This study appears to be the first in over 20 years to investigate a large series of patients undergoing any level of lower extremity amputation in an Australian population . The findings align with international studies indicating that people undergoing amputations were more likely to be older (mean 62 years), male (by a ratio of 2:1) and have diabetes or peripheral arterial disease (approximately 80%) [1, 4, 5, 9, 13, 14, 16]. Also the major to minor amputation ratio (1.19:1) seems to be consistent with international literature [1, 10, 36].
Whilst, this study’s results align mostly with those already reported in the international literature, it did identify some interesting findings with regard to: proportions of key conditions associated with amputation, mean age at amputation for key conditions, differences within key conditions, and first amputation status.
The proportions of key conditions associated with amputations in this study were mostly within ranges reported in existing international literature [1, 4, 5, 9, 14]. Although, interestingly the major key condition associated with lower extremity amputation found in this study, diabetes, was not reported as a key condition associated with lower extremity amputation in the last similar Australian study published in 1990 . The proportions of diabetes-related (type 1 and type 2) (60.2%) and malignancy-related (5.4%) amputations appear to be disproportionately high compared to the prevalence of these conditions in the general Australian population (diabetes (4.0%) and malignancy (1.6%)) . Furthermore, diabetes-related amputations seem to remain disproportionately high even when compared with similar ‘higher risk’ populations observed in this study; i.e. in adults (> 18 years) with multiple comorbidities in Australia (diabetes prevalence 5.3 – 19.2%) , yet, malignancy-related amputations become more proportionate in those ‘higher risk’ populations (malignancy prevalence 3.2 – 7.0%) . In contrast, the proportions of peripheral arterial disease-related (18.3%) and trauma-related (8.1%) amputations seem to be proportionate to the prevalence of similar conditions in the general Australia population (cardiovascular disease (16%) and injuries (12.0%)  and even in the ‘higher risk’ populations (cardiovascular disease 6.1 – 20.9) .
Furthermore, this study identified that around 80% of amputation cases were associated with complications of conditions defined by the Australian Institute of Health and Welfare, and World Health Organisation to be preventable chronic diseases (diabetes, peripheral arterial disease and malignancy) [40, 41]. Additionally, nearly 10% resulted from trauma or accidents. Thus, it may be suggested that most amputations are associated with a potentially preventable condition, complication or circumstance.
Age at amputation was significantly different for different groups of key conditions. Amputations associated with type 1 diabetes in our study occurred at significantly younger ages (52 ± 12 years) than type 2 diabetes (67 ± 10 years) or peripheral arterial disease (68 ± 18 yrs). Younger age has previously been reported for amputations due to type 1 diabetes (45 – 62 years) [4, 9, 31], compared to amputations due to type 2 diabetes (68 – 73 years) [4, 9, 14], and peripheral arterial disease (70 – 79 years) [9, 14]. Furthermore, the proportion of amputations due to type 1 diabetes made up approximately 10% of the total diabetes associated amputations in this study, which is comparable to previous literature [4, 9]. This may be associated with the different age of onset for the diagnoses of these various conditions  and a similar proportion of the total population of people with diabetes having type 1 compared with type 2 diabetes . Our findings suggest that having type 2 diabetes does not confer any additional risk for amputation compared with type 1 diabetes. Not surprisingly, amputations associated with trauma demonstrated a much younger age of onset (36 ± 10 yrs) than the average age of onset for all lower extremity amputation (62 ± 16).
Differences within key conditions associated with lower extremity amputation included that patients having an amputation due to malignant tumours were more likely to be female and need a major amputation . However, the numbers included in this group were small, and included six carcinomas and four sarcomas, thus, the small numbers and differing development of these different categories of malignant tumours should be interpreted with caution. All amputations due to trauma were first amputations, only occurred in male participants and were frequently caused by motor vehicle and work place accidents, which aligned with similar findings from the literature . Lastly, significantly more major amputations occurred in the peripheral arterial disease (non-diabetes) group which reflects existing international literature  and the more proximal nature of the disease that occurs in patients suffering from non-diabetes peripheral arterial disease .
The results for the proportion of first ever amputation (56%) in this overall series, however, seems to be at the lower range of those previously reported (50 – 86%) [1, 10, 13]. This may be the result of investigating a major tertiary referral hospital that houses the only vascular surgery department for a geographically vast region containing over 35 other hospitals in southern Queensland. Anecdotally, the authors observed in the clinical records many historical first amputations occurring in the patients’ local hospital with more complex cases requiring amputation or subsequent amputation procedures being transferred to the Princess Alexandra Hospital for more specialist care.
