Trial design and any changes after trial commencement

The study was a parallel-group, randomized, active-control trial with a 1:1 allocation ratio. There were no changes after trial commencement.

Participants, eligibility criteria, and setting

Approval for this randomized clinical trial was obtained from Peking University Medical Science Research Ethics Committee (IRB00001052-07094). Participants were recruited from the Department of Orthodontics at Peking University, based on the following inclusion criteria: (1) 7 to 13 years old before treatment with a midface soft tissue deficiency; (2) fully erupted maxillary first molars, Class III malocclusion, and anterior crossbite; and (3) ANB less than 0°, Wits appraisal less than −2 mm (corrected cephalometric tracing technique was applied for patients with a functional shift ), and distance from Point A to nasion perpendicular less than 0 mm. The exclusion criteria were (1) previous orthodontic treatment; (2) other craniofacial anomalies, such as cleft lip and palate; and (3) maxillary dentition unsuitable to bond a hyrax expander. Consent was obtained from the parents or guardians of the patients before their recruitment.

Interventions

Banded and soldered hyrax palatal expanders were used for the patients in both groups ( Fig 1 ). Two maxillary first molars and 2 deciduous molars were banded. For the patients whose maxillary first premolars had fully erupted, 2 maxillary first molars and 2 maxillary first premolars were banded. The anterior extension arms were bonded to the lingual surfaces of the maxillary anterior teeth to help palatal expansion. The protraction hooks were designed around the maxillary canine area. All expanders were manufactured by the same orthodontic technician.

Hyrax expander.

After the expanders were banded, the parents or guardians of the patients were taught how to activate and deactivate the expanders. Patients in the control group (RPE) were treated with RPE for 1 week. The expander was activated 4 times per day (1 mm per day) for 7 days. Patients in the intervention group (RPE/C) were treated with RPE/C for 7 weeks. Both the expansion and the constriction rates were 4 times per day (1 mm per day). The sequence was 7 days of expansion, 7 days of constriction, 7 days of expansion, 7 days of constriction, 7 days of expansion, 7 days of constriction, and then a final 7 days of expansion. The patients were instructed to come to the office for follow-up every week to ensure correct operation. The width of the expanders was measured during the weekly visits. Sometimes, inadequate expansion or constriction was found. Then the parents or guardians would be tested on their skill of activation or deactivation, and further guidance would be provided at the same time. Any damaged expander was required to be repaired (or replaced) and rebanded within 2 days.

After RPE or RPE/C, the patients in both groups were treated with facemask maxillary protraction. A 1-piece facemask with an adjustable anterior wire and hooks for elastics was used ( Fig 2 ). The elastic direction was 15° to 30° downward from the occlusal plane, delivering a force between 400 and 500 g per side. The patients were instructed to wear the facemask for at least 14 hours a day. The patients’ records of hours showed that facemask wearing time was about 11.5 hours on average in both groups. The patients were instructed to follow up every month. There were no purposive differences between the 2 groups during follow-ups, such as instructions. The protraction force was tested each time and adjusted if needed, as was the protraction direction. Any damaged facemask was replaced immediately. Any damaged expander was required to be repaired (or replaced) and rebanded within 2 days. The patients who did not wear their facemasks for enough time were educated by one of the authors (W.L.) during their monthly follow-ups. The same cooperation and education process was used for these patients and their parents or guardians, including an explanation of the importance of wearing the facemask and guidance about living normally with the facemask.

One-piece facemask.

The treatment completion criterion was a positive overjet with a Class II or a Class I molar relationship. No retention appliances were used after treatment. All clinical treatments were performed by the team of W.L. and Y.Z.

Outcomes (primary and secondary) and any changes after trial commencement

The primary outcome was the degree of maxillary forward movement after treatment (A-VRL). The secondary outcomes were the changes of the other cephalometric variables and the treatment time.

Lateral cephalometric radiographs were taken at the beginning and end of treatment. These radiographs were hand-traced and measured by an investigator (W.L.). Traditional cephalometric measurements were used to describe changes before and after treatment. An x-y coordinate system was also set up. The horizontal axis (CFH) was the sella-nasion line rotated 7° downward, whereas the vertical axis (VRL) was constructed by passing a line perpendicular to the horizontal axis through sella. The distances from several landmarks to the horizontal and vertical axes were measured. A total of 22 cephalometric variables were used to evaluate the craniofacial, dental, and soft tissue changes ( Fig 3 ). There were no outcome changes after trial commencement.

