Background

A significant proportion of dental service resources are dedicated to the placement of restorations, all of which have a finite lifespan. Factors influencing the failure of dental restorations include patient and clinician variables, as well as the properties of the restorative material used.1,2,3,4 The cause of restoration failure is frequently considered to be multifactorial, and it is often difficult to determine the most significant factor contributing towards this unfavourable outcome. Despite this however, the most commonly cited diagnosis for restoration failure is the presence of recurrent or secondary caries.5 The adequate repair or replacement of failed restorations is crucial because with each intervention, the likelihood of tooth tissue loss increases and the tooth in question is further weakened. Furthermore, it is important to avoid adverse sequaelae such as pain, root canal treatments or extraction, which are unwelcome and financially burdensome for patients. Dental clinicians can manage failed restorations either conservatively by repairing them, or more radically by replacing the entire restoration. Previous studies have shown that various dentist and restoration related factors influence the treatment decision made by dentists.6,7,8,9,10 Such studies have reported that dentists are more likely to repair rather than replace a failed restoration if it is associated with a posterior molar tooth or, if the tooth only has a single restored surface.11 It has also been found that dentists are more likely to repair a failed restoration if they themselves have placed the original restoration.10,11

An estimated 50% of restorative work carried out by dentists' accounts for the management of failed dental restorations.8 Despite this high incidence, there are currently no guidelines advising dental clinicians about the most appropriate method of managing patients presenting with failed restorations, and more specifically if and when such restorations should be repaired or replaced. While there is evidence to implicate the influence of restoration-related and dentist-related factors on the decision of repairing vs replacing failed restorations,10 there is very little evidence regarding the impact of such a decision on patient experience. There is therefore a need for further research to provide a greater understanding of dentists' clinical decision making in these circumstances, and to explore the impact of repair or replacement decisions on patients, in particular, the level of pain, anxiety and distress experienced. Recent systematic reviews have sought to address this clinical dilemma by evaluating the outcomes of repair vs replacement of defective resin composites12 and amalgam13 restorations. Both reviews found evidence in the subject area to be limited and incomplete, and as such, no firm conclusions could be drawn about whether to repair or replace a failed restoration. Despite this however, there was weak evidence from one study14 that suggested that the short term survival of repaired vs replaced failed restorations are similar.

The overall conclusion of the systematic reviews was that high quality randomised controlled trials should be undertaken to compare the outcomes of repair vs replacement of failed dental restorations, in order to improve the evidence base on which such treatment decisions are made.

However, conducting such randomised controlled trials is expensive and time consuming.15 Therefore, before any trial is conducted, it needs justification and at the very least, a feasibility study to inform trial design and to assess the variability of outcome measures to inform power calculations.

Based on the above, the purpose of this study was to:

  • Investigate both clinical and patient factors that influence treatment decisions

  • Conduct a pilot study to form the basis for a review of the protocol, or potentially a further pilot study, before more extensive investigation into the repair vs replacement of failed restorations is undertaken

  • Identify potential outcome measures for a future trial.

Materials and methods

Study setting and population

The study settings and population have been described in detail elsewhere.16 Briefly, all general dental practitioners (n = 300) in Salford, Trafford and East Lancashire, based in the North West of England, were invited to enrol into the study. Ethical approval for the study was obtained by the National Health Service (NHS) and the University of Manchester ethics committees and a total of 38 general dental practitioners (GDPs) consented to participate (13%). The dentists were asked to recruit adult patients (above the age of 16 years) attending the practice for routine treatment, and appropriate information sheets explaining the purpose of the study were sent to these patients before they attended their appointment. The dentists were then instructed to explain the study to patients once they attended for their appointments, as well as answer any questions they had before obtaining informed consent. A total of 881 patients in the eligible age group were contacted and 508 (57.7%) of these patients were recruited into the study. Of these, a total of 103 (20.3%) patients were diagnosed with failed restorations which required repair or replacement.

Study measures

Operative factors

Dentists were asked to diagnose failed restorations and decide on whether to replace or repair them based on their own clinical judgement and usual practice. The replacement of a failed restoration was defined as the complete removal of the existing restoration and any diseased tooth tissue, followed by placement of an entirely new restoration. Repair of a failed restoration was defined as the removal of part of the existing restoration and any diseased tooth tissue. As secondary caries is the most common cause of restoration failure,5 dentists were asked to record the depth of caries in cases where the failing restoration was judged to be caused by this pathological process. Caries depth was recorded using a four level index based on D1 representing pre-cavitated (for example, white spot lesions), D2 enamel lesions, D3 dentinal lesions and D4 pulpal lesions.