Lastly, this study found that hospital coding reliability for variables associated with lower extremity amputation (diabetes status, trauma status and amputation site) were categorised as almost perfect; above 0.8 Kappa values or 90% across all variables tested. This result seems to align with other methodologically similar Australian studies and adds further weight to the reliability of Australian discharge datasets in collecting data around lower extremity amputation procedures [32, 33]. However, these results are in contrast to a previous Australian study investigating the hospital coding accuracy of diabetic foot complication pathology which reported an accuracy rate as low as 34% . This result is not entirely unexpected as the previous study had significant methodological differences and its primary focus was diabetic foot pathology and not specifically amputations as was the primary focus of this study .
There are a number of methodological differences in this study when compared to other similar studies. Thus, the comparative results of this study should be viewed with some caution. Firstly, this study did not follow the protocol for such audits as defined by the Global Lower Extremity Study Group . This group recommend that at least two data sources are used to cross-check accuracy of data and estimate levels of case ascertainment; including hospital discharge data, operating theatre records, limb fitting centre records, amputation registries, and/or foot clinic records . Our study only used the one source from coded hospital discharge dataset records as other sources were not available or accessible at the time. This limitation may also impact our findings on the reliability of hospital coding as only those amputations originally coded have been audited and additional amputations that may have been missed by hospital coders remain unknown as identified in other studies . However, further Australian study’s’ findings investigating the reliability of lower extremity amputation coding reflected our results, potentially due to the more formal standard documentation that occurs for a surgical procedure than clinical assessment [32, 33].
Secondly, the definition of major amputations used in this study was different to that defined by the Global Study Group. The group defines a major amputation as that through, or proximal to, the tarso-metatarsal joint . However, the definitions used in our study aligned with those in commonly used international clinical guidelines , ICD-10-AM codes  and other similar large studies [3–5, 11, 13, 36] defining a major amputation being a resection proximal too, not through, the mid-tarsal level. This may have contributed to our reported lower proportions of major amputations compared to studies that have used different definitions of major and minor amputations [1, 10].
Thirdly, our study only interrogated medical records for a single principle key condition precipitating the amputation, thus, multiple key conditions were not recorded for each case as performed by other studies [1, 10, 13]. Also this study did not use standard co-morbidity measures to capture and analyse co-morbidities  and instead reported those conditions already stated in the literature to increase the risk of amputation rather than mortality as per other similar international studies [1, 3, 5, 10, 20]. Thus, other standard co-morbidities were not captured like other Australian studies have done; for example, those that are risk factors for chronic diseases like hypertension, dyslipidaemia or long term smoking . These conditions were frequently observed throughout this case series.
Fourthly, our study prioritised the diagnosis of diabetes over other conditions associated with amputations and this may have potentially over inflated diabetes as the key condition association with amputation. However, other studies using similar methodology to ours in this regard found similar results [3, 5]. Lastly, typical of a single site, retrospective study the quality of data obtained from medical records can lack rigour and generalisability. However, the site used in this study may provide more generalisable results than otherwise, as it was the major tertiary referral site for a third of the population of Queensland, performs over 20% of all Queensland lower extremity amputations , and its resultant amputation rate of 13.1 per 100,000 for the 1.5 million people it services  is comparable to that of international amputation rates [1, 11].
Whilst these differences may decrease comparability with some other studies, the results of this study align with the results obtained in most international studies. At the very least, robust data on the proportions of key conditions associated with all levels of lower extremity amputations in Australia, mean age at amputation, sex influence, and amputation site ratios is provided by this study. However, it would be recommended that for any future Australian studies investigating lower extremity amputations that the Global Study Group protocol be employed prospectively across multiple sites to improve the rigour and generalisability of results; with the exception of aligning amputation definitions with standard ICD code definitions and international guidelines.
Finally, this study supports an existing Australian recommendation stating that the reporting of standard annual national lower extremity amputation rates should be implemented in Australia ; including categories associated with at least type 1, type 2 and non-diabetes amputations [4, 9]. The reporting of these rates are best practice in other industrialised countries and have contributed to the reduction of both diabetes and non-diabetes amputations in these nations [1, 3, 4, 10, 11, 14, 15, 21, 28, 29]. The authors suggest that until such a standard lower extremity rate is reported across Australia the old adage of “you can’t improve what you can’t measure” will continue for this highly preventable complication.