Cephalometric measurements.

Sample size calculation

Because no studies in orthodontic literature used the same protocol as we did in this study, the sample size was estimated by G∗Power (version 3.0.8) according to the previous study on the 2-hinged expander RPE/C and intraoral maxillary protraction (95% power; 5% significance level; 2-tailed). The sagittal movements of Point A were 2.6 ± 1.5 mm in the RPE group and 5.8 ± 2.3 mm in the RPE/C group. A minimum sample size of 11 in each group was required to detect a significant difference between the groups. The sagittal movements of ANS were 2.1 ± 1.3 mm in the RPE group and 4.8 ± 2.5 mm in the RPE/C group. A minimum sample size of 16 in each group was required to detect a significant difference between the groups. Therefore, 16 was the minimum sample size of each group. The sample size was increased by 40%, resulting in 22 patents in each group, to account for dropouts and the use of a different protocol from the previous study.

Interim analyses and stopping guidelines

The study would stop if serious harm was observed during the treatment and research.

Randomization (random number generation, allocation concealment, implementation)

Patients were recruited and allocated to either the control group (RPE) or the intervention group (RPE/C) according to their participation sequence. Simple randomization was applied. Randomization was carried out using a random number table from a medical statistics textbook. Starting from the fourth line, first number of the table, a series of random numbers was obtained from left to right and up to down. The randomized numbers were divided by 4, and the remainder was taken. The remainder numbers 1 and 2 were allocated to the RPE and RPE/C groups, respectively, ignoring the other numbers. After all 44 numbers were obtained, there might be a size imbalance between the 2 groups. The following randomized number would be divided by 4 and the remainder (n) would be taken (if divided exactly, n = 4). For example, if n = 3, we would move the third subject of the larger group to the other group. We would repeat this process until the 2 groups were balanced. In this study, adjustment was not needed, since the 44 patients were assigned to the 2 groups with a 1:1 ratio. A table with recruitment sequential numbers, paired selected randomized numbers, and group numbers was prepared. Forty-four envelopes containing the subjects’ information were used to ensure allocation concealment from the researchers. Sequential numbers were written on the envelopes. One investigator (W.L.) was responsible for generating and implementing the random allocation process, enrolling participants, and opening the envelopes in sequence.

Blinding

Blinding of either orthodontists or participants was not possible because the treatment protocols in 2 groups were different. Operators and patients would easily know which group the patients were in. However, blinding was used during the cephalometric analysis. All cephalometric films were deidentified by opaque tape and replaced by research numbers, and then disarranged before tracing. The investigator who was responsible for measuring (W.L.) did not know the grouping of the cephalometric radiographs, which looked similar.

Statistical analysis (primary and secondary outcomes, subgroup analyses)

All statistical analyses were performed using software (version 18.0; PASW Statistics, Chicago, Ill). Descriptive statistics included the means and standard deviations of age, total treatment time, protraction time, cephalometric values, and changes of values in each group. The 1-sample Kolmogorov-Smirnov test was used to test the normality of the distributions for all times and differences of the cephalometric values. Independent t tests were used to compare the 2 groups, including total treatment time, protraction time, and changes of cephalometric values; t ′ tests were used for the values that showed “equality of variances not assumed” between the groups (SNA and UI-VRL). The level of significance was 0.05, with a 2-tailed test.

Error of the method

All point locations and measurements were double-checked carefully at the end of the initial tracing and measuring. Not all radiographs were traced twice. However, to evaluate the errors of tracing, 10 radiographs were chosen at random. Tracing, locating, and measuring were done twice, 2 weeks apart, and then subjected to Dahlberg’s formula. The method error was determined using Dahlberg’s formula, where n = 10, and d was the difference between the measurements of cephalometric values at 2 time points. The method error did not exceed 0.59° for any angular measurement or 0.66 mm for any linear measurement. The values indicate that this analysis was reliable compared with other estimations of technical error.

