From Volume 10, Issue 2, April 2017 | Pages 52-58
Discontinued and abandoned cases are an undesirable outcome of orthodontic treatment. Several patient and treatment variables have been identified as being associated with this treatment risk. The second paper of this two part series will describe a study that attempts to identify specific factors that are implicated in discontinued treatment within a specialist orthodontic practice.
The factors that are associated with discontinued and abandoned treatment are many and complex, and have been detailed in the first paper in this two part series. As the literature is sometimes conflicting and inconclusive, it would seem sensible to explore factors that may be relevant within a clinician's own practice environment. Specialist orthodontic practices provide a large pool of patients that allow for the investigation of variables that may be implicated in such cases. A study of completed and discontinued cases within an orthodontic practice in Northern England was conducted in order to identify factors that are associated with abandoned treatment.
The data source for this study were the patient records of two orthodontic specialists and an orthodontic clinical assistant working in a specialist orthodontic practice in Sunderland, Tyne and Wear, UK. The NHS in England and Wales recognizes four different treatment completion codes:
For the purpose of the study, the three termination codes were considered under the one umbrella of failed treatment. The schedules of six consecutive months, May to October 2014, for each of the three clinicians, were used to identify every patient that had either completed or failed his/her treatment. Of the 438 patients identified, one case was excluded as it was a transfer case from another practice, giving a total of 437 cases.
The clinical records of each patient were analysed to collect data for several variables (Table 1).
| Variables | Notes |
|---|---|
| Gender | |
| Age | At start of treatment |
| DMFT | Status of first molars |
| Oral hygiene | Three grades: Excellent; Improvement required; Poor |
| Index of treatment need | |
| Angle's molar classification | |
| Incisor classification | |
| Socio-economic status | Measured by Index of Multiple Deprivation (IMD) |
| Type of appliance used | All appliances that were provided to a patient. No distinctions made between different fixed appliance systems nor different removable or functional appliance designs |
| Extraction or non-extraction treatment | |
| Total number of appointments | Every visit including assessment, record-taking and active treatment until appointment when active treatment ceases and final records taken |
| Treatment length recorded | From start of active treatment to date active treatment is terminated. For patients that failed to return, the date of their final visit was recorded |
| Number of failed appointments | |
| Number of bracket failures | |
| Reason treatment is abandoned or discontinued |
IBM® SPSS® Statistics Version 22 software package www-03.ibm.com/software/products/en/spss-statistics was used for the data analysis. Data cleansing was performed and quality controlled for any missing or spurious values and typographical errors in data input.
The outcome variable (failed treatment, successfully completed treatment) is a binary (dichotomous) variable. To investigate the relationships between the dependent outcome variable and the independent variables (above), cross-tabulations were produced and unadjusted odds ratios (95% confidence intervals) calculated from binary logistic regression. Multiple regression analysis allows for the effect of multiple variables to be assessed simultaneously. The independent variables that were found to be clinically and/or statistically important were offered to a multivariable binary logistic regression model to produce adjusted odds ratios (95% confidence intervals) so factors that were independently associated with discontinued treatment would be identified.
A total of 437 (n) treatment cases were identified of which 56 were recorded as failed (discontinued or abandoned) treatments and 381 as successfully completed courses, a failure rate of 12.8% (Table 2). The discontinuation rate of 12.8% is consistent with rates that are reported in the literature, and exactly the same as Murray's (1989) finding in her Eastman Dental Hospital study.2 Discontinuation rates in NHS services in England and Wales appear to be at the lower end of those reported historically, which one may regard as a triumph for the quality of care provided by NHS orthodontic practices. However, the rate of completed and abandoned treatments is one of the key performance indicators that are assessed by commissioning teams to determine the quality of NHS orthodontic services, and can impact on the tendering of future contracts.3 It seems reasonable to suggest that this is a driver for practices to report low discontinuation rates, and that the true figures, both regionally and nationally, are in reality possibly higher than this. Additionally, a discontinued case does not necessarily equate to a total failure of treatment. In some circumstances, there is likely to be at least some improvement in the occlusion4 and a Dutch study has also suggested that patients whose adherence was poor are no less satisfied with their treatment outcome than the more adherent patient.5
| Frequency | Percentage | |
|---|---|---|
| Failed | 56 | 12.8 |
| Completed | 381 | 87.2 |
| Total | 437 | 100 |
The frequency of discontinued and completed cases for individual variables is summarized in Table 3. The unadjusted and adjusted odds ratios for individual variables are summarized in Table 4.
