Introduction

Foot ulceration presents a major threat to people with diabetes: it is a common complaint, healing is slow and uncertain, and the overall prognosis is poor [1, 2, 3, 4, 5]. It is also a source of considerable cost to healthcare agencies [3, 6, 7, 8, 9]. However, estimation of the total burden on people and budgets is difficult because the condition is managed by many different health professionals and frequently coexists with other morbidities. Moreover, there is no consensus on ulcer classification or agreement on measures of the effectiveness of disease management. It is against this background that strategic targets have been set that use the incidence of amputation as the principal marker of the quality of care of the diabetic foot [10, 11, 12, 13].

The advantage of using amputation as an endpoint is that it is superficially simple and easy to define, with assessment facilitated by the need for hospital admission. A large number of studies have investigated the incidence of amputation in many countries and cultures of the world, but comparison of the results has been made difficult by the lack of an agreed definition, the use of different numerators and denominators, and the effect of population selection. Crucially, the incidence of amputation is dependent not just on the severity of the disease and the quality of specialist care, but on many confounding medical, social and economic factors, including professional opinion and the organisation of local health services. The present article discusses the ways in which these factors may influence the incidence of amputation by reviewing the published literature on the subject.

Search methods

The material used for this review is based on the authors’ personal knowledge of the literature, cross-references from published work, and repeated searches on PubMed and Medline using diabetes, complications, foot ulcer and amputation as keywords. Unless there was a specific indication, papers referring to activity prior to 1990 were generally excluded.

Data sources in published surveys

Population-based data

Reports should ideally be based on comprehensive, population-wide, prospective capture of all relevant clinical activity, but this is not routinely possible. Publications have therefore required detailed work, with cross-checking of records, such as operating theatre and rehabilitation unit records [6, 14, 15], and the use of capture–recapture methods [16, 17]. This situation should improve with the increasing adoption of comprehensive community-wide registers [18].

Data derived from specialist units

Data from specialist units are more likely to be accurate and complete, but are usually biased by population selection, which is affected by geography, preference, lesion type and severity, and accessibility to other specialists (e.g. vascular, plastic and orthopaedic surgery teams, and podiatric, wound care, diabetes, geriatric and rehabilitation teams). The nature and extent of the selection process may change over time. Given that data from specialist units are selected by referral, interpretation of information on the incidence of amputation requires some indication of lesion type or severity. This has been missing from previous work, mainly because of the lack of an agreed system that is sufficiently specific. New systems have been described and validated in recent years [19, 20], but international consensus is awaited [21].

Incidence of amputation: numerator and denominator

Numerator

The numerator population may refer to all amputations, to the first in any individual, or to the latest. Data may be adjusted for age, or stratified according to race or gender. Other causes of amputation, such as trauma or malignancy, are usually (but not always) excluded. Different criteria are used to define major and minor operations (see below).

Denominator

The denominator may refer to the total (diabetic and non-diabetic) population, or restricted to those with diabetes (type 1, type 2 or both, either previously diagnosed or diagnosed at presentation). Data may be derived from a population at particular risk or a cohort of people selected for prospective study.

Incidence is variously expressed as a percentage of a cohort, either in person-years or per 103, 104 or 105 (total or diabetic) population, in crude or age-adjusted figures. Use of the total (diabetic and non-diabetic) population is necessary when the prevalence of diabetes is not well documented. If incidence is expressed in terms of the diabetic (‘at risk’) population alone, this figure is itself dependent on the prevalence of known diabetes. If a community adopts systematic screening for diabetes, an increased number of the total amputations will be identified as being diabetes-related. The increased effectiveness of such screening programmes may mask a decrease achieved by improvements in specialist care. In general, however, comparisons within and between centres are more meaningful if data are expressed in terms of the diabetic population. The value of expressing incidence in terms of the total population is limited to assessing the social and economic cost, and cannot be used to examine the effectiveness of clinical care.

Definition of amputation

Amputation and debridement

The term ‘amputation’ refers to the surgical removal of part of the lower limb by transection of the leg, the foot or a digit, and necessarily includes the removal of bone. However, there may be no distinction between minor amputation and aggressive debridement, especially in cases where a digit is necrotic or where bone is infected and capable of being removed piecemeal from a foot rendered anaesthetic by neuropathy. In addition, necrotic toes may occasionally auto-amputate. The loss of a digit during the course of debridement or by auto-amputation is not usually included in estimates of incidence.

