The prevalence of overhanging margins in posterior amalgam restorations and periodontal consequences
Angela R. C. Pack\ Leslie J. Coxhead^ and Barry W. McDonald^ 'Department of Periodontology, School ot Dentistry, University of Otago, Dunedin, New Zealand; ^Department of Statistics, La Trobe University, Melbourne Australia
Pack ARC, Coxliead LJ and McDonald BW: The prevaleticc of overhanging margins in posterior amalgam restorations attd periodontal consequences. J Clin Pcriodontol 1990; 17: 145-152. Abstract. The purpose of this study was to determine the prevalence of overhanging margins and associated periodontal status in 100 patietits who had received completed treatment by final year dental students. Pockets, bleeding on probing and clinically detectable overhanging tnargins were recorded on all posterior teeth. Overhanging margins on approximal restorations were detected by use of bitewing radiographs. 1319 teeth were examined in which 2117 restored surfaces were evaluated. Of these, 1186 (56%) had overhanging trtargins. 62% of all approximal restorations had overhanging rnargins while 35% of buccal and 40% of lingual restorations had overhanging tnargins. 59% of new approximal restorations placed in previously unrestored surfaces had overhanging rnargins, and 595 overhanging rnargins identified on pre-treatment radiographs were still present on post-treatnient radiographs. 64.3% of pockets adjacent to overhanging margins were > 3 tntn, cotnpared with 23.1 % of pockets adjacent to unrestored surfaces and 49.2% of pockets adjacent to restorations without overhanging margins. A similar association existed between restorative status and bleeding. 32% of pockets adjacent to overhanging margins bled on probing compared with 10.5% of pockets adjacent to unrestored surfaces and 21.6% of pockets adjacent to restorations without overhanging margins. Periodontal disease was more severe when overhangs were present. However, when approxitnal overhanging margins were adjacent to an edentulous space, the periodontal effects were lessened. When adjacent to neighbouring teeth, overhanging rnargins also significantly affected the periodontal status of those teeth.
Accepted for publication 9 March 1989
Overhanging margins are said to be responsible for trtuch iatrogenic periodontal disease. As long ago as 1912, Black included faulty tnargins of restorations among the causes of gingivitis. In 1953, Mosteller recommended radiographic evaluation of all restorations after placetnent because overhanging margins were datnaging to the adjacent soft tissues. Many investigators since then have reported upon the adverse effects of poor restorations on the health of the adjacent periodontal tissues (Waerhaug 1956, 1960, Zander 1956, 1957, 1958, Wright 1963, Trott & Sherkat 1964, Alexander 1967, 1968, Bjorn, Bjorn & Grkoric 1969, Gilmore & Sheihatu 1971, Leon 1976, Coxhe-
ed or eliminated, and may be a significant factor in the aetiology of periodontal disease. The study reported in this paper was designed to detemiine: (a) the prevalence of overhanging tnargins of restorations in patients treated by final year dental undeigraduate students in a University Clinic; (b) whether the prevalence of overhanging margins in patients from the University Clinic was significantly different from that found in patients from general dental private practices in New Zealand; (c) the status of periodontal tissues adjacent to tooth surfaces which were either unrestored, restored without overhanging margins, or restored with overhanging margin.
ad, Robertson & Simpson 1978, Jeffcoat & Howell 1980, Lang, Kiel & Anderhalden 1983 & Coxhead 1985). High proportions of restorations have been reported as having overhanging margins: 57% noted by Wright (1963), 75% by Bjorn et al. (1969), 32% by Gilmore & Sheihatn (1971), 52% by Coxhead et al. (1978), 74% by Gorzo, Newman & Strahan (1979), 48% by Hakkaranein & Aitiatno (1980) and 76% by Coxhead (1985), although methods of detection varied between these studies. Despite recent technological advances in restorative dentistry and dental materials, it appears that overhanging tiiargins of restorations are still being created, are not always detect-
Key words: prevalence; overhanging margins; bleeding on probing; probing depth; Liang-Zeger statistical method.