| Variable | Fail(%) | Complete(%) | Significance P-value | ||
|---|---|---|---|---|---|
| Gender | Male | 18.5 | 81.5 | ||
| Female | 8.7 | 91.3 | 0.003 | ||
| Age | <12 years | 23.8 | 76.2 | ||
| 12–14 years | 10.5 | 89.5 | |||
| >14 years | 10.2 | 89.8 | 0.05 | ||
| DMFT | No caries | 13.9 | 86.1 | ||
| Caries | 9.3 | 90.7 | 0.247 | ||
| Oral Hygiene | Sub-optimal | 18.1 | 81.9 | ||
| Optimal | 10.9 | 89.1 | 0.047 | ||
| IOTN | 3 | 9.3 | 90.7 | ||
| 4 | 10.5 | 89.5 | |||
| 5 | 21 | 79 | 0.026 | ||
| Molar | I | 5.4 | 94.6 | ||
| II | 20.6 | 79.4 | |||
| III | 4.8 | 95.2 | <0.001 | ||
| Incisor | I | 4.4 | 95.6 | ||
| II/1 | 18.8 | 81.2 | |||
| II/2 | 18.7 | 81.3 | |||
| III | 5.9 | 94.1 | <0.001 | ||
| IMD | 1 | 9.5 | 90.5 | ||
| 2 | 7.6 | 92.4 | |||
| 3 | 5.8 | 94.2 | |||
| 4 | 18.8 | 81.2 | |||
| 5 | 13.7 | 86.3 | 0.044 | ||
| Appliance | Fixed | 4.6 | 95.4 | ||
| Functional | 42.3 | 57.7 | |||
| Removable | 20 | 80 | <0.001 | ||
| Brackets | No bracket failures | 5.4 | 94.6 | ||
| ≥1 bracket failure | 5.3 | 94.7 | 0.987 | ||
| Extraction | Non-extraction | 14.8 | 85.2 | ||
| Extraction | 8.3 | 91.7 | 0.060 | ||
| FTA | 0 | 5.5 | 94.5 | ||
| 1 | 11.2 | 88.8 | |||
| ≥2 | 28.8 | 71.2 | <0.001 | ||
| Appointments | 1–12 | 55.9 | 44.1 | ||
| 13–24 | 7.1 | 92.9 | |||
| ≥25 | 2.4 | 97.6 | < 0.001 | ||
| Treatment Length (months) | 1–12 | 38.9 | 61.1 | ||
| 13–24 | 7.8 | 92.2 | |||
| ≥25 | 7.4 | 92.6 | <0.001 |
| Variable | Unadjusted odds ratio with 95% confidence interval | p-value | Adjusted odds ratio with 95% confidence interval | p-value | |
|---|---|---|---|---|---|
| Gender | Female | Reference | |||
| Male | 2.38 (1.34, 4.23) | 0.003 | 2.54 (1.23, 5.25) | 0.012 | |
| Age | <12 years | 2.75 (1.34, 5.64) | 0.006 | 1.36 (0.53, 3.48) | 0.52 |
| 12–14 years | 1.04 (0.53, 2.06) | 0.915 | 0.88 (0.38, 2.05) | 0.76 | |
| >14 years | Reference | ||||
| DMFT | Caries | Reference | |||
| No caries | 1.53 (0.76, 3.23) | 0.247 | |||
| Oral Hygiene | Optimal | Reference | |||
| Sub-optimal | 1.81 (1.00, 3.25) | 0.047 | 1.73 (0.82, 3.66) | 0.15 | |
| IOTN | 3 | 0.87 (0.29, 2.60) | 0.803 | 1.54 (0.43, 5.45) | 0.51 |
| 4 | Reference | ||||
| 5 | 2.26 (1.23, 4.14) | 0.009 | 0.94 (0.43, 2.07) | 0.88 | |
| Molar | I | Reference | |||
| II | 4.49 (2.25, 8.98) | < 0.001 | 1.43 (0.52, 3.89) | 0.49 | |
| III | 0.87 (0.1, 7.08) | 0.895 | 0.71 (0.05, 9.61) | 0.79 | |
| Incisor | I | Reference | |||
| II/1 | 5.02 (2.06, 12.24) | < 0.001 | 2.01 (0.61, 6.61) | 0.25 | |
| II/2 | 5.00 (1.68. 14.92) | 0.004 | 1.92 (0.34, 10.88) | 0.46 | |
| III | 1.35 (0.33, 5.63) | 0.677 | 1.75 (0.42, 7.25) | 0.44 | |
| IMD | 1+ 2 | Reference | |||
| IMD | 3 | 0.70 (0.18, 2.78) | 0.616 | 0.67 (0.14, 3.18) | 0.61 |
| 4 | 2.66 (1.33, 6.28) | 0.025 | 1.82 (0.64, 5.20) | 0.27 | |
| 5 | 1.82 (0.78, 4.25) | 0.164 | 1.55 (0.54, 4.42) | 0.42 | |
| Appliance | Fixed | Reference | |||
| Functional | 15.26 (7.48, 31.16) | < 0.001 | 10.91 (4.13, 28.79) | <0.001 | |
| Removable | 5.21 (2.28, 11.95) | < 0.001 | 3.89 (1.47, 10.30) | 0.006 | |
| Bracket Failures | ≤1 | Reference | |||
| 2 to 5 | 0.44 (0.13, 1.47) | 0.182 | |||
| 6 to 10 | 0.62 (0.16, 2.38) | 0.490 | |||
| > 10 | 1.87 (0.59, 5.90) | 0.288 | |||
| Extraction | Non-extraction | 1.93 (0.96, 3.86) | 0.060 | ||
| Extraction | Reference | ||||
| FTA | 0 | Reference | |||
| 1 | 2.18 (0.94, 5.03) | 0.068 | 1.44 (0.56, 3.71) | 0.45 | |
| ≥ 2 | 6.99 (3.43, 14.25) | < 0.001 | 4.95 (2.14, 11.44) | <0.001 | |
| Appointments | 1–12 | 52.04 (11.68, 231.80) | <0.001 | ||
| 13–24 | 3.15 (0.72, 13.73) | 0.126 | |||
| ≥25 | Reference | ||||
| Treatment Length | 0–12 | 7.93 (3.65, 17.22) | <0.001 | ||
| 13–24 | 1.06 (0.48, 2.33) | 0.89 | |||
| ≥25 | Reference |
There are more females (n = 253) than males (n = 184) within the sample, and this agrees with historic patterns of a larger uptake of orthodontic treatment by females compared with males.6 Of all the failed cases, 60.7% were male and 39.3% female. This statistically significant finding (p-value 0.003) agrees with the weight of evidence that suggests females have better adherence to treatment7 and may also be a reflection of the increased demand for treatment from females.
