New Ulceration, New Major Amputation, and Survival Rates in Diabetic Subjects Hospitalized for Foot Ulceration From 1990 to 1993: A 6.5-year follow-up

Between 1990 and 1993, 115 diabetic subjects were hospitalized for foot ulceration at our center(19); they were monitored after discharge until 31 December 1998. The reported data concerned the new ulceration, the new major amputation, and the survival rates of these subjects. We evaluated prognostic factors for these events.

The clinical characteristics investigated between 1990 and 1993 were reported in Table 1. During that period, 27 subjects (23.5%) underwent major amputation (above the ankle), and two of these subjects died in the hospital from septic shock. A total of 88 subjects (76.5%) were considered healed, 53 of whom recovered with and 35 without minor amputation. Of the 115 patients, 95 (82.6%) had sensorimotor diabetic neuropathy and 97 (84.3%) had peripheral vascular disease(ankle-brachial index [ABI] <0.9, trancutaneous oximetry on the dorsum of the foot ≤50 mmHg, and a peripheral angiography positive for occlusive disease). A total of 18 (15.7%) patients were considered purely neuropathic and 97 (84.3%) were ischemic or neuroischemic. Of these patients, 29 (29.9%)underwent vascular procedures (19 peripheral transluminal angioplasties and 10 bypass grafts).

All 115 subjects were monitored until 31 December 1998. The subjects from our region (57.4%) were examined approximately every 2 months, and patients not living near our center (42.6%) were examined once a year. Patients and families were instructed to visit the center if any sign of lesion appeared.

All of the subjects who had undergone a major amputation were given prostheses ∼3 months after discharge. Approximately 1 month after discharge, all of the healed subjects were provided with extra-deep rocker shoes with soft thermoformable leather (Buratto, Crocetta del Montello, Italy)and customized insoles, shaped by a cast, in Alkaform (derived from Plastazote) and Alcapy (derived from Professional Protective Technology, Deer Park, NY). In addition, all patients, including amputees, received a therapeutic shoe for the unaffected foot. The insoles were modified or changed every 6 months, and the shoes were modified according to wear.

New ulceration was defined as any ulceration at the same or different site of a previous ulcer or an ulcer in the contralateral foot. Major amputation refers to that performed above the ankle.

We examined 196 limbs: 2 limbs in 81 healed subjects, 1 limb in 7 subjects who had undergone a major amputation before 1990, and the remaining limbs of the 27 subjects who had undergone a major amputation during hospitalization from 1990 to 1993. All of the following were recorded: homolateral and contralateral new ulceration, major amputation of the limb healed during hospitalization, reamputation of the same limb in the subjects undergoing major amputation from 1990 to 1993, major amputation of the contralateral limb, and date and cause of death.

Table 1 shows the association between the incidence of events and the considered variables. However, because we did not consider HbA_lc and because procedural methods for and frequency of the determination of variables differed for subjects living far from the center, the data are not reliable.

From the time of hospital discharge (time 0), Kaplan-Meier survival curves were calculated for all causes of death, and the crude percentage of survivors was reported according to the modalities of discrete variables or classes of continuous variables observed at the cutoff points reported in the tables. All patients were exposed to risk of death for at least 5 years.

We used the log-rank test for comparing groups with respect to survival,and we performed a univariate analysis using Cox's regression model to identify the prognostic role of the considered variables. P < 0.05 was sufficient to statistically prove the association with patient survival time. We did the same analysis with a multivariate Cox's regression model and a stepwise selection of variables to identify the prognostic factors independently associated with new ulceration and survival time. The entry criterion was P ≤ 0.10, and the permanence criterion was P ≤ 0.05. For dichotomous variables, the constants estimate the mortality hazard ratios and the 95% CIs of the hazard ratios. For age, the hazard ratio and its CI refer to the unitary increment (1 year).

The proportional hazard assumptions were tested graphically for all of the variables with the plot of —log {—log [survivors proportion(t)]} versus survival time (t). Elaboration was performed with STATA 5.0 (Stata, College Station, TX) survival analysis routines.

The mean follow-up was 78.3 ± 15.3 months(minimum 60 months and maximum 106 months).

According to the Wagner classification system, there were 25 (12.75%)episodes of new ulceration: 4 grade IV, 3 grade III, 4 grade II, and 14 grade I episodes. All 25 episodes, 13 homolateral and 12 contralateral in terms of the first ulceration, occurred in 23 of the 88 subjects who recovered in the period between 1990 and 1993. No ulceration occurred on the stumps of the amputees. A total of 12 patients were hospitalized, 13 were treated in the outpatients' department, 3 underwent amputation below the knee, 5 underwent amputation of one or two toes, and 17 recovered without amputation. The average period of time the subjects were free from new ulceration was 32.7± 23.8 months. At univeriate analysis, no variable in Table 1 was significantly associated with new ulceration.

No reamputation of the same limb was carried out in the 27 patients undergoing major amputation in the 1990-1993 period. In the 88 subjects healed without major amputation, three major amputations were performed: one of the contralateral limb and two of the same limb healed in the 1990-1993 period. The average period of time the subjects were free from amputation was 9.3± 3.2 months: 7.5 months for amputation of the same limb and 13 months for amputation of the contralateral limb. Because of the small number of major amputation events, we decided that the possible association with the variables in Table 1 was not significant.

