The natural history of diabetic peripheral neuropathy determined by a 12 year prospective study using vibration perception thresholds

JOCN-157.QXD

10/13/01 5:09 PM

Page 520

Journal of Clinical Neuroscience (2001) 8(6), 520–524 © 2001 Harcourt Publishers Ltd doi: 10.1054/jocn.2001.0893, available online at http://www.idealibrary.com on

Clinical study

The natural history of diabetic peripheral neuropathy determined by a 12 year prospective study using vibration perception thresholds D. V. Coppini1, A. Wellmer2, C. Weng1, P. J. Young3, P. Anand2 FRCP, P. H. Sönksen1 1

Division of Medicine, United Medical and Dental Medical Schools, St. Thomas’ Hospital, London, UK, 2Peripheral Neuropathy Unit, Division of Neuroscience and Psychological Medicine, Imperial College School of Medicine, Hammersmith Hospital, London, UK, 3Department of Health Sciences, University of York, York, UK

Summary The development and long term progression of diabetic peripheral neuropathy was studied using vibration perception threshold (VPT) as a validated measure. Three hundred and ninety-two patients had a normal age corrected VPT (12.1<3.7 volts) at baseline, with an age corrected logarithmic VPTscore :12. 19.9% developed an abnormal VPT over a 12 year period, increasing from 14.2<3.7 volts (VPTscore 10.4<0.6) at baseline to 35.9<9.5 volts (VPTscore 12.6<0.45) at follow up (P:0.0001), and from 10.1<3.7 volts (VPTscore 9.4<0.8) to 14.2<4.7 (VPTscore 9.8<0.8) in the rest. Over 80% thus retained a ‘normal’ VPT after a mean diabetes duration of 16 years despite only average glycaemic control, suggesting that non-ideal long term glycaemic control leads to neuropathy in a subset of predisposed patients. VPT was correlated in 123 diabetic patients with definitive criteria for neuropathy and a range of quantitative sensory and autonomic tests. 62/63 patients with abnormal VPT fulfilled neuropathy criteria; of patients with normal VPT who fulfilled neuropathy criteria, all had at least one abnormal thermal threshold test result. We conclude that a combination of log-transformed VPT values (VPTscore 910.1) and thermal thresholds can identify diabetic patients at risk of developing peripheral neuropathy and select patients likely to benefit from prophylaxis in clinical trials. © 2001 Harcourt Publishers Ltd Keywords: diabetes mellitus, neuropathy, vibration perception threshold (VPT), VPTscore, risk factors

INTRODUCTION Adequate epidemiological information about the incidence and natural history of peripheral neuropathy in diabetic patients is still lacking, although it is a common and clinically significant complication. This is mainly due to different methods used to define neuropathy and the length of follow up required. Even prevalence data varies widely in the literature.1–3 The prospect of prophylaxis or new treatments of diabetic neuropathy makes such information more important, and critical to the design of long term clinical trials. The Consensus Development Conference on Standardised Measures in Diabetic Neuropathy recommended the following four measures for the diagnosis of diabetic neuropathy: clinical measures, morphological and biochemical analysis, electrodiagnostic assessment and quantitative sensory testing.4 These proposals, however, are impractical in most routine clinical settings and are mainly intended for research purposes. We therefore studied the role of vibration threshold as a predictor of future neuropathic complications in a long term and large scale study in diabetic patients.5,6 We also determined the validity of vibration perception abnormalities as markers of peripheral neuropathy by correlating these with standard Dyck neuropathy criteria7 and with a range of other quantitative sensory and autonomic tests. The main aims of the study were to estimate the rate of development

Received 31 October 2000 Accepted 19 January 2001 Correspondence to: Prof. Praveen Anand, Peripheral Neuropathy Unit, Division of Neuroscience and Psychological Medicine, Imperial College School of Medicine, Hammersmith Hospital, Du Cane Road, London W12 0NN, UK. Tel.: ;44 (0)181-383 3309 / 3319; E-mail: P. [email protected]