146
Pack et al. bitewing radiographs and re-evaluated them independently of the original data. Subsequent analysis showed that 92.8% of radiographic scores on the second occasion were identical to those recorded on the first occasion. In these cases, a total of 431 overhanging margins were scored on the first occasion and 425 were scored on the second occasion. This indicates a very high level of reproducibility of radiographic interpretation. Comparison of the pre- and posttreatment radiographs, permitted identification of new restorations placed in previously unrestored mesial or distal surfaces. Similarly, the number of overhanging margins left by general dental practitioners were compared with those left by students. This was achieved by examining pre-treatment radiographs of those patients who had attended the clinic for their first course of treatment and those of patients who had attended for more than one year. When attempting to show the relationship of an edentulous space adjacent to a restored approximal surface to either the prevalence of overhanging restorations, or to the periodontal status, only data from tooth surfaces between the mesial of the second molar and the distal of the first premolar were considered. Initially, computations for the contingency table analysis (;K') were performed using the statistical package SPSSX (Nie 1983) on a Vax 11/780 computer at the University of Otago, Dunedin, New Zealand. Subsequently, computations for analysis using the method of Missing teeth were also noted in both Liang and Zeger (1986) were performed the clinical and radiographic assess- on a Data General MV 20000 computer at La Trobe University, Melbourne, ments. 6 months after the survey was com- Australia. The programme was written pleted, one author (L.TC) randomly re- using SAS/IML statistical language and selected 30 of the 100 sets of posterior a copy is available upon request. All statistical analyses were performed by the third author (B McD). A paper reviewing concepts determining statistical analysis of dental data including the application of the Liang-Zeger statistical method, is published elsewhere in this issue of the .fourtutl (McDonald & Pack 1989).
treatment at the School of Dentistry were recorded. At completion of the clinical examination, posterior bitewing radiographs Material and Methods were taken. These, together with the 100 subjects (47 tnen and 53 women) pre-treatment bitewing radiographs, were recruited from patients who had were then examined by the other clincotnpleted a course of routine dental ician (L.IC) without reference to the treatment of not less than 3 months dur- clinical data. Radiographs were viewed ation by any one of the 55 final year under standardised conditions using a dental students in the Restorative Den- constant light source and no magnifital Care Clinic of the School of Den- cation. tistry, Suitable subjects had to have posOverhanging margins were recorded terior teeth present in at least 2 quad- on mesial or distal surfaces if the radiorants and have onfileposterior bitewing graphic image of the approxirnal resradiographs taken immediately before toration showed a step or ledge extenthe last course of treatment began. Pa- ding beyond the nortnal smooth profile tients with complicated medical con- of the tooth, or a "bevelled" appearance ditions were excluded. Subjects were in- at the base of the approximal restoravited to attend for an intra-oral examin- tion, attributable to an overhanging ation in which premolar and molar teeth margin present in a concavity on the (excluding third molars) were evaluated surface of the tooth (Coxhead et al. for probing depth, and bleeding on 1978) (Fig. 1). No attempts were made probing at 6 sites around each tooth. to classify the size of overhanging marFor this examination, a Michigan perio- gins. dontal probe was used. Subsequently, Approximal surfaces were scored all mesial, distal, buccal and lingual surfrom the radiographs as follows: faces of the same teeth were assessed using a fine sharp sickle probe and 0 = unrestored surface scored for the presence or absence of 1 = restoration without visible overhanging margin overhanging margins: 2 = restoration with visible overhanging margin in a tooth adjacent to an0 = unrestored surface other tooth 1 = restoration within 1 mm of the gingival margin or below, but without 3 = restoration with visible overhanging margin in a tooth adjacent to an clinically detected overhanging edentulous space margin. 8 = unable to make a radiographic as2 = restoration within 1 mm of the ginsessment owing to either overlapped gival margin or below, but with a approximal surfaces, or tiiissing surclinically detected ledge indicative of face on the radiograph. an overhanging margin.
when such teeth were either in an intact arch, or next to an edentulous space.
If the clinician was uncertain whether an overhanging margin was present, a score of 1 was allocated. All clinical assessments were made by one clinician (ARCP). The age of each subject and the years of attendance for routine
Results
Fig. I. Showing post-treatment posterior bitcwiiig r;idiogr;iphs of a subject with overhangs: (A) obvious overhangs: (B) bevelled appearance of restoration due to overhang: (C) overhang within the profile of the tooth.
