Non-traumatic major lower-extremity amputations (LEA) are a common consequence of long-term diabetes mellitus. Other well-known risk factors for LEA include coronary heart disease, cerebrovascular disease, or peripheral arterial disease (Crawford et al, 2015). A major lower limb amputation is defined as the surgical removal of a part or whole limb proximal to the ankle (Ajibade, 2013). They are performed in cases of excessive tissue loss, sepsis or if there are no further surgical or endovascular options for revascularisation (Rümenapf and Morbach, 2014). Further reasons for amputation are following trauma or accident, cancer or tumour, or orthopaedic deformity (Lazzarini et al, 2011). They are additionally classified as major and minor depending on whether the amputation is above (major) or below (minor) the ankle joint.
Prevalence rates help to identify the extent and depth of a problem. Establishing how prevalence rates may have changed over time is important, as it allows public health professionals to target their resources, implement preventative strategies and plan their services on a local, regional and national level (Hoffmann et al, 2013).
Published data globally demonstrates considerable variation in the incidence of major LEA ranging from 5.6–600 per 100,000 in the population with diabetes and from 3.6–58.7 per 100,000 in the total population (Moxey et al, 2010). More recent studies are within the range reported by Moxey et al (2010), with the average reported major amputation incidence (from 20 international reports worldwide) of 14.5 per 100,000 in the total population (Kolossvary et al, 2016). Fortunately, in England, the rates of major amputation have reduced overall by approximately 18% over the past 10 years, but still remain six times higher in the diabetic population (Ahmad et al, 2016). The excess risk of amputation in the diabetic compared with the non-diabetic population has been reported as ranging between 7.4 and 41.3 times higher with the variation explained by differences in study design and definitions used (Narres et al, 2017). Most variation in the prevalence of any condition in a population can, however, be explained by age, sex and social class (Bhopal, 2008). It is important, therefore, that any explanation of differences, for example, excess amputation rates in people with diabetes, are compared using the same definitions of diabetes, amputation, as well as across the same age, sex and social class groups.
In order to compare rates across populations and time, it is imperative epidemiological principles in calculating and reporting prevalence are followed. These include using the same numerator and denominator populations and age standardising the overall population rate. Further, age and gender specific rates should be provided as they are far more accurate than the age standardised rate and, additionally, allow variations to be explored.
Aim
The aim of this systematic review was to highlight all epidemiology studies based in England that describe the population prevalence of major lower-limb amputation in those with and without diabetes. The authors describe time trends across populations in those studies where such comparison is possible.
Methods
Systematic searches were performed using PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) guidelines, from a comprehensive range of databases: PubMed, Cumulative index to Nursing and Allied Health Literature (CINAHL), Excerpta Medica Database (EMBASE), Medical Literature Analysis and Retrieval Online (MEDLINE), Cochrane database, NHS Digital, Diabetes UK and Healthcare Quality Improvement Partnership (HQIP). The search included studies performed in England, published in English, between January 1, 1988 until December 31, 2018.
The key search terms (combined by Boolean operators) were: (amputation) AND (Lower limb)OR (lower extremity) AND (prevalence) OR (Incidence) OR (epidemiology) OR (frequency) NOT (cancer) NOT (trauma) AND (England) OR (UK) OR (British).
A forward citation search using Google Scholar was performed to identify new research that has yet to be indexed in any database. A grey literature search (research that is either unpublished or has been published in non-commercial form such as government reports, fact sheets, theses and dissertations) was performed to identify any further publications. In addition to the electronic searches, reference lists of the included articles were used as a source to hand-search for relevant citations.
The preliminary search yielded a total of 207 journal articles conducted by the primary author (Maria Davies; MD) and repeated by a secondary author (Lauren Burdett; LB). After removing duplicates and those papers published prior to 1988, the two screeners (MD and LB) reviewed the titles and abstracts for their eligibility based on the defined inclusion and exclusion criteria (Table 1); this led to the exclusion of 162 studies. The full-texts of the remaining articles were then assessed for their eligibility, resulting in a further 31 studies being excluded. Any disputes were settled by a third screener (Naseer Ahmad; NA). Eleven studies were deemed to meet the eligibility criteria and included in the review (Table 2). Eight of the papers provided results specific to both the diabetic and non-diabetic population, the remaining three did not differentiate between the two populations.
Quality assessment was performed on the remaining studies using the STROBE (STrengthening the Reporting of Observational studies in Epidemiology) checklist.
Results
Eleven studies reporting on amputation rates from within England between 1988 and 2018 were included in the analysis. Table 3 summarises the main methodological weaknesses of these studies.
