Outcome Measures in Facial Plastic Surgery: Title and subTitle BreakPatient-Reported and Clinical Efficacy Measures

The term outcomes study has become ubiquitous during the past 2 decades and has become increasingly confusing to physicians and researchers. The working definition of an outcomes study often depends on the context of its application; practically all studies that involve measurement of any kind could be termed an outcomes study. As described in past works, outcomes studies can be loosely classified into 1 of 3 broad categories: patient-reported outcomes, clinical or efficacy outcomes, or actuarial or financial outcomes.¹ The first 2 of these categories are perhaps more familiar and applicable to the physician-researcher and will be the focus of this study.

Patient-reported outcomes studies have become increasingly prevalent during the past decade, usually involving a prospective, observational study design using validated quality-of-life (QOL) instruments. With the increasing number of available QOL instruments, which can be either disease specific or more global, QOL instruments have become the gold standard for patient-reported outcomes and have largely replaced the traditional more simplistic ad hoc patient satisfaction questionnaires (eg, improved, not improved, or worsened) as they relate to the health-related condition.

Clinical efficacy outcomes studies are investigations that range from case studies to randomized controlled trials (RCTs) and attempt to measure in some objective fashion the effectiveness of the proposed intervention. The use of objective assessments, ranging from physician-rated scales to other tools or machinations, is common for many studies. However, objective assessments do not always undergo rigorous tests of reliability and are open to interpretation in terms of universal acceptance and availability. The purposes of the present study were to conduct a thorough review of existing outcome measures that have been used in patients undergoing facial plastic surgery, evaluate the validity and reliability of these measures, make recommendations on how to integrate existing measures into future studies, and, finally, discuss challenges in this evolving field.

Concept mapping was performed to determine significant issues related to the scope and topic of facial plastic surgery outcomes. From this mapping, a set of targeted interventions was developed (a condition for which each intervention might be used is given in parentheses), including blepharoplasty (facial aesthetic surgery), brow-lift (facial melanoma and nonmelanoma skin cancer), rhinoplasty (surgical scars), face-lift (facial disfigurement), soft tissue filler agent injection (facial nerve dysfunction), and botulinum toxin injection (hyperkinetic facial lines). Other interventions included facial rejuvenation, facial reanimation, facial cutaneous reconstruction, facial laser treatment, and facial trauma soft tissue and bony repair. Because of the broad scope of this topic, specific procedures and conditions were arbitrarily chosen; however, efforts were made to include the most common conditions and interventions in facial plastic and reconstructive surgery. Popular aesthetic interventions were identified from published national plastic surgery procedural statistical trends.² To narrow the scope of the review, procedures and conditions that deal with pediatric congenital anomalies, burns, and hair removal and restoration were excluded.

Once targeted interventions and conditions were identified, an OVID (OVID Technologies, New York, New York) software electronic search was performed to find relevant instruments in the following databases:MEDLINE (1950-September 2007), CINAHL (Cumulative Index to Nursing & Allied Health) (1982 to September 2007), and PsychINFO (1806 to September 2007) using keywords developed from this list. Limits were placed on the search to exclude non–English-language citations and nonhuman subjects. The OVID “remove duplicates” function was used to remove duplicate entries where the same study was listed in multiple databases.

Studies were examined for the presence of 1 or more of the targeted outcome instrument types. An instrument or measure was defined as a tool or system used to collect information about patient outcomes, which included questionnaires, scales, grading systems, or other tests that sought to quantify a health state or condition. All potentially relevant abstracts were evaluated, and full-text articles were evaluated if the abstract mentioned the use of an instrument but did not specify further information. Instruments were also screened for any kind of assessment of validity or reliability and whether they were specific for a particular intervention or condition.

Instruments were then analyzed for attempts to formally assess the reliability and/or validity of the measure in a population with 1 of the targeted conditions or interventions. Minimal validation was considered if an attempt was made to evaluate either the content (content validity) or performance of the measure against a specified standard or criterion (criterion-based validity). Examples of content validation included the use of literature reviews, expert consensus, or input from other health care professionals or focus groups to assess the measure. Examples of criterion-based testing included attempts to measure predictive testing, concurrent testing using case-controls or similar disease states, and/or convergent-discriminant testing against similar previously validated measures. An assessment of reliability was defined as an attempt to measure the internal consistency or reproducibility, to include interobserver and/or intraobserver reliability, or to include test-retest reliability. In addition, a notation was made if some assessment was performed for the sensitivity of the instrument to change or intervention. Informal ad hoc surveys, patient questionnaires, and gross observer descriptive reports were excluded. General health measures and scales that were not specific to targeted diseases or conditions were also excluded. Finally, instruments were included regardless of validation if they were part of established consensus guidelines, had been adopted by major medical or surgical organizations, or were widely used by multiple investigators.

