Polydactyly is a congenital malformation characterized by extra digits of the hands or feet and can be described as preaxial, postaxial, or central polydactyly1. A report on the prevalence in the Netherlands showed 8.4 patients per 10,000 births with polydactyly, with only 0.4 patients per 10,000 births with preaxial polydactyly of the foot, also known as medial polydactyly (involvement of the medial side of the foot). Forty percent of these patients were diagnosed with a syndrome2.
Medial polydactyly of the foot is not extensively studied, and cohorts are usually small. A reason for the deficit in the literature may be the low prevalence and the relatively minor related functional problems2-4. However, the hallux is important for pressure distribution and directional control during walking5-7. Furthermore, Phelps and Grogan reported the necessity of treatment in most patients with polydactyly due to shoe-fitting problems and an unsatisfactory aesthetic appearance8. Consequently, the preservation of foot function and the reduction of shoe-fitting problems with medial polydactyly are challenging and require individualized treatment based on the anatomical and clinical appearance of the foot. The use of a clear, uniform classification system will result in simplified communication and improved comparison of different features9,10.
Ideally, classification systems can help to guide management of treatment and provide prognosis. However, developing a classification system is challenging. The system should be easy to use and allow for adaptation and extension of the system9. Furthermore, it must be valid and reliable11. Reliability is easy to measure with an analysis of interrater and intrarater agreement. Validity is a broader concept, with content validity, construct validity, and criterion validity being the most important types. Content validity refers to the extent to which a classification represents all important factors of a condition. When items are selected carefully and within reason, the content validity is higher. Construct validity refers to the extent to which a test measures the construct it claims to be measuring. For example, when a classification system is unable to distinguish between different types, it has a low construct validity. Criterion validity refers to the degree to which the classification correlates with other measures or outcomes. For example, if the classification corresponds to treatment results, then it has a high degree of criterion validity11.
Unfortunately, current classification systems for medial polydactyly do not fulfill the important factors for a good classification system (Table I). The comparison of feet of different phenotypes is difficult with some classification systems because of the impossibility of classifying all types. For example, the classifications of Seok et al.12 and Venn-Watson4 do not describe polydactyly of the distal phalanx. Likewise, by the classification system of Masada et al., no distinction between polydactyly of the distal or proximal phalanx is made, which also results in difficulty in classifying feet13. Moreover, Watanabe et al. described medial polydactyly on the basis of their own patient population and distinguished between tarsal, metatarsal, proximal phalangeal, and distal phalangeal types14. This resulted in 15 groups for medial polydactyly that were specific to that population and without a clear analogy among the different types. The drawings used to describe specific characteristics of the feet, such as hypoplastic rays and deviation, lack universal properties and are not easy to use. Furthermore, medial and lateral polydactyly of the foot are sometimes grouped together in classification systems; both Blauth and Olason15 and Seok et al.12 combined medial and lateral polydactyly in 1 classification system. We think that this decreases the content validity of the classification of medial polydactyly because Venn-Watson described a difference in the anatomical properties of medial polydactyly compared with lateral polydactyly4. In addition, it is known that the hallux has a more important function than the fifth toe, which may also result in a decrease in criterion validity5,6.
We believe that the present classification systems do not provide a comprehensive description of medial polydactyly. Therefore, the aim of this study was to develop a more valid, reliable, and easy-to-use classification system. In order to improve the content validity of this new classification, we performed a literature review and held a consensus meeting to determine all of the important contributing factors to be considered in the classification of medial polydactyly. Next, we tested the usability of the developed descriptive classification system by assessing our own population, and an agreement analysis was performed among 5 observers to test the reliability of the classification system.
Materials and Methods
Development of the Classification System
An Embase literature search (see Appendix) for classification systems for polydactyly of the foot was performed in 2014, in order to develop a representative list of contributing factors of medial polydactyly. The article titles and abstracts were reviewed to identify studies about classification systems. All classifiable aspects of medial polydactyly were extracted and reviewed by an experienced plastic surgeon (C.A.v.N.) and the principal investigator (E.B.B.). Relevant categories were chosen on the basis of occurrence of the category in the literature and the influence on function and aesthetic outcome according to the literature.