The mean age at the start of all treatments was 13.7 years. This would appear to be higher than expected. It could be explained by patients being put on a treatment waiting list lasting several months following their initial assessment, plus the increased demand from older children seeking orthodontic treatment while they still qualify for NHS funding. The highest proportion of failed treatments were recorded in patients under the age of 12 when compared to children in older age ranges, a finding that was of statistical significance (p-value 0.005). This may be due to the more frequent prescription of functional appliances (FnA) in this age group, although this would contradict Banks et al, who suggested that children less than 12.3 years of age were three times more likely to comply with FnAs.8 It is possible that higher motivation is present in older patients who have requested referral for treatment, having become more conscious of the aesthetics of their dentition.9
Over 75% of the subjects in the study were assessed as having optimal oral hygiene (OH) at the time of their initial assessment. A higher failure rate of treatment was noted in patients initially assessed as having sub-optimal OH compared with the failure rate in patients with optimal OH, 18.1% and 10.9%, respectively. This difference was found to be statistically significant (p-value 0.047). The results agree with the suggestion that poor OH is associated with poor adherence, although having optimal OH doesn't preclude the possibility of a patient failing treatment as the failure rate is still reasonably high. Similarly, 81.8% of patients deemed to have sub-optimal OH still went on to complete treatment successfully, which agrees with the findings of El-Mangoury10 that poor OH doesn't necessarily imply poor appliance maintenance. It is clearly bad practice to embark on active orthodontic treatment if patients have anything less than optimal OH and, ideally, a non-cariogenic diet; in this situation it is the practice's policy either to refer patients back to their GDP or to a hygienist to have an intensive course of OH instruction and diet advice prior to considering any treatment. The high failure rate would suggest that this measure has not been sufficient to improve patient adherence. Indeed, a recently conducted literature review concluded that evidence suggests orthodontic OH instruction confers only a short-term improvement in plaque scores and gingival health.11
There were no papers identified that recorded discontinued rates in relation to the DMFT status of the first molar teeth, so there is no evidence to compare with the findings of the study. However, it seems reasonable to assume that OH status and DMFT are closely linked. Of patients, 75.5% had no caries experience in any of their first molar teeth as assessed from an OPT. This figure was very similar to the proportion of patients with optimal OH (73.5%). Patients with 1 and 2 cariously affected first molars accounted for a further 12.8% and 8% of cases, respectively. Interestingly, those patients with no caries experience had slightly increased odds of discontinuing treatment compared to patients with past caries experience in their first molars: OR 1.526 (CI 95% (0.763, 3.233), but this was not significant (p-value 0.217). There was no significant association for treatment outcome in relation to different levels of DMFT (p-value 0.573). An explanation for the lack of association with past caries experience and a positive treatment outcome is that it could reflect a more motivated patient who has successfully improved his or her OH and diet and hence demonstrating the type of controlled personality needed to negotiate a course of orthodontic treatment successfully.