Dentists were asked to record whether they used local anaesthetic (LA) or not in every case of restoration failure they treated. Dentists were also asked to record the overall time taken, in minutes, to complete the entire procedure.

Finally dentists were asked to record the type of dental material used to treat the failed dental restoration, and these included amalgam, resin composite or glass ionomer cement.

Patient factors

Dental anxiety was measured using the Corah Dental Anxiety Scale. This scale is based on four questions, each with five possible answers to describe increasing levels of dental anxiety which can be categorised as 'no anxiety', 'some unease', 'anxious' and 'very anxious'. Participants were asked to score their dental anxiety using this scale before receiving treatment while waiting in the waiting room of the dental practice.

Self-reported pain experienced by participants during the procedure was based on pain intensity. The scale used was the McGill short form pain questionnaire which includes overall pain intensity options; 'no pain', 'mild', 'discomforting', 'distressing', 'horrible' or 'excruciating'.

Participants in the study were asked to complete the pain scale straight after the procedure was conducted. They were requested to complete this inside the dental practice in the waiting area and not in the dental surgery where they could be influenced by the presence of the treating dental practitioner. Participants were reassured that any data collected would be confidential and dentists and practice staff would not be able to identify individual responses. Participants were also asked to complete the pain scale the day after the procedure (24 hours post-operatively) and send their responses to the research team using pre-prepared envelopes provided.

Statistical analysis

Data obtained from both operative and patient factors was used to form frequency distributions. Chi-squared tests and t-tests were used to compare the differences in proportions of the parameters being investigated between participants in the repair group compared to those in the replacement group. All the data was analysed in SPSS version 16 and STATA version 9.

Results

Demographics of the study population

Overall, of the 508 patients recruited into the study, 103 were diagnosed with failed restorations. Gender and age data for all 103 participants was available, and is demonstrated in Table 1 and Table 2 respectively. The percentage of males in the repair group was 43.2%, and in the replacement group 43.9%, with no statistical difference between the number of males and females in the repair and replacement group (p = 0.946).

Table 1 Gender distribution of patients in the repair versus replacement group
Full size table
Table 2 Age distribution of patients in the repair versus replacement group
Full size table

The mean age of patients in the repair group was 50.92 years (S.D. 15.6), similar to the replacement group which was 46.48 years (S.D. 14.9), and no statistical difference was observed between the ages of participants in the repair vs replacement group (p = 0.880).

Operative factors

Of the 103 patients with failed restorations, 37 (35.9%) of these underwent repair of the failed restoration, while 66 (64.1%) patients had replacement. The number of failed restorations undergoing replacement was significantly more than those which underwent repair (p = 0.004).

The results of data relating to whether failed restorations were treated on a private or NHS basis were recorded (Table 3). In total, 61 (59.2%) failed restorations were treated under the NHS, and 42 (40.8%) under the private sector. The proportion of failed restoration undergoing repair or replacement in either the NHS or privately did not demonstrate a statistically significant difference (p = 0.971).

Table 3 Provision of treatment under the NHS or privately for patients in the repair versus replacement group
Full size table

The proportion of these restorations which were repaired or replaced is illustrated in the bar chart in Figure 1.

Figure 1
figure 1

A bar chart demonstrating the treatment of failed restoration under NHS vs private care

Full size image

The data obtained for other operative factors (Table 4) shows the mean time taken to repair failed restorations was 21.7 minutes compared to 25.6 minutes taken for replacement. The increased time taken to replace compared to repair failed restorations was statistically significantly (p = 0.044).

Table 4 Operative factors relating to the repair versus replacement of failed restorations
Full size table

LA was used in 48.7% of patients who underwent a repair compared to 86.4% who had total replacement of the failed restoration. LA use was significantly greater in the replacement group compared to the repair group, p = 0.000.

