gms | German Medical Science

The popularity of aesthetic breast augmentation has been increasing worldwide, and millions of women have undergone breast implantation for aesthetic or medical reasons [1], [2]. Breast implant selection and outcome analysis is commonly conducted based on the surgeon’s subjective experience [3]. The same applies to aesthetic and breast reconstruction procedures [4]. Increasingly more diverse types of implants with multiple variations in sizes and shapes are becoming available, and how well surgeons are doing in choosing implants by subjective estimation remains unclear. Additionally, objective assessments of treatment outcomes are lacking [3].

Some manufacturers have offered implant sizers for temporary insertion ahead of the final implantation to help surgeons choose the most appropriate implants. However, these sizers are expensive and not readily available. Therefore, other types of implant sizers, even surgical gloves, have been described [5]. Other aids include tissue expander implants that can be adjusted as necessary and exchanged for permanent implants at a later stage [4]. Because the current subjective approach is based on physical examination and visual estimates of breast size, the overall success of breast reconstruction has reportedly remained impaired by the inability of plastic surgeons to objectively judge breast volume and shape [4]. Modern imaging technology has offered newly developed software programs to support surgeons’ performances and implant choices [6]. Although three-dimensional imaging has been judged as a valuable adjunct to tissue-expander breast reconstruction [4], it remains costly, cumbersome, and limited in application.

Overall, the degree of error associated with the subjective method of breast volume estimation remains unknown [7]. In the clinical setting, surgeons are frequently challenged to determine the most appropriate implant size preoperatively and estimate whether the chosen size might meet the surgical objective intraoperatively. Postoperatively, patients often complain about undesired volume differences between the two breasts; however, how well surgeons are doing in subjectively judging breast implant volumes remains unknown.

In the field of breast surgery, a volume of about 50 cc has commonly been viewed as the volume difference just visible with the human eye based on subjective examination [8]. Studies on subjective error assessment have been published [9], [10]; however, the degree of error in subjective implant volume estimation remains to be determined.

Surgeons must consider multiple factors beyond volume when performing breast surgery. Shape and symmetry are equally important. Although mathematically based morphological shape analysis has been described [11], this technique has not been introduced to the clinical setting because of the complexity of the calculations.

Shape analysis was beyond the scope of the current study, which focused on volume. The benefit of studying error in implant volume estimation lies in the availability of an accurate measure for comparison and the novelty of quantifying this error. Therefore, the present study was performed to investigate how well surgeons are doing in estimating breast implant volumes by a subjective blinded method. The accuracy and reproducibility of these estimations were examined.

Nine silicone breast implants of known volume were used by 16 examiners for subjective, blinded implant volume estimation. The implants (Mentor Worldwide LLC, Santa Barbara, CA, USA) were anatomically shaped and chosen based on their availability (all were used as demonstration tools in the clinic). The examiners were allowed to take the implants in their hands, but were not allowed to look at the stamp on the reverse side of the implants. The figures on the stamps were taken as the true breast implant volumes and used for comparison with the subjective estimations. The examiners were divided into two groups of equal size: experienced examiners (consultant breast surgeons) and unexperienced examiners (trainees). The experienced examiners had passed their specialist exam and had experience in dealing with breast surgical cases for around ten years or more. The unexperienced examiners were all still in their training with limited exposure to breast surgical cases who had not jet conducted breast surgical cases on their own. Fourteen examiners (seven in each group) repeated the tests about 2 weeks later. Two examiners dropped out of the study because of pregnancy. The data obtained from the subjective estimations were recorded twice to assess the reproducibility of the subjective estimations. The examiners did not receive notice about their previous extimations and their differences to the true volumes.

Statistical evaluation was conducted using STATISTICA software (StatSoft, Inc., Tulsa, OK, USA). The mean, minimum, maximum, and standard deviation (SD) of all data were assessed, and estimations were compared with true volumes (Table 1 [Tab. 1]). Mean and total volume differences were calculated in milliliters and percentages. The first and second estimation tests were termed M1 and M2, respectively. The correlation and regression coefficients were calculated to determine the size of the deviation between estimations and true implant volumes. A correlation coefficient (R) of 1 (range, −1 to +1) was taken as a perfect correlation, and larger R values indicated stronger correlations between two variables. The Wilcoxon test was used to determine whether two variables were significantly different from each other. The Mann-Whitney U-test was used to compare the experienced and unexperienced examiners.