520

of peripheral neuropathy in subjects with normal age corrected VPT at baseline and to identify the risk factors. PATIENTS AND METHODS All patients (n:985) first attending the diabetes clinic at St. Thomas’ Hospital between 1 January 1982 and 30 September 1985 had a structured and comprehensive standardised first visit record entered on our computer database (Diabeta). This clinic is a general diabetic clinic in South London, and is not attached to a specialist foot clinic. Clinical examination of the lower limbs and vibration threshold measurements at the great toe using a Biothesiometer (Biomedical Instrument, Newbury, Ohio, USA) were performed on all patients at first visit, irrespective of symptoms or complications. The voltage reading from the Bio-thesiometer was entered into the computer which calculates a ‘VPTscore’ by comparing the patient’s value with an age-related normal population. The VPTscore is calculated by adding a value of ‘10’ (hence producing a positive number) to the number of standard deviations that an individual’s reading is away from the mean of normal subjects of the same age. VPT has been shown to vary with age.8,9 In normal subjects there is a linear relationship between the logarithm of the Bio-thesiometer reading against age. The ‘VPTscore’ was calculated by the equation: (log10 bioo) – (log10 biom) (a) VPTscore:10; ᎏᎏᎏ s where bioo is the observed reading on the Bio-thesiometer (in volts), biom is the normalised value (mean of normal subjects) and s is the residual standard deviation of the normalised value. The normalised value was calculated from a linear fit to the logarithm of normal data of Bio-thesiometer reading against age in years.

JOCN-157.QXD

10/13/01 5:09 PM

Page 521

The natural history of diabetic peripheral neuropathy 521

(slope X age);intercept (b) (log10 biom ): ᎏᎏᎏ s Substituting for (log10 biom) in equation (b)

(log10 bioo)9(slope X age);intercept (c) VPTscore:10; ᎏᎏᎏᎏ s The values for the slopes, intercepts and standard deviations (s) at various sites were derived from Bloom et al.8 A VPTscore 912 corresponds to the 95th percentile, and thus scores above 12 are taken to be abnormally high. Details of lower limb examination were also entered into a structured template in the patient’s respective file in Diabeta. Eye fundus examination (through dilated pupils where possible), HbA1 (Corning method), dipstick urine test results (Albustix), and smoking and alcohol history were also entered. Entry of height and weight are automatically transformed into a body mass index (BMI). Retinopathy was present if patients had one or more microaneurysms, exudates or preretinal haemorrhages (background retinopathy) or had proliferative changes (proliferative retinopathy). Glycaemic control over 12 years was based on an average of 15 HbA1 tests per patient during that period. Smoking risk was assessed by comparing outcome in smokers and non-smokers at baseline, and by calculating the average lifetime smoking years in patients with and without neuropathy events at review. The effect of alcohol consumption in patients at baseline was analysed using a cut-off of 6 units weekly. Four hundred and twenty-three patients (43%) were successfully recruited to the follow up study. We have previously shown that this group of patients was adequately representative of the original cohort of patients.5 All these patients in the follow up study (n:423) were reviewed by one observer (DVC) in 1995 and VPT measurements were repeated. The Bio-thesiometer was balanced vertically on the pulp of the great toe to measure the vibration threshold (average of both toes) and a mean of three readings was used to derive the value on each patient. The same instrument was used to test all patients. Sensory testing using a 5.07 Semmes-Weinstein monofilament was also performed on the dorsum of the great toe. Eight correct responses out of 10 applications indicated normal sensation; one to seven correct responses indicated reduced sensation (score of 1); and no correct responses translated into absent sensation (score of 2). All patients also underwent a screening examination using the Michigan Neuropathy Screening Instrument (MNSI). A score 92 has been shown to identify patients with diabetic neuropathy.10 A history of both positive (tingling, burning or increased sensitivity in feet or legs) or negative (reduced feeling in feet) symptoms was also recorded. A family history of diabetes was recorded as positive only if present in parents or siblings. A history of hypertension was recorded if patients had been treated for this. A finger-prick capillary blood sample was used to measure total cholesterol, triglycerides and creatinine levels (Colorimetry method, Cobas Mira analyser, Roche Diagnostics, UK). A first morning urine sample was tested for proteinuria (Albustix, Bayer Diagnostics, UK). In patients with a negative test, an albumin-creatinine ratio (ACR) was measured on the same sample (Turbidimetry method, Cobas Mira analyser, Roche Diagnostics, UK). Microalbuminuria was defined as an ACR93. We defined VPT abnormality as a VPTscore at review 912. Subjects with a VPTscore :12 were considered as ‘normal’ for VPT. Thirty-one (7.3%) patients (males 62%; caucasian 87%) who had abnormal VPT by this criterion at the baseline visit were