Once a candidate instrument was identified, a search of references was performed to identify other instruments not detected in the initial search and follow-up searches were performed as needed to identify studies in which instruments were first described. An attempt was then made to find studies using identified instruments with the OVID “find citing articles” function. This was not meant to be an exhaustive search but rather to assist with a general assessment of the popularity and scope of applications for a particular instrument in the literature. Emphasis was placed on instruments that had been used more recently than 1995. The instruments were subdivided into 2 categories: patient-reported and clinical efficacy measures (observer-reported and objective measures). Observer-reported measures were defined as instruments in which the researcher or someone other than the patient was asked to grade a physical change or outcome. Objective measures were defined similarly to definitions given in previous reviews³ as methods that can numerically quantify physical changes using photographic or radiographic imaging, computer-assisted analysis, and/or anthropometric data.

We found 1911 articles and identified 68 different instruments for inclusion: 23 patient-reported, 35 observer-reported, and 10 objective measures. Three instruments exhibited some overlap between category type. Specifically, the Patient and Observer Scar Assessment Scale⁴ combined both patient and observer perceptions into a final result. This scale was classified with the patient-reported measures. In addition, 2 of the facial nerve grading scales, the Burres-Fisch Linear Measurement Index⁵ and the Nottingham Facial Grading System,⁶ used objective measurements to quantify results. The Nottingham scale further combined these measurements with an observer-reported component to describe synkinesis and secondary effects. These instruments were classified with other observer-reported facial nerve grading scales.

Twenty-three specific measures were identified in this category (Table 1). Nineteen (76%) of these were developed after 1997. An abundance of identified studies met the inclusion criteria for the final grading set; these studies used patient-reported outcomes such as nonvalidated patient satisfaction surveys or informal reporting of patient expressions of satisfaction.

Of the studies that displayed some validation or reliability testing, a wide variation was found in the degree of validation across measures. Some measures had undergone a rigorous process of item generation, reduction, and principal components analysis with pilot studies and subsequent modification to establish content and criterion-based validation, whereas others displayed only basic content validation. Likewise, reliability assessments were mostly made based on internal consistency measurements, usually using the Cronbach α, whereas others assessed stability using validation techniques such as test-retest results. Most of the patient-reported instruments were QOL assessments. Measures that were not QOL instruments included the Facial Appearance Sorting Test,²³ the Cassileth Scar Questionnaire,⁷ the Facial Line Outcomes Questionnaire,³⁷ and the Patient-Rated Facial Disfigurement Analogue Scale Questionnaire.⁶³ These instruments primarily focused on gauging the patient's perception of appearance. The House-Brackmann facial grading questionnaire⁵¹ is a translation of the House-Brackmann scale for grading facial nerve function into a self-reported questionnaire suitable for mail-out surveys.

Observer-reported scales and instruments (Tables 2, 3, 4, and 5) were more numerous than the other categories, with 35 instruments identified. Eighteen (51%) of these were developed since 1997. Similar to the patient-reported measures, many nonvalidated observer-reported outcome measures were found. Typically, these outcome measures involved surveys of independent observers or the independent subjective grading of photographs by observers or researchers. Observer scales that dealt with targeted interventions tended to be more specific, and although some sought to measure global outcomes, most focused on particular aspects of the condition they sought to measure. Several recurring subcategories emerged from this group: instruments to grade facial nerve function, facial wrinkles and hyperkinetic facial lines, and wound grading scores. These instruments are detailed separately in addition to the general list of observer scales. Twelve observer scales that did not fall into any of these categories were identified. Instruments and descriptions are given in Tables 2, 3, 4, and 5.

Facial nerve grading scales (Table 3) were generally well validated and differed primarily around the concept of a gross vs a regional assessment of different independent areas of facial function. There were 4 gross measurement scales and 8 regional scales. Three of the regional scales were also weighted to emphasize the relative importance of certain functions such as eye closure, and 5 scales evaluated synkinesis as an independent entity when deriving a final score. Although no data were available regarding formal validation of the Iowa Repaired Facial Nerve Recovery Scale,¹⁴⁹ this scale has been adopted by the 2003 Consensus Meeting on Systems for Reporting Results in Vestibular Schwannoma¹⁸² as a standard for facial nerve grading after facial nerve repair and was included for completeness.

