The Choice of Diabetic Foot Ulcer Classification in Relation to the Final Outcome

Introduction

Diabetes mellitus-related foot ulceration is very common. As a result of neuropathy, peripheral vascular disease, and infection, patients with diabetes are prone to develop diabetic foot problems that may eventually require a lower-extremity amputation. Of all individuals with diabetes mellitus, 15 percent will be affected by ulceration at least once in their lifetime.1,2 The presence of foot ulceration increases the hospitalization duration by 59 percent in people with diabetes mellitus. Because diabetic foot ulceration is a serious problem and because ulcers are heterogeneous in terms of etiology, anatomic location, depth of tissue involvement, and associated circumstances, including the presence or absence of infection, classification is needed in order to predict ulcer outcome and conduct clinical trials.1

In the literature, several classification systems for diabetic foot ulcers have been proposed. These classification systems have to comply with certain characteristics, such as precision, flexibility, specificity, and simplicity. They also must be applicable for education and communication between all care providers, including nurses, general practitioners, and specialists. They can be of great help for the assessment of treatment schemes. Classifications are also useful in standardization and analysis of multicenter research. The classification most frequently used analyzes one or more of the following elements: infection, neuropathy, vasculopathy, and the extent (surface and depth) of the ulcer.

The best known and widely available classifications are the following: Meggit/Wagner, Gibbon’s, Frykberg’s and Coleman’s, Forrest’s, Knighton’s, the Texas Diabetic Wound Classification, and the Ten-Level Seattle Wound Classification System.2–11 Each of these classifications was developed to accomplish a particular objective, utilizes different criteria, and categorizes lesions according to different rationales. Only a few of these classifications were evaluated for the assessment of the prognosis on salvage of the ulcerated or dysvascular diabetic limb.

The aim of this study was to evaluate a new classification, the Van Acker/Peter classification (VA/P). This is a two-dimensional classification on the vertical axis parameters of physiopathology and on the horizontal axis parameters of depth of the ulcer and the presence or absence of infection. Outcome measures studied are healing with or without amputation and the time until healing.

Patients and Methods

Study population. Patients were selected in the Antwerp Diabetic Foot Clinic. All patients who visited the diabetic foot clinic during the period of January, 1992, and December, 1997, were selected on the basis of having a specific health insurance file code. This code is the identifying mark of the health insurance organization LCM, the most important insurance system in Belgium. The reason of this selection was that this group of patients received follow up in an economical evaluation in a long-term study in our clinic. All files contained photographic images of the ulcers and careful clinical descriptions according to the standard procedure of the foot clinic. In order to eliminate interpretation bias by different investigators, the same investigator (KVA) coded all ulcers. The total number of ulcers with a Wagner classification greater than 0 was 303. Only the files with data on healing outcome were used for analysis (n = 253).

The Meggit-Wagner classification. This system is based on three features: depth of the ulcer, the degree of infection, and the presence or absence of gangrene and its extent. Grades 1 to 3 are mainly based on neuropathy, while grade 4 and 5 represent mainly ischemic lesions.

The Van Acker/Peter classification. This classification is based upon the Texas Wound Classification, the first bidimensional classification. This new system is established on clinical experience and gives, through different shades of gray, an estimate of the clinical expected risk of amputation in relation to foot pathology. The criteria for the new classification were ease of use by clinicians and clinical investigators and the covering of physiopathology as well as clinical picture, including the depth of ulcers, the degree of infections, and the presence or absence of osteomyelitis. The system includes 25 separate classes. This classification is represented as a black/white scale (Figure 1).

Definitions. In the analyses, ulcer healing with or without amputation was used as the main outcome variable. Ulcer healing was defined as a total closure of the skin with a normal appearance of the skin without callus. The date at which this stage was reached was used as time of endpoint. The time of onset for many ulcers was unknown; therefore, the date at which the patient was initially included in the records was used as time zero for calculation of healing times. Healing time was expressed in days.

