Which limb-factors predict amputations?

A new paper from the longest ongoing diabetic foot study in the world, the Seattle Diabetic Foot Study, has reported on the limb-specific factors that predict amputation over 20 years. They found that peripheral neuropathy and peripheral arterial disease were the leading predictors of amputation in people with diabetes. “So what, we already knew that”, we hear you say; but did we already know that?

What do we know about predictors of amputations?

As we outlined in our summary of a huge NZ study investigating risk factors for amputations, there’s been an enormous number of studies investigating factors that predict amputation. For example, the New Zealand study found that many person-specific factors such as different socio-demographic, medical history and foot disease history factors predicted amputation. And even this Seattle Diabetic Foot Study has already investigated amputation predictors finding that people with peripheral neuropathy, peripheral arterial disease and foot ulcers are more likely to have amputations.

So what’s the difference with this new Diabetes Care paper we hear you ask? Well instead of diagnosing various factors in each patient and following patients to see who had an amputation regardless of which limb; this paper reports the first ever study to specifically diagnose these various factors in each limb (of each patient) and the followed each limb to see which limb had an amputation.

What did this new paper do then?

As you’ve probably now gathered, the Seattle Diabetic Foot Study is a bit of a big deal. It’s the longest prospective study into diabetic foot disease. It’s a like the “seven up” television series for foot disease – the “seven up” series followed a bunch of seven year old kids from 1964 every seven years to see what happened to them over the course of their lives. The Seattle Diabetic Foot Study started in 1992 – led by one of the world’s leading diabetes epidemiologists, Professor Ed Boyko – recruiting ~1,500 patients from a large veterans’ diabetes clinic in Seattle and following them periodically for up to 22 years now.

In short, they recruited patients with diabetes that did not have a current foot ulcer. They collected a whole bunch of data from these patients at baseline. First, they interviewed them for data on their demographics, diabetes history, smoking history and foot disease history. Next, they examined them for their weight, blood pressure and to diagnose eye disease, kidney disease, peripheral neuropathy, peripheral artery disease (PAD) and Charcot foot, amongst other conditions. Last, they collected bloods to look at HbA1c, eGFR etc. To diagnose peripheral neuropathy they used a 10g monofilament and for PAD an Ankle Brachial Index (ABI) and transcutaneous O2 tension (TcPO2).

They then followed up all these patients at their clinic until 2002 and after that using hospital records until 2012. If they identified an amputation they audited the hospital records to determine which limb the amputation was performed on. Then instead of using people as the denominator they used limbs to predict which “limb-specific” factors predicted amputations using some complex statistical models that allowed them to control for all the other factors diagnosed for that limb.

What did they find?

First, they originally recruited 2,893 limbs (1,461 patients) without a foot ulcer. Average age was 62 years at recruitment and all were male as they were recruited from a veteran’s diabetes clinic. Over the following 22 years, 136 limbs had an amputation (two thirds were major amputations above the ankle).

Second, they found when they controlled for all factors, that peripheral neuropathy (inability to feel 10g monofilament on one site), critical PAD (ABI <0.5) and mild-moderate PAD (ABI >1.3 or 0.5-0.9) were the only independent limb-specific predictors for amputations; these factors also had the greatest hazard ratios of any predictors found. Other person-specific predictors found were eye disease (poor vision scores), kidney disease (low eGFR), younger age (<70 years) and lower body weight.

Last, they also looked at interactions between these various factors to see if that changed the predictors found. They found very little change; except interestingly in the sub-group of limbs with an ABI >1.3 they found that also having a TcPO2 <26mmHg significantly increased the predicted risk of amputation.

What was good or bad about this study?

Whilst this large long-term study had many, many strengths – it rigorously examined limbs with diabetes and no current foot ulcer using gold standard clinical examinations we still use today, followed them for >20 years and used complex statistical modelling to determine which of those factors independently predicted amputation – like also studies this also wasn’t without limitations.

These limitations included: i) all limbs (patients) were male because of the clinic in which they were assessed, so we don’t know if gender influenced these findings; ii) >800 patients died during the follow up period, which probably influenced some finding especially that younger age predicted amputations; iii) they relied on hospital records after 2002 to identify limbs having amputations, so they may have missed some amputations; and iv) whilst they recruited a large number of limbs (patients) it was from only one US city meaning their findings may not be as applicable to other cities or nations.

What does that all mean?

Regardless, the authors rightly concluded that peripheral “arterial disease and neuropathy emerged as the only limb-specific risk factors for amputation, but these and several person-level factors may be amenable to prevention or treatment interventions to reduce diabetic amputation risk.”

In short, this large long study basically confirms what we thought we knew, that we need to continue to examine each limb (and patient) with diabetes using a 10g monofilament (for peripheral neuropathy) and ABIs (for PAD) to identify the main risk factors for amputation. Also, it supports the need to use further non-invasive vascular tests if an ABI is high (>1.3), such as TcPO2 or Toe Pressures as recommended in the Lower Extremity Threatened Limb Classification WIfI system.

Overall, these findings reinforce the critical need to screen all people with diabetes for peripheral neuropathy and PAD. And if these limb-specific factors are identified, then patients need to be closely monitored by appropriately trained clinicians to prevent them developing foot disease and amputations in future. Coincidentally this is also outlined in the Australian diabetes-related foot disease strategy ;).