INTRODUCTION

The Herbst appliance is widely used to correct Class II malocclusions.1–9 In addition to distal movement of the upper teeth and mesial movement of the lower teeth, the appliance stimulates growth of the mandible and maxilla and probably also corrects the direction of their growth.6,10,11

The effect of Herbst appliance treatment has previously been investigated by means of the Pancherz cephalometric method.5 However, unlike more conventional cephalometric measures, this specific cephalometric approach does not address dimensional changes or the inclination and location of each dental and skeletal part of the facial complex. Consequently, the effects exerted by the Herbst appliance on each dentoskeletal structure are largely unknown and difficult to compare with the effects of other orthopedic devices that are generally evaluated with more conventional cephalometric parameters. Last, the outcome of Herbst treatment in terms of sex differences is unknown and most of the maxillary and mandibular basal bone modifications arising from its use have not been compared with European standard growth samples.

The aims of this study were to determine (1) the effects of Herbst treatment by means of conventional cephalometric variables and eliminating the effects of natural growth and (2) whether gender affects response to treatment.

SUBJECTS AND METHODS

A retrospective sample of 50 patients, 25 males (aged 10.7–14.5 years) and 25 females (aged 10.7–14.3 years) was selected among Class II malocclusions treated with the Herbst appliance by one author (HP) between 1977 and 1985 at the Orthodontic Department, University of Lund (Sweden). All patients presented a full molar Class II relationship at the beginning of treatment (mean ANB = 6.7 degrees, ranging from 2.5 degrees to 10.5 degrees) and a full molar Class I (or even Class I overcorrection) at the end of 6–8 months of Herbst therapy. Preformed A4-sheets were superimposed on pre- and posttreatment lateral head films, and 50 selected cephalometric landmarks were marked on each sheet. The sheets were entered into a computer system and analyzed by means of software expressly developed by one of the authors (CM).12

To assess the combined error of the method (manual digitizing, telefax transmission of data, and computer processing of the cephalometric landmarks), the whole procedure was performed twice by 2 different operators in 12 (6 pre- and 6 posttreatment) female and 12 (6 pre- and 6 posttreatment) male randomly selected patients. The following formula was used to determine the method error (ME) as ME = (Σd2/2n)^1/2, where d is the difference between 2 recordings of a pair and n is the number of double recordings. The mean error was 1.24 mm for linear and 1.10 degrees for angular measures. Radiographic magnification of 7% was corrected for all linear cephalometric measurements.

The mean, standard deviation (SD), and standard error of all cephalometric variables were calculated. Pretreatment and posttreatment variations of both linear and angular measures were statistically analyzed by means of a paired-samples t-test. To counteract the effect of growth, a second paired-samples t-test was performed with the means of pretreatment and posttreatment z-scores of every cephalometric variable compared with the appropriate age- and sex-matched norm of Bhatia-Leighton standard.13 Posttreatment values following more than 6 months of treatment were standardized on the next age-matched norm.

The z-score is probably the most adequate statistical definition for the reader. However, it is important to note that this definition is a little different from what we really did in customizing our software. Indeed, while a z-score is generally considered the distance from the midpoint of normal range, in our software, it was the distance from the first standard deviation, in other words, the distance from the upper (or lower) border of the normal range.

When the data are transformed into this standardized form by our customized software, the amount of growth effect on the results was automatically excluded. While the first t-test estimated total variation of cephalometric variables (including growth effects), the second one estimated that part of variation mainly due to Herbst appliance (excluding or at least reducing growth effects). The levels of significance were P < .001 (***), P < .01 (**), and P < .05 (*); P = .05 was not considered statistically significant.

RESULTS

Males

There was no significant total variation between pre- and posttreatment maxillary sagittal measurements (Table 1) while z-score variations (Table 2) were at the border of significance (P^{z-Sc BA-ANS} = .038; P^{z-Sc SNA} = .053). Conversely, while vertical maxillary skeletal changes were observed in terms of total variation (P^UAFH = .046; P^UPFH = .056), no such changes were observed in terms of z-scores.

TABLE 1. Paired Samples t-Test of Pre- and Posttreatment Cephalometric Variables for Males^a

View Fullscreen

TABLE 2. Paired Samples t-Test of Pre- and Posttreatment z-Scores for Males^a

View Fullscreen

Mandibular ramus height and mandibular corpus length were significantly increased (P^CD-GO < .001; P^AR-GO < .002; P^GO-ME < .001; P^Xi-PM < .001), but these changes were not confirmed in terms of z-scores (P^{z-Sc CD-GO} > .09; P^{z-Sc GO-ME} > .13; P^{z-Sc Xi-PM} > .09). Only AR-GO z-score appeared to be increased (P^{z-Sc AR-GO} = .03).

There were significant changes in total variation of SNB and ANB angles (P < .001). ANB and Xi-CF-PTV angles were also significantly changed in terms of z-score (P < .001). No significant changes were detected for the gonial angle.