In thisstudy of 100 subjects with a mean age of 39.22-fl 3.42 S.D. (min: 20 years, max: 73 years), 1319 teeth were assessed in which 2117 gingivally located margins of restorations were found. The
Overhangs and periodonlal status mean number of restored tooth surfaces per subject was 21.17+ 10.81 S.D. (min: 3, max: 47). The mean number of tnissing posterior teeth (excluding 3rd mo.lars) was 2.81+2.88 S.D. with 32 subjects having no tnissing teeth. 21 subjects had attended the School of Dentistry for only one course of treatment and 16 had attended for more than 11 years, (mean = 5.86+ 7.75 S.D. rnedian = 3).
restored surfaces without overhanging tnargins was 9.31+6.22 S.D. per subject. Of all the 2117 restored surfaces evaluated, 1186 (56%) had overhanging margins. 978 overhanging margins were found clinically and/or radiographically on approxitnial restorations. This was 62% of all approxinial restored surfaces examined. Clinical data alone revealed that 42% of distal and 36% of mesial restored surfaces had overhanging tnargins, whereas evaluation of the radioOverhanging margins graphs taken on the satne day gave figures of 44% and 47%, respectively. Prevalence Combining clinical and radiographic Table 1 shows the number of overhandata, every subject was found to have ging margins detected clinically and/or at least 1 overhanging margin present radiographically. Furthermore, 595 apat the completion of dental treatment. proxitnal overhanging tnargins (61% of 3 subjects had 1 overhanging margin; 11 the 978 identified on the second radiosubjects had more than 20 overhanging graph) were also identified on the premargins, and the mean was 11.86 + 6.33 treattnent radiograph and had not been S.D. (median 12). The mean number of eliminated.
Tdhlc !. Showing that there were no statistically significant differences between the prevalence of overhangs in the recall patients and the newly enrolled patients in 1984 Correlation model Regression parameters
-0.2172 0.1600
Test statistics mean effect regressor effect
-3.0421 0.8322 N.S.
estimate
0.0714 0.1922
S.E.
-0.2030 0.1190
0.0715 0.1976
-2.8427 0.6030 N.S.
S.E. = standard error. N.S. = not significant. Table 2. Comparing the restored status of the different posterior tooth surfaces No. of sites Status
distal
mesial
hiiccal
lingual
unrcstoied restored, no overhang restored with overhang
495 308 516
496 328 495
1044 178 97
1125 116 78
no. restorations
824
823
275
194
X-= 1168.83; df=6; f 3 mtn
41.5 34.5 24.8
58.5 65.5 75.2
58.1 63.2 61.8
41.9 36.8 38.2
r = 3H.2 ; d f = 2 significance: />< 0.0001* * Significant. NS: not significant.
When the pre-treattnent radiographs of recall patients were cotnpared with those of newly enrolled patients, the nutnber of overhanging tnargins did not differ significantly. In the statistical analyses used, the response variables were those approxitnal sites in all patients which had restorations on the pretreatment radiograph. The response was positive if the restoration had an overhanging tnargin and negative if there was no overhanging tnargin. The regressor variable was whether or not the patient was newly enrolled in 1984. The standard equi-correlation structure was assumed for the responses producing a very stnall estimated correlation (/• = 0.041). This indicated a very low level of association between restored sites so that overhanging tnargins occurred virtually independently of each other. Table 2 shows that whether or not the patient was newly enrolled in 1984 (the regressor) was not a significant predictor of the probability that a patient would have an overhanging tnargin at a restored site.
