandibular canal). The length of bifid mandibular canal was measured to be the starting point of separation from the main canal to the tip point that resides in the mandible. The superior and inferior angles between canals were also measured using the CBCT sys-tem’s own software again in sagittal or panoramic recon-structed images. The superior angle was considered to be the angle between the main canal and superior wall while the inferior angle was the angle between the main canal and inferior wall of bifid mandibular canal (Fig.2).
Bifid mandibular canals were classified by location into one of four main groups, namely, forward, retromolar, buccolingual, or dental. Forward canals were further sub-divided into those with and without confluence, dental canals were subdivided into first-, second- and third-molar canals and in addition buccolingual canals were subdivided into either buccal or lingual canals separately following Naitoh et al.’s classification [15] (Table1). All measure-ments were done three times by the same observer and the mean of these measurements were noted for analysis. The observer also performed the study two times with an interval of 2 weeks to detect intra-observer variability.
Fig. 1 Reconstructed a axial, b cross-sectional, c sagittal, d panoramic CBCT images used for detection and measurement of the bifid mandibular canals (arrows)
Fig. 2 aSagittal images showing the bifid mandibular canal with and without length measurement of the canal in the same patient. Note that software allows observer to measure the canal in curved fashion. b Sagittal image also showing the measurement of superior and inferior angle of the bifid mandibular canal. Note that superior angle is the angle between the main canal and superior wall while the inferior angle was the angle between the main canal and inferior wall of bifid mandibular canal
Statistical methods
Statistical analyses were carried out using SPSS 12.0.1 (SPSS, Chicago, IL, USA) software program. To assess intra-observer reliability, Wilcoxon matched-pairs signed-ranks test was used for repeat measurements of the observer. Pearson Chi square and Student t test was per-formed for statistical analysis among gender, localization and measurements (p \ 0.05).
Results
Bifid mandibular canals were identified in 225 out of 484 sides (46.5%) and in 161 out of 242 patients (66.5%). Bifid mandibular canals were observed in 88 females (54.7%) and 73 males (45.3%). The incidences of different types of bifid mandibular canals are given in Table1. The most frequently observed type of bifid canal was the forward canal [n = 86, 46 right sides (20.4%) and 40 left sides (17.7%) (Fig.3)], followed by the retromolar canal [n = 78, 45 right sides (20%) and 33 left sides (14.6%) (Fig.4a)], the buccolingual canal [n = 40 sides; 20 right sides (8.8%) and 20 left sides (8.8%) (Fig.4b, c)] and the dental canal [n = 21; 10 right sides (4.4%) and 11 left sides (4.8%) (Fig.4d)]. Of the 86 forward canals identified, 27 (5.58%) occurred with confluence and 59 (12.18%) without confluence. Of the 21 dental canals identified, 8 (1.68%) extended to the root apex of the first molar, 1 (0.2%) to the second molar and 12 (2.43%) to the third molar. Of all 40 buccolingual canals, 23 (4.71%) identified as buccal and 20 (3.48%) as lingual canal (Table1). The retromolar canals ended in the retromandibular region behind the teeth (if exist 3rd molar, if not 2nd molar). All
dental canals were inserted into the teeth from the apex of the root. Forward canals with confluence were inserted again to the main canal. Forward canals without confluence
Table 1 Prevalence and rate of bifid mandibular canal according to gender, type and localization Classification Of all patients (%) n = 242 Of all sides (%) n = 484
Left side Right side Male n (%) Female n (%) Total n (%) Male n (%) Female n (%) Total n (%) Type 1 retromolar canal 23.10 16.10 13 (5.7) 20 (8.8) 33 (14.6) 22 (9.7) 23 (10.2) 45 (20) Type 2 dental canal type 6.60 4.30 6 (2.6) 5 (2.2) 11 (4.8) 5 (2.2) 5 (2.2) 10 (4.4)
