Trial design

We report a 3-arm parallel-group, randomized, single blinded, clinical trial with an allocation ratio of 1:1:1 conducted in the United States. The Institutional Review Board at the University of Connecticut Health approved the study on December 12, 2011 (number 12-025-2), and written informed consent was received from all parents, guardians, and children. No changes to the methodology occurred after trial commencement. The study follows the Declaration of Helsinki guidelines The trial is registered at ClinicalTrials.gov ( NCT02745626 ). The presentation of this report is according to the CONSORT guidelines for reporting trials. All methods were performed in accordance with the approved guidelines and regulations.

Participants, eligibility criteria, and settings

The participants were recruited from the outpatient orthodontic clinic, University of Connecticut Health, Farmington, Conn, between December 2011 and March 2014. Eligibility criteria for the participants included the following: (1) physically healthy with no relevant allergies or medical problems, (2) older than 12 years of age at commencement of treatment, (3) permanent dentition, (4) less than 5 mm of anterior crowding or spacing with adequate overjet and overbite, (5) ability to maintain adequate oral hygiene, and (6) optimum dental health without immediate need for any allied dental procedure. The exclusion criteria were (1) skeletal anteroposterior discrepancies between the maxilla and the mandible (ANB, ≥5°), (2) centric relation-centric occlusion discrepancies greater than 3 mm; and (3) active periodontal disease.

Interventions

Participants were randomly assigned to 1 of the 3 treatment groups based on the choice of intervention: (1) Clear Aligners (CLA) (Invisalign, Align technology, San Jose, Calif); (2) preadjusted edgewise fixed appliance with self-ligated brackets (SLB) (Carriere, Carlsbad, Calif); and (3) preadjusted edgewise fixed appliance with elastomeric ligated brackets (ELB) (Ortho Organizers Inc., Carlsbad, CA). In addition to the consent form for routine orthodontic care currently used in the University of Connecticut Health orthodontic clinic, the parent or guardian of each subject was given a written explanation of the background of the study, its objectives, and their involvement. The study called for no additional treatments or procedures not normally performed during routine oral care for initial bonding or orthodontic adjustment visits.

Participants were requested to refrain from eating, drinking, or brushing 1 hour before the sampling appointments. They were given oral hygiene instructions and oral hygiene supplements at regular intervals throughout the treatment. The supplements consisted of a toothbrush, toothpaste, interdental brush, dental floss, and an information brochure on oral hygiene maintenance. The orthodontic treatment was performed in all patients by treating both arches simultaneously. All teeth were bonded with a direct technique using Transbond XT (3M, St. Paul, Minn). The first molars were banded. Participants with aligners were instructed to wear them at least 20 hours per day. The aligners were replaced every 2 to 3 weeks with a new set, which had been previously developed according to the treatment plan of each patient.

Outcomes (primary and secondary)

Periodontal measurements were recorded specifically for the maxillary second premolar before treatment (T0), after 9 months of treatment (T1), after 18 months or completion of treatment (T2), whichever came sooner. Clinical assessment of the periodontal health status was achieved by measuring plaque index (PI), gingival index (GI), and papillary bleeding index (PBI) ( Fig 1 ). All clinical measurements were performed by examiners (A.C., S.A., M.U.), calibrated to a single examiner. At least 90% intraexaminer agreement was achieved for each index before the study to ensure reliability of clinical evaluations.

CONSORT diagram showing the flow of subjects through the trial.

The PI was assessed by visual inspection of the accumulated plaque in the gingival area and was classified into 1 of 4 grades of the modified index of Loe ( Fig 1 ). The GI was assessed referring to color and tissue consistency according to the severity of existing inflammation of the marginal gingiva. The PBI was assessed by carefully inserting a periodontal probe (Hu-Friedy, Chicago, Ill) into the gingival sulcus. The data were registered 20 seconds after probing. The original indexes were modified because all evaluations took place during regular control visits.

The primary outcome measure was dental plaque surrounding the bracket of the maxillary second premolar at the 3 time points, whereas the secondary outcomes were gingival health and papillary bleeding. There were no outcome changes reported after the commencement of the trial. Adverse events during the study were also recorded.

