Volume 77 • Number 8

Clinical Reliability of the ‘‘Furcation Arrow’’ as a Diagnostic Marker David E. Deas,* Alan J. Moritz,† Brian L. Mealey,‡ Howard T. McDonnell,† and Charles A. Powell†

Background: The radiographic entity known as the ‘‘furcation arrow’’ has long been used in practice even though little is known about its usefulness as a clinical indicator. The definitive study of the furcation arrow suggests that its presence on a radiograph reliably predicts furcation invasion, but this has not been confirmed in an in vivo investigation. The purpose of this study was to evaluate the furcation arrow in a clinical setting, testing the assertion that the furcation arrow image is an accurate predictor of furcation invasion. Specifically, we sought to determine the following. First, what is the prevalence of furcation arrow images in the radiographs of maxillary molars with periodontitis? Second, what is the interexaminer agreement on what constitutes a furcation arrow? Third, how does the presence or absence of a furcation arrow correlate with the true clinical status of the furcation? Fourth, what is the sensitivity and specificity of the furcation arrow as a diagnostic indicator? Methods: Eighty-nine patients requiring surgical treatment of periodontitis in the maxillary molar regions were included in this study. Before surgery, one of five calibrated examiners viewed periapical and bitewing radiographs of the surgical site and recorded the presence or absence of a furcation arrow at each proximal furcation. Before administering anesthesia, the same examiner recorded a Hamp index value of each proximal furcation, with a second Hamp index taken after flap reflection and debridement. After surgery, each of the four remaining examiners independently reviewed the radiographs for furcation arrows. Descriptive statistical analysis was performed to correlate the appearance of the furcation arrow image to the actual degree of furcation invasion as determined by the intrasurgical Hamp index. Results: A total of 164 maxillary molars were examined, providing 328 interproximal furcations; 111 (33.8%) furcations were determined at surgical debridement to have a furcation invasion of Hamp degree 1 or greater. Of the 111 furcation invasions, 43 (38.7%) were predicted by a furcation arrow image seen by at least three of the five examiners. When comparing the appearance of the radiographic image to the extent of furcation invasion, 20 of 64 (31.3%) Hamp 1 furcation invasions and 23 of 47 (48.9%) Hamp 2 and 3 furcation invasions were predicted by furcation arrows observed by at least three of five examiners. The multirater k statistic for interexaminer agreement on the presence or absence of the image was 0.489. The sensitivity of the furcation arrow image as a diagnostic marker was 38.7%, and the specificity was 92.2%; the positive predictive value of the image was 71.7%, and the negative predictive value was 74.6%. Of the 324 furcations used to compare clinical indices, the agreement of preanesthesia and postdebridement Hamp indices was 0% for degree 3, 83.7% for degree 2, and 98.4% for degree 1 furcation lesions. Conclusions: These data suggest that the furcation arrow has limited usefulness as a diagnostic marker of furcation invasion. The image is difficult to interpret and highly subjective and can correctly predict furcation invasions only ;70% of the time when present on the radiograph. In addition, when furcation invasions are truly present, the furcation arrow is seen in <40% of sites. J Periodontol 2006;77:1436-1441. KEY WORDS Furcation defects; radiography; x-ray diagnosis; x-ray image.

* Currently, 48th Dental Squadron, Royal Air Force Lakenheath, U.K.; previously, U.S. Air Force Periodontics Residency, Wilford Hall Medical Center, Lackland Air Force Base, TX. † U.S. Air Force Periodontics Residency, Wilford Hall Medical Center, Lackland Air Force Base, TX. ‡ Currently, Department of Periodontics, University of Texas Health Science Center at San Antonio, TX; previously, Wilford Hall Medical Center.