Independence model
S.E.
estimate
mean effect regressor effect
147
r = 0.50 ; df=2 significance: P>0J7NS
Distribution on tooth sttrfaces When comparing the restorative status of mesial, distal, buccal or lingual surfaces, very significant differetices existed in the distribution of status across the sites (Table 3). A tnuch greater proportion of approxitnal restorations had overhanging tnargins cotnpared with buccal or lingual restorations {/} = 85.86, df=l,;;3 mm ( / = 702, df= 2,/7< 0.0001). Highly significant differences in pocket depth existed between unrestored and restored sites and also between sites with and without overhanging margins (Z, = 13.25, p
Angela R. C. Pack\ Leslie J. Coxhead^ and Barry W. McDonald^ 'Department of Periodontology, School ot Dentistry, University of Otago, Dunedin, New Zealand; ^Department of Statistics, La Trobe University, Melbourne Australia
Pack ARC, Coxliead LJ and McDonald BW: The prevaleticc of overhanging margins in posterior amalgam restorations attd periodontal consequences. J Clin Pcriodontol 1990; 17: 145-152. Abstract. The purpose of this study was to determine the prevalence of overhanging margins and associated periodontal status in 100 patietits who had received completed treatment by final year dental students. Pockets, bleeding on probing and clinically detectable overhanging tnargins were recorded on all posterior teeth. Overhanging margins on approximal restorations were detected by use of bitewing radiographs. 1319 teeth were examined in which 2117 restored surfaces were evaluated. Of these, 1186 (56%) had overhanging trtargins. 62% of all approximal restorations had overhanging rnargins while 35% of buccal and 40% of lingual restorations had overhanging tnargins. 59% of new approximal restorations placed in previously unrestored surfaces had overhanging rnargins, and 595 overhanging rnargins identified on pre-treatment radiographs were still present on post-treatnient radiographs. 64.3% of pockets adjacent to overhanging margins were > 3 tntn, cotnpared with 23.1 % of pockets adjacent to unrestored surfaces and 49.2% of pockets adjacent to restorations without overhanging margins. A similar association existed between restorative status and bleeding. 32% of pockets adjacent to overhanging margins bled on probing compared with 10.5% of pockets adjacent to unrestored surfaces and 21.6% of pockets adjacent to restorations without overhanging margins. Periodontal disease was more severe when overhangs were present. However, when approxitnal overhanging margins were adjacent to an edentulous space, the periodontal effects were lessened. When adjacent to neighbouring teeth, overhanging rnargins also significantly affected the periodontal status of those teeth.
Accepted for publication 9 March 1989
Overhanging margins are said to be responsible for trtuch iatrogenic periodontal disease. As long ago as 1912, Black included faulty tnargins of restorations among the causes of gingivitis. In 1953, Mosteller recommended radiographic evaluation of all restorations after placetnent because overhanging margins were datnaging to the adjacent soft tissues. Many investigators since then have reported upon the adverse effects of poor restorations on the health of the adjacent periodontal tissues (Waerhaug 1956, 1960, Zander 1956, 1957, 1958, Wright 1963, Trott & Sherkat 1964, Alexander 1967, 1968, Bjorn, Bjorn & Grkoric 1969, Gilmore & Sheihatu 1971, Leon 1976, Coxhe-
ed or eliminated, and may be a significant factor in the aetiology of periodontal disease. The study reported in this paper was designed to detemiine: (a) the prevalence of overhanging tnargins of restorations in patients treated by final year dental undeigraduate students in a University Clinic; (b) whether the prevalence of overhanging margins in patients from the University Clinic was significantly different from that found in patients from general dental private practices in New Zealand; (c) the status of periodontal tissues adjacent to tooth surfaces which were either unrestored, restored without overhanging margins, or restored with overhanging margin.
ad, Robertson & Simpson 1978, Jeffcoat & Howell 1980, Lang, Kiel & Anderhalden 1983 & Coxhead 1985). High proportions of restorations have been reported as having overhanging margins: 57% noted by Wright (1963), 75% by Bjorn et al. (1969), 32% by Gilmore & Sheihatn (1971), 52% by Coxhead et al. (1978), 74% by Gorzo, Newman & Strahan (1979), 48% by Hakkaranein & Aitiatno (1980) and 76% by Coxhead (1985), although methods of detection varied between these studies. Despite recent technological advances in restorative dentistry and dental materials, it appears that overhanging tiiargins of restorations are still being created, are not always detect-
Key words: prevalence; overhanging margins; bleeding on probing; probing depth; Liang-Zeger statistical method.