Sample size

The power analysis was performed using the computer application G-Power to detect the mean differences among the treatment groups in the PI at T2. The effect size for analysis of variance was given by the following equation.

where n _i = number of observations in group i where (i = 1, 2, 3; N = total number of observations; p i = n i N 😉 μ _i = mean of group I; μ = grand mean; and σ ² = error variance within groups.

It was assumed that all 3 groups would have the same sample size, ie, p ₁ = p ₂ = p ₃ = 1/3. The mean PI at T2 in the ELB group was assumed to be 1.0 with the 3 groups sharing a common standard deviation of 0.5. The mean PI at T2 in the CLA or SLB group if ≤0.5 or ≥1.5 was considered clinically significant: ie, μ ₁ = 1, μ ₂ = 1, μ ₃ = 0.5, or μ ₁ = 1; μ ₂ = 0.5, μ ₃ = 0.5 or μ ₁ = 1, μ ₂ = 1; μ ₃ = 1.5 or μ ₁ = 1, μ ₂ = 1.5, μ ₃ = 1.5, all leading to f = 0.471. To detect this effect size by an analysis of variance F test, 16 patients per group were needed to achieve 80% power at the 5% significance level. It may be optimistic to assume that the PI would follow a normal distribution across the treatment groups. To maintain the same power, when the parametric assumptions are violated and a nonparametric test is used, a rule of thumb is to add 15% to the sample. This finalized the sample size to 19 per group.

Interim analyses and stopping guidelines

Not applicable.

Randomization (random number generation, allocation concealment, implementation)

The randomization sequence was generated by the statistician (C.L.K.) with a computer-generated list of random numbers by using an online software program ( http://mahmoodsaghaei.tripod.com/Softwares/randalloc.html#Random%20Allocation%20Software ), independent of the clinical investigators (A.C., S.A) involved in recruitment. To ensure a 1:1:1 allocation ratio, block randomization was done with random block sizes of 15, 6, 9, and 3. For all participants once the clinical investigators had obtained consent, they telephoned or texted a contact (M.U.) who was independent of the recruitment process for allocation consignment. Therefore, different persons performed sequence generation, allocation concealment, and implementation.

Blinding

Blinding of the clinical investigators and participants to the intervention in each group was impossible. Therefore, blinding was done for outcome assessment only. Unique identifiers that were created for each patient coded the extracted data appropriately. This was maintained in a secure, locked location separate from the collected data. The coding of the data was broken after the end of the analysis, and no breach of blinding was identified.

Interrater reliability

The examiners underwent training and calibration exercises for the collection of the clinical data (determination of PI, GI, and PBI). The calibration exercise consisted of 2 steps. The first was a theoretical step procedure that involved discussion of the indexes by analysis of photographs of the conditions. A periodontist with considerable experience coordinated this step, training 2 examiners in how to perform the examination. The second step was the clinical part, in which the examiners examined 20 randomly selected participants. Interexaminer agreement was tested. To examine measurement reliability, concordance correlation coefficients were calculated for the various parameters. Data analysis used the Cohen kappa coefficient on a tooth-by-tooth basis. Very good interexaminer agreements were obtained where all values of kappa were greater than 0.81.

Statistical analysis (primary and secondary outcomes)

Demographic and clinical characteristics were summarized with conventional descriptive statistics. Since most patients had periodontal scores of 0 or 1 across the 3 time points, the scores for the outcome variables (PI, GI, and PBI) were dichotomized as 0 and 1+ (1, 2, 3, or 4). A between-group comparison within the framework of a logistic regression model was conducted. The sample size was not sufficiently large for us to fit a logistic mixed effects model or its alternative of generalized estimating equations (GEE) model with the mixed effects of treatment group, time, and their interaction. Based on our graphic exploration of the relevant data, the effect of time was not found to be linear—ie, treatment groups and the differences among the groups varied with time. Therefore, time was modeled as a categorical variable to include the interaction between time and treatment groups. Logistic regression models were used to model the 3 outcomes at T0, T1, and T2 separately. We were able to include age and sex for adjustment to improve precision. In each model, the ELB group was considered as the reference group and compared with another treatment group via odds ratio. A P value smaller than 5% was considered statistically significant. All statistical analyses were performed in R 3.3.1 (R Core Team 2016; https://www.R-project.org ).