Classification of Our Population
We searched the hospital database of the congenital hand team of our department for patients with medial polydactyly of the foot seen between 1993 and 2014. Patients with Apert syndrome were excluded from this initial search because foot anatomy is evidently different in these patients. Patients without preoperative radiographs and clinical photographs were also excluded. Foot anomalies were classified according to the new classification system by the principal investigator (E.B.B.). The occurrence of each type was analyzed.
An analysis was performed in order to test intrarater and interrater agreement. The principal investigator (E.B.B.) randomly chose 30 feet from our database. Five raters, who included 2 plastic surgeons (S.E.R.H. and C.A.v.N.), 1 orthopaedic surgeon (B.J.B.), 1 medical student, and the principal investigator, classified the presented clinical and radiographic images according to the developed classification system. The principal investigator was excluded in the intrarater agreement analysis because we expected a biased outcome due to multiple case views by the principal investigator during the study. We chose to involve the 2 specialties because patients consult both plastic and orthopaedic surgeons. Radiographs and clinical photographs were presented via LimeSurvey, a protected computer program used for questionnaires and assessments. Two classification rounds were performed within approximately 4 to 6 weeks. Raters performed the classification independently and were blinded to the clinical information of the patients.
The intrarater and interrater agreement was assessed with use of the Cohen kappa (κ) statistic. Guidelines of Landis and Koch were used for interpreting κ values16. The κ value of intrarater agreement was calculated using the classification for each pair of observations of the 4 observers in the first and second rounds and then averaged to provide a single κ value. The average percentage of agreement among the observers was also calculated. Interrater agreement was calculated by comparing the first-round classifications among the different raters. Again, an average of κ and the percentage of agreement were calculated. Because no gold standard for the classification was present, the agreement among the observers was analyzed using the classification that was most chosen by the 5 observers as the gold standard. Statistical analyses of κ values were performed using SPSS software (version 22.0; IBM). The percentages of agreement were calculated using Excel software (2010; Microsoft).
The study was approved by the Institutional Ethical Review Board of the Erasmus Medical Centre, Rotterdam (mec-2014-263), and is in accordance with the Declaration of Helsinki.
Development of the Classification System
The search of the literature resulted in 650 articles. After our review of the titles and abstracts, 46 articles remained. We compiled a list of contributing factors that were mentioned and the number of articles describing each (Table II). Ray involvement was the second-most mentioned factor in the literature; however, our classification system is exclusively for polydactyly of the first ray, so we excluded this factor. Consensus was reached regarding the other 4 most mentioned categories: duplication level, syndactyly, hypoplastic ray, and deviation.
Duplication level was included because the surgeon should be informed about duplicated osseous structures. Syndactyly was included because planning for incision and the need for tissue grafting are influenced by the presence of syndactyly. Furthermore, expected deviation after surgery depends on the presence of syndactyly between the preserved hallux and second toe. Hypoplastic ray was added to the classification because the choice of excision side may be influenced by the presence of a hypoplastic ray. Deviation of the hallux was included because it influences the surgical techniques used to achieve correct orientation of the hallux. As the classification system is only for medial polydactyly, involvement of rays other than the first ray was not included.
The proposed classification system is illustrated in Figure 1. The appearance of a floating hallux or polydactyly without osseous structures is classified as type 0. The duplication type is scored by Roman numerals, corresponding to the Rotterdam classification system for radial polydactyly and initially derived from the Wassel classification, starting with type I, for distal phalangeal duplication, and proceeding to type VIII, for duplication of the tarsal bones17,18. Even numbers represent complete duplication of the osseous structure, and odd numbers represent incomplete duplication.