Major and lower extremity (lower limb) amputation

The term ‘major amputation’ has no agreed definition, but generally refers to the loss of a normally functional lower limb as a result of surgery at, above (usually transfemoral, rarely at the hip) or just below (transtibial) the knee. However, some definitions include operations performed below the ankle but proximal to the tarso-metatarsal joints, including Syme’s procedure and ankle disarticulation, or any operation that excludes the toes [16]. The terms ‘lower extremity amputation’ and ‘lower limb amputation’ are used with imprecision, with no distinction in practice between the terms ‘extremity’ and ‘limb’. Both are generally used to refer to all amputations of the lower limb. In the recently reported World Health Organization (WHO) multinational study of vascular disease in diabetes, the term ‘lower extremity amputation’ was also extended to include unoperated gangrene [22, 23].

Factors influencing the incidence of amputation

Population

The incidence of amputation may vary with race [24], with the prevalence of earlier major amputation being very high in certain groups. For example, 4.6% of all diabetic Mexican Americans aged >65 years have lost a leg [25]. This may be truly racial and reflect metabolic differences between different groups [26]; however, this is difficult to confirm due to the influence of diet, lifestyle, cultural beliefs and socio-economic factors [27, 28, 29, [30], 31]. If a racial group constitutes an ethnic minority in a multicultural society, then differences between groups may reflect varying access to healthcare services. Such factors may contribute to reported differences between the incidence of amputation in the US (where amputation in more common in those of Afro-Caribbean origin) and the UK (where it is less common in those of Afro-Caribbean origin, especially in men) [32]. Racial differences in the incidence of amputation between Caucasians and South Asians in the UK have been shown to be entirely attributable to differences in smoking and the prevalence of peripheral vascular disease and neuropathy [32, 33].

Aims of amputation

The aims of major and minor amputation differ. Transfemoral and transtibial operations are undertaken when the lower part of the leg is non-viable, or when it is in the patient’s best interest (because of pain or incapacity, or because function is never likely to be restored). Major amputations represent a mutilating admission of failure in the face of otherwise untreatable disease. In contrast, minor amputation is intended to limit the extent of the presenting problem in order to maintain reasonable function of the limb. The two approaches should not be regarded simply as different ends of the same spectrum; combining them will mask any changes that may occur in only one approach. Some authors have noted that a decrease in the incidence of major amputations may be accompanied by an increase in the incidence of minor ones [34].

Minor amputation

The indications for undertaking minor amputation are subject to variation and, to a certain extent, depend on the specialist training of the clinician in charge. If the clinician is not surgically trained, minor amputation requires referral to another specialist, which may constitute a significant administrative barrier. The decision to opt for early minor surgery is also influenced by prevailing medical orthodoxy. For example, minor amputation for osteomyelitis of the forefoot is regarded as routine or recommended practice in some medical cultures [35, 36], whereas the conclusion drawn from numerous observational studies is that there is a good case for managing the condition conservatively [37].

Major amputation

The incidence of major amputation is influenced by many factors, including the prevalence of diabetes, lesion severity at the time of specialist referral, the options available, and agreement between the patient and the expert that loss of the limb is (or is not) the best option. A high incidence could therefore reflect greater disease prevalence, late referral, and limited resources, but may also indicate a particularly interventionist approach by a specialist team. It is known that the incidence of major amputation varies between four-fold [38, 39] and eight-fold [40] within the same country. In the UK, this has been shown to be associated with varying professional opinion to an appreciable extent [41]. The decision of whether or not to operate may also be influenced by the system for professional and institutional funding.

The corollary of this is that a low incidence of amputation may indicate a low level of disease prevalence or disease severity, with particularly good management of diabetes in primary and secondary care, and good management of established ulcers. However, it can also reflect an inappropriately conservative approach by an expert team, and such an approach might enhance suffering by condemning a person to months of incapacity before they die with their ulcer unhealed.

Patient’s beliefs and wishes

The attitude of the patient (and/or family and carers) is a major factor in the decision of whether to amputate. Some patients will not consider the option of amputation. Faced with limited life expectancy, some will prefer to have a non-viable limb removed as soon as possible, while others will prefer to avoid major surgery and cope as best they can. Such attitudes are influenced by the advice of their carers and by the relative weight attached to professional and patient choice in the prevailing medical culture.

Reported incidence of amputation

All of the aforementioned factors influence the reported data on incidence listed below; therefore, apparent differences have to be assessed with great care [40, 42]. Because of the considerable changes that have taken place in all aspects of healthcare over the last 20 years, limited reference has been made to the older literature. When centres have reported incidence more than once, that provided in the Tables is the most recent.

