Independent Observational Analysis of Ulcer Outcomes for SINBAD and University of Texas Ulcer Scoring Systems

Diabetes-related foot ulcers are associated with complex health needs, often resulting in prolonged admissions and high health care costs in the U.K. and the U.S. (1,2). This is important, as it is estimated that one in four patients with diabetes will develop a foot ulcer (3). Clinical systems that help direct resources to those at greatest need are important to help organize an effective and efficient service. Clinical systems that classify foot ulcers are primarily designed to identify patients at greatest risk of nonhealing, thus guiding treatment, although they have other benefits such as communication between professionals, clinical audit, and others (4). A large number of ulcer classification systems exist (5). Some of the best validated systems include Meggitt-Wagner (6–8), University of Texas (UT) (9,10), Size (Area, Depth), Sepsis Arteriopathy, and Denervation (S[AD]SAD) (11) and subsequent Site, Ischemia, Neuropathy, Bacterial Infection, and Depth (SINBAD) (12), Perfusion, Extent, Depth, Infection and Sensation (PEDIS) (7), Wound, Ischemia, and Foot Infection (WiFi) (13), Saint Elian Wound Score System (SEWSS) (14), Diabetic Ulcer Severity Score (DUSS) (15), and others (16). The mostly widely recognized worldwide are probably Meggitt-Wagner, UT, and SINBAD. The SINBAD score has recently been endorsed by the International Working Group on the Diabetic Foot (IWGDF) (4). This endorsement is based on a balance of evidence, and pragmatism, although it is recognized that there is a paucity of evidence comparing ulcer grading systems (5). The Meggitt-Wagner system does not include criteria on infection or ischemia, while the WiFi system has a specific vascular focus and requires the use of sophisticated equipment that is generally not available routinely to many clinics worldwide. Many of the clinical systems that describe and predict ulcer outcomes have been validated internally and/or externally (5,17). There are few studies that compare the outcomes of the ulcer clinical systems within the same group of patients. Meggitt-Wagner and UT have been compared most often (6–8,18,19), while in some studies other systems have been added into the comparison such as S(AD)SAD (7,8,11), PEDIS (7), SINBAD, DUSS, and DEPA (depth, extent of bacterial colonisation, phase of healing, and associated aetiology) (17,19). The SINBAD system, as recommended by the IWGDF, has been validated (12) but has only been involved in small comparator studies of 137 (19) and 293 patients (20). The Scottish Intercollegiate Guideline Network (SIGN) score is routinely used in Scotland to determine the risk of ulcer from clinical criteria of foot examination (21), but it has also been shown to predict outcomes of ulcer (17,22). Although no ulcer criteria are used in the score, it is one of the few systems that now routinely incorporate renal function in assessment.

The two systems used most commonly in clinical practice are the SINBAD and the UT scores. The SINBAD ulcer scoring system uses six criteria from ulcer and foot examination, comprising ulcer site, area, depth, and presence of ischemia, neuropathy, or infection. Each criterion, if present, is added to create a scale of 0–6 reflecting increasing severity (12). The UT score creates a binomial table dependent on ulcer depth to score the “grade” and presence of infection and/or ischemia to score the “stage,” which creates an ulcer score from A1 to D3 (9,10).

The UT ulcer score (9) is recommended for ulcer grading across Scotland. However, with the recent recommendations by IWGDF to use SINBAD, we aimed to compare both systems to see which best predicts foot ulcer healing in people with diabetes.

This was a cohort study using data from all patients referred to Ninewells Hospital outpatient diabetes foot clinic with a foot ulcer. All patients who had diabetes and a foot ulcer that was not showing signs of healing within 4 weeks were recommended for referral to the hospital clinic from the community of Dundee. Patients attending from 2006 until the end of 2018 were included in the study. All ulcers included in the study were followed up until a defined outcome, death, or the end of the study. Clinical data were used to categorize ulcers. Ischemia was defined as absence of both pedal pulses in at least one foot. Loss of protective sensation was defined as an inability to detect ≥3 of 10 sites on the plantar aspect of both feet (5 sites on each foot). Lack of both dorsalis pedis and posterior tibial pulses defined ischemic feet for the purposes of this study. Infection was defined as presence at any time during treatment of soft-tissue inflammation (e.g., red, hot, swollen, tender) or presence of pus, or proven osteomyelitis, in line with the Infectious Diseases Society of America guidelines (23). Site of ulcer was registered as either forefoot (on the toe, metatarsal head) or hind foot (e.g., heel). Ulcer depth was categorized as superficial, deep to the level of muscle or tendon, or deep to the level of bone. Ulcer diameters were measured, and size was categorized as <1 cm² or >1 cm². All data points were available except for two data points missing on depth and 224 missing on size, which were equally distributed between the healed and nonhealed groups.