Trial design and any changes after trial commencement

The study was a parallel-group, randomized, active-control trial with a 1:1 allocation ratio. There were no changes after trial commencement.

Participants, eligibility criteria, and setting

Approval for this randomized clinical trial was obtained from Peking University Medical Science Research Ethics Committee (IRB00001052-07094). Participants were recruited from the Department of Orthodontics at Peking University, based on the following inclusion criteria: (1) 7 to 13 years old before treatment with a midface soft tissue deficiency; (2) fully erupted maxillary first molars, Class III malocclusion, and anterior crossbite; and (3) ANB less than 0°, Wits appraisal less than −2 mm (corrected cephalometric tracing technique was applied for patients with a functional shift ), and distance from Point A to nasion perpendicular less than 0 mm. The exclusion criteria were (1) previous orthodontic treatment; (2) other craniofacial anomalies, such as cleft lip and palate; and (3) maxillary dentition unsuitable to bond a hyrax expander. Consent was obtained from the parents or guardians of the patients before their recruitment.

Interventions

Banded and soldered hyrax palatal expanders were used for the patients in both groups ( Fig 1 ). Two maxillary first molars and 2 deciduous molars were banded. For the patients whose maxillary first premolars had fully erupted, 2 maxillary first molars and 2 maxillary first premolars were banded. The anterior extension arms were bonded to the lingual surfaces of the maxillary anterior teeth to help palatal expansion. The protraction hooks were designed around the maxillary canine area. All expanders were manufactured by the same orthodontic technician.

Hyrax expander.

After the expanders were banded, the parents or guardians of the patients were taught how to activate and deactivate the expanders. Patients in the control group (RPE) were treated with RPE for 1 week. The expander was activated 4 times per day (1 mm per day) for 7 days. Patients in the intervention group (RPE/C) were treated with RPE/C for 7 weeks. Both the expansion and the constriction rates were 4 times per day (1 mm per day). The sequence was 7 days of expansion, 7 days of constriction, 7 days of expansion, 7 days of constriction, 7 days of expansion, 7 days of constriction, and then a final 7 days of expansion. The patients were instructed to come to the office for follow-up every week to ensure correct operation. The width of the expanders was measured during the weekly visits. Sometimes, inadequate expansion or constriction was found. Then the parents or guardians would be tested on their skill of activation or deactivation, and further guidance would be provided at the same time. Any damaged expander was required to be repaired (or replaced) and rebanded within 2 days.

After RPE or RPE/C, the patients in both groups were treated with facemask maxillary protraction. A 1-piece facemask with an adjustable anterior wire and hooks for elastics was used ( Fig 2 ). The elastic direction was 15° to 30° downward from the occlusal plane, delivering a force between 400 and 500 g per side. The patients were instructed to wear the facemask for at least 14 hours a day. The patients’ records of hours showed that facemask wearing time was about 11.5 hours on average in both groups. The patients were instructed to follow up every month. There were no purposive differences between the 2 groups during follow-ups, such as instructions. The protraction force was tested each time and adjusted if needed, as was the protraction direction. Any damaged facemask was replaced immediately. Any damaged expander was required to be repaired (or replaced) and rebanded within 2 days. The patients who did not wear their facemasks for enough time were educated by one of the authors (W.L.) during their monthly follow-ups. The same cooperation and education process was used for these patients and their parents or guardians, including an explanation of the importance of wearing the facemask and guidance about living normally with the facemask.

One-piece facemask.

The treatment completion criterion was a positive overjet with a Class II or a Class I molar relationship. No retention appliances were used after treatment. All clinical treatments were performed by the team of W.L. and Y.Z.

Outcomes (primary and secondary) and any changes after trial commencement

The primary outcome was the degree of maxillary forward movement after treatment (A-VRL). The secondary outcomes were the changes of the other cephalometric variables and the treatment time.

Lateral cephalometric radiographs were taken at the beginning and end of treatment. These radiographs were hand-traced and measured by an investigator (W.L.). Traditional cephalometric measurements were used to describe changes before and after treatment. An x-y coordinate system was also set up. The horizontal axis (CFH) was the sella-nasion line rotated 7° downward, whereas the vertical axis (VRL) was constructed by passing a line perpendicular to the horizontal axis through sella. The distances from several landmarks to the horizontal and vertical axes were measured. A total of 22 cephalometric variables were used to evaluate the craniofacial, dental, and soft tissue changes ( Fig 3 ). There were no outcome changes after trial commencement.