First molar 2.53 1.63 2 (0.8) 2 (0.8) 4 (1.8) 1 (0.4) 3 (1.3) 4 (1.7) Second molar 0.31 0.2 1 (0.4) 0 (0) 1 (0.4) 0 (0) 0 (0) 0 (0) Third molar 3.80 2.43 3 (1.3) 3 (1.3) 6 (2.6) 4 (1.7) 2 (0.8) 6 (2.7) Type 3 forward canal 26.90 17.80 17 (7.5) 23 (10.2) 40 (17.7) 20 (8.8) 26 (11.5) 46 (20.4)
With confluence 8.43 5.58 4 (1.7) 8 (3.6) 12 (5.3) 6 (2.6) 9 (4.0) 15 (6.7) Without confluence 18.40 12.18 13 (5.7) 15 (6.7) 28 (12.4) 14 (6.2) 17 (7.5) 31 (13.8) Type 4 buccolingual canal 9.90 8.20 10 (4.4) 10 (4.4) 20 (8.8) 11 (4.8) 9 (4.0) 20 (8.8)
Buccal canal 5.68 4.71 5 (2.2) 5 (2.2) 10 (4.4) 7 (3.1) 6 (2.7) 13 (5.7) Lingual canal 4.20 3.48 5 (2.2) 5 (2.2) 10 (4.4) 4 (1.8) 3 (1.3) 7 (3.1)
Fig. 3 a Forward canal without confluence (arrows). Note that the forward canal ending with an accessory mental foramen (arrow head). b 3D image of the same patient showing clearly the accessory mental foramen (arrow head). c Forward canal with confluence which bifurcated from the mandibular canal and then joined up with the main canal
were extended to the premolar region and located in buccal plate of the mandible. Sometimes these canals were opened to an accessory mental foramnen in the premolar region (Fig.3a, b). All Buccal and linguals canals were located in the ramus of the mandible (Fig.4b, c). Repeated evaluation and measurement of CBCTs indicated no significant intra-observer difference (p [ 0.05). Intra-intra-observer consistency was rated at 95% between two measurements.
The mean length of bifid canals located in right-side mandibles was 13.6 mm, as compared to 14.1 mm for those in left-side mandibles. When looked by type, the mean length of bifid retromolar canals was 13.5 mm (right side 13.4 mm; left side 13.6 mm), as compared to 8.3 mm for dental canals (right side 8.1 mm; left side 8.4 mm),
20.1 mm for forward canals (right side 19.3 mm; left side 21 mm) and 3.8 mm for buccolingual canals (right side 3.4 mm, left side 4.1 mm). Mean superior angles were 139° on the right side and 141° on the left side, whereas mean inferior angles were 38° on the right side and 32° on the left side. When looked at by side and type, mean superior angles of right-side bifid retromolar, dental and forward canals were 128°, 143.4° and 149.4°, respectively, whereas mean inferior angles of of right-side bifid retro-molar, dental and forward canals were 48.2°, 35° and 29.5°, respectively. Mean superior angles of left-side bifid retromolar, dental and forward canals were 133.9°, 151.5° and 145.2°, respectively, whereas mean inferior angles of left-side bifid retromolar, dental and forward canals were 40.4°, 24.7° and 29°, respectively. No statistically signifi-cant differences were found in either lengths or angles between the right and left sides and also for gender (p \ 0.05). Figure 5 shows the photographs of a surgical third molar extraction of a patient with bifid mandibular canal. The nerve bundle clearly appears in the images which sent to histopathology examination and confirmed as to be the nerve tissue.
Fig. 4 a Panoramic reconstructed image showing retromolar canal type which bifurcated from the main canal to retromolar region. bLingual. c Buccal canals bifurcated from the main canal to lingual and buccal side of the mandible. d Dental canal which bifurcate from the mandibular canal and coursed to the apex of the third molar
Fig. 5 Photos showing third molar extraction of a 26-year old female with bifid mandibular canal. Note that arrows showing the nerve bundle (arrows)
Discussion
Although generally considered to be a single structure, since 1973, bifid and trifid variations of the mandibular canal have been detected using panoramic radiography [5,
11, 13, 17, 20], CT imaging [13, 14, 19] and CBCT imaging [9, 10,15, 16, 18,19]. Studies using panoramic radiography have reported bifid mandibular canals at rates ranging from 0.08 to 0.95% [5, 11, 17, 20]; however, identification of bifid canals using panoramic radiography is complicated by ghost shadows created by the opposing semi-mandible, pharyngeal airway, soft palate and uvula [22]. In a cadaver study conducted by Klinge et al. [9], the mandibular canal could be identified in only 63.9% of panoramic radiographs, and Lindh et al. [12] found that the mandibular canal was clearly visible in only 25% of pan-oramic radiographs.