There was considerable variation in the reported prevalence of major LEA, ranging from 0.7 to 332.4 per 100,000 in the diabetic population and 3.0 to 76.1 per 100,000 in the general population.
There were four different definitions describing major LEA in the literature. Canavan et al (2008), Leggetter et al (2002), Unwin (2000) and Vamos et al (2010a; 2010b) defined major LEA at a level through or distal to the ankle joint. The remaining studies agreed that any amputation through or distal to the ankle joint is classified as a minor amputation.
Moxey et al (2010) and Vamos et al (2010a) made a mistake with the denominator populations by using the population of England as the numerator but the entire population of Great Britain as the denominator. Moxey (2010) was the only paper to publish enough methodological data to check the prevalence rate was calculated correctly. They provided rates per 10,000 rather than 100,000. They also used only 1-year population data (2008) and not five-year averages. If calculated correctly then they would have reported a national prevalence rate of 51/100,000 not 5.1/100,000.
Canavan et al (2008), Holman et al (2012), Leggetter et al (2002), Moxey et al (2010), Rayman et al (2004) and Vamos et al (2010a) did not define the denominator population age groups, while Rayman et al (2004) and Vamos et al (2010a; 2010b) did not provide the source of the denominator population.
There were inconsistencies in the age groups studied, ranging from all ages, including children, to only those in the age bracket 50–84 years. Leggetter et al (2002), Moxey et al (2010), Rayman et al (2004) and Vamos et al (2010a) did not define the age group studied, and Leggetter et al (2002), Moxey et al (2010), Rayman et al (2004) and Vamos et al (2010a; 2010b) did not age standardise their results. Results were rarely presented with age and gender specific breakdowns, with only Ahmad et al (2016), McCaslin et al (2007) and Unwin (2000) providing age specific results and Ahmad et al (2014; 2016), Unwin (2000) and Vamos et al (2010b) providing gender specific results.
Cavanan et al (2008), Rayman et al (2004) and Unwin (2000) gathered their numerator data from medical records. The remaining eight studies by Ahmad et al (2014; 2016), Holman et al (2012), Leggetter et al (2002), McCaslin et al (2007), Moxey et al (2010) and Vamos et al (2010a; 2010b) extracted amputation numbers from the Hospital Episode Statistics (HES) database using the Office of Population Censuses and Surveys — 4th Revision (OPCS-4), Classification of Surgical Operation Codes to record the level of amputation. Leggetter et al (2002), McCaslin et al (2007) and Vamos et al (2010a) published OPCS-4 codes that did not correlate with their written definition.
Canavan et al (2008), Leggetter et al (2002), and Unwin (2000) identified people with diabetes through their medical records, as did the prospective study by Rayman et al (2004). The World Health Organization’s International Classification of diseases (ICD-10) is used to code conditions such as diabetes mellitus. The remaining studies by Ahmad et al (2014; 2016), Holman et al (2012), McCaslin et al (2007), Moxey et al (2010) and Vamos et al (2010a; 2010b) used ICD-10 codes to identify people with diabetes mellitus. Vamos et al (2010a; 2010b) excluded three (E12–E14) of the five codes that the other papers included.
Discussion
There was wide variation in those studies presenting amputation prevalence in England. Generally rates were higher in the diabetic than non-diabetic population but the range of presented rates varied such that there was significant overlap. The variation stemmed from differences in the definitions of major LEA and denominator populations, different amputation capture techniques and lack of both age standardisation of overall rates and presentation of age and gender specific rates. These variations also made comparison of amputation rates over time difficult.
The only way to describe time trends is if two papers calculate prevalence in the same way. Only one paper by Ahmad et al (2016) provided both age standardised and age-sex specific results over a 10-year period. This found major amputation rates to be falling overall, faster in the population with diabetes than those without and interestingly minor amputation rates (defined as those below the ankle) to be rising in both populations but at a faster rate in those without diabetes and also men compared with women. The effect of age standardisation was also shown in Germany where Spoden et al (2019) investigated amputation rates between 2005–2015. There was an overall increase in lower-limb amputations, however, after age-sex standardisation an overall decrease of 11% was revealed. The increase in amputation numbers is attributed to the aging German population.
The same can be seen in the study by Ahmad et al (2016); amputation numbers increased, yet after age-sex standardisation there was an overall decrease in major LEA by 40% in the population with diabetes and 26% in the population without diabetes. This may explain why McCaslin et al (2007) and Vamos et al (2010b) reported lower rates of major LEA in people with diabetes, however, without the demographic data this cannot be proven.
Conclusion
Significant variation in methodology and reporting of results hinders evaluation of time trends of amputation rates. A standard method of reporting amputation rates is urgently required. The next step is to create a standard to which future amputation data should be presented.