© 2001 Harcourt Publishers Ltd

excluded from subsequent analysis leaving 392 of the original 423 patients eligible for analysis in the follow up. Insulin-dependent diabetes was defined as onset of diabetes : age 40 years. To evaluate the validity of VPT as a useful measure of neuropathy, 123 diabetic patients from a separate cohort were studied at the Royal London Hospital. The age range of patients in this group was 25 to 69 years (mean 48 years) and 65% were male. 70% had insulin dependent diabetes and the mean HbA1c in the cohort was 8.1%. On the basis of neuropathic symptoms, neurological examination, quantitative sensory and autonomic tests and nerve conduction studies, neuropathy was defined according to Dyck’s criteria.7 The methods of quantitative sensory and autonomic function testing have been described previously.11 In brief, foot sole thermal thresholds were tested with computer-driven Marstock thermal threshold testing system (Somedic Thermotest, Stockholm). The device was set at a rate of rise in temperature of 1°C/sec. Thermal thresholds were determined for cool and warm sensation, and heat pain, as a mean of four consecutive tests. VPT was measured using a Bio-thesiometer, and light touch was assessed using a set of Semmes-Weinstein monofilaments using the same methods as for the prospective study described above. Peak axon-reflex sweating was measured with an Evaporimeter (Servomed, Stockholm) after intradermal injection of nicotine (0.8 ␮g in 0.2 ml) into the right lateral calf skin. Axon-reflex vasodilatation (ARV) was determined by measurement of increased flux using a laser Doppler device (Perimed PF4, Stockholm) after intradermal injection of capsaicin (0.5␮g in 10 ␮l) into the right lateral calf skin. Sural and peroneal nerve conduction was measured using standard techniques. All studies were approved by the St. Thomas’ and Royal London Hospital Ethics Committees. STATISTICAL ANALYSIS Analyses were conducted with logistic regression, using the NLIN procedure of the SAS system.12 Sensitivity and specificity values were also obtained using the CTABLE option with prior probabilities set to the observed frequency of cases. In the validation study, the frequency of abnormality on VPT was compared with the frequency of abnormality on other tests using the chi-square test. RESULTS Baseline clinical characteristics of patients entering the prospective study are presented in Table 1. Normal ankle reflexes were present in 93% of patients and cotton wool sensory testing was intact in all but one patient. None of the patients had a history of foot ulceration at the baseline visit. Seventy-eight patients (19.9%) developed peripheral neuropathy over a mean 12–year period. VPT in these patients increased from 14.2<3.7 volts (VPTscore 10.4<0.6) at baseline to 35.9<9.5 volts (VPTscore 12.6<0.45) at follow up (P:0.0001) and from 10.1<3.7 volts (VPTscore 9.4<0.8) to 14.2<4.7 (VPTscore 9.8<0.8) in the rest of the patients (P:0.0001) (Table 1). Fifteen (3.8%) patients (all but one patient in the abnormal VPT group) developed lower extremity complications. Seven patients (1.8%) had active plantar foot or toe ulcers at the time of review and eight patients (2%) had either toe or more proximal amputations over the 12-year period. There was no significant difference in the number of patients with impalpable pedal pulses in the two groups. Retinopathy at baseline, male gender, tall stature and Caucasian ethnicity were all independent positive predictors for peripheral neuropathy (Table 1). Glycaemic control (average HbA1 over 12 years) was worse in patients developing an abnormal VPT Journal of Clinical Neuroscience (2001) 8(6), 520–524

JOCN-157.QXD

10/13/01 5:09 PM

Page 522

522 Coppini et al.

Table 1

Baseline characteristics of patients with and without future neuropathy

P value

Patients developing neuropathy (n:78)

‘Normal’ patients (n:314)