We identified 5 scar and wound grading systems (Table 4). These instruments specifically dealt with scar appearance and have seen applications in laceration repair assessment and scar evaluation for facial defect repair after Mohs micrographic surgery. Most of the scales we identified were used to compare outcomes between tissue adhesives and traditional suture techniques for laceration and surgical wound repair.

Six scales dealt with the assessment of facial wrinkles and hyperkinetic facial lines (Table 5). The primary applications for these scales were in the assessment of the efficacy of minimally invasive procedures, such as botulinum toxin injection and soft tissue fillers. Scales were typically graded as a single numeric value, although there were subtle distinctions in both the areas of the face assessed and whether the face was assessed in repose with or without expression. All included scales had robust reliability testing.

Ten instruments (Table 6) sought to quantify outcomes using objective measures with photography-based anthropometry, computerized analysis of digitized photographs, and 3-dimensional computer reconstructions. Three measures, computer-assisted photographic analysis of rhytidectomy,¹⁸⁴ harmonized profiloplasty,¹⁹⁰ and laser facial mapping,¹⁹² were used to assess surgical outcomes with patients. Two instruments, the Phi Mask Analysis¹⁸⁸ and FacePrints¹⁸⁹ software, attempted to use computer analysis to achieve objective measures of attractiveness. Of the listed instruments, only the Phi Mask Analysis system had undergone reliability testing, and none of the measures had undergone formal validation assessments.

As attention has focused on improving patient outcomes, the use of validated reliable measures has become increasingly important for physicians and researchers wanting to accurately assess results and make clinical decisions. The use of validated tools allows for true comparisons among different interventions or different techniques within a single intervention. Such tools can also reliably assist in identifying good surgical candidates and approaches, as well as identifying patients unlikely to benefit from surgery. Finally, they can serve to help demonstrate treatment efficacy and establish legitimacy for third-party payers and government oversight bodies charged with the allocation of resources.¹⁹³

Our comprehensive review of the literature for existing outcome measures for an assortment of common facial plastic and reconstructive procedures has revealed a growing abundance of well-constructed and validated outcome measures. However, a wide range in the rigor of validation for some measures and controversy in the universal acceptance of some clinical efficacy measures still remain.

In the realm of facial plastic surgery, outcomes are often based on subjective impressions of the patient, which can span multiple psychological and psychosocial domains, making patient-reported outcome measures especially relevant. Traditionally, surgical patient-reported outcomes for facial aesthetic and reconstructive procedures relied on nonstandardized satisfaction questionnaires or other nonvalidated measures. Our review has demonstrated a substantial increase in available standardized validated QOL questionnaires—condition specific or more global—during the past 10 years. Some of these scales have been more rigorously psychometrically tested than others. Instruments such as the Nasal Obstruction Symptom Evaluation Scale,⁵⁵ the Facial Disability Index,²⁷ the Derriford Appearance Scale,¹⁷ the Patient and Observer Scar Assessment Scale,⁴ and the Skin Cancer Index⁶⁸ have undergone rigorous multistage validation using pilot testing and refinement. Other studies that described novel instruments limited testing to assessments of interobserver agreement or other limited assessments of reliability.

Our review has revealed that QOL instruments have become an important primary outcome measure of the success for certain procedures and have become the gold standard for health-related patient-reported outcomes. With the use of these instruments, investigators have attempted to identify demographic or clinical predictors for higher rates of success or failure. Moving forward, these scales should be attempted to be incorporated into future facial plastic outcomes studies if relevant to the goals of the study. If possible, investigators should attempt to first use existing instruments before embarking on a path to create a new measure. The creation of new QOL instruments is a laborious task; a well-constructed and psychometrically valid instrument often takes years to develop. The choice of the appropriate QOL instrument depends on the research question asked. If available, a condition-specific instrument tailored for a specific intervention may be more useful because such instruments are generally more sensitive to small but clinically relevant changes in outcomes. However, a more global scale, such as the Derriford Appearance Scale, may allow for comparisons among various conditions (eg, facial paralysis vs unfavorable scarring).

Furthermore, with future increased use of QOL scales, historical cohorts can potentially be created for possible comparison cohort studies. Other future studies using QOL instruments may consider the next steps in the interpretation of these scores (eg, delineating minimally important differences in scores and anchoring scores to objective measures or other meaningful measures of change).

