How accurate are non-invasive peripheral arterial disease tests in people with diabetic foot ulcers?

A new study published in Diabetic Medicine from the team of globally renowned vascular surgeon Professor Rob Hinchliffe has found that toe-brachial index (TBI) and Doppler waveforms are more useful for the exclusion of peripheral arterial disease (PAD) in patients with diabetic foot ulceration (DFU) compared to pulse palpation or ankle-brachial index (ABI). But, is that the full story?

What do we know about non-invasive vascular testing in people with diabetes?

Well we know that PAD is a big deal in people with DFUs as PAD can have a huge bearing on if a DFU heals or needs an amputation. Thus, we need to test for PAD in all people with DFU to determine if they have PAD. If they don’t have PAD that typically means we don’t have to worry about their perfusion to heal and can continue with conservative best practice care. But if they do have PAD that means we do have to worry about referring to a vascular surgeon to see if they can improve their perfusion.

To test for PAD in our clinics we have traditionally used clinical observations and non-invasive vascular tests. But, we know that these clinical observations alone are very poor at detecting PAD, such as noticeable hair loss and capillary refill time. We also know that all of the non-invasive tests we frequently use have been shown to have varying levels of accuracy in people with diabetes, such as Ankle Brachial Indexes (ABIs), Toe Brachial Indexes (TBIs) and Doppler waveforms.

This highlights the difficulty of clinical assessing for PAD in people with diabetes. Many suggest this varied accuracy is due to reasons such as the presence of medial arterial calcification (MAC) with PAD, the distal anatomical distribution of PAD  and alterations to skin blood flow secondary to arteriovenous shunting as a result of autonomic neuropathy in people with diabetes. But, while studies have tested the accuracy of these tests in diabetes patients, hardly any have tested this in DFU patients. And this is exactly where this new study from Professor Rob Hinchliffe comes in.

What did this new study do then?

In short, firstly they recruited 60 patients with new DFUs in one diabetic foot clinic. Secondly they performed a comprehensive range of clinical observations and non-invasive vascular tests on each of these patients. The clinical observations they tested for included: examining for hair loss, muscle atrophy, cool skin, dependent rubour, abnormal pedal pulses (one absent), abnormal venous filling time (>15 seconds) and abnormal capillary refill time (>2 seconds) were present.

The objective non-invasive vascular tests they tested for included: ankle pressure, ABI, toe pressure, TBI, TcPO2 and Doppler waveform analysis. The cut-off they used to diagnose PAD in these tests were: <70mmHg for ankle pressures, <0.9 or >1.3 for ABIs, <50mmHg for toe pressures, <0.75 TBI, <60mmHg TcPO2, and presence of a monophasic Doppler waveform in any vessel.

Finally, they performed colour duplex ultrasound on all patients as the gold standard benchmark to determine if PAD was actually present or not. This was performed on all abdominal, femoral and lower limb arteries and PAD was confirmed if they found >50% stenosis in any one of these arteries.

They then could test the various accuracy of the clinical observations and non-invasive vascular tests against the gold standard colour duplex. In this study to determine accuracy they relied on the statistical tests of negative likelihood ratio (NLR) which indicates if a test can reliable exclude the disease (<0.1 means its accurate for this) and positive likelihood ratios (PLR) which indicates if a test can reliably confirm the presence of the disease (>10 means its accurate for this).

So what did they find?

Firstly, the 60 patients with DFU had similar characteristics to what we’d expect in any outpatient DFU clinic: mean age 66 years, 75% male, 12% smokers, 73% had superficial ulcers, 40% were infected. They then confirmed that 20 (33%) of these patients had PAD using the gold standard colour duplex ultrasound test.

Secondly, when they compared the clinical observations against the gold standard colour duplex, they found none of the clinical observations could reliably exclude or confirm PAD, including palpating pulses. NLRs ranged from 0.46-1 and PLRs from 0-3.9.