146
Pack et al. bitewing radiographs and re-evaluated them independently of the original data. Subsequent analysis showed that 92.8% of radiographic scores on the second occasion were identical to those recorded on the first occasion. In these cases, a total of 431 overhanging margins were scored on the first occasion and 425 were scored on the second occasion. This indicates a very high level of reproducibility of radiographic interpretation. Comparison of the pre- and posttreatment radiographs, permitted identification of new restorations placed in previously unrestored mesial or distal surfaces. Similarly, the number of overhanging margins left by general dental practitioners were compared with those left by students. This was achieved by examining pre-treatment radiographs of those patients who had attended the clinic for their first course of treatment and those of patients who had attended for more than one year. When attempting to show the relationship of an edentulous space adjacent to a restored approximal surface to either the prevalence of overhanging restorations, or to the periodontal status, only data from tooth surfaces between the mesial of the second molar and the distal of the first premolar were considered. Initially, computations for the contingency table analysis (;K') were performed using the statistical package SPSSX (Nie 1983) on a Vax 11/780 computer at the University of Otago, Dunedin, New Zealand. Subsequently, computations for analysis using the method of Missing teeth were also noted in both Liang and Zeger (1986) were performed the clinical and radiographic assess- on a Data General MV 20000 computer at La Trobe University, Melbourne, ments. 6 months after the survey was com- Australia. The programme was written pleted, one author (L.TC) randomly re- using SAS/IML statistical language and selected 30 of the 100 sets of posterior a copy is available upon request. All statistical analyses were performed by the third author (B McD). A paper reviewing concepts determining statistical analysis of dental data including the application of the Liang-Zeger statistical method, is published elsewhere in this issue of the .fourtutl (McDonald & Pack 1989).
treatment at the School of Dentistry were recorded. At completion of the clinical examination, posterior bitewing radiographs Material and Methods were taken. These, together with the 100 subjects (47 tnen and 53 women) pre-treatment bitewing radiographs, were recruited from patients who had were then examined by the other clincotnpleted a course of routine dental ician (L.IC) without reference to the treatment of not less than 3 months dur- clinical data. Radiographs were viewed ation by any one of the 55 final year under standardised conditions using a dental students in the Restorative Den- constant light source and no magnifital Care Clinic of the School of Den- cation. tistry, Suitable subjects had to have posOverhanging margins were recorded terior teeth present in at least 2 quad- on mesial or distal surfaces if the radiorants and have onfileposterior bitewing graphic image of the approxirnal resradiographs taken immediately before toration showed a step or ledge extenthe last course of treatment began. Pa- ding beyond the nortnal smooth profile tients with complicated medical con- of the tooth, or a "bevelled" appearance ditions were excluded. Subjects were in- at the base of the approximal restoravited to attend for an intra-oral examin- tion, attributable to an overhanging ation in which premolar and molar teeth margin present in a concavity on the (excluding third molars) were evaluated surface of the tooth (Coxhead et al. for probing depth, and bleeding on 1978) (Fig. 1). No attempts were made probing at 6 sites around each tooth. to classify the size of overhanging marFor this examination, a Michigan perio- gins. dontal probe was used. Subsequently, Approximal surfaces were scored all mesial, distal, buccal and lingual surfrom the radiographs as follows: faces of the same teeth were assessed using a fine sharp sickle probe and 0 = unrestored surface scored for the presence or absence of 1 = restoration without visible overhanging margin overhanging margins: 2 = restoration with visible overhanging margin in a tooth adjacent to an0 = unrestored surface other tooth 1 = restoration within 1 mm of the gingival margin or below, but without 3 = restoration with visible overhanging margin in a tooth adjacent to an clinically detected overhanging edentulous space margin. 8 = unable to make a radiographic as2 = restoration within 1 mm of the ginsessment owing to either overlapped gival margin or below, but with a approximal surfaces, or tiiissing surclinically detected ledge indicative of face on the radiograph. an overhanging margin.
when such teeth were either in an intact arch, or next to an edentulous space.
If the clinician was uncertain whether an overhanging margin was present, a score of 1 was allocated. All clinical assessments were made by one clinician (ARCP). The age of each subject and the years of attendance for routine
Results
Fig. I. Showing post-treatment posterior bitcwiiig r;idiogr;iphs of a subject with overhangs: (A) obvious overhangs: (B) bevelled appearance of restoration due to overhang: (C) overhang within the profile of the tooth.
