Mortality in Diabetic and Nondiabetic Patients After Amputations Performed From 1990 to 1995: A 5-year follow-up study Free

We reviewed all nontrauma, nonneoplasm-related amputations (ICD9-CM codes 84.10–84.18) (13) performed at the Manchester Royal Infirmary between 1 January 1990 and 31 December 1995. Participants in the diabetic group (n = 102) were patients who had been diagnosed with diabetes according to their medical records. In addition, patients with unknown diabetes but with fasting serum glucose values ≥126 mg/dl measured twice were considered as having diabetes. Nondiabetic amputees (n = 154) were patients without a history of diabetes and with fasting serum glucose levels <126 mg/dl. Duration of diabetes was from the time of diagnosis until the first event amputation. Presence of diabetic nephropathy, retinopathy, and history of smoking were noted from the medical records. Nephropathy was defined as either serum creatinine values >130 μmol/l for both sexes and/or 24-h urine protein >300 mg/dl. The demographic and clinical characteristics of the study patients are shown in Table 1.

Major amputation refers to any amputation above the midtarsal level (14). In addition, the exact level of amputation was reported. The second amputation refers to amputation performed to the same or the contralateral leg after healing of the first amputation. Reamputation refers to a revision of an amputation to a higher level during the same hospitalization. Absence of at least two of four peripheral pulses in both feet and a history of intermittent claudication, ischemic rest pain, gangrene, or any previous revascularization procedure (bypass graft or percutaneous transluminal angioplasty) at the peripheral arteries were all considered an indication of the presence of peripheral vascular disease. Ischemic heart disease (IHD) was determined by 12-lead electrocardiogram, hospital records of confirmed myocardial infarction, and definite history of angina or coronary revascularization procedure. Cerebrovascular disease was assessed by history, clinical examination, and hospital records of definite stroke. The cause of the amputation was noted from the medical records. Follow-up was from the time of the first amputation until death or 31 December 2001, whichever came first. Five patients (two with and three without diabetes) were excluded from the analysis because their survival status at the end of the study was not known. Patients who had an amputation before 1990 and a new amputation in the study period were also excluded from the study. When multiple amputations were performed on the same patient during the same hospitalization, only the most proximal amputation was used in the data analysis. Perioperative mortality refers to deaths that occurred in the first 28 days after the first amputation.

All patients had blood pressure, serum glucose, creatinine, and cholesterol values measured as well as plasma hemoglobin and white blood cell count (WBC) on admission to the hospital. Hypertension was defined according to older World Health Organization criteria (systolic >160 and/or diastolic >95 mmHg or on antihypertensive treatment on admission) (15). The majority of the diabetic patients had HbA_1c values determined during their hospitalization.

Statistical analyses were performed using programs available in the SPSS 10.0 statistical package. The Student’s t test and ANOVA were used to assess differences in continuous variables, while the χ² test was used for categorical variables. The all-cause mortality hazard ratios (HRs) for both continuous and categorical variables were based on Kaplan-Meier estimates and compared by log-rank test. Cox proportional univariate and multivariate hazards regression analyses were performed to examine the variables predictive of all-cause mortality. The values of serum glucose, creatinine, and total cholesterol, as well as of hemoglobin and WBC on admission were used in the analyses. The multivariate models included those variables found to have an association (P < 0.05) or suggestive of an association (P < 0.20) in univariate analysis. Serum creatinine values were log transformed because the underlying data were skewed. The HR for this variable, therefore, corresponds to a 10-fold increase in concentration. The results are presented as HRs.

Of the 257 amputations performed during 1990–1995, 39.7% were in patients with diabetes. Nondiabetic patients were older than diabetic patients, while the proportion of women was higher among the diabetic patients, as was the prevalence rates of IHD, hypertension, and nephropathy (Table 1). More nondiabetic patients had major amputations in comparison with diabetic patients. Mean duration of follow-up was comparable between the two groups; however, the time to death was shorter in the nondiabetic than in the diabetic patients (Table 1).

Altogether 61% of the diabetic patients and 54.3% of the nondiabetic patients died during follow-up. Deceased diabetic patients had a longer known duration of diabetes, had higher serum creatinine levels, and were more likely to have undergone revascularization procedures in the leg arteries (Table 2). Deceased nondiabetic patients were older, with a higher prevalence of IHD, an absence of pedal pulses, and major amputations. They also had higher WBCs and serum creatinine values on admission (Table 2). Perioperative mortality was 5.9% in diabetic patients and 9.1% in nondiabetic patients (P = 0.34).

