Diabetic peripheral neuropathy is a complex disorder in which the disease process may affect different sets of nerve fibers to different degrees in different individuals. Thus, one individual may have an abnormality of large-fiber sensory function, which could be detected by measuring the vibration perception threshold (VPT), whilst another may have a predominantly small-fiber neuropathy that can only be detected by measuring the thermal perception threshold (TPT). This feature of neuropathy can cause problems in the selection of a single test with which to screen a population. The issue of measurement of neurological function is further complicated by the nature of the tests. Many are psychophysical tests, in which the subject is required to interpret the nature of an external stimulus (usually applied to the foot). This subjectivity can lead to relatively poor reproducibility of tests such as VPT, in which the subject has to differentiate light touch from vibratory sensation.
This lack of certainty over the value of individual tests for the assessment of diabetic neuropathy has led to recommendations that several different tests should be performed, and that diabetic neuropathy should only be diagnosed when more than one is abnormal [1]. Whilst this may make the diagnostic process more rigorous in any individual or in an individual study, it can make comparisons between studies more difficult. Such recommendations have been only patchily adopted, and where they have been put into practice, the selection of tests has not been uniform. Thus, we are faced with having to compare the prevalence when neuropathy is determined by a single test with the prevalence when neuropathy is diagnosed when any two out of perhaps three to five tests are abnormal. Increasing the number of tests that are performed will automatically increase the number of individuals in whom an abnormality is found, while requiring that more than one abnormality is present will then tend to decrease the prevalence. The overall effect is thus complex. The impact of varying the diagnostic testing procedure can be seen in the Diabetes Control and Complications Trial (DCCT) data (Table 3.1), where the prevalence of distal symmetrical polyneuropathy (DSP) at baseline in the conventional therapy cohort varied from 0.3% (abnormalities of reflexes, sensory examination, and neuropathic symptoms) to 21.8 % (abnormal nerve conduction in at least two nerves) [2]. Confirmed clinical neuropathy (abnormal history or examination, confirmed by abnormal nerve conduction or autonomic function) was the gold standard for the study, and was found in 2.1% of this cohort. This 73-fold difference in the prevalence of DSP, between the two extremes, but within a single population, highlights the difficulties of comparing studies with differing diagnostic criteria.
The influence of test selection on the understanding of etiological factors associated with neuropathy is likely to be considerably smaller than that on prevalence. As long as each diagnostic process is indeed measuring some aspect (or aspects) of neuropathy, it is likely that those who are rated as having neuropathy are genuinely more severely affected than those who are not so rated. Thus, associations with factors such as age or glycemic control may be easier to compare across different studies than are prevalences, which are highly dependent not only on the test (or tests) selected, but also on the diagnostic threshold that is used. For example, the groups selected either by absent ankle reflexes or by elevated VPT are both likely to be older and have poorer glycemic control than the groups who are normal on these respective tests. However, unless loss of ankle reflexes represents the same degree of neuropathy as does an abnormal VPT (according to the threshold selected - see below), the prevalences generated from the same population by these two tests could be quite different.
Table 3.1 Prevalence of polyneuropathy within the DCCT conventional therapy primary prevention cohort, according to diagnostic test
| Diagnostic criteriaa | Neuropathy prevalence (%) |
|---|---|
| Sensory examination, reflexes, and symptoms | 0.3 |
| Symptoms and reflexes | 0.5 |
| Sensory examination and symptoms | 1.1 |
| Symptoms or sensory examination, confirmed by abnormal nerve conduction or autonomic functionb | 2.1 |
| Sensory examination and reflexes | 2.1 |
| Abnormal autonomic function | 2.4 |
| Symptoms alone | 3.7 |
| Any two out of symptoms, sensory examination, reflexes | 4.5 |
| Reflexes alone | 5.8 |
| Sensory examination alone | 12.2 |
| Nerve conduction (abnormal in at least 2 nerves) | 21.8 |
Data adapted from [2]
aAbnormalities required in the fields listed
bCold standard measure of neuropathy for the study
Even when the same test has been used, the methods and the selection of thresholds have not been uniform. VPT is widely used, but some studies use agerelated thresholds (based on UK normal data) [3], while others have ignored age and used a single cutoff [4]. More recently, locally derived normal ranges have been applied, although in one study these were derived separately for different age groups [5], while another compared all diabetic subjects to a young nondiabetic population [6].
It is obvious from the discussion above that there is a long way to go in finding a definition of neuropathy that can be widely used, although it is clearly needed for accurate epidemiological study. This definition needs to be appropriate for the clinical forms that the disease may take (e.g., small fiber or large fiber involvement), related to hard endpoints, such as foot ulceration, agreed upon widely, and useful for a range of both clinical and research purposes. Multiple testing of different neurological functions, although recommended as the gold standard [1], can be difficult and expensive, especially in large field studies. Furthermore, recent data suggest complex patterns of sensation in DSP with decreased VPT (hypoesthesia) and heat stimulus-induced hyperesthesia (low thresholds), both being characteristic of mild DSP as they correlate with neuropathic symptoms and deficits, whereas panmodality hypoesthesia is typical of severe DSP [7]. The only studies that relate measures of neurological function (in a prospective manner) with subsequent foot ulceration and amputation have used simple quantitative sensory tests, such as VPT or monofilament sensitivity [8,9].
The second important obstacle in determining the prevalence of neuropathy is the selection of people who are going to be tested. Hospital clinic-based studies will usually be biased towards those with more serious problems, and are likely to overestimate the prevalence of neuropathy in the general diabetic population. Nevertheless, the value of hospital-based prevalence studies should not be underestimated. Whilst they do not directly address the issue of what the general risk of having neuropathy is in a person with diabetes, they precisely answer the question of how common neuropathy is amongst hospital clinic patients. This in itself may be a very important and valuable issue. The problem, however, is that one hospital-based population may not be the same as another - an issue that can be partially overcome with multicenter studies [10]. Another problem is that temporal trends in referral patterns from primary care can make today's hospital population quite different to tomorrow's.
Studies using registers of people with diabetes