Trial design

We report a 3-arm parallel-group, randomized, single blinded, clinical trial with an allocation ratio of 1:1:1 conducted in the United States. The Institutional Review Board at the University of Connecticut Health approved the study on December 12, 2011 (number 12-025-2), and written informed consent was received from all parents, guardians, and children. No changes to the methodology occurred after trial commencement. The study follows the Declaration of Helsinki guidelines The trial is registered at ClinicalTrials.gov ( NCT02745626 ). The presentation of this report is according to the CONSORT guidelines for reporting trials. All methods were performed in accordance with the approved guidelines and regulations.

Participants, eligibility criteria, and settings

The participants were recruited from the outpatient orthodontic clinic, University of Connecticut Health, Farmington, Conn, between December 2011 and March 2014. Eligibility criteria for the participants included the following: (1) physically healthy with no relevant allergies or medical problems, (2) older than 12 years of age at commencement of treatment, (3) permanent dentition, (4) less than 5 mm of anterior crowding or spacing with adequate overjet and overbite, (5) ability to maintain adequate oral hygiene, and (6) optimum dental health without immediate need for any allied dental procedure. The exclusion criteria were (1) skeletal anteroposterior discrepancies between the maxilla and the mandible (ANB, ≥5°), (2) centric relation-centric occlusion discrepancies greater than 3 mm; and (3) active periodontal disease.

Interventions

Participants were randomly assigned to 1 of the 3 treatment groups based on the choice of intervention: (1) Clear Aligners (CLA) (Invisalign, Align technology, San Jose, Calif); (2) preadjusted edgewise fixed appliance with self-ligated brackets (SLB) (Carriere, Carlsbad, Calif); and (3) preadjusted edgewise fixed appliance with elastomeric ligated brackets (ELB) (Ortho Organizers Inc., Carlsbad, CA). In addition to the consent form for routine orthodontic care currently used in the University of Connecticut Health orthodontic clinic, the parent or guardian of each subject was given a written explanation of the background of the study, its objectives, and their involvement. The study called for no additional treatments or procedures not normally performed during routine oral care for initial bonding or orthodontic adjustment visits.

Participants were requested to refrain from eating, drinking, or brushing 1 hour before the sampling appointments. They were given oral hygiene instructions and oral hygiene supplements at regular intervals throughout the treatment. The supplements consisted of a toothbrush, toothpaste, interdental brush, dental floss, and an information brochure on oral hygiene maintenance. The orthodontic treatment was performed in all patients by treating both arches simultaneously. All teeth were bonded with a direct technique using Transbond XT (3M, St. Paul, Minn). The first molars were banded. Participants with aligners were instructed to wear them at least 20 hours per day. The aligners were replaced every 2 to 3 weeks with a new set, which had been previously developed according to the treatment plan of each patient.

Outcomes (primary and secondary)

Periodontal measurements were recorded specifically for the maxillary second premolar before treatment (T0), after 9 months of treatment (T1), after 18 months or completion of treatment (T2), whichever came sooner. Clinical assessment of the periodontal health status was achieved by measuring plaque index (PI), gingival index (GI), and papillary bleeding index (PBI) ( Fig 1 ). All clinical measurements were performed by examiners (A.C., S.A., M.U.), calibrated to a single examiner. At least 90% intraexaminer agreement was achieved for each index before the study to ensure reliability of clinical evaluations.

CONSORT diagram showing the flow of subjects through the trial.

The PI was assessed by visual inspection of the accumulated plaque in the gingival area and was classified into 1 of 4 grades of the modified index of Loe ( Fig 1 ). The GI was assessed referring to color and tissue consistency according to the severity of existing inflammation of the marginal gingiva. The PBI was assessed by carefully inserting a periodontal probe (Hu-Friedy, Chicago, Ill) into the gingival sulcus. The data were registered 20 seconds after probing. The original indexes were modified because all evaluations took place during regular control visits.

The primary outcome measure was dental plaque surrounding the bracket of the maxillary second premolar at the 3 time points, whereas the secondary outcomes were gingival health and papillary bleeding. There were no outcome changes reported after the commencement of the trial. Adverse events during the study were also recorded.