The use of objective measures as a surrogate measure for success is not only appealing to physicians but also necessary as the foundation for potential RCTs or other interventional studies. One of the difficulties in measuring outcomes for many disease processes or conditions is the lack of a gold standard objective test for any given condition. However, in conditions in which specific outcome reporting has become an accepted or required practice (eg, facial nerve paralysis) or Food and Drug Administration efficacy trials (eg, facial wrinkles), we have observed a proliferation of similar instruments. This finding of focused clusters of outcome measures would seem to provide several benefits. First, as competing measurement systems are refined, they will better serve to define the gold standard for objective measurement. Second, the presence of multiple robust instruments allows physicians to answer different questions applied to the same condition or procedure. For example, the regional facial nerve grading scales may be more appropriate for monitoring facial nerve rehabilitation progress compared with other similar facial paralysis scales.¹⁹⁴

In contrast to the patient-reported outcome measures, still more voids and challenges exist in this area of outcome measurement. In addition, the clinical meaningfulness of any observer-based or objective surrogate measures for patient outcomes remains controversial for many of the common facial plastic and reconstructive procedures. As an example, blood pressure measurements, as an outcome measure for comparing 2 groups of patients receiving a placebo vs a new antihypertensive medication, would likely be accepted as an acceptable clinical efficacy outcome measure by most physicians and researchers. In contrast, how best to quantify the amount of acceptable lift for a certain forehead lift technique is often left to debate. Nevertheless, validated instruments are becoming more available for some conditions, although more validated clinical efficacy measures are needed and necessary to serve as common foundations for quantifying and accurately communicating conditions or specific results.

An exciting area of research and perhaps the next frontier in clinical efficacy measures is the use of computer-assisted measures for facial anthropometric features. Some of this research has been conducted for patients with facial paralysis, but now researchers have become interested in applying this technology to facial aesthetic surgery. For this portion of the study, we focused our efforts on 2 primary fronts rather than an exhaustive search for all objective measures that would apply to all of the possible conditions.

The first area relates to the quantification of facial anthropometric measurements and changes. These methods have used computer image analysis to establish reliable ways to digitize facial features^{186 - 187} and quantify changes.^{185 ,187} Few of these instruments have focused on surgical patients or outcomes. These measures offer the prospect of providing specific, accurate, and reproducible measures; however, most continue to be encumbered by reliance on special software packages and time-consuming techniques. Methods that have focused on outcomes, such as harmonized profiloplasty,¹⁹⁰ suggest that computer-assisted analysis can potentially affect decision making and may improve beneficial outcomes. Ultimately, the creation or acceptance of a gold standard objective test that is widely available and practical to administer would create outcomes measures that could potentially be used for future comparison cohort studies or RCTs. In addition, correlation of these objective tests with patient-reported symptoms (via QOL scales) would help establish the clinical meaningfulness of both of these outcome measures.

A second, more esoteric, area of focus has been on efforts to use computerized facial analysis to define objective aesthetic ideals, such as beauty and attractiveness, based on facial anthropometrics.^{188 - 189} Computer-assisted analysis offers the promise of preintervention modeling against known standards similar to methods that have been used by professionals in other fields, such as architecture and engineering. However, with regard to facial aesthetics, these methods are still largely conceptual. Much controversy remains regarding the definition of such ideals and, in fact, whether such ideals can be objectively formulated. Furthermore, little work has been done toward bridging the divide between what can be altered on a computer model and what is surgically achievable.

Our study has several limitations that warrant further discussion. Although we have attempted to be as inclusive as possible in our search, there may be measures or instruments that may have been inadvertently filtered out by our search variables. Also, we realize that some subjectivity exists in our grading scale of the existing measures. Finally, our study likely falls short in being an exhaustive and all-inclusive list of every existing outcome measure for facial plastic and reconstructive surgery; rather, the purpose of the study was to review the existing literature in a targeted fashion to highlight the voids and accomplishments of outcome measures in our specialty.

In conclusion, assessing outcomes in facial plastic surgery can be difficult because of the subjective nature of aesthetics and of quantifying certain outcomes. However, our literature review has demonstrated that validated instruments have become more prevalent for many facial plastic surgery procedures during the past decade, measuring both patient-reported and clinical efficacy outcomes. Ideally, future outcome studies will attempt to incorporate the best objective outcome measures that are clinically meaningful and correlate these findings with validated patient-reported outcome measures.

Correspondence: John S. Rhee, MD, MPH, Department of Otolaryngology and Communication Sciences, Medical College of Wisconsin, Milwaukee, WI 53226 (jrhee@mcw.edu).

Accepted for Publication: December 10, 2007.

Author Contributions:Study concept and design: Rhee and McMullin. Acquisition of data: McMullin. Analysis and interpretation of data: Rhee andMcMullin. Drafting of the manuscript: Rhee and McMullin. Critical revision of the manuscript for important intellectual content: Rhee and McMullin. Administrative, technical, and material support: Rhee and McMullin. Study supervision: Rhee.

Financial Disclosure: None reported.