Study
Lesiona
No lesion
Healing (%)
Effect on healing
Strindberg (1956)
2/4
9/11
11/15 (73%)
Overall 19% reduction (although lower when lesion is present)
Grahnén and Hansson (1961)
NR
NR
NR
No effect
Ingle and Glick (1965)
NR
NR
NR
No effect
Engström et al. (1964)
NR
NR
6/9 (67%)
No effect
Engström and Lundberg (1965)
0/0
5/5
5/5 (100%)
No effect
Grossman (1969)
9/19
42/47
51/66 (77%)
Reduced only when lesion is present
Crump and Natkin (1970)
27/29
21/24
48/53 (91%)
No effect
Fox et al. (1972)
NR
NR
93/100 (93%)
Reduced only when lesion is present
Bergenholtz et al. (1979)
NR
NR
NR
Reduced only when lesion is present
Kerekes and Tronstad (1979)
NR
NR
9/11 (82%)
Reduced only in teeth with necrotic pulps
Cvek et al. (1982)
3/4
NA
3/4 (75%)
Not stated specifically for fractured files
Sjögren et al. (1990)
NR
NR
9/11 (82%)
Not discussed
Molyvdas et al. (2001)
8/11
32/35
40/46 (87%)
Reduced only when lesion is present
Spili et al. (2005)
51/56
62/63
113/119 (95%)
No effect
Imura et al. (2007)
NR
NR
8/11 (73%)
Not discussed
Fu et al. (2011)
NR
NR
3/8 (38%)
Reduced due to perforation
Ng et al. (2011)
NR
NR
18/27 (67%)
Not reported
Ungerechts et al. (2014)
NR
NR
13/23 (57%)
No effect
Total (%)
100/123 (81%)
171/185 (92%)
430/508 (85%)
The landmark outcome-based paper by Strindberg (1956) was the earliest research to look at the impact of fractured files on clinical and radiographic outcomes. This comprehensive long-term follow-up study of factors related to the results of pulp therapy was the first published work to report the influence of retained fractured instruments (or what he referred to as “file breakage”) on the prognosis of endodontic treatment. Using strict criteria for healing (i.e., “incomplete” or “uncertain” healing were categorized as “failure”) and observation periods of 4–10 years of his own cases, Strindberg (1956) included 15 cases with fractured instruments (five in single-rooted teeth without apical periodontitis, two in single-rooted teeth with apical periodontitis, six in multi-rooted teeth without apical periodontitis, and two in multi-rooted teeth with apical periodontitis). Four failures occurred among the 15 teeth with fractured instruments present (27%) compared with 42 of 453 (9%) teeth without fractured files. Despite the small numbers of fractured instruments associated with periapical lesions, Strindberg (1956) concluded that, while the presence of fractured files would always reduce the prognosis of RCT, the effect would be more profound if there was a preoperative lesion present. Strindberg (1956) considered instrument fracture a serious problem, and although he was usually unaware of the bacterial status of the root canal prior to file breakage, he surmised that prognosis would be poorer in the presence rather than in the absence of infection (i.e., a periapical radiolucency). Further, he speculated that in cases where there was intracanal infection apical to the retained fragment, subsequent conservative therapy alone would probably not eradicate such infection or eliminate its potential consequences.
Using the clinical and radiographic methods described by Strindberg (1956), Grahnén and Hansson (1961) calculated the failure frequency of pulp and root canal therapy on adult patients treated by students. They analyzed 763 teeth (1277 roots) with a review period of 4–5 years and claimed that the failure rate of cases with fractured files was no different from that of cases without retained file fragments even when the preoperative periradicular status was considered. However, they did not actually specify the number of fractured file cases, although the overall failure rate was 12%. The 4–5-year follow-up investigation by Engström et al. (1964) of 306 conservatively root-filled teeth revealed no statistically significant difference between fractured instrument cases with (2 of 4 failures) or without (1 of 5 failures) pretreatment positive bacterial culture. The following year, Engström and Lundberg (1965) also published a 3.5–4-year radiographic follow-up study of teeth conservatively root-filled following pulpectomy; hence, there were no cases with lesions. All five fractured instrument cases, which yielded negative cultures before obturation, were classified as successes. However, contemporary concepts question the validity of culturing (Sathorn et al. 2007).