Syndactyly, abbreviated as the letter “S,” is classified as S0, S1, S2, or S1S2. S0 represents the presence of a broad hallux, without the appearance of syndactyly. S1 indicates the presence of syndactyly between medial and lateral (duplicate) halluces. S2 indicates the presence of syndactyly between the lateral hallux and the second toe. S1S2 indicates syndactyly between duplicate halluces and between the lateral hallux and the second toe. No distinction was made between complete or incomplete syndactyly. In cases in which no syndactyly was present, no “S” classification was made.
A hypoplastic ray is abbreviated as the letter “H” followed by “M” or “L,” representing a medial or lateral hypoplastic structure. In the classification of a hypoplastic ray, all hypoplastic osseous structures are taken into account. In addition, a preaxial ray with both lateral and medial hypoplastic structures can be scored as HLHM. In cases in which no hypoplastic structure was present, no “H” classification was made.
Deviation of the hallux, abbreviated as the letter “D,” is noted when the alignment of the hallux, which we determined on the basis of clinical photographs, is not in alignment with the rest of the foot.
Classification of Our Population
A total of 64 patients with medial polydactyly of the foot were identified from the hospital database. Twenty-one patients were excluded; 11 patients did not have preoperative radiographs because of conversion to electronic patient systems, 3 did not have clinical photographs, and 7 did not undergo surgery in our hospital. The available radiographs or clinical photographs of these 21 patients were reviewed, and no foot types different from those of our included population were suspected in this group. Therefore, selection bias is not likely. Of the included 43 patients, 16 were male and 27 were female (Table III). Radiographs and clinical photographs were made between the age of 6 and 18 months in the majority of cases. Medial polydactyly of the foot was seen in 73 feet: 39 right feet and 34 left feet. In most (70%) of the cases, bilateral involvement was noted.
All of the feet could be classified according to the classification system (Table IV). Seven types of duplication were noted. Duplication type VI was most frequently seen (in 29% of the feet). Partial duplication was mostly seen in the distal phalanx (duplication type I; 15%). Twelve feet (16%) showed a broad hallux (S0) without the expression of 2 nails. A hypoplastic ray was seen in 75% of the feet (medial, n = 21; lateral, n = 34). Deviation of the hallux medially was seen in the majority (73%) of the cases.
With respect to duplication in combination with syndactyly, we found that a larger proportion of feet with duplication type II (61%) and type IV (79%) had syndactyly between 2 halluces (S1) and a larger proportion with duplication type VI (57%) had syndactyly between a lateral hallux and the second toe (S2) (Fig. 2). Furthermore, duplication type VI never involved a medial hypoplastic ray, whereas phalangeal duplications (type II and type IV) involved a medial hypoplastic ray in the majority of feet.
The combined classification showed a wide variety of presentations of medial polydactyly. Fourteen foot types were scored only once. Larger groups of 1 specific type were present in type-IV and type-VI duplications. Eight feet were classified as IV S1 HM D, and 11 feet were classified as VI S2 HL D.
Intrarater agreement and interrater agreement of the Rotterdam foot classification are shown in Table V. Classification of all categories showed moderate to good reliability. Intrarater agreement had the lowest κ value for syndactyly (κ = 0.59), with an agreement of 69%. The other 3 categories showed a mean κ value that was >0.7. The mean κ value for interrater reliability in all 4 categories was lower than that for intrarater reliability.
Medial polydactyly of the foot is a rare congenital anomaly2. The diversity of the anomaly requires individualized treatment to diminish shoe-fitting problems and improve aesthetic appearance3,4. In order to plan and evaluate surgical treatment, the classification of medial polydactyly is useful3. However, current systems are not able to classify all feet and are less feasible to use in clinical practice. Therefore, the aim of the current study was to develop a reliable and valid classification system for medial polydactyly of the foot.