doi: 10.1902/jop.2006.060034

1436

J Periodontol • August 2006

I

n the treatment of maxillary molar teeth with periodontitis, an accurate diagnostic assessment of furcation invasions is paramount to the ultimate success of therapy. Important topographical information gathered at the periodontal examination may include the degree of furcation involvement, root and furcation morphology, and the nature of any associated osseous defects. Traditionally, radiographic assessment in conjunction with clinical probing using a curved explorer or furcation probe have been the chief diagnostic methods used for detecting and characterizing furcation involvement and, as with all diagnostic tests, are associated with inherent error. Using radiographs to diagnose proximal maxillary molar furcation defects has been the subject of much clinical discussion but few published articles. Prior studies have investigated the relationship between the size of an osseous lesion and radiographic evidence of destruction but did not address the subject of bone loss in furcations.1-3 Rees et al.4 examined the radiographs of various osseous defects in dried skulls and cadavers to determine the predictability of diagnosing alveolar defects on the basis of radiographic appearance. They found that furcation defects on facial and lingual surfaces of molar teeth could be identified with a high degree of accuracy on radiographs. They identified three types of radiographic patterns associated with furcation bone loss but did not specifically address the subject of proximal furcation defects in maxillary molars. Additional mention of specific radiographic patterns accompanying furcation invasions was made by Prichard5 when he characterized a ‘‘subtle shadow’’ over the mesial root of maxillary first molars that pointed toward the furcation entrance; however, no further discussion of this finding was made relative to furcation diagnosis. In the most extensive study using radiographs to diagnose maxillary molar furcation invasions, Ross and Thompson6 evaluated 303 maxillary molars in 72 treated periodontal patients. After clinical and radiographic examination, they found that 90% of these teeth had furcation invasions and that the diagnosis for 22% of the teeth was based on the radiographic appearance alone. The authors also described a certain pattern of bone loss between mesio-buccal, distobuccal, and palatal roots as their criteria for radiographic evidence of furcation invasion but did not elaborate further. More recently, to document a specific radiographic entity observed by many clinicians, Hardekopf et al.7 proposed the term ‘‘furcation arrow’’ to describe the small, triangular radiolucent shadow sometimes seen across the mesial or distal roots of maxillary molars. They noted that this shadow was empirically associated with proximal furcation involvement and at-

Deas, Moritz, Mealey, McDonnell, Powell

tempted to assess its clinical value as a diagnostic indicator. Using dry human skulls, the authors graded 162 proximal maxillary first and second molar furcation invasions using the Hamp index8 and then obtained radiographs of these same teeth. The radiographs were projected and examined by six dentists for the presence of a furcation arrow image. The authors reported that the association of the furcation arrow with degree-2 or -3 furcation defects was significant compared to 120 uninvolved furcations used as controls. Within the abstract of the article, the authors further stated, ‘‘because the furcation arrow seldom appears over uninvolved furcations, the appearance of the image indicates that there is proximal bony furcation involvement.’’ Despite this assertion, they were careful to point out that they did not observe the arrow in a large number of furcations that had degree-2 or -3 involvement. Although subsequently described in several review articles concerning furcation management,9-11 the furcation arrow image has apparently not been the subject of direct clinical investigation, and its association with furcation invasions has yet to be evaluated in vivo. The purpose of the present study was to evaluate the furcation arrow in a clinical setting, with emphasis on testing the assertion that the radiographic presence of a furcation arrow reliably identifies a furcation invasion. This study sought to answer the following questions. First, what is the prevalence of furcation arrow images in the radiographs of maxillary molars with periodontitis? Second, what is the interexaminer agreement on what constitutes a furcation arrow within a group of experienced clinicians? Third, how does the presence or absence of the furcation arrow image correlate with the clinical status of the furcation as determined by intrasurgical measurements? Fourth, what is the sensitivity and specificity of the furcation arrow as a diagnostic indicator? MATERIALS AND METHODS The experimental protocol was approved by the Institutional Review Board for human studies, Wilford Hall Medical Center (WHMC), Lackland Air Force Base. This study evaluated 89 consecutive patients referred to the Department of Periodontics at WHMC between February 2004 and June 2005 for treatment of moderate to advanced periodontitis that required surgical access to at least one maxillary posterior sextant. Inclusion in the study required at least one preoperative diagnostic periapical and bitewing radiograph of each maxillary molar in the proposed surgical site. Demographic data were not deemed important for the study and, therefore,were not recorded. Before beginning the study, each of the five investigators, all of whom were board-certified periodontists, 1437