The 2D nature of panoramic radiography may result in the appearance of thin, cortical outlines in images of the mandible that may simulate the presence of a bifid canal. False images may also be produced by radiological os-teocondensation caused by the insertion of the mylohyoid muscle into the internal mandibular surface [1]. Some authors [1, 20] have suggested that certain characteristic radiographic features may aid in identifying suspected bifid or trifid mandibular canals on panoramic radiographs. For example, Auluck et al. [1] has suggested that in cases where a panoramic radiograph shows the cortical outlines of different canals joining together to form a triangular island of bone, the vertex of the triangle may represent the point of separation between canals. The authors recom-mended panoramic radiography over 3D imaging for the diagnosis of bifid mandibular canals due to reduced costs as well as reduced radiation exposure [1]. However, Naitoh et al. [14] found that panoramic radiographs were able to detect the presence of bifid mandibular canals in only two of five semi-mandibles in which bifid canals could be identified using multislice CT images. Rouas et al. [19] have also stated that 3D imaging is required to reveal the anatomical truth, and they recommended CBCT examina-tion as a low-cost method that has an effective radiaexamina-tion dose only slightly higher than panoramic radiography, and less than CT imaging. Other studies [7, 13–16] note that not only does CBCT impart less radiation, it also produces images of higher quality than CT imaging. Futhermore, multi-slice CT images in the retromolar region have been found to be negatively effected by artifacts from metal restorations and crowns [16].
Different researchers [11, 15, 17] have used different classifications in describing bifid mandibular canals. Our study also classified bifid mandibular canals into four main types according to location using CBCT imaging in line with a CBCT study conducted by Naitoh et al. [15] with
emphazing subgroups of that study. Both our study and that of Naitoh et al. found overall similar prevalence rates for bifid mandibular canals (66.5% of the population and 46.5% of sides in our study, as compared to 65 and 43%, respectively, in Naitoh et al.), and both these studies, as well as a study by Bilecenog˘lu and Tuncer [2] conducted with dry mandibles, reported similar prevalence rates for retromolar bifid canals (28.1% of patients and 16.1% of sides in the present study, as compared to 25.4 and 13.5%, respectively, in Naitoh et al. and 25 and 15%, respectively, in Bilecenog˘lu and Tuncer). However, reported prevalence rates differed for other types of bifid mandibular canals. Moreover, in our study, the forward canal was found to be the most commonly occurring type of bifid mandibular canal (29.8%), whereas the least common type was found to be the dental canal (8.3%); Naitoh et al. also found the forward canal to be the most common type of bifid man-dibular canal (44.3%), but their study found the least common type to be the buccolingual canal (1.6%). Although some authors [17, 20] have reported a slightly higher incidence of bifid mandibular canals among women, our study found no significant difference between men and women in the incidence of bifid mandibular canals.
The presence of a bifid mandibular canal has clinical implications that are of particular importance in surgical procedures involving the mandible, such as dental implant treatment, impacted third molar extraction and sagittal split ramus osteotomy. Failure to accurately localize a bifid mandibular canal may result in damage to the canal and other complications such as traumatic neuroma, paresthe-sia, anesthesia and bleeding during surgery [4, 21]. With regards to different types of bifid mandibular canals, a retromolar canal may be particularly at risk of damage during surgery for an impacted third molar due to its location near the third molar. Retromolar canals may also be at risk of damage when harvesting bone blocks, since this region is commonly used as a donor site; therefore, preoperative CBCT imaging may be needed to safely harvest bone blocks from this area. A bifid dental canal, on the other hand, has implications for root canal treatment as well as extraction [15].
Regardless of type, bifid mandibular canals may be associated with increased difficulty in obtaining inferior alveolar nerve block, especially in cases where there are two mental foramina. Whereas anesthesia of the soft tissue around the injection site, but not of the ipsilateral lip and chin, may be an indication of local anaesthesia failure, anesthesia of the lips and chin, but not the teeth, may indicate the presence of a bifid mandibular canal or other anatomical variation [4,21].
Bifid mandibular canals may also cause pain and dis-comfort in patients with mandibular prostheses due to additional pressure placed on the neurovascular bundle [4,
21]. Also, in cases of trauma, all mandibular fractures should be handled with care to ensure precise alignment of the neurovascular bundle to avoid impingement when the fracture is healed, and alignment of fragments becomes much more difficult in cases where a second neurovascular bundle is located in a different plane [4].