A search of the literature resulted in 15 potential categories. We included the 4 most mentioned terms, with the exception of ray involvement, because it is plausible to presume that these terms are important in the description of medial polydactyly. The classification of our own population resulted in a variety of groups, indicating the diversity of anomalies in medial polydactyly. However, more specific description of the appearance of the feet also revealed frequent combinations. For example, syndactyly between lateral and medial halluces was mostly seen with the duplication of the proximal or distal phalanx, while syndactyly between the lateral hallux and the second toe was more frequently seen in metatarsal duplication. In addition, in all cases with a lateral hypoplastic ray, metatarsal duplication was present. Comparable results were also seen in the population of Watanabe et al.14. Moreover, medial hypoplastic rays were mostly identified with duplication of the proximal and distal phalanges, as was previously reported by Masada et al.13. Group sizes for distal phalangeal (I and II), proximal phalangeal (III and IV), and metatarsal (V and VI) duplications were comparable in our population, while for other populations in the literature3,14, more proximal phalangeal and metatarsal duplications compared with distal phalangeal duplications have been reported. In contrast to the patient populations of Seok et al.12, Belthur et al.3, and Masada et al.13, our population more frequently showed an incomplete distal phalangeal duplication. Both of these differences in findings might be the result of the absence of clear external appearance of polydactyly and the inability to classify distal phalangeal duplication with their classification systems. In our study, no feet with type-III and type-VII duplication were present. However, in thumb polydactyly, these duplication types do exist, and in the study by Watanabe et al., 2 feet with type-III duplication are described14. This prompted us to include these types in the classification system. All feet in our population could be classified with the proposed system. However, it is known that phenotypic variations are present in different parts of the world; other types of medial polydactyly may be present. By conducting a review of the literature, we tried to include all described types in our classification. However, triplications and triphalangism are also described in polydactyly15,18. We did not include these categories because of the rarity of these anomalies. Therefore, we suggest describing rare types in more detail by adding specific terms to the classification system. Moreover, if frequently occurring categories are found to be missing, the classification system allows adaptation.
Classification by all observers was performed with use of digital clinical photographs and digital radiographs, the latter also being used in clinical practice. Therefore, duplication level and hypoplastic rays were assessed during the classification sessions in a manner similar to that of clinical practice.
However, parts of the osseous structures of the foot are not visible on radiographs between 6 and 18 months19. Consequently, the original duplication level can be different from the observed. Despite this lack of precision, the assessment of duplication level in this study is comparable with assessment of duplication in clinical practice. Syndactyly was assessed using digital clinical photographs, while assessment in clinical practice is conducted by physical examination. As syndactyly is not always clearly visible on photographs, this may have led to increased uncertainty for the observers compared with the observation of syndactyly in clinical practice. This could have resulted in the lower κ value found for this category.
It is not yet clear whether or not the categories of the Rotterdam system can effectively be used in the planning of treatment and the prediction of postoperative functional outcome. However, Belthur et al. compared the modified Venn-Watson classification and the more extensive Watanabe classification for surgical planning3. They concluded that the comprehensive Watanabe classification is more useful because of the complete description of the foot, including the presence of a hypoplastic ray, deviation, and tarsal duplication. Unfortunately, the Watanabe classification system only describes types on the basis of their patient population and does not contain a consistent and uniform description of medial polydactyly14. Furthermore, it requires drawings for complete description. The Rotterdam foot classification is based on types described in the literature and evaluated in our own population. Furthermore, the 4 categories make specific description easier and more analogous between observers.
In this study, we developed a new classification system for medial polydactyly of the foot that is based on a review of the literature and clinical experience. The results of the application of the classification system to our own population and the agreement analysis showed that the Rotterdam foot classification is a usable system to describe medial polydactyly. Therefore, we recommend this system for the description of medial polydactyly of the foot. In the future, we hope to demonstrate how this classification system can contribute to surgical planning and evaluation.
Details of the parameters used in the Embase literature search are available with the online version of this article as a data supplement at jbjs.org.
NOTE: The authors thank A.J. Pieterse for her contribution to the agreement analysis.
Investigation performed at Erasmus Medical Centre, Rotterdam, the Netherlands
Disclosure: Funding was received for this study from the Esser Foundation, which is associated with the Department of Plastic, Reconstructive and Hand Surgery, Erasmus Medical Centre, Rotterdam, the Netherlands. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article.