Outcome was assessed as follows: 1) healed without amputation, which was defined as wound closure without the need for any dressing (12); 2) healed after minor amputation defined as surgery distal to the ankle (most proximal being a ray excision); 3) major amputation, defined as surgery above the ankle (all were below knee or above knee); 4) death; or 5) a chronic nonhealing wound. Only ulcers that healed without any form of amputation were considered as a healed outcome. Outcomes were determined at the “completion of a consultant event.” The consultant event was complete when all ulcers had healed, including ulcers that may have developed while the first ulcer was being managed. If there were multiple ulcers, the most proximal ulcer, e.g., heel ulcer, was recorded as the study ulcer. Patients may have been represented more than once if they developed recurrent ulcers after a previous completed consultant episode.

We used the UT scores and the SINBAD scoring system. The UT system uses a binomial matrix, with depth of ulcer graded on the horizontal axis and presence of ischemia and infection staged on the vertical axis. We looked at ulcer outcome according to the grade and to the stage of the ulcer. The UT system is sometimes difficult to follow sequentially, as it is reported as a binomial matrix with no integrated unified score. To combine both aspects of the ulcer assessment, we created an a priori “unified” UT score. The value of the ulcer depth/grade (superficial = 1, tendon/capsule = 2, bone = 3) was multiplied by the stage (no additional risk factors = 1, one risk factor of infection or ischemia = 2, both infection and ischemia = 3). This created an overall score from 1 to 9 for someone with a foot ulcer.

All records from patients seen at the foot clinic are sent independently to the Health Informatics Centre for health care research (https://www.dundee.ac.uk/hic) of the University of Dundee. All data for analysis were anonymized using standard operating procedures and data protection approved by Tayside Medical Ethics Committee and the Tayside Caldicott Guardians.

ANOVA and χ² tests were used to compare means and frequencies among groups, respectively, and nonparametric tests were used where appropriate. Tests of association with diabetic foot ulcer outcome (coded as binary: not healed [case] vs. healed [control]) were performed with unconditional logistic regression models. Regression models were used to estimate odds ratios (ORs) of patients’ adverse outcomes by clinical score systems (SINBAD and UT). The performance of the individual scoring systems was estimated by receiver operating characteristic (ROC) curves derived from the regression models. Nonhealing incident cases were defined as ulcers that resulted in any amputation (minor or major) and chronic nonhealing ulcers with or without death. Controls were ulcers that healed without any form of amputation. Univariate and multivariate logistic models were considered in the analysis of patient characteristics associated with ulcer outcome (sex, age, ulcer site/size/depth, ischemia, neuropathy, infection, deformity, and previous ulcer). Quartiles of age were used in the multivariate logistic models for age adjustment. ORs of clinical score systems (SINBAD and UT) for risk of being a case were calculated including and excluding ulcer deaths from the cases group. The performance for classification of risk models was assessed by ROC curves and the c-statistic (area under the operator curve) as a discrimination index. Statistical analyses were conducted with Stata/SE, version 13.1, software (StataCorp, College Station, TX), and all hypotheses tested were two sided and a P value <0.05 was considered statistically significant.

Outcomes were available for 1,645 ulcers in 1,068 patients, and baseline details are shown in Table 1. Of these ulcers, 1,182 were in males (72%) and 1,108 (67%) healed. Follow-up of the ulcers showed that in 174 cases the patient died (10.6%) before healing occurred and 144 (8.8%) patients underwent a major amputation, 177 (10.8%) underwent a minor amputation, and 42 (2.6%) had chronic ulcers or were lost to follow-up. Median follow-up time was 3 months (interquartile range 2–6) for healed ulcers and 4 months (2–8) for other outcomes.

Nonhealing ulcers were more common (Table 1) in older patients, in males, and where ischemia, neuropathy, and infection were present in those who had experienced a previous ulcer but not in those with deformity. Ulcer site, ulcer size, and depth were related to healing.

Table 2 shows the proportion of ulcers healed for each subsection of each scoring protocol. These are reported as proportion of ulcers healing when death is included as a nonhealing outcome and where deaths have been excluded as an outcome.