The majority (43.8%) of the failed restorations which subsequently underwent repair were associated with minimal caries depth, that is, D1. Of the failed restorations which were replaced, only 9.8% in this group were associated with D1 level of caries. The majority of the failed restorations which were replaced were associated with D2 (54.1%) or D3 (36.1%) levels of caries depth. There was a statistically significant difference between the repair vs replace group in relation to the type of material used (p = 0.001).

The majority of repairs carried out were done using composite resin (40.5%) or glass ionomer cement (35.1%), whereas amalgam was most frequently used in total replacements (53.0%). There was a significant difference in type of material used between the two treatment groups, p = 0.001.

Patient factors

Dental anxiety before procedure

A number of patients in both treatment groups reported feeling 'uneasy' or 'nervous' at the thought of the dental procedure although no patients described themselves as being 'phobic'. Overall, there was a statistical difference between the two groups, with significantly more participants being anxious in the replacement group than the repair group (p = 0.008) (Table 5).

Table 5 Patient factors based on the repair versus replacement of failed restorations
Full size table

Pain intensity during the procedure

A small proportion of patients who had repair of a failed restoration experienced mild (16.2%) or discomforting (5.4%) pain during the procedure. This was similar to the replacement group in which 19.7% of patients reported experiencing mild and 9.1% experienced discomforting pain. The difference between the two groups in overall intensity of pain during the procedure was not statistically significant (p = 0.692).

Pain intensity at 24 hours

The overall intensity of pain experienced 24 hours post-operatively in both groups was similar, with no statistically significant difference between the two groups (p = 0.766). In the repair groups 77.8% experienced no pain, and in the replace group 79.1% experienced no pain.

Discussion

To our knowledge this is the first study to report the impact of repair vs replacement of failed restorations on, not only clinical, but also patient related outcome measures. For patient measures, there were significant differences for levels of pre-treatment dental anxiety associated with having repair vs replacement. On the whole, repairs were associated with less anxiety. There were no significant differences in pain intensity between the procedures. For clinical measures, local anaesthetic use was significantly greater in patients having replacement fillings. Repairs were significantly quicker, involved greater use of composites or glass ionomer cements and were performed where the depth of caries was minimal.

We found there to be a statistically significant difference between the types of material used in the repair vs the replacement treatment groups. The most common material used for replacement of restorations was amalgam. Composite or glass ionomer cements were more often used in repairs. Clinically, this may be due to the adhesive properties offered by composites and glass ionomer cements which make them easier to use in instances of repair where little or no preparation of a tooth may be required. Amalgam on the other hand requires mechanical retention and often necessitates the need for tooth or restoration preparation, which may further weaken the restored tooth. Additionally, this may also have the disadvantage of increasing clinician and patient chair-side time. Furthermore we found that fewer participants undergoing repair of restorations required local anaesthetic compared to those in the replacement group. This may be explained by the reduced need for tooth/restoration preparation in repairing restorations in contrast to total replacement. From a clinician viewpoint, this is evidently more time-efficient, which is beneficial under dental remuneration services which are often target driven. Moreover, this may also reduce the cost of materials used within the dental surgery.

The average time taken to complete the repair of restorations was significantly lower than replacements. This may be attributed to factors discussed above such as the reduced use of LA in the repair group, as well as the reduced amount of time required to prepare a tooth or cavity for repair compared to replacement of a failed restoration. Additionally, as repairs were more frequently carried out when caries depth was minimal, the size of repaired restorations was likely to be smaller and hence account for the reduced time taken to carry out the procedure.

Despite these findings, previous research indicates that failed restorations are more likely to be replaced than to be repaired in general dental practice.9,10,11,12,13,14,15,16,17 This finding was also reflected in our study. However, the most important clinical indicator that favours replacement over repair is the extent of recurrent or secondary caries. Indeed our study found evidence indicating that as the depth of recurrent caries increased, dentists were more likely to replace a restoration than to repair it. This was an expected finding as clinically, if recurrent caries has progressed into dentine, it is often difficult to determine the extent of its spread and removal of the entire restoration may be warranted to ensure that no caries is left in situ. However, given the patient and clinical benefits of repairing found in our study, it may be prudent to consider repairing vs replacing a failed restoration where the former is a feasible option for example, in minimal or no caries affected failures.