1. Malik S. Polydactyly: phenotypes, genetics and classification. Clin Genet. 2014 ;85(3):203–12. Epub 2013 Oct 18.
2. Vasluian E, van der Sluis CK, van Essen AJ, Bergman JE, Dijkstra PU, Reinders-Messelink HA, de Walle HE. Birth prevalence for congenital limb defects in the northern Netherlands: a 30-year population-based study. BMC Musculoskelet Disord. 2013 ;14:323.
3. Belthur MV, Linton JL, Barnes DA. The spectrum of preaxial polydactyly of the foot. J Pediatr Orthop. 2011 ;31(4):435–47.
4. Venn-Watson EA. Problems in polydactyly of the foot. Orthop Clin North Am. 1976 ;7(4):909–27.
5. Chou SW, Cheng HY, Chen JH, Ju YY, Lin YC, Wong MK. The role of the great toe in balance performance. J Orthop Res. 2009 ;27(4):549–54.
6. Barca F, Santi A, Tartoni PL, Landi A. Gait analysis of the donor foot in microsurgical reconstruction of the thumb. Foot Ankle Int. 1995 ;16(4):201–6.
7. Tanaka T, Hashimoto N, Nakata M, Ito T, Ino S, Ifukube T. Analysis of toe pressures under the foot while dynamic standing on one foot in healthy subjects. J Orthop Sports Phys Ther. 1996 ;23(3):188–93.
8. Phelps DA, Grogan DP. Polydactyly of the foot. J Pediatr Orthop. 1985 ;5(4):446–51.
9. Tonkin MA, Tolerton SK, Quick TJ, Harvey I, Lawson RD, Smith NC, Oberg KC. Classification of congenital anomalies of the hand and upper limb: development and assessment of a new system. J Hand Surg Am. 2013 ;38(9):1845–53. Epub 2013 May 17.
10. Williams BA, Matsumoto H, McCalla DJ, Akbarnia BA, Blakemore LC, Betz RR, Flynn JM, Johnston CE, McCarthy RE, Roye DP Jr, Skaggs DL, Smith JT, Snyder BD, Sponseller PD, Sturm PF, Thompson GH, Yazici M, Vitale MG. Development and initial validation of the Classification of Early-Onset Scoliosis (C-EOS). J Bone Joint Surg Am. 2014 ;96(16):1359–67.
11. Kimberlin CL, Winterstein AG. Validity and reliability of measurement instruments used in research. Am J Health Syst Pharm. 2008 ;65(23):2276–84.
12. Seok HH, Park JU, Kwon ST. New classification of polydactyly of the foot on the basis of syndactylism, axis deviation, and metatarsal extent of extra digit. Arch Plast Surg. 2013 ;40(3):232–7. Epub 2013 May 16.
13. Masada K, Tsuyuguchi Y, Kawabata H, Ono K. Treatment of preaxial polydactyly of the foot. Plast Reconstr Surg. 1987 ;79(2):251–8.
14. Watanabe H, Fujita S, Oka I. Polydactyly of the foot: an analysis of 265 cases and a morphological classification. Plast Reconstr Surg. 1992 ;89(5):856–77.
15. Blauth W, Olason AT. Classification of polydactyly of the hands and feet. Arch Orthop Trauma Surg. 1988;107(6):334–44.
16. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977 ;33(1):159–74.
17. Wassel HD. The results of surgery for polydactyly of the thumb. A review. Clin Orthop Relat Res. 1969 ;64:175–93.17.
18. Zuidam JM, Selles RW, Ananta M, Runia J, Hovius SE. A classification system of radial polydactyly: inclusion of triphalangeal thumb and triplication. J Hand Surg Am. 2008 ;33(3):373–7.
19. Harty MP. Imaging of pediatric foot disorders. Radiol Clin North Am. 2001 ;39(4):733–48.