The classic “Washington study ” described (but not actually published) by Ingle and Glick (1965) in the first edition of Ingle’s textbook also concluded that treatment outcome was unaffected by a retained fractured instrument. During the eight years of the study, which also provided the caseload for Crump and Natkin (1970), a great number of instruments were fractured, yet only one case out of the 104 failures from 1229 cases at the 2-year recall could be attributed to a broken instrument. The authors hypothesized that a broken instrument itself could serve as “an adequate root canal filling,” which was to be later supported by Fox et al. (1972). Ingle and Glick (1965) concluded that even though fractured instruments were not “favored,” they were unlikely to affect prognosis and were amenable to surgical treatment if found in the apical third.
Grossman (1968, 1969) conducted a 5-year survey of patients in a university clinic to assess the effect of fractured files on prognosis. With an average follow-up period of 2 years, the data (n = 66) included 19 cases with lesions and 47 without (31 of the latter having vital pulps). The outcomes were then compared with a sample of “normal” controls (presumably cases without fractured files, although this was not specified). No difference was found between vital cases and necrotic cases without preoperative periapical lesions; however, there was a 39% reduction in success (47% vs. 86%) when “rarefaction” was present; if “doubtful” the cases were considered failures. Grossman (1968, 1969) claimed this to be a significant difference compared with vital cases and cases without a periapical lesion; however, he did not provide any statistical analysis or further details on the “normal” cases. Additionally, an unspecified number of teeth in this study were obturated with silver cones. The study design of this investigation could be considered a case series. Like most other outcome studies, including those that evaluated the prognostic impact of a retained fractured instrument, it highlights the limitations or weaknesses inherent with such a research design. This view that the presence of periapical pathosis rather than the fractured instrument per se was of greater impact was supported in subsequent papers by Fox et al. (1972) and Molyvdas et al. (1992), finding that fractured files had reduced prognosis in the presence of a periapical radiolucency.
The interesting study by Fox et al. (1972) reported similar conclusions to those of Grossman (1968, 1969). In their case series, of 304 teeth with retained carbon steel or SS files, fractured either accidentally (n = 100; 32.9%) or intentionally (n = 204; 67.1%), the overall “failure” rate noted was 6.25% (n = 19). However, for the accidentally fractured cases, the failure rate was 7%. Interestingly, these authors described a technique of intentionally filling root canals with SS instruments that were cemented in place with root canal sealer. On the other hand, in the case of accidental fracture, no attempt was made to bypass or remove the instruments; rather the remainder of the canal was filled with gutta-percha and sealer. Teeth with preoperative periapical radiolucencies increased the probability of failure by threefold. Similarly, Molyvdas et al. (1992) found that all cases (n = 23) with preoperative diagnosis of pulpitis were categorized as successes, whereas only 75% of 12 necrotic cases and 73% of 11 teeth with periapical pathosis were successes; “compromised” cases were considered failures. Importantly, the latter authors found that bypassing the instrument fragment in 22 cases resulted in 95% success.