There was a higher rate of failure in cases that are rated the highest IOTN category 5 (p-value 0.026) compared with IOTN cases of 3 and 4. IOTN 5 cases were just over two times more likely to fail compared with IOTN 4 cases, and this was statistically significant (p-value 0.009). There does appear to be a clear association between an IOTN of 5 and discontinued treatment. These findings can be explained by the evidence from studies into the impact of indices of treatment need and treatment uptake. Patients with a low IOTN of 3 are possibly more likely to have sought treatment due to an aesthetic concern, hence being more motivated for treatment12,13 and more likely to adhere to treatment. The higher failure rate in the higher IOTN category could be due to the referral being primarily influenced by the opinion of the referring dentist based on an objective treatment need that may not necessarily match the desires or concerns of the patient.14,15
The treatment outcomes in both molar and incisor classification categories follow a very similar pattern, with Class II relationships having nearly a 20% discontinuation rate. The Class III cases had a low failure rate of around 5%, similar to the Class I cases. Class II treatment may involve the use of a functional appliance, which will certainly impact on the adherence of patients if their overjet and overbite is of little concern; this would agree with the findings of studies that relate a patient's self-perceived treatment need (and hence his/her potential motivation) more strongly to the aesthetic rating of malocclusion compared to the dental health component.16 As with the case of IOTN, it would appear from this study and most of the evidence, that the level of a patient's own aesthetic concern is important for predicting adherence regardless of the incisor or molar classification.
A chi-squared test of the failure rate in relation to Index of Multiple Deprivation (IMD) is statistically significant (p-value 0.044). Higher failure rates were recorded for IMD 4 and 5 at 18.8% and 13.7%, respectively. The increased odds of failure in the patients from areas of lower socio-economic status (SES) concurs with the findings of studies that suggest that these patients are at increased risk of treatment discontinuation.17 The most successful treatment group was the middle ranking IMD 3, with a failure rate of 5.8%. This mirrors the results of a Danish study that found that patients from the highest and lowest end of the SES scale were less likely to follow maintenance instructions.18
FnA cases make up the majority of failed treatments (53.6%) when compared with removable appliance (RA) and fixed appliance (FA) cases. In all FnA cases, they were used to treat Class II malocclusions. Twin block appliances are used in the majority of cases (the medium opening activator is used in a small minority of patients, depending on the stage of dental development, although the study fails to differentiate between the two types); the failure rate of 42.3% is much higher than that reported in the literature, where a maximum rate of 34% for twin blocks is reported in one seminal paper.19 However, one paper did report an almost identical failure rate of 42%, albeit with a different type of FnA.20 The high failure rate of FnAs highlights the importance of establishing patient motivation before embarking on orthodontic treatment to ensure that they are fully committed to wearing each type of appliance that is required.
The high failure rate of RAs (20%) also concurs with the majority of published evidence that suggests appliances that can be removed by the patient are more likely to facilitate non-adherent behaviour.21 Compared to the overall failure rate of 12.8%, FAs have a much lower failure rate of 4.6%, which suggests that, if a fixed appliance alternative exists, this may be preferable to a RA or FnA.
Patients with one or more bracket failures were no more likely to fail treatment than patients with no bracket failures: OR 1.01 (CI 95% (0.36, 2.83)). There was a slightly higher chance of failure in patients with a high bracket failure of 10 or more brackets: OR 1.87 (CI 95% (0.59, 5.9)) and, although this is statistically insignificant (p-value 0.288), it does demonstrate the notion that bracket failures equate to poor patient adherence that can potentially result in treatment abandonment.