54 (69.2) 46.8<9.6 26 (33.3) 61 (78.2) 12 (15.3) 14.2<4.7 10.4<0.6 8 (10.2) 1 (1.3)

153 (48.7) 48.3<10.3 92 (29.3) 201 (64.1) 88 (28.0) 10.1<3.7 9.4<0.8 25 (8.0) 2 (0.6)

0.002 0.40 0.30 0.02 0.02 0.0001 0.0001 0.62 0.53

18 (23.1) 7 (9.0) 12.6<2.6 28.1<5.2 169.7<8.2 14 (17.9) 49 (62.8)

42 (13.4) 15 (4.8) 12.1<2.7 27.2<4.4 165.9<7.2 79 (25.2) 186 (59.2)

0.02 0.12 0.20 0.20 0.001 0.30 0.50

Male Age* IDDM Caucasian Afro-Caribbean Toe VPT (volts)* VPTscore* Absent ankle reflexes Absent cotton wool sensation (feet) Retinopathy (any grade) Proteinuria (.1;) HbA1* BMI (kg/m2)* Height (cm)* Smokers Alcohol (96 units weekly) Data are n (%) or otherwise specified. *mean
Characteristics of patients with and without neuropathy at review

Age (years)* Duration* Toe VPT (volts)* VPTscore* Abnormal monofilament Perception (score 1or 2) MNSI (score 92) Absent pedal pulses Plantar ulceration Amputations Neuropathic symptoms Hypertension history Retinopathy (any grade) Microalbuminuria (ACR93) Proteinuria (.1;) Average HbA1 over 12 years* Total cholesterol (mmol/L)* Triglycerides (mmol/L)* Creatinine (␮mol/L)* Family history of diabetes Lifetime smoking years † Insulin treatment Tablet treatment Diet treatment

P value

Patients with neuropathy n:78

‘Normal’ patients n:314

58.0<10.0 17.6<7.3 35.9<9.5 12.6<0.45 36 (46.1)

60.0<11.8 16.1<6.1 15.5<6.6 9.8<0.8 37 (11.8)

0.30 0.10 0.0001 0.0001 0.0001

42 (53.8) 11 (14.1) 6 (7.7) 8 (10.2) 35 (44.8) 14 (17.9) 48 (61.5) 17 (21.8) 25 (32.1) 11.0<1.4 5.5<0.9 2.4<1.2 103.0<24.7 25 (32.0) 17.2 (0-57) 43 (55.1) 30 (38.5) 5 (6.4)

33 (10.5) 25 (8.0) 1 (0.3) 0 (0) 68 (21.6) 80 (25.5) 125 (39.8) 85 (27.0%) 75 (23.8) 10.4<1.4 5.4<1.0 2.2<1.1 97.0<21.4 75 (27.1) 13.8 (0-61) 138 (44.0) 155 (49.4) 21 (6.6)

0.0001 0.06 0.002 0.0007 0.0001 0.20 :0.01 0.20 0.01 0.01 0.40 0.60 0.10 0.50 0.10 0.20 0.08 0.60

Data are n (%) or otherwise specified. Paired statistical tests were performed using logistic regression. * mean
(11.0<1.4% v. 10.4<1.4%, neuropathy v. normal; P:0.01) (Table 2). First visit (baseline) HbA1 was not significantly different in the two groups. Abnormal clinical tests (monofilament testing and MNSI 9 2) and lower extremity complications were all strongly associated with an abnormal VPT at review (Table 2). A VPTscore cut-off 9 10.1 showed overall better (P:0.02) sensitivity (77.3%) and specificity (72.8%) in predicting outcome than raw VPT values (sensitivity and specificity 63.6% and 68%, respectively, for VPT of 10 volts). It is interesting to note that a VPTscore of 10 is the mean value of normal subjects in an age group. Thus the cut off value to predict an abnormal value (VPTscore912) at the follow up coincides with the mean value of normal subjects. In the second (validation) study, there was a highly significant correlation of other tests with VPT for abnormal values (Table 3). Journal of Clinical Neuroscience (2001) 8(6), 520–524