Kerekes and Tronstad (1979) investigated the outcome of a standardized treatment protocol performed by dental students on 647 roots (in 478 teeth). There were only 11 instances of instrument fractures with six occurring in vital cases and five in necrotic cases. Of these cases, all of the vital teeth were considered to have successful outcomes, while two of the necrotic cases resulted in failure. The criteria used to analyze the radiographs were such that anything larger than a “slight” radiolucent zone around the gutta-percha was considered uncertain or failure. Although the low prevalence of instrument fracture did not allow statistical analysis, the data did support the finding of the other studies described above. Bergenholtz et al. (1979) conducted a radiographic follow-up to assess the effect of over-instrumentation and over-filling only on retreated root canals. They observed 11 retained file fragments that were fractured during the retreatment of 660 cases subsequently followed up for 2 years. They concluded that file fracture did not seem to influence prognosis in those cases retreated purely for technical reasons but did reduce prognosis for retreated cases with preoperative periapical pathosis.
Cvek et al. (1982) evaluated the treatment outcome of 54 endodontically treated non-vital maxillary and mandibular incisors with post-traumatically reduced pulpal lumens and preoperative periapical lesions. In this study, four file fractures were noted, all of which occurred when the smallest observable lumen diameter was 0–0.1 mm, which was measured by comparing with an orthodontic wire of 0.1 mm diameter; 0.1 mm was found to be the smallest width discernible in the radiograph with acceptable precision. Of these four teeth only one showed signs of “osteitis ” at 4 years following treatment. In the Sjögren et al. (1990) outcome-based study of 356 teeth, retained instruments were present in 11 roots, two of which subsequently showed periapical lesions. However, there was no information on the preoperative status of these teeth, and as with Kerekes and Tronstad (1979), they considered roots rather than teeth. In a study of the outcome of endodontic retreatment, Van Nieuwenhuysen et al. (1994) reported 10 (1.6%) cases of fractured instruments from 612 retreated roots but did not clarify whether these instruments were retained or retrieved following fracture. Their findings indicated that complications during retreatment, such as file fracture, resulted in a reduced retreatment outcome, but no further details were provided.
More recently, a retrospective study of 2000 cases (Imura et al. 2007) treated in a single private practice over 30 years found that teeth without intraoperative complications (instrument fracture, perforation, and flare-up) healed at a higher rate than those with such complications (91.9% vs. 72.6%). Complications occurred in 51 cases, but file fractures only accounted for 11, of which only three resulted in failure; the actual type of instruments was not specified. With such a low-fracture prevalence, statistical analysis was not feasible without being combined with other complications. This is a common theme in much of the outcome literature with the earlier pooled phases of the Toronto study (de Chevigny et al. 2008) also encountering 11 fractured files (in 373 teeth), but the authors only reported the change in healing between teeth with and without complications (including pulp chamber cracks, aberrant anatomy, perforation, and non-negotiable canals) rather than specifically for fractured files alone.
Another more recent study that attempted to report the effect of fractured files on treatment outcome (Fu et al. 2011) reexamined 102 teeth with fractured instruments present 12–68 months after treatment. Using PAI scores to measure periapical disease and a dichotomized description of root fillings as either adequate (including fractured instruments in the apical third) or inadequate (nonhomogeneous appearance or not ending at either the point of obstruction or within 2 mm of the apex), they were able to follow up 66 cases, of which 58 had the fragments successfully removed and eight still had the fractured instrument present at the time of review. Of these eight cases, five were deemed failures. In these five cases, two instrument fragments were pushed through the apex during the attempt at removal. Though the authors concluded that a failure to remove a fractured instrument reduced prognosis, it is difficult to establish whether the attempted removal (which resulted in perforation in three of the five teeth) may have actually contributed to the rate of failure. Interestingly, the only other factor that significantly impacted the prognosis of these teeth was the quality of the root canal filling, which may be interpreted to suggest that control of intraradicular infection rather than file removal per se is the key to obtaining favorable outcomes as indicated by Fox et al. (1972). Interestingly, one of the more robust prospective outcome studies of recent times (Ng et al. 2011) did analyze the impact of fractured instruments on prognosis as an independent variable, recording 15 instrument fractures (of 1155 roots) in primary treatment and 12 (of 1302) in retreatment cases. There was only a significant difference in healing in the retreatment cases (50% healing vs. 80% in primary treatment cases). However, despite this finding, the authors pointed out that the type of fractured instrument as well as its fate was in the same confounding pathway as the ability to obtain patency. Hence, the inference was that the presence of the fractured file itself was unlikely to be the true cause of persistent disease but rather has a negative impact because of its interference with the ability to gain patency.