There is a greater failure rate of non-extraction cases compared with extraction cases, but this is not statistically significant (p-value 0.06). The higher failure rate of non-extraction treatment is probably a reflection of the number of FnA cases that failed. Extraction cases also require a higher level of commitment from the patient given that unsightly spaces will result at the extraction sites which can only be closed adequately through successful FA treatment. There will likely be a greater desire from both the patient and the clinician to complete this type of treatment without leaving any spaces.
The results show that a statistically significant increase in the chance of treatment failure exists if a patient fails on two or more occasions (p-value <0.001); this agrees with previously published work.2,22 The degree of patient motivation is strongly linked to adherence,23 and it is not unreasonable to associate FTAs with a patient that lacks motivation. It has been suggested that the provision of cost-free treatment is linked to an increase in FTAs,24,25 which may have a bearing within UK NHS services.
The mean treatment length of all completed cases was 22.85 months, which compares well with much of the previous evidence on treatment length with regard to FA treatment.26 The mean number of appointments for completed cases was 19.88. Of failed treatment, 60% occurred within 12 or less visits. Of the failed cases, 50% did so within 12 months, much fewer than the mean length of 21.6 months for all cases. The figures indicate that, in the majority of failed cases, unnecessary appointments and wasted clinical hours are reduced by terminating these cases prematurely, which is in agreement with the best practice suggested by a US-based study.27
Patients failing to return for treatment was the most common reason for treatment abandonment, accounting for 60% of the discontinued cases featuring a FnA and 57% of the cases utilizing a RA. The figure was lower for treatment featuring FA cases (33.3%), which is perhaps not surprising given the greater difficulty a patient may have in removing such an appliance.
For FnA cases, the other failures were due to the patient being unable to tolerate the appliance, with just one requesting that the treatment is terminated (as opposed to the clinician steering this outcome due to a patient's inability to adhere to wearing the appliance). This is a similar pattern to the RA cases. These figures highlight the importance of patients being able to comprehend fully the difficulties that may be encountered with FnAs and RAs, and for them to be able to demonstrate this knowledge whilst simultaneously showing sufficient motivation before they are consented for treatment.
Just under 50% of the discontinued FA cases were because of poor OH and maintenance issues. This would suggest that patients with suspect OH need to be considered very carefully prior to agreeing to provide treatment. Maintenance issues alone make up a small proportion of failed FA cases (14.4%), but are more difficult to predict. It may be worth establishing the presence of fingernail-biting and pen-chewing habits28 early in the assessment stage to eliminate the chances of treatment failing because of these habits.
The factors that were found to be statistically significant as a result of hypothesis testing were:
All the variables, with the exception of the number of appointments and treatment length (the number of appointments and treatment length are consequences of treatment rather than factors that may determine outcome, so including them in a multivariable binary regression model for treatment outcome would result in spurious values) were included in a multivariable binary regression model to identify those factors that are independently associated with abandoned treatment.
The study concludes that three factors are identified as being associated with abandoned or discontinued treatment:
Although these factors in themselves are not absolute predictors of non-adherent behaviour, they may at least serve to help a clinician gauge the potential adherent behaviour of a patient. In attempting to reduce rates of discontinuation, establishing patients' motivation for treatment as early as their referral appears to be a significant factor that determines future adherence. Therefore, the early identification of poorly motivated patients could eliminate them from treatment provision. Patients with a good comprehension and understanding of their treatment regimen are also more likely to show good adherence, so it is vital that they are given appropriate and relevant information at the outset and throughout the whole course of treatment. The current literature on adherence suggests only possible associations with different variables, and in many cases, the evidence for individual factors is conflicting. There is also a growing opinion that, rather than attempting to explain non-adherence through the identification of different variables, many of which are immutable, the importance of effective clinician-patient communication could be of much more relevance. This concurs with the more modern concept of adherence to treatment protocol through a collaborative approach between clinician and patient, as opposed to the more passive patient role of compliance. This would help with the maintenance of good patient adherence throughout treatment, and lessen the chances of premature termination of treatment.
There are currently no available indices or tools to allow a clinician to predict non-adherence in patients readily. Future studies of discontinued treatment should perhaps concentrate more on patient-clinician interactions and on the development of psychological-based questionnaires that help to determine patients' attitudes and motivations for treatment.