Clinical Reliability of the ‘‘Furcation Arrow’’ as a Diagnostic Marker

participated in a calibration exercise using full-mouth radiographs of five patients being treated for moderate to advanced periodontal disease with known furcation involvement of maxillary molars. The purpose of this exercise was to ensure recognition and standardization of what was to be called a ‘‘furcation arrow’’ in the study. Medical and dental histories were reviewed, and no contraindications to surgery were noted. Each patient received a comprehensive periodontal evaluation and treatment planning session. After their acceptance of the initial treatment plan, each patient was taken through a course of initial therapy that included detailed oral hygiene instructions and subgingival scaling and root planing with anesthesia. Four to 6 weeks after the completion of initial therapy, patients were seen for reevaluation. A surgical treatment plan was proposed if indicated by the clinical parameters and only if the patient was consistently able to maintain a modified O’Leary plaque index12 of at least 80%. Informed consent was obtained from all patients who entered into the study, and each patient was assigned a random number identifier for use on study forms and for data analysis. Surgical procedures were accomplished under local anesthesia with or without intravenous or oral sedation at patient request. Clinicians performing the surgeries included both faculty and residents of the Department of Periodontics at WHMC. Before examining the patient, one of the five investigators (not the surgeon) reviewed the radiographs of the maxillary molars in the surgical site and determined the presence or absence of a furcation arrow image at each proximal furcation. The number of radiographs on which a furcation arrow was visible and the specific films in which a furcation arrow was seen were recorded onto a number-coded study form. Next, before the administration of anesthesia, the same examiner recorded a Hamp index8 value for each proximal furcation using a color-coded probe.§ According to this index, a degree-1 furcation demonstrates horizontal loss of periodontal tissue support <3 mm, a degree-2 furcation demonstrates horizontal loss of periodontal tissue support >3 mm but not encompassing the total width of the furcation area, and a degree-3 furcation demonstrates horizontal through-and-through destruction of the periodontal tissues in the furcation.8 After flap reflection and debridement, the examiner recorded a second Hamp index at each proximal furcation. Once the initial examiner recorded the clinical data, each of the four remaining examiners was asked to independently review the radiographs and determine the presence or absence of a furcation arrow at each proximal site and the specific films on which the image was noted. Radiographs were reviewed on view boxes without magnification in an effort to recreate normal 1438

Volume 77 • Number 8

Table 1.

Furcation Involvement Determined After Surgical Debridement Hamp Index

Furcations (N)

Total Furcations (%)

Degree 0

217

66.2

Degree 1

64

19.5

Degree 2

43

13.1

Degree 3

4

1.2

Total

328

clinical conditions. Each additional examiner was blinded to the findings of the clinical examination and to the radiographic findings of the other examiners. All clinical and radiographic data were recorded on separate, numerically coded data sheets. Descriptive statistical analysis was then performed to correlate the appearance of the furcation arrow image to the actual degree of furcation invasion as determined by the intrasurgical Hamp index recording. RESULTS A total of 164 maxillary molars were examined in the study, providing a total of 328 interproximal furcations for analysis (Table 1). Of the 328 total furcations, 111 (33.8%) were determined after surgical debridement to have a furcation invasion of Hamp degree 1 or greater. Most maxillary molars in the surgical sites (66.4%) had intact furcations. Two teeth did not have furcation assessments recorded before surgery; therefore, the agreement between preanesthetic and postdebridement Hamp index recordings was calculated using 324 sites. The agreement of the preanesthetic and postdebridement assessments of the Hamp index for each furcation ranged from 0% for degree-3 to 98.4% for degree-1 furcation lesions (Table 2). Of the 111 surgically confirmed furcation invasions, 43 (38.7%) were predicted by a radiographic furcation arrow image seen by at least three of the five examiners (Table 3). This value decreased to 24 (21.6%) when the threshold was four of five examiners and 16 (14.4%) when the standard was agreement by all five examiners. There were 64 Hamp degree-1 furcation invasions, of which 20 (31.3%) were determined to have a furcation arrow by at least three of the five examiners. There were 47 Hamp degree-2 or -3 furcation invasions, of which 23 (48.9%) were deemed to have a furcation arrow by at least three examiners. The number of furcation arrow images reported by individual examiners varied from 49 to 80; § Nabers 2N, Hu-Friedy, Chicago, IL.