A very recent study (Ungerechts et al. 2014) analyzed the outcome of treatment by students at a Norwegian university dental clinic focusing on the impact of instrument fracture. Fractured instruments occurred in 38 of 3854 treated teeth and mostly comprised SS hand files and lentulo-spiral burs (81.6%) as well as several NiTi instruments (18.4%). Ten of these instruments were removed prior to obturation, and the other 28 were left in situ. As with Fu et al. (2011), the authors found higher rates of success associated with teeth that had the fragments removed prior to obturation (71.4% vs. 56.5%) as well as those teeth with preoperative diagnosis of vital pulps compared with those that were necrotic or previously treated (72.7% vs. 58.3% vs. 42.9%, respectively). However, none of these findings reached statistical significance, likely because eight of the 38 fractured instrument cases could not be followed up. Unfortunately, the fractured instrument cases were not matched to “normal” controls nor was any information provided about the periapical status of the teeth in question. As a result, limited information can be gathered from this paper about the impact of fractured instruments on prognosis.
In summary, the lower-level evidence on the prognosis for fractured instruments seems to suggest that, in cases without preoperative lesions, the presence of a fractured instrument has no impact on prognosis. However, most of these early papers offer little insight into the actual impact of instrument fracture on the prognosis of modern endodontic treatment. This is because of the inherent issues in study design, including a lack of matched controls and a small sample size of fractured instruments, and the questionable relevance of the techniques and instruments to contemporary practices. Consequently, the conclusions of the authors of many of these papers were often subjective, contradictory, and made unsubstantiated statements based on insufficient sample size, inappropriate or no control groups, poor or no inclusion/exclusion criteria, lack of blinding leading to observer bias, unsatisfactory or undefined outcome measures and criteria, uncontrolled confounding factors, and especially unsatisfactory statistical analyses. Further, a major shortcoming of most of these studies was recognized by Strindberg (1956), who stated the following when summarizing the limitations of the published studies in his survey of the literature: “The effect of any one factor on the results has been studied without regard for other factors”—in other words, most studies failed to perform logistic regression analysis to account for possible associations among various potential prognostic (independent) variables and treatment outcome (the dependent variable).
8.4 Case-Controlled Studies
Case-controlled studies offer the greatest level of insight into the impact of instrument fracture on prognosis by allowing comparison of outcomes in teeth which differ only in the presence or absence of a retained instrument but are similar in all other respects. Crump and Natkin (1970) provided the first of such studies, searching through 8500 cases treated by dental students at the University of Washington between 1955 and 1965. They identified 178 retained fractured instrument (carbon steel or SS) cases and matched them to a selection of 400 controls by tooth type, canal number, material, and the presence of absence of a lesion (but not pretreatment pulpal status, medicament used, or quality of root filling). All teeth were required to have had at least a 2-year review, and new recalls were made for study patients and matched controls for clinical and radiographic evaluation. Clinically, the presence of signs or symptoms of persistent periapical disease was assessed, and radiographs were taken to categorize the teeth as either “success” (the complete absence of any discernible periapical lesion), “uncertain” (a questionable clinical sign or a periapical lesion reduced in size by more than 75% or the presence of PDL thickening up to 1 mm where there was an initial diagnosis of normal apical tissues), or “failure” (the presence of definitive clinical signs or symptoms, less than 75% reduction in lesion size, or appearance of a new lesion). A total of 53 matched pairs could be recalled and reviewed. No significant differences could be found between the outcomes of teeth with and without fractured instruments whether they were analyzed in three groups (success, failure, uncertain) or two groups (with uncertain considered as success or failure). In order to show that the negative result was not a consequence of unmatched variables, the authors analyzed the distribution of these variables (including lateral canals, voids, unfilled canals, root resorption, and root perforation) and showed they were evenly distributed among controls and fractured instrument cases as well as between successes and failures. Based on these findings, Crump and Natkin (1970) suggested a conservative approach to the management of fractured files.