Kaplan-Meier survival curves showed that mortality rates at 1, 3, 5, 7, and 9 years were 17, 37, 44, 56, and 68% in the diabetic patients compared with 21, 35, 44, 48, and 68% in the nondiabetic patients (Fig. 1), respectively. Mean (95% CI) survival time was 6.5 (5.6–7.3) years in the diabetic patients and 6.7 (5.9–7.5) years in the nondiabetic patients (P = 0.49).

Mean (95% CI) survival time after the first amputation was not different between men and women with diabetes [6.6 (5.4–7.8) and 5.9 (4.6–7.2) years, respectively, P = 0.61] or between nondiabetic men and women [6.7 (5.8–7.7) and 6.9 (5.5–8.5) years, respectively, P = 0.98].

When analysis was performed using the Kaplan-Meier survival curves and after adjustment for the level of amputation (major versus minor), survival status was not different between diabetic and nondiabetic patients (χ² = 0.17, P = 0.67); the same was valid after controlling for the cause (ischemia versus infection) of amputation (P = 0.72). In addition, survival was better in the younger than in the older (≤65 vs. >65 years of age at the first amputation) nondiabetic patients (P = 0.001). No age differences in survival were found in the diabetic group (P = 0.37).

Univariate Cox regression analysis in diabetic patients showed that duration of diabetes, log serum creatinine, and a history of stroke were each associated with increased all-cause mortality (Table 3). The level of amputation was not associated with the survival status, even when above-knee amputations were compared with amputations limited to the foot [HR 1.07 (0.51–2.21), P = 0.42]. Multivariate Cox regression analysis, after controlling for duration of diabetes, presence of hypertension, history of stroke, serum creatinine level, and previous revascularization procedures, demonstrated that duration of diabetes [1.50 for an increase of 5 years (1.20–1.86), P = 0.05], a history of stroke [1.85 (1.06–3.20), P = 0.02], and log of serum creatinine levels [11.16 (3.77–33.02), P < 0.0001] were independently associated with increased mortality in diabetic amputees.

Univariate Cox regression analysis in the nondiabetic patients showed that age, a history of stroke, serum creatinine level, and a higher value of WBC on admission were associated with increased mortality (Table 3). Multivariate Cox regression analysis, after controlling for age, serum creatinine concentration, level of amputation (major versus minor), history of stroke, and WBC (>10.0 × 10⁹/l vs. <10.0 × 10⁹/l), demonstrated that a history of stroke [HR 1.62 (1.01–2.61), P = 0.04], log of serum creatinine levels [7.65 (1.85–31.61], P = 0.005], and a higher WBC on admission [1.76 (1.09–2.84), P = 0.02], were independently associated with increased mortality in nondiabetic patients.

Ischemia (rest pain, gangrene, and nonhealing ischemic ulcer) was the cause of amputation in 69.6% of the diabetic patients and in 82.5% of the nondiabetic patients. Infection (severe soft tissue infection and osteomyelitis) was the cause of amputation in 28.4% of the diabetic patients and in 8.4% of the nondiabetic patients (P < 0.0001). In addition, the number of revascularization procedures performed during the study period was not different between diabetic and nondiabetic patients; 14.7% of the diabetic patients had two or more revascularization procedures compared with 15.6% of the nondiabetic patients (P = 0.84).

Prevalence of amputations by age strata for diabetic patients and nondiabetic patients is shown in Fig. 2. Increasing prevalence was observed from the younger to the older age-group. The peak prevalence was observed in the age-group 67–76 years in both diabetic and nondiabetic patients. In all age-groups, major amputations were more common in the nondiabetic than in the diabetic patients.

Altogether, 40.0% of the diabetic patients and 29.8% of the nondiabetic patients had a second amputation to the same or contralateral foot in a mean (95% CI) time of 16.2 (8.6–23.8) and 12.3 (7.3–17.2) months, respectively (P = 0.37), after the first amputation. More often diabetic patients had two or more second amputations during the study period in comparison with nondiabetic patients (54.9 vs. 36.4%, respectively, P = 0.003). Reamputation rates (3.9 and 3.2%; P = 0.78) and perioperative complications after the first amputation (42.1 vs. 37.7%, respectively; P = 0.47) were similar in the diabetic and nondiabetic patients.

The median (interquartile range) time of hospitalization after the first amputation was similar in diabetic and nondiabetic patients [28 (16.7–50.5) vs. 25.5 (13.2–48.7) days, respectively, P = 0.45].