Deas, Moritz, Mealey, McDonnell, Powell

J Periodontol • August 2006

Table 2.

Agreement Between Preanesthetic and Postdebridement Hamp Index Recordings Made by Clinical Examiner at Surgical Appointment Hamp Index

Preanesthetic Furcation Assessment

Postdebridement Furcation Assessment

Difference Between Assessments (N)

Degree 0

213

216

3

1.4

98.6

Degree 1

62

63

1

1.6

98.4

Degree 2

49

41

8

16.3

83.7

Degree 3

0

4

4

324

324

Total

Table 3.

Agreement Between Examiners on Presence of Furcation Arrow Sites in Agreement With Surgical Findings (N)

Total Furcations in Agreement With Surgical Findings (%)

At least 3

43/111

38.7

At least 4

24/111

21.6

All 5

16/111

14.4

Examiners in Agreement (N)

Table 4.

Two-By-Two Contingency Table for Agreement of Three of Five Examiners: All Furcations Furcation Invasion at Surgery? Yes

No

Totals

Furcation arrow on x-ray? Yes

43

17

60

No

68

200

268

111

217

328

Totals

Sensitivity = 38.7%; specificity = 92.2%; positive predictive value = 71.7%; negative predictive value = 74.6%.

this represents a range of ;15% to 24% of all furcations examined. The multirater k statistic for interexaminer agreement for the presence or absence of the furcation arrow image was 0.489. Although there was a trend for furcation arrows to be seen more often on periapical radiographs than on bitewing films for all examiners, the difference was significant for three of the five examiners (range, P <0.003 to P <0.016).

Difference Between Assessments (%)

100

Agreement Between Assessments (%)

0

The sensitivity and specificity of the furcation arrow image as a clinical diagnostic marker is presented in Table 4. If agreement by at least three out of five examiners is used to confirm the radiographic image, the furcation arrow as a diagnostic marker in this study had a sensitivity of 38.7% and a specificity of 92.2%. Also, based on these data, the positive predictive value of the furcation arrow is 71.7% and the negative predictive value is 74.6%. DISCUSSION The radiographic image known as the furcation arrow is a widely used diagnostic marker that has only occasionally been mentioned in the periodontal literature. Hardekopf et al.7 named this radiographic entity but did not describe it in sufficient detail for clinicians to know exactly when it is present and when it is not. As a result, there still seems to be confusion as to what extent the radiolucent image needs to be present or overlap the furcation area before it can be called a furcation arrow (Figs. 1 and 2). The shape of the roots, superimposition of the palatal root, thickness of the alveolar bone, exposure settings of the x-ray unit, and variations in horizontal angulation of the x-ray tubehead are just a few of the variables that may affect the appearance of the radiographic image.7 In this study, despite the fact that all films were taken and processed with the same equipment and after a calibration exercise, five board-certified periodontists with a combined average of >20 years of clinical experience varied widely in the interpretation of furcation arrow images. This is evidenced by the fact that all five examiners agreed on the presence of the radiographic image at only 16 of the 111 sites in which a furcation lesion was actually present. Even when at least three of the five examiners agreed on the presence of a furcation arrow, this agreement only correctly coincided with 38.7% of actual furcation invasions. Overall, the interexaminer agreement in 1439

Clinical Reliability of the ‘‘Furcation Arrow’’ as a Diagnostic Marker

Figure 1. Maxillary right second molar, distal furcation: furcation arrow positive by five of five examiners; first molar, distal furcation: furcation arrow positive by three of five examiners.