Because of the low number of ulcers in Wagner grade 5 (n = 2), this category was aggregated with grade 4. In order to reduce the number of categories with low numbers in the VA/P classification, the horizontal classes 4 and 5 were aggregated as well as the vertical classes B+C and D+E, respectively. As a result, a modified VA/P classification with 15 categories was used in the analyses.

Patients were considered to have vascular insufficiency with chronic critical ischemia when one or more of the following criteria were fulfilled:
• Systolic ankle blood pressure (Doppler) less than 50mmHg
• Ankle/brachial systolic blood pressure index less than 0.9
• Tissue percutaneous oxymetry (TcpO2) less than 30mmHg.
For the diagnosis of osteomyelitis in case of an ulcer, three of the following criteria had to be present:
• Cellulitis
• A positive bacteriological culture from deep tissue
• Radiological and/or scintigraphic signs compatible with osteitis
• Histological diagnosis
• Probing to bone.

Patients were considered to suffer from neuropathy in case of an abnormal Semmes-Weinstein monofilament and/or abnormal test of vibration perception, a loss of deep sensation using a tendon hammer (international consensus group 1996), or an EMG confirming the presence of polyneuropathy by a latent F-wave response or a latent Hoffman reflex.

Statistical analysis. Differences between the subgroups of healing with or without amputation were tested by Student’s t-test in case of continuous data and by Chi-square test or Fisher’s exact test (low numbers) in case of categorical data. The degree of association between the scores of the different classification systems was assessed by Spearman’s rank correlation coefficients. Time until ulcer healing was calculated by Kaplan-Meier survival analyses. Because all ulcers reached their endpoint (healing), no dates were censored. Differences in healing times between categories in the Wagner and Van Acker/Peter classifications were tested by logrank tests. All tests were two-sided, and p values less than 0.05 were considered to be statistically significant. Statistical analyses were performed using SPSS 9.0 for Windows (SPSS Inc., Chicago, Illinois).

Results

The study population consisted of 121 patients with 253 ulcers. Of these, 223 (88.1%) healed without amputation and 30 (11.9%) healed after amputation. Table 1 shows the characteristics of the patients in the study, stratified according to outcome status: healing without or with amputation (WOA or WA, respectively). In the latter group, mean duration of diabetes was significantly higher (25.5 years vs. 19.3 years).

No significant differences were seen in other characteristics. Patients healed with amputation showed more frequent cardiovascular complications, cerebrovascular disease, and retinopathy. No significant differences in smoking history existed between both outcome categories. Also therapeutic characteristics were analyzed. More frequent antibiotic treatment (71.3% in WOA vs. 96.8% in WA; p = .002), extensive surgical debridement (16.2% in WOA vs. 36.7% in WA; p = .007), and more frequent revascularization techniques (2.7% in WOA vs. 12.9% in WA; p = .023) were used for the patient group healed with amputation. This group contained significantly more ulcers with gangrene, as could be expected (5.8% in WOA vs. 48.4% in WA; p = .000).

The distribution of the ulcers in the Wagner and in the VA/P classification system is presented in Table 2. The distribution of diabetes foot ulcers according to Wagner classification was Wagner 1, n = 145 (57.3%); Wagner 2, n = 40 (15.8%); Wagner 3, n = 43 (17.0%); Wagner 4, n = 25 (9.9%); total 253 (100%). According to the VA/P classification, 97 (38%) ulcers were of vascular origin (DE) from which 74 percent in Wagner 1 through 3 and the remaining 26 percent were classified as Wagner 4. About 60 percent of the ulcers were located at the toes, half of which were at the first toe. These toe localizations accounted for 83 percent of the amputations.