All amputations (major and minor) in diabetes

The reported total incidence of amputation (major and minor) in diabetes ranges from 2.8 to over 40 per 105 total inhabitants (Table 1). When expressed per 103 people with identified diabetes, the incidence ranges from 0.46 to approximately 7 in the majority of industrialised countries, although very much higher values have been reported in some ethnic and socially disadvantaged groups (Table 2). The incidence in the US is approximately double that in The Netherlands [1]. Within Europe, the incidence varies by a factor of five to ten, with rather higher figures in Germany and Switzerland [22] compared with most of the UK and The Netherlands. Whether expressed in terms of total inhabitants or at-risk population, the incidence reported for Madrid (Spain) is the lowest [6, 16].

Table 1 Total incidence of amputation (major and minor) in diabetes expressed per 105 total (diabetic and non-diabetic) population
Full size table
Table 2 Total incidence of amputation (major and minor) in diabetes expressed per 103 at-risk (diabetic) population
Full size table

Major amputation in diabetes

Reported data on the incidence of major amputation alone (however defined and/or selected) are shown in Tables 3 and 4. If high-risk populations are excluded, the incidence of major amputation per 103 people at risk is generally less than 4.0 (Table 4). One group studied the incidences of different types of major procedure (per 103) and found that the incidence of transtibial amputation was 1.80 in The Netherlands and 2.21 in California (USA), while the corresponding values for the incidence of transfemoral amputation were 0.36 and 0.65, respectively [1].

Table 3 Incidence of major amputation in diabetes expressed per 105 total (diabetic and non-diabetic) inhabitants
Full size table
Table 4 Incidence of major amputation in diabetes expressed per 103 at-risk (diabetic) population
Full size table

Reported changes in amputation rate

All diabetic and non-diabetic amputations

The incidence of non-traumatic amputation has been closely monitored in an attempt to evaluate the massive increase in the number of angioplasty and other revascularisation procedures performed. However, the results have been conflicting. Tunis and colleagues reported no change in the total incidence of amputation in Maryland, USA between 1979 and 1989 [55], while Feinglass and colleagues reported that there was a small decrease in the incidence of major amputation in the US in the early 1990s that was not maintained [56]. In contrast, the incidence of major amputation is reported to have decreased in the UK (although the fall largely preceded the increase in angioplasty) [57] and in Denmark [58]. A reduction in all amputations was reported by a study on Veterans Health Administration facilities (a predominantly male, lower-income study population) [59]. One Finnish study reported a decrease in the incidence of major amputation [60], whereas a different group from Finland failed to observe a decrease in the total incidence of amputation. It should be noted that in some of the cases where a decrease was observed, the baseline and final incidences were higher than those in equivalent populations.

Amputation in diabetes

Following the introduction of targeted education and care programmes in high-risk populations in the US, a decrease in the total incidence of amputation in diabetes has been reported by some groups [26, 49, 62], but not all [63]. In the Veterans Health Administration population, the decrease in incidence was confined to the non-diabetic subjects [59]. The overall incidence of amputation for diabetes in the US has not changed [64].

The data from Europe are conflicting. Various groups have reported a decrease in the total incidence of amputation in diabetes [16, 18, 44, 53]; however, in some cases the baseline incidence was high [18, 44]. Nevertheless, the decrease observed across Denmark was from a more modest baseline [53], while that in Madrid was already low [6]. In contrast, two community-based studies from Germany failed to demonstrate any decrease over periods of 8 and 5 years respectively [47, 48]. Significant decreases in the incidence of major amputation have also been reported from Denmark [34, 58] and Sweden [44], but the baseline was high in each case. The estimated decrease over 11 years in Tayside, Scotland is subject to the same limitation [18]. More encouragingly, recent nationwide data from The Netherlands have indicated a significant decrease in the incidences of amputations at all levels between 1991 and 2000, even though the baseline was not high [65]. In Madrid, the low baseline incidence of major amputation has been further reduced [16].

Conclusions

There are many reasons why there is wide variation in the reported incidence of amputation. These relate to medical, cultural and social issues, and are very dependent on the structure of healthcare services, as well as on the attitudes of the professionals involved. However, reported figures are also critically dependent on population selection and on the way in which the data are collected and analysed. Future analyses and comparisons will not be meaningful unless considerable efforts are made to optimise and harmonise the methods used. Until this occurs, the incidence of amputation must be regarded as a flawed measure of the quality of clinical care. Ultimately, however, the aim should be to achieve not only a reduction, but also an overall low incidence of amputation. Furthermore, a low incidence of major amputation needs to be qualified by measures of the morbidity and mortality of those who do not undergo surgery so as to ensure that their suffering is not increased. There is a clear need for a consensus panel to be established under the auspices of the WHO or the International Diabetes Federation to determine how the quality of care of the diabetic foot can be best monitored.