Figure 2. Maxillary left first molar, distal furcation: furcation arrow negative by five of five examiners.

identifying furcation arrow lesions was quite low given a multirater k statistic value of 0.489. Using the agreement of at least three of the five examiners for the presence of a furcation arrow, the sensitivity of the image to identify an actual furcation invasion was 38.7% in this study, whereas the specificity was 92.2%. This indicates that most actual furcation invasions were not associated with a furcation arrow: a high number of false negatives. The absence of a furcation arrow had a much higher likelihood of not having an accompanying furcation invasion present: a low number of false positives. The positive predictive value (71.7%) and the negative predictive 1440

Volume 77 • Number 8

value (74.6%) in this study also indicate the limited value that the furcation arrow alone has for predicting the presence of furcation bone loss. In the only other study to specifically examine the furcation arrow image, Hardekopf et al.7 made radiographs of 66 proximal furcation lesions found in 45 dried human skulls and compared them with images of 120 uninvolved furcations as controls. Radiographs were projected and evaluated by a team of six examiners; a furcation arrow was identified when at least four of the six examiners agreed on its presence. They found that the prevalence of furcation arrows was greater with increasing depth of the furcation invasion; the image was found to be present at 19% of Hamp degree-1 sites, 44% of Hamp degree-2 sites, and 55% of Hamp degree-3 sites. Interestingly, the authors also reported seeing furcation arrows at 18% of the sites with no furcation invasion. Overall, the furcation arrow was seen in 40 of 96 furcations diagnosed as Hamp degree 2 and 3, resulting in a sensitivity of 42%. The furcation arrow was absent in 159 of 186 furcations having no furcation involvement or only incipient involvement for a specificity of 85%.9 These results may be interpreted as similar to those in the current study, where the sensitivity and specificity of all degrees of furcation invasion were pooled for analysis. For the current in vivo study, sites were selected on the basis of treatment needs rather than preexisting furcation status. In addition, radiographs were examined without projection or magnification in keeping with normal clinical practice. Only four degree-3 furcation invasions were discovered; therefore, these sites were combined with degree-2 sites for analysis. Although differences in the methodology of this study and that of Hardekopf et al.7 may make direct comparison difficult, the results are similar; in the current paper, furcation arrows were seen at 31.3% of Hamp 1 sites and 48.9% of Hamp 2 and 3 sites. In addition, furcation arrows were noted by three or more examiners at 17 of the 217 sites with no furcation invasion (7.83%). Although the value of the furcation arrow image to predict furcation lesions was found to be limited, the agreement between preanesthetic and postdebridement assessments of the Hamp index8 at individual sites was quite good. Despite the fact that none of the four degree-3 lesions was correctly detected before anesthesia, the preanesthetic assessment correctly identified the status of the furcation in 83.7% of degree-2 lesions, 98.4% of degree-1 lesions, and 98.6% of the time when no furcation invasion was actually present. Used in conjunction with radiographs, probing is still the most important diagnostic tool available to the clinician in assessing furcation invasions, albeit with limitations.

Deas, Moritz, Mealey, McDonnell, Powell

J Periodontol • August 2006

Several studies have compared the accuracy of preoperative and intrasurgical probing of furcations. Perhaps the one most applicable to the present report was conducted by Anderegg et al.13 In that study, using the Glickman index to assess the extent of furcation invasion, the authors found that preoperative probing without anesthetic correctly identified the true status of the furcation 62.38% of the time, with most of the errors representing underestimation of the degree of invasion. It is possible that the better agreement in our study was due to the nature of the index used. The Hamp index requires more precise discernment in distinguishing between degree-1 and -2 furcation invasions compared to differentiating between grade-1 and -2 lesions using the Glickman index. This may have improved the accuracy in the current study. Zappa et al.14 also compared preoperative probing to intraoperative assessment of 426 furcations using the Ramfjord index15 and the Hamp index.8 Preoperative probing was found to result mainly in overestimation and some underestimation of the surgically determined furcation classification, which prompted the authors to conclude that furcation diagnosis through clinical probing is of limited validity. Regardless, the Zappa et al.14 study, the Anderegg et al.13 study, and our study did not evaluate the use of preoperative bone sounding with anesthetic. Mealey et al.16 compared preanesthetic probing and bone sounding with anesthesia to direct measurement of furcation involvement at surgery in 274 furcations. Underestimation of surgical furcation depths was much more common with preanesthetic probing than overestimation, and sounding improved the accuracy of measurement in all furcation types. The authors concluded that postanesthesia bone sounding has greater diagnostic value in furcation assessment than preanesthetic probing. Clinical probing, postanesthetic sounding, and ultimately surgical access provide more diagnostic data about furcations to the clinician than radiographs. However, the current study attempted to clinically determine whether the long-held assertion of the diagnostic value of the furcation arrow image has merit. The findings of the present study suggest that the furcation arrow image has limited usefulness as a diagnostic marker in a clinical setting. Due to variations at individual sites, the image is often difficult to interpret and, when present, has the potential to correctly predict a furcation invasion only ;70% of the time. This study showed that significant variation among clinicians with regard to what constitutes a furcation arrow image also limits its usefulness. For this reason, the diagnosis of furcation invasions using radiographs in conjunction with probing or sounding remains the most reliable method for the assessment of furcation sites.