In order to evaluate the characteristics of the VA/P classification, the prevalence of amputations according to the horizontal axis scores (clinical assessment of depth and grade of infection) and the vertical axis scores (physiopathological background) was examined separately and is presented in Table 3. Prevalence of amputation increased significantly with increasing extent of grade of infection (Chi-square for linear trend, p < .001), whereas on the vertical axis the prevalence of amputation was related to vascular etiology (class DE) in particular (Chi-square for linear trend, p = .002).

For comparison, the distribution of amputations between the different Wagner classes are presented in Table 4 showing also a strong positive relationship between prevalence of amputations and Wagner scores (Chi-square for linear trend, p < .001). The association between all four classification systems, evaluated by Spearman’s correlation coefficients, is shown in Table 5. All classifications correlated well with one another, except for the two axes of the VA/P classification (correlation coefficient not significant). The strongest correlation was seen between the Wagner and vertical (clinical) axis of the VA/P classification.

Amputations registered in this study population were distributed as follows: 26 toe, 2 forefoot, 3 lower leg, and 2 upper leg amputations. The amputations considered are from patients with complete ulcer healing after amputation. Table 6 shows not only the distribution of the ulcers between all classes of the VA/P and Wagner classification, but also for each combination the proportion of amputation. Within each pathological category (A, BC, DE) low scores of the VA/P agreed with low scores of the Wagner. However, all Wagner class 4 ulcers (gangrene) appeared in the vascular (DE) category. In the Wagner categories 1, 2, and 4, the probability of amputation was not associated with the clinical extent of infection, though for different reasons.

In Wagner 1 and 2 classifications, amputations were rare (2.0% and 2.5%, respectively), whereas in Wagner 4 the prevalence of amputations was 52 percent and confined to the pathological category DE only. However, within this group no relation existed between the number of amputations and the clinical extent of wound infection (DE1–4). In contrast, in Wagner 3 the prevalence of amputations was higher in accordance with increasing clinical extent of infection as well as with the pathological background.

In the second part of the analyses, we investigated whether the time until healing was associated with different ulcer categories. Figures 2, 3, and 4 show the proportion of healed ulcers over time according to the Wagner, the clinical axis of the VA/P, and the pathological axis of the VA/P, respectively.

Between the Wagner categories, median healing time increased from 51 days in class 1 to 174 days in class 4 (logrank test for linear trend, p < .001) (Figure 2). Healing times of classes 3 and 4 were significantly longer as compared to class 1. Of the clinical axis of the VA/P classification, category 2 had the lowest healing time (50 days), followed by categories 1, 3, and 4 (69, 116, and 103 days, respectively) (logrank test for linear trend, p = .065) (Figure 3). When compared to class 1, no significant differences were seen between these classes. Regarding the pathological axis of the VA/P classification, a significant trend in healing times could be seen from class A-BC-DE (60, 76, and 91 days, respectively) (logrank test for linear trend, (p = . 033) (Figure 4). However only class DE, representing vascular pathology, had a significantly longer healing time when compared to class A.

Discussion

In this study we investigated the Van Acker/Peter classification system, a new diabetic foot ulcer classification. We compared its performance with that of the existing clinically validated Wagner classification.12,13 The classifications were assessed by evaluating clinical outcome, e.g. healing with or without amputation and duration of healing.

Our results show that in the VA/P classification the prevalence of amputation increases significantly with increasing extent of grade of infection and is related to vascular origin. Also, in the Wagner classification, a strong positive relationship exists between prevalence of amputation and Wagner scores.12 According to amputation rate, Wagner class 3 seems to be a heterogeneous group. In addition, both VA/P and Wagner classifications show a similar association with duration of healing time, although the best linear trend in healing times was seen in the Wagner classification.

The VA/P classification is based upon the Texas Classification and takes into account not only clinical features, such as depth and grade of infection (horizontal axis), but also the physiopathological background (vertical axis), where ischemic components play a role.