In addition to operative factors, our study also looked at the influence of repair vs replacement of failed restorations on patient-rated outcomes such as self-reported pain intensity and anxiety experience. We found that patients were more uneasy and anxious at the prospect of having a whole restoration replaced compared to those who were having a repair, a factor which is of great significance in dental practice. The general public are often fearful of dentistry and associate dental procedures with pain. It is well recognised that psychological factors such as anxiety have a major influence on increasing the perception of pain.18,19 Dentistry has progressed significantly over the years and it is the duty of dental practitioners to strive to demolish the 'stereotypical' perceptions of the public regarding dental treatment. Patients who display higher levels of anxiety may also be more likely to fail their appointment. As a result, any method of reducing patient anxiety levels should be explored further and appropriately incorporated into patient management.

We discussed above, the potential benefits of repair vs replacement regarding the reduced need for LA administration in the repair group from a clinician perspective; however, this factor may also have the potential to improve patient experience. Patients have traditionally associated dental treatment with pain and discomfort,20 and there is little evidence to show that such views are changing, despite recent advances.18 The levels of anxiety experienced vary considerably,21 but in some patients these levels are so high that they act as a barrier in seeking dental care. Based on previous research conducted,16 it has been shown that patients who require LA for a dental procedure have a two-fold increase odd of reporting pain during the procedure. The reduced need for LA administration in the repair of restorations and therefore the potential to reduce the levels of associated pain may be further indicative of the benefits of repairing vs replacing failed restorations. Surprisingly, this study found no statistically significant difference between levels of patient self-reported pain intensity experience at different time intervals between the two groups, despite there being a significant difference in LA administration.

Overall the findings of our study tend to indicate that repair of failing restorations, except in cases where failure is due to deep recurrent caries, would be preferential to the replacement of the entire restoration. However, there are some methodological limitations to consider. Firstly, clinicians were asked to record the depth of recurrent caries if this was the cause of the failed restoration. A drawback of such a method is the subjective nature of caries depth assessment. That said, any inconsistent errors are likely to have been the same in both groups (repair vs replace) and are therefore unlikely to have biased the results. Secondly, a further drawback of our study was that it did not allow for long-term data collection looking at the longevity of the failed restorations which were repaired vs those which were replaced. Whilst the repair of failed restorations based on this study alone may be looked upon as a more favourable option, further studies looking at the longevity of these restorations would be appropriate to determine whether repaired restorations last longer than those that have been replaced or vice versa. Thirdly, the study may be prone to selection bias as only 38 (13%) of the 300 GDPs contacted, agreed to participate. Our study included an equal number of NHS (50%) and private dental practices (50%). We assessed selection bias by assessing the difference in repair vs replacement by the type of practice that patients attended. Based on previous studies22 we may expect GDPs working in NHS practices to repair rather than replace failed restorations due to the reduced clinical time required. The results of this study however, found that in NHS dental practices 30% and 60% of failed restorations were repaired and replaced respectively. This was comparable to private dental practices, where 36% of failed restorations were repaired and 64% were replaced.

We have explored the potential benefits that repairing failed restorations can have for both patients and clinicians alike, particularly in cases where recurrent caries is minimal. Based on the results of this pilot study, it is apparent that where clinically appropriate, dentists are currently opting to repair rather than replace a failed restoration. Previous research has shown the short-term outcomes of repair vs replaced failed restorations to be similar;14 however there is currently no high quality evidence comparing long-term outcomes.

Conclusion

The results of our pilot study support the notion that failed restorations should be repaired where clinically possible. Repairs are quick, less anxiety provoking and require less local anaesthetic when compared to total replacement of restorations. Based on the findings of our study, conducting a large scale randomised controlled trial using clinical and patient outcome measures such as those used in our study would be difficult to 'sell' to a funding body because adverse short term outcomes are rare and because of the major role that clinical expertise plays in assessing all of the clinical variables which determine whether a patient with a failed restoration is treated by repair or replacement. The complexity of this decision making process would make fixing decisions on inclusion and exclusion criteria for a trial very difficult from an ethical and medico-legal perspective, also tight inclusion and exclusion criteria would mean that any trail would be unlikely to reflect 'real life' clinical practice. Therefore the research focus should be to provide high quality prospective data to evaluate the longevity of repaired vs replaced restorations, to determine whether repaired failed restorations offer comparable long-term outcomes to those that are replaced.