The proportion of ulcers healing varied from 100% in the SINBAD category 0 to 49% in SINBAD category 6, with a stepwise decrease in proportion of ulcers healing throughout each of the other categories. Foot ulcers classified in SINBAD category 4 were the most frequent (28%), with SINBAD 0 (0.7%) and SINBAD 1 (4.5%) being the least common to present to the foot clinic.

For the UT score superficial ulcers were most common (45%) with bone being exposed in 23% of ulcers. Ulcers with a combination of infection and ischemia were most common (43%), while 24% had only infection and 20% had only ischemia and while only 13% of ulcers had neither. The proportion of ulcers healing gradually declined as the ulcer became deeper and also with an increasing combination of infection and ischemia (Table 2). When infection was superimposed on an ischemic foot the outcomes deteriorated.

Tables 3 and 4 show the OR of a nonhealing ulcer in comparison with ulcers in the lowest scoring categories. For SINBAD, the risk of a nonhealing ulcer was 114- or 82-fold greater for category 6 than for category 1, depending on whether death was or was not included in the outcomes (Table 3). There was a stepwise increase in risk for each intermediary category. For the UT score, there was a graded increase in risk of an ulcer not healing by UT grade/depth and UT stage. With use of our unified score (multiplying grade and stage together), a D3 ulcer was 107 times less likely to heal than an A1 ulcer (Table 4).

The c-statistic (area under the operator curve) was 0.72 (95% CI 0.69–0.75) for SINBAD, 0.64 (0.61–0.67) for UT stage, and 0.67 (0.65–0.71) for UT grade. This improved to 0.71 (0.68–0.74) when the UT combined score was used. SINBAD was better at predicting ulcer outcome than UT stage but similar for UT grade (depth) or UT combined score.

The accuracy of predicting foot ulcer outcomes with use of the UT and SINBAD scores was similar, with a c-statistic of 0.71 and 0.72, respectively. A value of >0.7 suggests a good fit, which will be useful in clinical practice. The UT score showed decreasing rates of healing by depth and also with presence of infection, ischemia, or both. An ulcer probing to bone was five times less likely to heal than a superficial ulcer. An ulcer with infection and ischemia was 16 times less likely to heal than one without either risk factor. SINBAD also showed an impressive stepwise regression of the proportion of ulcers healing, as its scale increased from 0 to 6. The proportion of ulcers healing ranged from 100% (score 0) to 49% (score 6) with a progressive reduction of rate in between. The OR for nonhealing in comparison with SINBAD score of 1 ranged from a 5-fold decreased rate of healing with a score of 2 to an 82-fold decreased rate of healing with a score of 6.

There are a number of limitations to our study. Our study only examined patients who were referred to the hospital foot clinic and may not be relevant to those with ulcers successfully managed in the community. The outcomes in this study did not compare how long it took ulcers to heal and what impact this may have had on the ulcer grading systems. Retrospectively we wondered about assessing other systems, but the data collected were insufficient for examination of other ulcer classification systems. Data on gangrene or depth of infection were unavailable, which meant it was not possible to calculate scores for Meggitt-Wagner, PEDIS, or other systems. Lack of equipment and time in the clinic meant it was not possible to collect information for WiFi. Also, we did not undertake a reliability assessment, but UT and SINBAD have both been demonstrated to be reliable and reproducible (5). In addition, we compromised with the UT system by using foot pulses rather than Doppler. Armstrong et al. (10) did show that patients with ischemic limbs were 36 times more likely to have absent foot pulses, and absent or diminished pulses have been shown to correlate with low ankle-brachial pressure index (24). In practice, many clinicians, especially in a community setting, rely on checking pulses rather than noninvasive tests, and our results help to validate such a pragmatic approach, especially at a time when there is little consensus as to which noninvasive test is best. We could be criticized for using a unified, unique, and previously untested score for the UT system, which combines grade and stage of foot ulcer. However, we have analyzed the data using depth and ulcer stage in addition to the unified UT score. The unified score may be of use going forward, enhancing the ability to follow UT scores sequentially and numerically; however, it requires further validation in a variety of different populations. Data were missing for 2 patients on ulcer depth and 224 for ulcer size. As 61% of ulcers were >1 cm², we estimate that 137 (8%) ulcers were possibly underscored by 1 point on SINBAD. The missing data were equally distributed between healed and nonhealing ulcer groups, and we do not believe this has introduced any significant bias. A strength of our study is the large number of ulcers (1,645) analyzed. A small number of observers were used to record the data (n = 3), but previous studies have shown that there is a strong interobserver reliability for UT and SINBAD (5,25). The collection of data in a busy routine clinic is a drawback in that extensive data items could not be collected, but it also makes the study generalizable in other busy diabetes foot clinics. Also, apart from two small previous studies (19,20) this is the first large, independent, and external study comparing the outcomes of ulcer staging in SINBAD with those in UT, which are possibly the most widely used in clinical practice.