By 31 December 1998, 64 (55.7%) subjects were alive, and 51 (44.3%) had died (24 of 31 women [77.4%] and 27 of 84 men [32.1%]). Mortality concerned 20 of the 27 (74.1%) subjects who underwent major amputation in the 1990-1993 period and 31 of the 88 (35.2%) healed subjects with a highly significant difference (hazard ratio 3.020, 95% CI 1.71-5.33, P < 0.0001). Table 2 shows the causes of death. Figure 1 shows the cumulative survival curve and the survival curve by major amputation carried out during hospitalization from 1990 to 1993.

The results of univariate analysis are shown in Table 3. The multivariate analysis of the variables that were found to be associated with death during univariate analysis confirmed the independent role of ABI ≤0.5 (hazard ratio 2.29, 95% CI 1.29-4.08, P = 0.005), age (hazard ratio by increase of 1 year 1.05, 95% CI 1.02-1.08, P = 0.003), and the female sex (hazard ratio 1.96, 95% CI 1.08-3.56, P = 0.027). In Fig. 2, the relative survival curves are illustrated. The association with major amputation carried out during hospitalization from 1990 to 1993 is alternative to ABI ≤0.5.

Published data concerning new ulceration are scarce. One study (20) during the 1960s showed a 23% increase in rate of ulceration of the opposite extremity within a year in diabetic subjects amputated for gangrene. Only 6%had an intact contralateral limb after 5 years. Recurrence rates for ulcers in neuropathic subjects were estimated at 32%(21), and the incidence of ulcers at the same or different sites in a foot with prior ulceration was estimated at 50% over a period of 2-5 years(22).

In two recent studies(23,24),the use of special shoes significantly decreased the incidence of new ulcerations. The population considered in these studies was, however, an outpatient population of neuropathic subjects, whereas our population mostly comprised neuroischemic subjects hospitalized for severe foot ulcer.

All of our subjects had therapeutic shoes and received education on foot care; their families also received education on foot care. Our patients also received slippers with a rigid sole and thermoformable lining. The percent of new ulcerations in our population was very low, and the period of time the subjects were free from new ulceration was reasonably long.

The lack of significant association between new ulceration and the investigated variables surprised us. We believe that some variables that we did not record, such as social status, limited joint mobility, degree of patient compliance regarding the continual use of the orthesis, and educational learning(25,26,27,28),may play a role in episodes of new ulceration. We excluded alcoholism(29), because none of our patients consumed much alcohol. The lack of association with minor amputation seems particularly interesting; we believe that this indicates the effectiveness of orthesis, even in a deformed foot.

Available published data on ipsilateral and contralateral amputation is discouraging. Reiber(7,30)reported that 13% of amputees in California and 9% in New Jersey would undergo reamputation within a year of initial amputation, and he estimated that 30-50%of the amputees would undergo contralateral limb amputation within 1-3 years of initial amputation. In the 1980s, two studies(8,9)reported an incidence rate of contralateral amputation of 44.3% and ipsilateral amputation of 23.1% after a 4-year follow-up and an incidence rate of contralateral amputation of 26.4% after a 5-year follow-up.

Considering these data, we believe that the number of new ulcerations and new major amputations in our follow-up was very low. We reached this goal by providing instructions and education regarding orthesis to our patients(31) and their families. It seems that the use of therapeutic shoes is recommended(32), but, to date, they are not used much among diabetic patients(33,34). It should be emphasized that there is no difference in new ulceration between patients monitored directly at our center and patients monitored infrequently at distant locations. Therefore, undergoing an intense informational program on admission to a hospital for ulceration and the fitting of therapeutic shoes may reduce the number of new ulcerations, an independent risk factor for amputation (35). Thus, we feel that the American Diabetes Association(36) should provide more specific information on and a more assertive recommendation for the use of therapeutic shoes for subjects with previous diabetic foot ulceration.

Because of the small number of new major amputations, we did not feel that it was necessary to evaluate an eventual association with the variables in question. However, we must specify that all three patients amputated in the follow-up had an ABI <0.5 and could not be revascularized.

Our data lead us to conclude that the current epidemiological data on new ulceration, new major amputation, and reamputation in diabetic subjects who have undergone major amputation or who have recovered from a severe foot ulcer are much more accurate than those previously reported. In these subjects,major amputation only occurred in those with severe occlusive disease, when revascularization was impossible.

The rate of survival in our amputees was similar to that found in other published studies. Reiber (30)reported a 39-68% mortality rate over a 5-year period. In the study by Deerochanawong et al. (17),the mortality rate was similar to ours, but the follow-up period was shorter. In addition, Frykberg et al.(12) reported a higher mortality rate, but their subjects were much older. The data reported by Rosemblum et al. (14)presented a better outcome than previously reported data, but Rosemblum et al. chose a population who had undergone vascular procedures under general anesthetic, presumably with a low recurrence of ischemic cardiopathy. Survival rates of our nonamputated subjects present a worse outcome than those reported in a recent study by Ramsey et al.(10), who analyzed the survival of diabetic patients after ulcerative episodes, but during a very short observation period. Furthermore, the presence of ischemic cardiopathy in the Ramsey population (7%) was much lower than that in our study (47%), in which it was the most frequent cause of death.

In addition, the protective value of smoking in our population was particularly surprising. These results were contingent on the fact that the women, who were older and had a higher mortality rate, were nonsmokers. However, when sex and age were not considered, smoking lost its significant association.

In conclusion, ischemic cardiopathy is the most frequent cause of death among diabetic subjects with foot ulceration. These data are well documented in nondiabetic and diabetic subjects with peripheral vascular disease(37,38),but little is known about diabetic subjects with foot ulceration who present with ischemic cardiopathy. Therefore, we modified our protocol to adopt a more aggressive diagnostic and therapeutic approach of ischemic cardiopathy in diabetic subjects who were admitted to our hospital for foot ulceration.