The implant volumes were underestimated by a mean ± standard deviation of 61.6 ± 58.1 cc (21.9% ± 20.1%) in M1 and 37.9 ± 59.1 cc (13.6% ± 20.7%) in M2. Absolute differences independent of volume overestimation or underestimation were 77.3 ± 36.9 cc (27.2% ± 12.8%) in M1 and 58.1 ± 41.3 cc (20.6% ± 14.6%) in M2. In terms of absolute figures, the experienced doctors showed significantly better results than did the unexperienced doctors in M1, at 52.6 ± 26.3 cc (18.8% ± 9.4%) versus 102 ± 28.9 cc (35.5% ± 10.3%), respectively. While the majority of the estimations were underestimations in view to the true figures the calculation of the absolute differences took any estimation differences from the true figures into account independent if the estimations were too low (negative prefixes) or too high (positive prefixes); in contrast to this the mean differences of the group of examiners resulted when mathematically the positive and negative prefixes were equalized and the resulting estimation differences from the true values for the group of examiners were obtained. Calculation of the correlation and regression coefficients revealed that overall, the unexperienced examiners showed increasing deviations with increasing implant volumes. One examiner showed this quite clearly, revealing a systematic error in the underestimation of the volumes (Figure 1a [Fig. 1]). A close correlation was demonstrated by an R of −0.9. This examiner underestimated the implant volume by an average of 60 cc for each additional 100 cc. Another examiner also demonstrated a clear but slightly weaker correlation with an R of –0.8. This examiner underestimated the implant volume by an average of 30 cc for each additional 100 cc. However, this systematic error was not true for all examiners. One experienced examiner showed an error distribution without any rule (Figure 1b [Fig. 1]). Therefore, the correlation and regression coefficients were smaller for this experienced examiner. Overall, both the experienced and unexperienced examiners performed better in M2, showing a decrease in estimation errors; however, this improvement was not statistically significant.

The data for both the experienced and unexperienced examiners are displayed in Table 2 [Tab. 2]. The results of comparison of both groups using the Mann-Whitney U-test are given in Table 3 [Tab. 3].

The mean difference between the estimated and true volumes in M1 was −84.3 ± 66.3 cc for the unexperienced doctors and −39.0 ± 40.9 cc for the experienced doctors (p=0.065) (Figure 2a [Fig. 2]). The absolute difference between the estimated and true volumes without consideration of volume overestimation or underestimation was 102.0 ± 28.9 cc for the unexperienced examiners and 52.6 ± 26.3 cc for the experienced examiners (p=0.010, gray shaded area in Table 3 [Tab. 3]) (Figure 2b [Fig. 2]). Similarly, the mean difference in percentage between the estimated and true volumes was 14.1% ± 14.6% for the experienced doctors and 29.6% ± 22.7% for the unexperienced doctors (p=0.082) (Figure 3a [Fig. 3]). The absolute difference in percentage without consideration of volume overestimation or underestimation was 18.8% ± 9.4% for the experienced doctors and 35.5% ± 10.3% for the unexperienced doctors (p=0.015, gray shaded area in Table 3 [Tab. 3]) (Figure 3b [Fig. 3]).

During M2, the differences (both cc and %) between the estimated and true volumes were smaller in both groups of examiners, and the improvement from M1 to M2 was clearer among the unexperienced doctors. Therefore, the differences were no longer statistically significant (all p>0.05). Again, the estimated volumes were more accurate among the experienced doctors. The mean difference between M1 and M2 was −25.2 ± 43.2 cc for the experienced doctors and −45.5 ± 79.6 cc for the unexperienced doctors (p>0.05) (Figure 4 [Fig. 4]). The mean absolute difference between M1 and M2 was 50.2 ± 15.3 cc for the experienced doctors and 76.1 ± 50.1 cc for the unexperienced doctors (p>0.05). The mean difference in percentage between M1 and M2 was −10.8% ± 17.4% for the experienced doctors and −16.7% ± 38.9% for the unexperienced doctors (p=0.80). Finally, the mean absolute difference in percentage between M1 and M2 was 19.9% ± 8.9% for the experienced doctors and 37.1% ± 20.6% for the unexperienced doctors (p=0.073).

The mean correlation coefficient between the estimation error and true implant volume was close to zero among the experienced doctors and −0.64 among the unexperienced doctors. This difference was statistically significant (p=0.038). This result indicates that among the unexperienced doctors, the estimation error increased quite reliably with increasing implant volumes (see Figure 1a [Fig. 1]); however, among the experienced doctors (see Figure 1b [Fig. 1]), there was no such correlation (Figure 5a [Fig. 5]).

The estimation error described in this context indicates the difference between the estimations and true volumes.

The regression coefficient among the experienced doctors was also close to zero (0.034); however, that among the unexperienced doctors was −0.28. This means that the increase in estimation error was much clearer among the unexperienced doctors (Figure 5b [Fig. 5]).