The SINBAD score is now used extensively in the England and Wales national audit. A SINBAD score of ≥3 results in the proportion of ulcers healing in 12 weeks being 35%, compared with 60% if the score is <3, with corresponding 6-month major amputation rates of 2.7% and 0.7%, respectively (26). SINBAD was validated in an international study of 1,340 patients in four centers: two from Europe, Tanzania, and Pakistan. Delayed healing was associated with a score of ≥3 of 6 (12). Ischemia, infection, and ulcer depth were the most consistent features associated with healing across all four centers. The predecessor, but more complex, ulcer scoring system of S(AD)SAD was validated (7,8,11) and used similar clinical criteria. Although these were not independent of the original investigator, subsequent studies have conformed the utility of S(AD)SAD (17). The strongest predictors of healing in S(AD)SAD and UT were depth and infection (6,7,10), which interestingly was reflected in the current study, yet depth was most important in a large U.S. epidemiological database study (27).

A few studies have compared UT with either S(AD)SAD (7,8) or SINBAD (19,20), and they demonstrated similar predictability of outcomes. The two studies comparing UT and SINBAD were external and independent but relatively small in size (n = 293 and 137). In the one that reported a c-statistic, it was 0.85 for SINBAD, 0.81 for UT stage, and 0.89 for UT depth (19), which again are similar to each other. These results are higher than seen in our study: 0.72 and 0.71 for SINBAD and UT, respectively. More extensive data were collected in a more controlled setting with Jeon et al. (19), while our data were collected in a much larger cohort within a busy clinical setting and not within a research environment. However, this may make our results more representative of a real-world clinical scenario. Patients in the study by Jeon et al. (19) had a worse overall outcome than patients in our study with an amputation rate of 49% compared with 33%. However, the study by Jeon et al. (19) had more ulcers down to bone or joint (38% vs. 23%) and more infection (66% vs. 63%), though less ischemia (59% vs. 69%).

The current study demonstrates that the UT and the SINBAD scores are equally useful in routine clinical practice for prediction of which foot ulcers are unlikely to heal when they first present. This is one of the first truly independent large studies to look at the SINBAD score in a clinical setting, and the results help support the position statement of the IWGDF (4) recommending the use of the SINBAD score. The UT score, however, is equally useful for clinicians. It is important to have a system that is easy to use in a setting where most foot ulcers are managed. Detailed systems may have their place in more specialist units. The UT system uses a 4 × 4 bidimensional matrix, which can make it more difficult to remember. However, the unified UT score presented here appears to be a useful addition to the current stage and grade scores of UT and is easier to use. Further research is required to see whether either of these systems could be adapted to enhance their predictive ability. In addition, other ulcer grading scores need further evaluation for assessment of whether they may be equally good and possible to implement in a routine clinical setting. Other, more complex, grading schemes, e.g., WiFi and SEWSS, can be helpful and important at a later stage in the clinical evaluation, e.g., in a tertiary center, especially in the attempt to define treatment pathways.

In conclusion, within the real-world setting of a diabetes foot clinic, the UT and SINBAD systems have similar utility in predicting ulcer healing when compared directly. This is by far the largest comparator study between UT and SINBAD, which are popular, internationally recommended, and widely used systems in current clinical settings. In addition, we validate a new way of integrating the UT matrix system into a unified numerical score.

Acknowledgments. The authors thank Dr. Alasdair Mackie, Vicky Green, Brian McMurray, and Joanna Donaldson, NHS Tayside, for helping in the clinic to collect data.

Funding. No funding was obtained for this study.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. G.P.L. collected the data, had the original idea for the study, examined the data, and wrote and revised the manuscript. E.S.-P. analyzed the data, helped write the manuscript, and revised the manuscript. C.S. collected the data, advised on the study, and revised the manuscript. G.P.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.