Sample size

The power analysis was performed using the computer application G-Power to detect the mean differences among the treatment groups in the PI at T2. The effect size for analysis of variance was given by the following equation.

where n _i = number of observations in group i where (i = 1, 2, 3; N = total number of observations; p i = n i N 😉 μ _i = mean of group I; μ = grand mean; and σ ² = error variance within groups.

It was assumed that all 3 groups would have the same sample size, ie, p ₁ = p ₂ = p ₃ = 1/3. The mean PI at T2 in the ELB group was assumed to be 1.0 with the 3 groups sharing a common standard deviation of 0.5. The mean PI at T2 in the CLA or SLB group if ≤0.5 or ≥1.5 was considered clinically significant: ie, μ ₁ = 1, μ ₂ = 1, μ ₃ = 0.5, or μ ₁ = 1; μ ₂ = 0.5, μ ₃ = 0.5 or μ ₁ = 1, μ ₂ = 1; μ ₃ = 1.5 or μ ₁ = 1, μ ₂ = 1.5, μ ₃ = 1.5, all leading to f = 0.471. To detect this effect size by an analysis of variance F test, 16 patients per group were needed to achieve 80% power at the 5% significance level. It may be optimistic to assume that the PI would follow a normal distribution across the treatment groups. To maintain the same power, when the parametric assumptions are violated and a nonparametric test is used, a rule of thumb is to add 15% to the sample. This finalized the sample size to 19 per group.

Interim analyses and stopping guidelines

Not applicable.

Randomization (random number generation, allocation concealment, implementation)

The randomization sequence was generated by the statistician (C.L.K.) with a computer-generated list of random numbers by using an online software program ( http://mahmoodsaghaei.tripod.com/Softwares/randalloc.html#Random%20Allocation%20Software ), independent of the clinical investigators (A.C., S.A) involved in recruitment. To ensure a 1:1:1 allocation ratio, block randomization was done with random block sizes of 15, 6, 9, and 3. For all participants once the clinical investigators had obtained consent, they telephoned or texted a contact (M.U.) who was independent of the recruitment process for allocation consignment. Therefore, different persons performed sequence generation, allocation concealment, and implementation.

Blinding

Blinding of the clinical investigators and participants to the intervention in each group was impossible. Therefore, blinding was done for outcome assessment only. Unique identifiers that were created for each patient coded the extracted data appropriately. This was maintained in a secure, locked location separate from the collected data. The coding of the data was broken after the end of the analysis, and no breach of blinding was identified.

Interrater reliability

The examiners underwent training and calibration exercises for the collection of the clinical data (determination of PI, GI, and PBI). The calibration exercise consisted of 2 steps. The first was a theoretical step procedure that involved discussion of the indexes by analysis of photographs of the conditions. A periodontist with considerable experience coordinated this step, training 2 examiners in how to perform the examination. The second step was the clinical part, in which the examiners examined 20 randomly selected participants. Interexaminer agreement was tested. To examine measurement reliability, concordance correlation coefficients were calculated for the various parameters. Data analysis used the Cohen kappa coefficient on a tooth-by-tooth basis. Very good interexaminer agreements were obtained where all values of kappa were greater than 0.81.

Statistical analysis (primary and secondary outcomes)

Demographic and clinical characteristics were summarized with conventional descriptive statistics. Since most patients had periodontal scores of 0 or 1 across the 3 time points, the scores for the outcome variables (PI, GI, and PBI) were dichotomized as 0 and 1+ (1, 2, 3, or 4). A between-group comparison within the framework of a logistic regression model was conducted. The sample size was not sufficiently large for us to fit a logistic mixed effects model or its alternative of generalized estimating equations (GEE) model with the mixed effects of treatment group, time, and their interaction. Based on our graphic exploration of the relevant data, the effect of time was not found to be linear—ie, treatment groups and the differences among the groups varied with time. Therefore, time was modeled as a categorical variable to include the interaction between time and treatment groups. Logistic regression models were used to model the 3 outcomes at T0, T1, and T2 separately. We were able to include age and sex for adjustment to improve precision. In each model, the ELB group was considered as the reference group and compared with another treatment group via odds ratio. A P value smaller than 5% was considered statistically significant. All statistical analyses were performed in R 3.3.1 (R Core Team 2016; https://www.R-project.org ).