Similar results for the correlation and trend compared with the true volumes were found in M2, with particular improvement noted among the unexperienced doctors.

In the current study, blinded examiners estimated breast implant volumes. Regarding the clinical benefit of this study, surgeons need to be aware of their own error when estimating volumes in breast implant surgery. Further, the quantification of this error (average underestimation of 50 cc in the present study) might give surgeons some consideration and guidance when performing this type of surgery. Younger surgeons in particular should be conscious of higher rates of errors associated with unexperience and increasing breast sizes and might wish to seek options to support their performances. Overall, knowledge of possible errors might help surgeons to plan their performances, evaluate their own results, or raise interest in alternative, supportive, and objective methods.

The present study showed large inter- and intra-examiner variation in subjective volume estimations. This negatively impacts the validity of the subjective method. In reality, surgeons in many hospital settings face the need to choose, order, and use implants under blind estimation of the size or type of implants they might need. A stock of various implant sizers is not always available. The data presented herein prove and quantify the weakness of subjective estimations, especially those performed by less experienced doctors.

The fact that the examiners did better in the second test, which was performed several weeks later, raises questions. Could this be interpreted as a learning curve? Upon questioning, the examiners stated that they became newly conscious of the subject of estimation errors in breast implant surgery through their participation in the study. They reported that they followed up on their previous estimations to improve their own performance in breast implant surgery. Even though the results of the first test of estimations versus the true volumes were not provided to the examiners it might have happened that they informed themselves about implant sizes as breast implants for demonstration purposes have been available in the outpatient clinic. To investigate this finding further a third examination might have been beneficial even though the differences between the two tests were found not to be statistically significant and standard deviations did not improve. Interestingly, in the majority of cases estimations were too low. Especially among the unexperienced examiners, the exposure to breast implants prior to this study was rather limited. Nevertheless, a possible clinical benefit might result from awareness of the errors of the subjective method among surgeons by raising the focus to this subject. Whether the time between the two tests should be lengthened could certainly be discussed; however, this would require a new examination. We found no information in the literature regarding possible recommendations for time requirements between repeated subjective tests. Similarly, whether an increase in the cohort size might lead to different results and a decrease in familiarities/bias could also be discussed. In the current study, 16 examiners, 14 of whom repeated the test in equally sized groups, were deemed sufficient for the statistical evaluation. Furthermore, some critics might argue that the ability to accurately estimate breast implant volumes might not be a normal index of the surgeon’s capability to choose the appropriate breast implant for a patient who undergoes breast augmentation. In the present study, however, breast implants with known but obscured volumes were utilized, and the known volumes served as a gold standard for comparison. This allowed us to elucidate the subjective skills of surgeons in terms of their capability to perform subjective volume estimations.

Our finding that expert examiners performed better than unexperienced examiners is in line with other previously published data [7]. In the current study, the calculated mean differences were obtained by mathematical equalization of the positive and negative prefixes of the estimations, and absolute differences therefore seem to display the variations in estimations more clearly. Based on this and the possible learning curve of the second test, it seems that the absolute differences of the first test most accurately reflect the results of the subjective estimation.

Patients must be advised about the existing calculated error in subjective breast assessments as measured and quantified versus a gold standard. When discussing possible asymmetries in breast surgery, surgeons and patients need to be aware that there are possible limits to what can be perceived by visual measures. Subjective estimations should not be used alone in breast surgery, but should instead be supported by objective methods such as the use of implant sizers or objective measurement methods if possible [12]. Nevertheless, discussions about the ability to correct breast asymmetry continue, and the role of three-dimensional imaging has been judged as questionable [13]. Therefore, a study such as that presented here, involving quantification of the error of the subjective method, appears to be even more important. While implant sizers were introduced to the market before the use of sometimes costly objective breast assessment methods such as three-dimensional imaging, they have not yet found widespread application in clinical practice. Especially for unexperienced surgeons, this newly emerging technology may be a helpful tool with which to support clinical breast assessments and plan and perform breast surgery, even more so when dealing with larger volumes as objectively calculated herein. Breast volume measurements have been considered to support better surgical planning and implant selection, and the preoperative application of magnetic resonance imaging (MRI) for assessment of the total breast volume and tumor volume has been suggested [14]. Additionally, determination of the “relative breast volume” in breast augmentation between the implant size and total breast volume has been suggested to play a role in, for example, consequent sensibility alterations [15]. The accuracy of MRI for total breast volume determination in comparison with mastectomy specimens has also been described [16]. Nevertheless, cost issues have prevented routine application of MRI to date. Alternatively, the BREAST-V has been propagated. This is a unifying predictive formula for volume assessment of breasts of various sizes and presents a less expensive alternative based on the input of 10 anthropomorphic breast measurements [17]. Time will reveal whether this tool is helpful and easy to use by various surgeons in routine clinical practice, not just research teams, and precision will need to be determined by independent validation.