Conclusion
In sum, this study, as well as other studies using CBCT, has found higher rates of bifid mandibular canals than studies using panoramic radiography, suggesting that CBCT is an extremely useful tool for the detection of secondary canals that uses less ionizing radiation than other 3D imaging systems.
Acknowledgments The authors wish to thank Assistant Prof. Dr. Doruk Kocyigit, Oral Maxillofacial Surgery Department, Kırıkkakle University, Kırıkkale, Turkey for providing the surgical images (Fig.5) of a patient with a bifid mandibular canal.
Conflict of interest The authors declare that they have no conflict of interest.
References
1. Auluck A, Pai KM, Shetty C (2005) Pseudo bifid mandibular canal. Dentomaxillofac Radiol 34:387–388
2. Bilecenoglu B, Tuncer N (2006) Clinical and anatomical study of retromolar foramen and canal. J Oral Maxillofac Surg 64:1493–1497
3. Chavez-Lomeli ME, Mansilla Lory J, Pompa JA et al (1996) The human mandibular canal arises from three separate canals innervating different tooth groups. J Dent Res 75:1540–1544 4. Claeys V, Wackens G (2005) Bifid mandibular canal: literature
review and case report. Dentomaxillofac Radiol 34:55–58 5. Grover PS, Lorton L (1983) Bifid mandibular nerve as a possible
cause of inadequate anesthesia in the mandible. J Oral Maxillofac Surg 41:177–179
6. Hassan B, van der Stelt P, Sanderink G (2009) Accuracy of three-dimensional measurements obtained from cone beam computed
tomography surface-rendered images for cephalometric analysis: influence of patient scanning position. Eur J Orthod 31:129–134 7. Kamburoglu Kilic K, Ozen T et al (2009) Measurements of mandibular canal region obtained by cone-beam computed tomography: a cadaveric study. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 107:e34–e42
8. Kiersch TA, Jordan JE (1973) Duplication of the mandibular canal. Oral Surg Oral Med Oral Pathol 35:133–134
9. Klinge B, Petersson A, Maly P (1989) Location of the mandibular canal: comparison of macroscopic findings, conventional radi-ography and computed tomradi-ography. Int J Oral Maxillofac Implants 4:327–332
10. Kuribayashi A, Watanabe H, Imaizumi A et al (2010) Bifid mandibular canals: cone beam computed tomography evaluation. Dentamaxillofac Radiol 39:235–239
11. Langlais RP, Broadus R, Glass BJ (1985) Bifid mandibular canals in panoramic radiographs. J Am Dent Assoc 110:923–926 12. Lindh C, Petersson A, Klinge B (1992) Visualisation of the
mandibular canal by different radiographic techniques. Clin Oral Implants Res 3:90–97
13. Milog˘lu O, Yılmaz AB, Cag˘layan F (2009) Bilateral bifid man-dibular canal: a case report. Med Oral Patol Cir Bucal 14:244–246
14. Naitoh M, Hiraiwa Y, Aimiya H et al (2007) Bifid mandibular canal in Japanese. Implant Dent 16:24–32
15. Naitoh M, Hiraiwa Y, Aimiya H et al (2009) Observation of bifid mandibular canal using cone-beam computerized tomography. Int J Oral Maxillofac Implants 24:155–159
16. Naitoh M, Nakahara K, Suenage Y (2010) Comparison between cone beam and multislice computed tomography depicting mandibular neurovascular canal structures. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 109:25–31
17. Nortje CJ, Farman AG, Grotepass FW (1978) Variation in the normal anatomy of the inferior dental (mandibular) canal: A retrospective study of panoramic radiographs from 3612 routine dental patients. Br J Oral Surg 15:55–63
18. Rouas P, Delbos Y, Nancy J (2006) Pseudo multipl and enlarged mandibular canals: the evidence-based response of come beam computed tomography. Dentomaxillofac Radiol 35:217–218 19. Rouas P, Nancy J, Bar D (2007) Identification of double
man-dibular canals: literature review and three case reports with CT scans and cone beam CT. Dentomaxillofac Radiol 36:34–38 20. Sanchis JM, Pen˜arrocha M, Soler F (2003) Bifid mandibular
canal. J Oral Maxillofac Surg 61:422–424
21. Wadhwani P, Mathur RM, Kohli M et al (2008) Mandibular canal variant: a case report. J Oral Pathol Med 37:122–124
22. White SC, Pharoah MJ (2004) Oral Radiology—principles and interpretation, 5th edn. Mosby, St. Louis