Not all tests could be performed on every patient, as the numbers indicate in the Table. Seventy nine of 94 patients who had all the appropriate tests fulfilled the Dyck criteria for neuropathy. Only one out of 63 patients (1.6%) with an abnormal VPT was not diagnosed as having neuropathy. The predominant symptom occurring in patients with neuropathy on Dyck criteria who had normal VPT was distal limb pain (in 14 out of 17 patients), and there were abnormalities in thermal thresholds in all such cases. There was a relatively high proportion of such cases as they were referred to a neuropathy clinic for management of symptoms. DISCUSSION To our knowledge, this is the first long term prospective study reporting on the natural history of peripheral neuropathy in relation © 2001 Harcourt Publishers Ltd

JOCN-157.QXD

10/13/01 5:09 PM

Page 523

The natural history of diabetic peripheral neuropathy 523

Table 3

Correlation of vibration perception threshold (VPT) abnormalities with abnormalities of other tests

Test (n)

sural nerve action potential (112) monofilaments (110) cool perception threshold (108) warm perception threshold (108) axon-reflex vasodilatation (73) Neuropathy defined stages 1–3 (94) normal – no neuropathy abnormal – neuropathy

VPT abnormal (n:76)

VPT normal (n:47)

P value ␹² test

Regression analysis (on 48 cases with all tests) Correlation of test values with VPT values (slope (95% CI)

P value

90.124 (90.203 to 90.045)

:0.003

normal % (n)

abnormal % (n)

normal % (n)

abnormal % (n)

31.9 (22)

68.1 (47)

79.1 (34)

20.9 (9)

:0.0001

35.8 (24)

64.2 (43)

74.4 (32)

25.6 (11)

:0.0001

0.275 (0.209 to 0.340)

:0.0001

47.8 (32)

52.2 (35)

85.4 (35)

14.6 (6)

:0.0001

0.204 (0.076 to 0.332)

:0.003

16.2 (11)

83.8 (57)

37.5 (15)

62.5 (25)

:0.05

0.140 (0.019 to 0.261)

:0.03

25 (12)

75 (36)

28 (7)

72 (18)

ns

45.2 (14)*

54.8 (17)†

<0.001

1.6 (1)

98.4 (62)

90.592 (90.926 to 90.258) 0.036 (0.023 to 0.049)

:0.001 :0.0001

* 4/14 (28.6%) and † 14/17 (82.4%) had pain in the feet (P:0.005; Chi-square test). * 4/14 (28.6%) and † 17/17 (100%) had abnormalities on warm threshold (P:0.0001; Chi-square test).

to quantitative vibration perception threshold testing. Use of age corrected toe vibration thresholds as a marker of diabetic neuropathy seems both reliable and practical. An abnormal VPT is clearly related to other abnormal quantitative tests, and we have previously shown that it is a useful predictor of lower extremity complications.5 Although clinical tests (cotton wool sensation, ankle reflexes) were normal in most patients (90%) at baseline, vibration thresholds and ‘VPTscores’ were then already significantly higher in patients who went on to develop an abnormal VPT. We have previously reported higher vibration threshold measurements at the time of diagnosis in a subset of insulin-dependent diabetic patients in our cohort who subsequently developed peripheral neuropathy.13 Vibration thresholds have also been shown to correlate strongly with clinical scores in other studies.1,2 These findings suggest that subclinical large nerve fibre damage may be present at an early stage in patients who subsequently develop neuropathy. Most amputations (7/8) were at toe or forefoot level, indicating that these were for reasons of neuropathy rather than ischaemia. In our validation study, 62 out of 63 patients with neuropathy by Dyck criteria had an abnormal VPT. The few patients with a normal VPT who had neuropathy by Dyck criteria presented with symptoms of small sensory fibre dysfunction, particularly distal limb pain, and all of these had at least one abnormal thermal threshold. Painful diabetic neuropathy may result from dysfunction of small unmyelinated sensory fibres, with or without dysfunction of large sensory fibres.14,15 In accord, tests of unmyelinated sensory fibre function were poorly correlated with VPT, especially axon-reflex vasodilatation. Therefore when careful history taking is combined with physical examination, VPT testing is a sensitive and simple way for detecting neuropathy, and a combination of VPT and thermal threshold appears to detect all cases of neuropathy defined by Dyck criteria as well as subclinical neuropathy. Patients with retinopathy at baseline were at increased risk of developing peripheral neuropathy. Impaired blood flow has been shown in some studies to be an important aetiological feature of both neuropathy and retinopathy.16,17 Mild retinopathy is easier to detect than early neuropathy in diabetic patients. Minor retinal changes in the absence of any significant neurological abnormalities may help identify patients at risk of future neuropathy. Taller diabetic patients seem more vulnerable to peripheral neuropathy, in keeping with the length-dependent presentation of © 2001 Harcourt Publishers Ltd