Females

None of the investigated sagittal maxillary measures, either in terms of total variation or in z-scores, significantly differed between pre- and posttreatment (Tables 3 and 4). Similarly, no differences were detected in vertical maxillary variables. Only mandibular ramus height increased (P^CD-GO = .018; P^AR-GO = .057). The z-score of the increased AR-GO was at the border of significance (P^{z-Sc AR-GO} = .056).

TABLE 3. Paired Samples t-Test of Pre- and Posttreatment Cephalometric Variables for Females^a

View Fullscreen

Also, females showed significant changes of total variation of SNB and ANB angles (P < .001). ANB angle significantly changed also in terms of z-score (P = .001). No significant changes were observed for the gonial angle.

DISCUSSION

In clinical orthodontics, the correction of the facial complex is the result of the final rearrangement of many cephalometric variables.14 Consequently, the positive or negative variation of any one measure can never explain by itself the correction of the whole face. Conventional cephalometrics is not sufficiently powerful to take into account the interactions between measurements. In addition, it does not allow an effective comparison between angular and linear measurements (as well as different intensity of variation between the same unit of measure) or the correlation between vertical and sagittal derangements. In other words, at the moment, the state of the art does not address the multivariated interpretation of cephalometric variables.15

The procedure described here overcomes the limits of conventional cephalometrics. We investigated the effects of Herbst appliances using a large number of conventional cephalometric parameters and a combination of several cephalometric techniques. Only those linear and angular skeletal measurements that have been previously studied in European samples were selected and incorporated in the procedure, expressly to compare the data of our treated sample with the European norms, matched for sex and age (Figure 1). Data were easily managed by means of expressly devised software, the output being in millimeters, in angular degrees, and in z-score units. The output of linear and angular variables in terms of z-scores was intentionally introduced in the procedure, with the aim of making them truly comparable. Indeed, different from any pre- and posttreatment variation of parameters, z-score changes are adimensional changes with which any kind of parameter may be compared and scaled to each other (irrespective of whether pounds, angle degrees, gallons, or inches are being compared). That is the essential precondition to be introduced to multivariated interpretation of these variables.

View Figure

• Download as Powerpoint
• Download Figure

In addition, pre- and posttreatment cephalometric variations evaluated in terms of z-score allowed us to establish to what extent changes in skeletal measures depended on growth. In other words, by matching our absolute values for sex and age with an acknowledged European standard sample like the Bathia-Leighton group, we also avoided (and to some extent excluded) the effects of natural growth on our results.

With this assumption, it could be argued that the effects of growth can appear constant for the sample, meaning that pre- and posttreatment z-scores should be the same if growth were the only factor. Actually we assumed that growth effects were constant, but only within each single age- and sex-group of norm. In other words, it is more appropriate to say that, if growth was the only factor, z-scores should always be null or zero but not always the same. Indeed, it can easily be understood that, because every age- and sex-group of a normal sample (as well as of a pathological one) preserves its different range of normal values with its own different means and own different minimum and maximum values, we can consider constant ordering of the sample by different age- and sex-classes or groups.

Transforming data in this standardized form of z-score, these class differences were absolutely respected and, on the basis of x-rays and birthdates of patients (±6 months), pretreatment and posttreatment values were automatically ordered by age- and sex-classes. Posttreatment values following more than 6 months of treatment were standardized on the next age-matched norm. Although approximate, this annualizing method of z-scores offered us many relevant advantages, listed in Table 5. We do not claim that this is the ideal method for clinical investigations, but it is a reliable tool to discuss in terms of a multivariated interpretation of cephalometric variables.15

TABLE 5. Advantages of Annualized z-Scores

View Fullscreen

Another objection to this method is that normal growth can be easily mistaken with treatment effects. The growth pattern of our treated skeletal Class II sample was significantly lower than normal growth, and this is the main reason why the increasing effect of normal growth on measurements could often overlap the treatment effects. To some extent, our sample measurements were not properly matchable or comparable with the Bathia-Leighton norms because of their different growth rate.

Otherwise, in every cephalometric variable of our Herbst-treated sample, the slope of pre- and posttreatment mean variation was always greater than that of the Bathia-Leighton standard (ie, always greater than the slope of any untreated skeletal Class II sample; see, eg, the Droschl16 Class II sample). It means that our sample treated with Herbst therapy, although interacting with the natural lower pathologic rate of growth of skeletal Class II patients, was able to activate more than a normal rate of growth, inducing a faster and unmistakable increase of bone. Such relevant facts can only be due to treatment effects and not to the naturally reduced growth in this kind of patient.

CONCLUSIONS

The analysis of changes occurring in maxillary and mandibular cephalometric variables, especially considered in terms of z-score, gives a measure of the selective skeletal effects consequent to Herbst treatment. The appliance favorably affected sagittal and vertical maxillary jaw base position only in males. A large mandibular forward repositioning occurred in both sexes and mandibular corpus length increased in a large part of the sample. Through the specific range of age in our sample, the Herbst appliance was an effective vertical posterior orthopedic activator of ramus height.