Spili et al. (2005) conducted a more recent, and the only other, case-controlled study. The study itself consisted of two distinct parts with the first assessing the incidence of instrument facture over a 13.5-year period and the second part comparing the outcome of treatment in cases with retained instruments with matched controls in order to determine the impact of instrument retention on prognosis. A total of 8460 cases treated between 1990 and 2003 were screened (with the transition from hand to rotary instruments occurring between 1996 and 1997) and coded for various variables with all cases, which had both the presence of a retained instrument and at least a 1-year clinical and radiographic follow-up identified. Teeth with previously fractured instruments, obviously defective restorations, or insufficient clinical or radiographic documentation were excluded. The radiographic observations were separated into signs of complete healing, incomplete healing, uncertain healing, and no healing, while the teeth were judged clinically as either having the presence or absence of clinical signs or symptoms. Success was then determined to be complete or incomplete healing in the absence of clinical signs or symptoms.
The results reported by Spili et al. (2005) showed 277 teeth with fractured instruments of which 146 had a greater than 1-year recall available. The total number of fractures accounted for 5.1% of the teeth with 4.4% being rotary NiTi instruments and 0.7% being SS files (in the period between 1997 and 2003 where hand instruments were used exclusively as pathfinders). For the case-control portion of the study, the overall rates of healing were 91.8% and 94.5% for cases and controls, respectively. When these results were divided according to the absence or presence of a radiographic lesion prior to treatment, the results were 96.8% compared with 98.4% for controls (without a lesion) and 86.7% compared with 92.9% for controls (with a lesion). These differences were not statistically significant, with the 95% confidence interval for the reduction in healing rate in the presence of a periapical lesion ranging from −3.0 to 15.3%. In fact, the only factor that was shown to have a statistically significant impact on prognosis was the presence or absence of a preoperative lesion. Like previous authors (Molyvdas et al. 2001), Spili et al. (2005) hypothesized that despite the positive results, the true impact of fractured instruments may depend on the stage of root canal preparation at which the fracture occurred, although the information required to be able to confirm this was not available from the study sample. Spili et al. (2005) concluded that, based on the results of the study, instrument fracture, when occurring in the hands of experienced endodontists, does not in itself affect prognosis.
8.5 Meta-Analysis
A literature review and meta-analysis was performed by Panitvisai et al. (2010), to answer the question “in adult patients who have had nonsurgical RCT, does the retention of a separated instrument, compared with no retained fractured instrument, result in a poorer clinical outcome?” Of the 17 studies retrieved, all but two were excluded for various reasons, mostly due to the fact that they were not case controlled. The two included studies were those already discussed above (Crump and Natkin 1970; Spili et al. 2005). Despite several differences between the two studies, namely, the different instruments and techniques employed in treatment as well as the difference in treatment setting, Panitvisai et al. (2010) combined the data through meta-analysis, with the main justifications being the similarity in study design and the fact that endodontic outcomes have not changed considerably in preceding three decades. When the data from the two case-controlled studies were combined, no significant difference was found in healing with or without the presence of a retained instrument, with a 95% confidence interval of −0.05 to 0.06. The authors pointed out that despite the relatively small sample size, due to the review being based on only two articles, the narrow confidence interval would suggest that larger samples would not alter the results. The authors concluded that, based on these findings, there was no significant reduction in prognosis when fractured endodontic instruments were retained in canals, although this may not be fully applicable to general practice dentistry.
Stay updated, free dental videos. Join our Telegram channel
VIDEdental - Online dental courses