ACKNOWLEDGMENTS The views expressed in this article are those of the authors and are not to be construed as official or as reflecting the views of the United States Air Force or Department of Defense. The authors gratefully acknowledge the contributions of Dr. Anneke Bush, 59th Clinical Research Squadron, 59th Medical Wing, Lackland Air Force Base, for statistical support and SSgt. Erika Haselhoff, 59th Dental Squadron, Lackland Air Force Base, for administrative support. REFERENCES 1. Bender I, Seltzer S. Roentgenographic and direct observation of experimental lesions in bone. J Am Dent Assoc 1961;:-. 2. Bender I, Seltzer S. Roentgenographic and direct observation of experimental lesions in bone II. J Am Dent Assoc 1961;:-. 3. Ramadan A, Mitchell D. Roentgenographic study of experimental bone destruction. Oral Surg Oral Med Oral Pathol 1962;:-. 4. Rees T, Biggs N, Collings C. Radiographic interpretation of periodontal osseous lesions. Oral Surg Oral Med Oral Pathol 1971;:-. 5. Prichard J. Advanced Periodontal Disease/Surgical and Prosthetic Management, 2nd ed. Philadelphia: W.B. Saunders; 1972:. 6. Ross I, Thompson R. Furcation involvement in maxillary and mandibular molars. J Periodontol 1980;:-. 7. Hardekopf J, Dunlap R, Ahl D, Pelleu G Jr. The ‘‘furcation arrow’’: A reliable radiographic image? J Periodontol 1987;:-. 8. Hamp S, Nyman S, Lindhe J. Periodontal treatment of multirooted teeth: Results after 5 years. J Clin Periodontol 1975;:-. 9. Muller H-P, Eger T. Furcation diagnosis. J Clin Periodontol 1999;:-. 10. Cattabriga M, Pedrazzoli V, Wilson T Jr. The conservative approach in the treatment of furcation lesions. Periodontol 2000 2000;:-. 11. Al-Shammari K, Kazor C, Wang H. Molar root anatomy and management of furcation defects. J Clin Periodontol 2001;:-. 12. O’Leary T, Drake R, Naylor J. The plaque control record. J Periodontol 1972;:. 13. Anderegg C, Metzler D, Gray J, McKeever B. Comparison of presurgical and surgical measurements of maxillary molar furcations: A preliminary study. J West Soc Perio Periodontal Abstr 1991;:-. 14. Zappa U, Grosso L, Simona C, Graf H, Case D. Clinical furcation diagnosis and interradicular bone defect. J Periodontol 1993;:-. 15. Ramfjord S, Ash M. Periodontology and Periodontics. Philadelphia: W.B. Saunders; 1979:. 16. Mealey B, Neubauer M, Butzin C, Waldrop T. Use of furcal bone sounding to improve accuracy of furcation diagnosis. J Periodontol 1994;:-. Correspondence: Col. David E. Deas, 48th Dental Squadron, Royal Air Force Lakenheath, PSC 41 Box 272; APO, AE 09464, U.K. Accepted for publication March 9, 2006.

1441