The Texas Diabetic Wound Classification is the first combined bidimensional classification published in 1996 by Lavery, Armstrong, and Harkless. Harkless described in an earlier publication a combined classification based on the Wagner stages (0–5).14,15 He created subclasses depending on the presence of ischemia. The ischemic or B-forms were described as presenting a weaker prognosis than the nonischemic A-forms. The actual classification by Lavery, Armstrong, and Harkless is based on three essential questions: depth of the ulcer, infection, and ischemia.

The Meggit/Wagner classification is probably the best known and the most frequently used. Nevertheless, it is described as very simple and, therefore, often considered to be inconveniently inaccurate. Sims, et al., noted that the scheme provides insufficient levels to discriminate between wounds that may benefit from nonsurgical rather than surgical management. The grading system was adapted in 1988 by Calhoun, et al., in order to combine medical and surgical elements of therapy to monitor the treatment of diabetic foot infection.16

Both Wagner and horizontal axis of the VA/P classifications correspond well regarding the prediction of healing with amputation. This was to be expected, as both are based mainly on the clinical picture. The vertical axis of the VA/P also is associated with the chance of healing with amputation. As no correlation exists between the horizontal axis and the vertical axis of the VA/P, these characteristics can be considered as complementary to one another and, therefore, may provide additional information.

However, because the probability of amputation in Wagner 1 and 2 is low in all pathological classes, and in Wagner 4 all ulcers are prevalent in the vascular (DE) class only, additional information is useful in the Wagner class 3 in particular. According to the VA/P classification, Wagner 3 appears to be a heterogeneous subgroup. Forty-two of 43 ulcers in Wagner 3 are classified as VA/P class 3 or 4, belonging to all three pathological classes. Likewise, of the 13 ulcers in Wagner 3 that healed with amputation, 12 were classified as VA/P class 4, with amputation rates rising from 30 to 50 percent over the pathological VA/P levels. Thus, within Wagner 3 the presence of osteomyelitis (VA/P class 4) is the determining factor related to healing with amputation, with vascular pathology (VA/P class DE) as a secondary risk factor. Unlike Wagner 3, Wagner 4 is a homogeneous subgroup regarding pathological background. This is not surprising, as gangrene is restricted to underlying vascular pathology. However, the risk of amputation is unrelated to the extent of the ulceration.

According to the VA/P classification, 97 ulcers (38%) were of vascular origin. However, of those, only 26 percent were categorized in the Wagner 4 group, whereas 74 percent were assigned to Wagner classes 1 through 3. By incorporating the physiopathological axis, the VA/P classification gives more detailed information concerning the ischemic origin of ulcers. Lavery, Armstrong, and Harkless already mentioned that ischemic ulcers present a weaker prognosis than nonischemic ulcers.14

Like previous studies,17 our study demonstrates the long duration of healing of diabetic foot ulcers. Our study also confirms the results from the Texas group14,15 that ischemic wounds (VA/P class DE) have a longer duration of healing compared with neuropathic wounds without bone deformities (VA/P class A). However, no differences were found between neuropathic wounds with (class BC) or without (class A) bone deformities. These findings warrant our approach for taking into account the physiopathological background in the classification system.

Conclusion

In conclusion, for daily clinical practice a simple, though accurate, classification system is necessary in order to give adequate therapy and to predict probable outcome. In his editorial, Levin suggested that the American Diabetes Association would provide a consensus statement on classification of diabetic foot ulcers. In general, the new ulcer classification VA/P, which takes into account the underlying pathology of the ulcers, gives more detailed information than the Wagner classification. In the intermediate category (e.g., Wagner 3) in which the risk of amputation is related to size/depth and the physiopathological characteristics of the ulcer, the new VA/P classification seems preferable. Furthermore, the VA/P classification characterized by more detailed information may also be useful in well-organized multidisciplinary diabetic foot clinics and in multicenter research. However, because of its simplicity, the Wagner classification remains a perfectly usable instrument in everyday clinical practice (e.g., in primary health care).