diabetic sensory polyneuropathy.18 Caucasian patients seem to be at higher risk of developing neuropathy than Afro-Caribbean patients. The risk of amputation in diabetic patients was 2.3 times greater for blacks than whites in South Carolina,19 and 1.4 times greater for non-whites in New Jersey20; other risk factors such as peripheral vascular disease and poor socio-economic status may play a more important role in foot complications in black patients in the USA. The number of Asian patients entering the study was probably too small (3.8%) to assess the overall risk of neuropathy in this group, but no increase in susceptibility was detected. The Diabetes Control and Complications Trial (DCCT)21 has established that a lowering of HbA1c in patients with insulindependent diabetes (IDDM) was associated with a reduction in subsequent development of clinical neuropathy.21 Although haemoglobin A1 levels at the baseline visit were not different, mean glycaemic control in our cohort of both insulin and non-insulin dependent diabetic patients was worse in patients subsequently developing an abnormal VPT, indicating the importance of metabolic control. Of great interest, however, was that over 300 patients in our cohort (80%) retained a ‘normal’ VPT after an average diabetes duration of 16 years despite only average glycaemic control (mean HbA1 10.4<1.4). In the DCCT, 9.6% of conventionally treated insulin-dependent patients (mean HbA1c 9.1%) who had no clinical neuropathy at baseline developed confirmed clinical neuropathy over 6.5 years.22 Although the DCCT represented a younger and more homogenous population of patients, glycaemic control in conventionally treated patients was fairly representative of that seen in our clinics. It seems likely that non-ideal long term glycaemic control (at least over a 12 year period) leads to neuropathy only in a subset of patients predisposed to this and possibly other complications. Although we would have liked to study risk factors for neuropathy for IDDM and NIDDM separately, the number of IDDM patients with neuropathy (n:26) was too small relative to the entire cohort to enable this. However, there was no association between diet, drug or insulin treatment and neuropathy. In summary, ‘VPTscores’ show overall better sensitivity and specificity than raw VPT in identifying patients at risk at an early stage and are likely to be more useful in regular (e.g. annual) monitoring of patients. A ‘VPTscore 910.1’ indicates increased risk of developing neuropathy, and may select patients suitable for clinical trials and those likely to benefit from potential treatments for polyneuropathy. Similar long term prospective studies with thermal Journal of Clinical Neuroscience (2001) 8(6), 520–524

JOCN-157.QXD

10/13/01 5:09 PM

Page 524

524 Coppini et al.

threshold tests may help identify patients with early selective small nerve fibre dysfunction, who may present with pain or autonomic features.23

10.

11.

