Response to Comment on Hoffstad et al. Diabetes, Lower-Extremity Amputation, and Death. Diabetes Care 2015;38:1852–1857

We thank van Netten et al. (1) for their kind words about our study (2) and we will try to address their concerns. With respect to Core Outcome Measures in Effectiveness Trials (COMET), as noted in our article, we used a medical record data source of more than 400,000 individuals with diabetes who on average were followed for 9 years (2). More than 6,000 of these individuals had a lower-extremity amputation (LEA). It is highly unlikely that this source or other population-based prospective data sources will ever be able to use the COMET initiative to collect information that might be more specific to diabetes, LEA, and death. We do encourage investigators that want to create cohorts with data collected in a more uniform manner and that is more centered on the lower-extremity setting. However, it is unlikely that these cohorts will ever be as large or generalizable or contain as many years of observation as the data source used in our study.

With respect to their concern about our exclusion criteria, our goal was not to evaluate the risk of death from surgery but the risk of death from medical factors that occur in those who had a LEA. We had hoped to better understand the medical risk factors that are associated with death and LEA such that some could be modified by medical or surgical care. In any event, excluding outcomes due to the procedure itself is often done in studies of surgical interventions and was a suggestion of the reviewers of our original submission. Including the first 2 weeks would have made the risk estimates even larger.

Finally, van Netten et al. (1) suggest that we should only evaluate those with a major amputation. First, we did provide information on the subgroup that had major LEAs. However, our goal was to investigate LEA and not just major LEA. As requested, those with a major LEA are more likely to die than those with only a minor LEA (hazard ratio 3.80 [95% CI 3.51, 4.11]). A significant proportion of individuals with minor LEAs will have a major LEA within the first year of the initial procedure, leading to methodological concerns with respect to how to classify individuals with respect to the general cohort of those with diabetes who were used as our comparator arm (3). Should we exclude those with minor amputations from the diabetes comparator arm (4.19 [3.69, 4.43]) or should those with minor amputation be part of the comparator arm (2.03 [1.98, 2.09])? The options result in very different risk estimates that are likely biased, and with respect to the former, it seems improper to exclude those with minor LEAs from a study that might provide information on modifiable risk factors. Furthermore, several investigations have shown that those with diabetes and a foot ulcer (the usual precursor to a LEA) are at increased risk of death (4,5).

We thank the authors for their interest in our study and we encourage them to help refine our understanding of why those with diabetes who have had a major amputation are more likely to die.