Thirdly, they found varying degrees of accuracy for the non-invasive vascular tests when compared against the gold standard. The NLR values for Doppler waveform analysis (0.15) and TBI (0.24) came very close to being able to accurately exclude PAD if it was actually there. However, all other non-invasive vascular tests did not with NLR ranges of 0.53-1.1.

Whereas the PLR values for toe pressures (17.55) indicated it could accurately confirm PAD if it was actually there. The other tests had PLR ranges from 0.8-2.25; however, the Doppler waveform analysis recorded perfect diagnostic PLR values because it used the same methods in this regard to the gold standard, so it could not be tested for PLR as it was essentially using the same test.

What was good and not so good about this study?

This study had many strengths, including: i) the first study to prospectively examine such a variety of testing methods, both clinical and objective non-invasive vascular tests, in patients with DFU; ii) non-clinical and non-invasive tests were performed by one trained operator which limits any inter-rater errors; iii) the use of diagnostic colour duplex ultrasound imaging is a relatively strong gold standard and iv) they used quite robust accuracy statistics of NLR and PLR.

However, this study wasn’t without limitations, including: i) a smallish sample size, a bigger sample may have meant the TBI and Doppler waveform analysis had the power to reach <0.1 NLR; ii) most of the DFU patients had small, shallow ulceration, so the generalisability of the results to people with more severe DFU may be limited; iii) the room temperature was controlled at 18-20 degrees, which is cooler than previous protocols, and may have influenced the performance of toe pressure and TBI which are highly influenced by environmental factors; iv) some of the cut-offs for PAD in non-invasive tests were different to what has been used in previous research, for example the low toe pressure NLR may be partially attributed to the diagnostic threshold used (<50 mmHg); and v) the sonographer who performed the waveform analysis was not blinded to the results of the colour duplex which may be a reason for why the Doppler waveform analysis had such high accuracy.

So what does that all mean?

In short, they found that not having a TBI <0.75 or a monophasic Doppler waveform in people with DFU were relatively accurate for excluding PAD.  This suggests that a clinician finding these TBI and Doppler waveform results in their patient with a DFU can have some confidence that they have the perfusion to heal their DFU and don’t have to refer to a vascular surgeon. Additionally, they found that a toe pressure of <50mmHg in people with DFU was relatively accurate for confirming PAD which suggests clinicians should refer those patients to a vascular surgeon. But, the authors rightly recommend that patients that have PAD excluded with these tests should still be closely monitored for PAD and particularly if the DFU deteriorates or doesn’t heal at all over the next month.

Conversely this study found that historically used clinical observations and other non-invasive vascular tests such as ABIs and ankle pressures had reduced accuracy to exclude or confirm PAD in patients with DFU. Toe pressures were also not accurate to exclude PAD, but were found to be relatively accurate for confirming PAD if it was actually there.

This study has added valuable data about the performance of clinical and non-invasive vascular testing methods in patients with DFU. Previous diagnostic accuracy studies have focused on diabetes populations, but not included such a large number of participants with DFU. However, does this mean that we should only be using Doppler waveforms and TBIs in our clinical tests for PAD in patients with DFU, and throw out other tests such as the good old ABI?

Based on this study alone, no, it is not. Whilst the evidence against the use of ABI as a screening tool for detection of PAD in diabetes populations is mounting, much of the evidence is still from small populations and of lower quality – so we can’t throw out one of the few objective tests we have just yet. What this study does highlight though is that we should be using a range of non-invasive tests to exclude PAD and this should at least include Doppler waveform analysis or TBI and perhaps toe pressures to confirm PAD. However, whilst Doppler performed very well in this population and seems an attractive option, it has varying levels of reliability, and requires a high level of operator skill, and it ultimately subjective in its interpretation.

What is our final word on this study?

This study really highlights the fact that our clinical observations are not good at detecting or excluding PAD, and our non-invasive vascular assessments are not perfect alone and should not be used in isolation. However, it does highlight that we should be using a combination of non-invasive testing methods to paint the picture of PAD and perhaps TBIs or Doppler waveform analyses should be at least one of those tests used to exclude PAD in people with DFU.