ACKNOWLEDGEMENTS This study was supported by Wyeth Laboratories, Maidenhead, England, UK; PA and AW gratefully acknowledge support of the Medical Research Council (UK) and Amgen, Inc., USA. REFERENCES 01. Franklin GM, Kahn LB, Baxter J, Marshall JA, Hamman R. Sensory neuropathy in non-insulin dependent diabetes mellitus. The San Luis Valley study. Amer J Epidemiol 1990; 131: 633–643. 02. Young MJ, Boulton AJM, Macleod AFM, Williams DRR, Sönksen PH. A multicentre study of the prevalence of diabetic peripheral neuropathy in the United Kingdom hospital clinic population. Diabetologia 1993; 36: 150–154. 03. Pirart J. Diabetes mellitus and its degenerative complications: a prospective study of 4400 patients observed between 1947 and 1973. Diabetes Care 1979; 1: 168–188. 04. Proceedings of a consensus development conference on standardised measures in diabetic neuropathy. Diabetes Care 1992; 15 (Suppl 3): 1080–1107. 05. Coppini DV, Young PJ, Weng C, Macleod AFM, Sönksen PH. Outcome on diabetic foot complications in relation to clinical examination and quantitative sensory testing- a case control study. Diabetic Med (in press). 06. Young MJ, Breddy JL, Veves A, Boulton AJM. The prediction of diabetic neuropathic foot ulceration using vibration perception thresholds: a prospective study. Diabetes Care 1994; 17: 557–560. 07. Dyck PJ. Quantitating severity of neuropathy. In: Dyck PJ, Thomas PK (Eds.) Peripheral neuropathy, 3rd ed. Philadelphia: Saunders, 1993: 686–697. 08. Bloom S, Till S, Sönksen P, Smith S. Use of a biothesiometer to measure individual vibration thresholds and their variation in 519 non-diabetic subjects. BMJ 1984; 288: 1793–1795. 09. Wiles PG, Pearce SM, Rice PJS, Mitchell JMO. Vibration Perception Threshold: Influence of age, height, sex, and smoking, and calculation of accurate centile values. Diabetic Med 1991; 8: 157–161.

Journal of Clinical Neuroscience (2001) 8(6), 520–524

12. 13.

14.

15.

16.

17.

18.

19. 20.

21.

22.

23.

Feldman EL, Stevens MJ, Thomas PK, Brown MB, Canal N, Greene DA. A practical two-step quantitative clinical and electrophysiological assessment for the diagnosis and staging of diabetic neuropathy. Diabetes Care 1994; 17: 1281–1289. Anand P, Terenghi G, Warner G, Kopelman P, Williams-Chestnut RE, Sinicropi DV. The role of endogenous nerve growth factor in human diabetic neuropathy. Nat Med 1996; 2: 703–707. Collett D. Modelling binary data. London: Chapman and Hall, 1991. Coppini DV, Sönksen PH. Subclinical peripheral neuropathy may be present at onset of IDDM in high risk patients. Diabetologia 1996; 3 [Suppl 1]: 976. Lanting P, Faes TJ, Ijff GA, Bertelsmann FW, Heimanns JJ, van der Veen EA. Autonomic and somatic peripheral nerve function and the correlation with neuropathic pain in diabetic patients. J Neurol Sci 1989; 94: 307–317. Tsigos C, White A, Young RJ. Discrimination between painful and painless diabetic neuropathy based on testing of large somatic nerve and sympathetic fibres. Diabet Med 1992; 9: 359–365. Tesfaye S, Harris N, Jakubowski JJ, Mody C, Wilson RM, Rennie IG, Ward JD. Impaired blood flow and arterio-venous shunting in human diabetic neuropathy: a novel technique of nerve photography and fluorescein angiography. Diabetologia 1994; 36: 1266–1274. Sebag J, McMeel JW. Diabetic retinopathy. Pathogenesis and the role of retinaderived growth factor in angiogenesis. Survey of Opthalmology 1986; 30: 377–384. Sosenko JM, Gadia MT, Fournier AM, O’Connell MT, Aguiar MC, Skyler MS. Body stature as a risk factor for diabetic sensory neuropathy. Am J Med 1986; 80: 1031–1034. Most RS, Sinnock P. The epidemiology of lower extremity amputations in diabetic individuals. Diabetes Care 1983; 6: 87–91. Miller AD, Van Buskirk AH, Verhoek-Oftedahl W, Miller ER. Diabetes related lower extermity amputations in New Jersey, 1979–1981. J Med Soc NJ 1985; 82: 723–726. Diabetes Control and Complications Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications of diabetes mellitus. New Eng J Med 1993; 329: 977–986. The Diabetes Control and Complications Trial Research Group. The effect of intensive diabetes therapy on the development and progression of neuropathy. Ann Int Med 1995; 122: 561–568. Wellmer A, Warner G, Kopelman P, Anand P. Quantitative sensory and autonomic testing in male diabetic patients with erectile dysfunction. Brit J Urology 1999; 83: 66–70.

© 2001 Harcourt Publishers Ltd