Cephalometric measurements.

Sample size calculation

Because no studies in orthodontic literature used the same protocol as we did in this study, the sample size was estimated by G∗Power (version 3.0.8) according to the previous study on the 2-hinged expander RPE/C and intraoral maxillary protraction (95% power; 5% significance level; 2-tailed). The sagittal movements of Point A were 2.6 ± 1.5 mm in the RPE group and 5.8 ± 2.3 mm in the RPE/C group. A minimum sample size of 11 in each group was required to detect a significant difference between the groups. The sagittal movements of ANS were 2.1 ± 1.3 mm in the RPE group and 4.8 ± 2.5 mm in the RPE/C group. A minimum sample size of 16 in each group was required to detect a significant difference between the groups. Therefore, 16 was the minimum sample size of each group. The sample size was increased by 40%, resulting in 22 patents in each group, to account for dropouts and the use of a different protocol from the previous study.

Interim analyses and stopping guidelines

The study would stop if serious harm was observed during the treatment and research.

Randomization (random number generation, allocation concealment, implementation)

Patients were recruited and allocated to either the control group (RPE) or the intervention group (RPE/C) according to their participation sequence. Simple randomization was applied. Randomization was carried out using a random number table from a medical statistics textbook. Starting from the fourth line, first number of the table, a series of random numbers was obtained from left to right and up to down. The randomized numbers were divided by 4, and the remainder was taken. The remainder numbers 1 and 2 were allocated to the RPE and RPE/C groups, respectively, ignoring the other numbers. After all 44 numbers were obtained, there might be a size imbalance between the 2 groups. The following randomized number would be divided by 4 and the remainder (n) would be taken (if divided exactly, n = 4). For example, if n = 3, we would move the third subject of the larger group to the other group. We would repeat this process until the 2 groups were balanced. In this study, adjustment was not needed, since the 44 patients were assigned to the 2 groups with a 1:1 ratio. A table with recruitment sequential numbers, paired selected randomized numbers, and group numbers was prepared. Forty-four envelopes containing the subjects’ information were used to ensure allocation concealment from the researchers. Sequential numbers were written on the envelopes. One investigator (W.L.) was responsible for generating and implementing the random allocation process, enrolling participants, and opening the envelopes in sequence.

Blinding

Blinding of either orthodontists or participants was not possible because the treatment protocols in 2 groups were different. Operators and patients would easily know which group the patients were in. However, blinding was used during the cephalometric analysis. All cephalometric films were deidentified by opaque tape and replaced by research numbers, and then disarranged before tracing. The investigator who was responsible for measuring (W.L.) did not know the grouping of the cephalometric radiographs, which looked similar.

Statistical analysis (primary and secondary outcomes, subgroup analyses)

All statistical analyses were performed using software (version 18.0; PASW Statistics, Chicago, Ill). Descriptive statistics included the means and standard deviations of age, total treatment time, protraction time, cephalometric values, and changes of values in each group. The 1-sample Kolmogorov-Smirnov test was used to test the normality of the distributions for all times and differences of the cephalometric values. Independent t tests were used to compare the 2 groups, including total treatment time, protraction time, and changes of cephalometric values; t ′ tests were used for the values that showed “equality of variances not assumed” between the groups (SNA and UI-VRL). The level of significance was 0.05, with a 2-tailed test.

Error of the method

All point locations and measurements were double-checked carefully at the end of the initial tracing and measuring. Not all radiographs were traced twice. However, to evaluate the errors of tracing, 10 radiographs were chosen at random. Tracing, locating, and measuring were done twice, 2 weeks apart, and then subjected to Dahlberg’s formula. The method error was determined using Dahlberg’s formula, where n = 10, and d was the difference between the measurements of cephalometric values at 2 time points. The method error did not exceed 0.59° for any angular measurement or 0.66 mm for any linear measurement. The values indicate that this analysis was reliable compared with other estimations of technical error.