In thisstudy of 100 subjects with a mean age of 39.22-fl 3.42 S.D. (min: 20 years, max: 73 years), 1319 teeth were assessed in which 2117 gingivally located margins of restorations were found. The
Overhangs and periodonlal status mean number of restored tooth surfaces per subject was 21.17+ 10.81 S.D. (min: 3, max: 47). The mean number of tnissing posterior teeth (excluding 3rd mo.lars) was 2.81+2.88 S.D. with 32 subjects having no tnissing teeth. 21 subjects had attended the School of Dentistry for only one course of treatment and 16 had attended for more than 11 years, (mean = 5.86+ 7.75 S.D. rnedian = 3).
restored surfaces without overhanging tnargins was 9.31+6.22 S.D. per subject. Of all the 2117 restored surfaces evaluated, 1186 (56%) had overhanging margins. 978 overhanging margins were found clinically and/or radiographically on approxitnial restorations. This was 62% of all approxinial restored surfaces examined. Clinical data alone revealed that 42% of distal and 36% of mesial restored surfaces had overhanging tnargins, whereas evaluation of the radioOverhanging margins graphs taken on the satne day gave figures of 44% and 47%, respectively. Prevalence Combining clinical and radiographic Table 1 shows the number of overhandata, every subject was found to have ging margins detected clinically and/or at least 1 overhanging margin present radiographically. Furthermore, 595 apat the completion of dental treatment. proxitnal overhanging tnargins (61% of 3 subjects had 1 overhanging margin; 11 the 978 identified on the second radiosubjects had more than 20 overhanging graph) were also identified on the premargins, and the mean was 11.86 + 6.33 treattnent radiograph and had not been S.D. (median 12). The mean number of eliminated.
Tdhlc !. Showing that there were no statistically significant differences between the prevalence of overhangs in the recall patients and the newly enrolled patients in 1984 Correlation model Regression parameters
-0.2172 0.1600
Test statistics mean effect regressor effect
-3.0421 0.8322 N.S.
estimate
0.0714 0.1922
S.E.
-0.2030 0.1190
0.0715 0.1976
-2.8427 0.6030 N.S.
S.E. = standard error. N.S. = not significant. Table 2. Comparing the restored status of the different posterior tooth surfaces No. of sites Status
distal
mesial
hiiccal
lingual
unrcstoied restored, no overhang restored with overhang
495 308 516
496 328 495
1044 178 97
1125 116 78
no. restorations
824
823
275
194
X-= 1168.83; df=6; f 3 mtn
41.5 34.5 24.8
58.5 65.5 75.2
58.1 63.2 61.8
41.9 36.8 38.2
r = 3H.2 ; d f = 2 significance: />< 0.0001* * Significant. NS: not significant.
When the pre-treattnent radiographs of recall patients were cotnpared with those of newly enrolled patients, the nutnber of overhanging tnargins did not differ significantly. In the statistical analyses used, the response variables were those approxitnal sites in all patients which had restorations on the pretreatment radiograph. The response was positive if the restoration had an overhanging tnargin and negative if there was no overhanging tnargin. The regressor variable was whether or not the patient was newly enrolled in 1984. The standard equi-correlation structure was assumed for the responses producing a very stnall estimated correlation (/• = 0.041). This indicated a very low level of association between restored sites so that overhanging tnargins occurred virtually independently of each other. Table 2 shows that whether or not the patient was newly enrolled in 1984 (the regressor) was not a significant predictor of the probability that a patient would have an overhanging tnargin at a restored site.
Independence model
S.E.
estimate
mean effect regressor effect
147
r = 0.50 ; df=2 significance: P>0J7NS
Distribution on tooth sttrfaces When comparing the restorative status of mesial, distal, buccal or lingual surfaces, very significant differetices existed in the distribution of status across the sites (Table 3). A tnuch greater proportion of approxitnal restorations had overhanging tnargins cotnpared with buccal or lingual restorations {/} = 85.86, df=l,;;3 mm ( / = 702, df= 2,/7< 0.0001). Highly significant differences in pocket depth existed between unrestored and restored sites and also between sites with and without overhanging margins (Z, = 13.25, p
