Indirect bonding may or may not improve the efficiency of orthodontic treatment

Today’s post focuses on indirect bonding once again. One of the most significant benefits of this technique is the increased efficiency in bracket placement, which can potentially lead to better treatment outcomes. This new trial examines these important factors. I will also use this paper to demonstrate how I critically appraise a trial.

In a post a few weeks ago, I discussed a new trial that looked at Direct v Indirect bonding and concluded that there were limited advantages to indirect bonding. This new study also sheds some light on these interesting clinical issues.

A team from Okayama, Japan did this study. The EJO published the paper.

It was great to see that this paper was open-access.  

When I evaluate a paper, my first consideration is its relevance to my clinical practice. This particular paper aligns with my criteria. If I think that a paper’s subject matter is not relevant to my practice, I usually choose not to read it.

What did they ask?

They did this study to 

“Evaluate and compare the direct and indirect bonding methods through comprehensive and detailed clinical evaluation”.

What did they do?

They did a single-centre, open-label, quasi-randomised controlled clinical trial.

The PICO was

Participants

I could not find any information on the patients in the study in the methods. But this was contained in a supplementary table. This revealed that the participants were over 12 years old and had a malocclusion.

Intervention

Indirect bonding

Control

Direct bonding

Outcomes

The primary outcome was total treatment time.  Secondary outcomes were level of discomfort during bonding, oral hygiene, and post-treatment occlusal index scores using the OGS index.

Further details of the method

Twenty-one orthodontists treated the patients in a university setting. Their clinical experience ranged from 1 to 6 years.  They used pre-adjusted plastic or metal brackets. When the crown length was short, the metal brackets were used only on the upper and lower premolars. Most cases had plastic brackets placed.

At this point, I am starting to think that complex variables (co-founders) may influence the results. These are multiple operators and different brackets.

The team randomly assigned the patients to the direct or indirect bracket group. One person used medical record numbers to allocate the interventions.

This rang an alarm bell for me. It means that the patients were not randomly allocated, as there could be bias inherent in the allocation of medical record numbers. The researcher may construct systems to increase the chance of achieving an outcome. This is a well-recognised problem. The authors did not say why they did not simply randomise the participants.

The team detailed their sample size calculation based on the primary outcome of total treatment time. They correctly used data from a previous retrospective study and wanted to power their study to detect a meaningful difference of 3.16 months between interventions. They calculated that they needed 50 participants per group if they wanted to take dropouts into account.

This was clear and when I repeated the sample size calculation I got the same result.

They used simple multivariate statistical tests to evaluate the data.

I am not 100% sure about this because this does not take into account the multiple operators. I would have liked to see a regression analysis.

What did they find?

They identified a possible sample of 1059 participants and excluded 906 of them. This was because they did not meet the inclusion criteria.  

They enrolled 72 patients into the indirect group and 81 into the direct.  However, 14 in the indirect group (19%) and 15 in the direct group (18%) dropped out.  This left a final sample of 58 in the indirect group and 66 in the direct bonding group.

This is a high number of dropouts.

Outcomes

Let’s have a look at the outcome measures

When they looked at the overall treatment time. This was 30.51 (SD=7.27)  months for the indirect group and 34.27 (8.87 months) for the direct group.  They stated this was significantly different.

The standard deviations for this data were rather large, so I calculated the 95% confidence interval. The mean CI for indirect bonding was 28.59-32.42. For the direct, this was 32.08-36.45.  When the Cis overlap it suggests that the difference is unlikely to be significant.  This was reinforced when I calculated the CI of the difference between the groups. This was 3.76 months with a 95% confidence interval of 0.85-6.66. This means we can be 95% confident that the true difference between the means is within this range.  This is very close to zero. We need to bear this in mind when we consider the results.  

You may think that this was a lot of work to calculate. However, it took me about 5 minutes using this calculator.  Of course, it would be easier for us all if the journals asked for confidence intervals!

When they looked at chair time. They defined this as the time it took for bonding and banding. They found that the time required to bond brackets from the mandibular incisors to the molars was 35.91 (SD=15.51) minutes for the indirect group and 58.51 (SD=25.4) minutes.

These are long! But they are similar to the paper that I discussed a couple of weeks ago.

The only other data that I want to consider is the number of bracket failures. This was 7.28 for the indirect and 2.92 for the direct group. The number of repositions was 3.2 for the indirect and 8.89 for the direct group.

This was a very high level of repositions and I wonder if this is an operator issue?

When they looked at the outcome of their treatment with the Objective grading system of AAO they found a mean of 47.31 (8.17) for the indirect and 53.74 (7.74) direct.  Again, this difference was statistically significant.

Although the results are statistically significant, I’m uncertain about their clinical significance. While I’m not very familiar with this scoring system, I understand that a score greater than 30 points typically indicates a failure in the examination. I wonder if these scores are on the higher side.

Their final conclusion was rather unclear and the text did not reflect the abstract.  I, therefore, took this from the abstract

“Indirect bonding may improve the efficiency of orthodontic treatment”.

This was rather vague.

What did I think?

I have put a lot of thought into the text (highlighted in red). While I have expressed several criticisms, we need to consider whether the issues I pointed out influence the results and conclusions. My next step is to analyze the impact of my most significant concerns.

Firstly, we lack information about the nature of the malocclusions and the treatment provided. For example, the distribution of non-extraction/extraction treatments. This makes it difficult to assess their relevance to my practice. My uncertainty regarding the findings is further compounded by the absence of details about the operators. Notably, 21 operators treated 112 patients, but we do not know how many patients each operator treated. Additionally, the operators varied in experience, ranging from one six years of experience. Consequently, we must assume that the results are primarily applicable to inexperienced operators. This may also explain the long duration of the bracket bonding procedure and the high occlusal index scores at the end of treatment.

I have another concern regarding the statistical analysis. First, if the team had conducted a multivariate analysis, they could have accounted for variations among the operators. Furthermore, when we examine the effect size and the broad confidence intervals, particularly for the duration of treatment, it’s important to note that they are close to zero. This indicates that we have a low level of confidence in the data, and the true effect may, in fact, be minimal.

Finally, the lack of information about the randomisation and concealment means that the trial is at high risk of bias.  I am not saying that the trial is biased but the risk of bias is high.

Final comments

When I consider all these issues my only conclusion is that there is a great deal of uncertainty in the findings of this study.  I think that the authors, perhaps, feel the same. Hence their rather vague conclusion.

I hope that you have found this post interesting and have picked up some tips on how to read a trial publication.