In the previous decade, mortality rates after an amputation had remained high in both diabetic and nondiabetic patients. In addition, this study has shown that survival, hospital stay, postoperative complications, and mortality were comparable between diabetic and nondiabetic amputees.

Diabetic amputees were noted to have high mortality rates in our study. Our data agree with previous reports of high mortality rates in diabetic patients with amputations (4,9–12). In addition, it was demonstrated that presence of nephropathy and a history of stroke were associated independently with increased mortality in both study groups. Furthermore, duration of diabetes was also associated with increased mortality in the diabetic group.

Renal status is a well-known factor affecting survival in diabetic patients (16). Prevalence of nephropathy was high in our study subjects with diabetes (53%), yet 5-year mortality was similar to other cohorts with less frequent nephropathy (11). A novel finding of this study was the independent association between a higher WBC and survival in the nondiabetic amputees. Leukocytosis is associated with increased in-hospital mortality after acute myocardial infarction, higher mortality after percutaneous coronary intervention (17), and poorer survival in patients with end-stage renal disease (18). Leukocytosis, and especially increases in neutrophils, may reflect an inflammatory process. Why a higher WBC is not associated with mortality in diabetic patients is not known.

The results of this study agree with previous reports (9–12) showing that amputations in individuals with diabetes are performed at a younger age and are more distal. Ischemia was a common cause for amputation in both study groups, yet infection as the cause of amputation was almost 3.5 times more common in diabetic than in nondiabetic patients. On the other hand, our findings do not agree with previous reports showing higher mortality rates in subjects with major amputations (11,12,19) and peripheral vascular disease (20, 21) in female subjects (11) or in older diabetic subjects (11). Interestingly, the number of revascularization procedures performed before the first amputation, which might be a better index of the aggressiveness of the vascular disease, was also not associated with increased mortality. However, recent studies have shown increased mortality rates among diabetic patients with neuropathy (21) and foot ulcers (21,22) in comparison with diabetic patients without these complications. These findings suggest that neuropathy and foot ulcers, conditions that precede amputation, may serve as markers of as-yet-unknown conditions increasing mortality.

Published data concerning comparison of survival and hospital stay between diabetic and nondiabetic patients are scarce. Some studies (12) found better survival rates in a 5-year period after the first amputation in diabetic amputees. We have also shown similar survival rates between diabetic and nondiabetic patients, even when differences between the groups studied, such as age and level of amputation, were taken into account. The reason for this may be that major cardiovascular disease was very common in both study groups. However, the diabetic patients had more often (almost by 30%) two or more amputations in the study period than the nondiabetic patients.

In the past, a longer hospitalization and a higher prevalence of perioperative complications was reported for diabetic amputees, both being the main determinants of the direct cost associated with amputations (23). In-hospital stay (28 days in our diabetic patients) was comparable with recent data from Australia (24.7 days) (24), less than the mean in the Netherlands (42 days) (23), and more than the mean in the U.S. (15.9 days) (10,23). It is of note that in multicenter studies (24), large regional variations in hospital stay have been reported, which probably reflects variations in clinical practice and access to health care services. In addition, the data from this study show that hospital stay and the perioperative complications were not different between diabetic and nondiabetic patients, possibly implying improvements in the care of patients with diabetes. Our data agree with recent reports from other countries, showing that the hospital stay is comparable between diabetic and nondiabetic patients (25).

This study, being retrospective and observational by design, potentially suffers from the limitations of such observations. Thus, factors that have been found to be associated with increased mortality, such as a low ankle-brachial index (11) and presence of neuropathy (21), were not studied. However, it has the advantage that parameters such as differences in surgeons, pre- and postoperative care, and availability of foot clinics could be controlled because the patients were from one area and were managed in the same hospital. Cause-specific mortality was not examined in this study. However, previous data has conclusively shown that cardiovascular disease is the leading cause of death in patients with amputations (10,12, 26).

In summary, this study confirmed the high mortality rates in both patients with and without diabetes after an amputation. In addition, we have shown that in the previous decade, mortality rate, in-hospital stay, and postoperative complications were not different between subjects with and without diabetes. No modifiable factors, except for the prevention of nephropathy, were found to improve survival in amputees. Peripheral vascular disease and neuropathy are the leading causes of amputations in both diabetic and nondiabetic patients; therefore, prevention of these complications is warranted to prevent amputation and subsequent high mortality.