Previous reports have addressed the weakness of subjective judgment methods. A review of the literature on subjective ratings of aesthetic breast outcomes showed that the use of qualitative subjective scales is not widespread because they are typically vague and have low intra- and interobserver agreement [18], [19]. Moreover, the lack of objective evaluation tools and the ongoing need for visual subjective breast assessments in breast surgery have been described [12]. However, qualitative rating scales and quantitative evaluation of errors in subjective estimations of breast implant volumes must be differentiated, and the latter has been newly presented in this study. While Sigurdson and Kirkland [8] stated that 50 cc is the breast volume discernable by the human eye and accounts for asymmetry in breast evaluation based on subjective assumption, we objectively calculated that the volume discernable by the human eye is actually around 70 cc with a volume underestimation of approximately 50 cc in breast implant assessment. Worse results were obtained by less experienced examiners. Knowledge of the limitations of visual volume estimation puts the process of subjective judgments of breast symmetry into perspective.

Subjective and objective breast assessment were previously examined by subjectively judging overall breast symmetry by comparing the Harris score [20] with an objectively calculated asymmetry score. The Harris score is used for qualitative assessment, and results are judged as excellent, good, fair, or poor results (score of 4 to 1, respectively). In that study, the errors, limitations, and application of the subjective method were presented. Weaknesses were comparable to those in the currently presented study, adding to the previous findings in the quantitative field.

In current national and international audits, outcome reporting mechanisms are gaining increasing importance, and robust breast device registries are being demanded [21]. However, outcome data that indicate the degree to which the aesthetic demands have been met by the surgical objective (meaning, for example, how well the desired volume, shape, or symmetry has been achieved in breast augmentation surgery according to objective measurements) are also required [3]. The current study data revealed considerable error in subjective estimations. Surgeons seem to have difficulty visualizing the final appearance of the breast and determining the optimal volume required to achieve the desired result before finally deciding which implants should be used. Besides the breast volume, the overall breast appearance is also determined by shape. However, shape cannot be as easily quantified and used for comparison with the subjective method. A different study would have been required to compare the objective measurements of shape with subjective estimations, which was beyond the scope of the present investigation. Additionally, the present investigation was not performed to determine the optimal criteria for surgical planning, but to determine the error in subjective implant volume estimation. Implant sizers currently appear to remain necessary before performing breast implant surgery, and their cost must be taken into consideration when planning surgery. The increasing estimation error with increasing implant volumes among the unexperienced doctors in this study indicates the even greater weakness of the subjective method in this group, and alternative objective methods of surgical planning and outcome evaluation are therefore particularly valuable for unexperienced doctors. In this context, we predict that three-dimensional imaging will become useful as an adjunct to the subjective method in daily surgical practice as soon as issues with costs, user friendliness, and precision are improved.

In summary the limitations and weaknesses of the subjective method of volume estimations are the limited accuracy and reproducibility, intra- and interobserver errors, varying and increasing errors with increasing volumes for estimation and lesser experience of the examiners. A connection of the content of the current study with a study investigating breast volume estimations in patients would be desirable. The later has recently been investigated, however in a group of patients after autologous breast reconstruction without implant insertion [22]. With the findings of the current study being available the next investigation would be to establish a connection between the estimation of breast implant volumes with breast volumes in patients after implant insertion in a consecutive study. Patients’ desires, conclusions from the investigation and surgical recommendations should be investigated.

These blinded subjective estimations of breast implant volume exhibited limited accuracy and reproducibility. Estimation errors were quantified and increased with increasing implant volumes. Experienced examiners performed considerably better than unexperienced examiners, and results of the second test were better in both groups. The subjective volume estimations deviated from the true volumes by around 70 cc (underestimated by around 50 cc).

Surgeons must be aware and patients advised about the error in subjective breast assessments as quantified versus a gold standard. Subjective estimations should not be used alone in breast surgery, but should be supported by objective methods such as the use of implant sizers or objective planning and measurement methods.

The authors acknowledge all study participants for kindly providing their estimations for data analysis. The authors also acknowledge Dr. Wolfgang Reimers for performing the statistical evaluation of the data and the Springer language editing service for providing help with language revision.

The authors declare that they have no competing interests.

The work was self-funded. The implants that were used were from the own stock of implants.

This article does not contain any studies with human participants or animals performed by any of the authors.