Drn
ORIGINAL ARTICLES
-
The Distribution and Severity of Diabetic Foot Disease: a Community Study with Comparison to a Non-diabetic Group D.P. Walters’, W. Gatling”, M.A. Mulleeb, R.D.
Hill”
aDepartment of Diabetic Medicine, Poole General Hospital, Poole, Dorset and bDepartment of Medical Statistics and Computing, University of Southampton, UK
A surveillance programme was undertaken t o identify all diabetic patients with foot disease i n a defined population with the same age and sex structure as that of the UK. Of 1150 diabetic patients identified, 1077 were reviewed either at home or i n hospital. The presence of foot deformity, amputation, and foot ulceration was determined. The site, depth, and duration of ulcers were recorded and any previous ulceration noted. All feet with ulcers were X-rayed. A non-diabetic comparison group of 480 age- and sex-matched individuals were also examined by the same observer. The prevalence of past or present foot ulceration was 7.4 (95 % CI 5.8-9.0) % in diabetic patients and 2.5 (95 % CI 1.1-3.9) % in the non-diabetic group, yielding an odds ratio of 2.94 (95 % CI 1.58-5.48) ( p < 0.001) for the occurrence of foot ulceration in diabetic patients. O f the ulcers found on examination, 39.4 % were neuropathic, 24.2 % were vascular, and 36.4 % were mixed. Multiple logistic regression analysis of selected variables revealed that duration of diabetes, absent light touch, impaired pain perception, absent dorsalis pedis pulse, and the presence of any retinopathy were significant predictors of the presence of foot ulcers. The prevalence of amputation i n diabetic patients was 1.3 (95 % CI 0.6-2.0) %, but there were no amputations in the non-diabetic group. KEY WORDS
Diabetes mellitus Complications
Foot problems Amputation
10 general practices were identified using previously published methods.6 The study population has a similar Diabetic patients are prone to ulceration and gangrene age and sex structure to that of the UK and covered rural of the lower limb and are up to 70 times more likely to and urban areas around the Poole area of East Dorset. have an amputation than non-diabetic subjects.’ Apart From a general population of 97034, 1150 diabetic from being a serious cause of morbidity, diabetic foot patients were identified. Of these, 19.7 % were classified disease is economically of great importance. Connor2 as having Type 1 diabetes (defined as previous docuestimated the cost of a major amputation to be f8544 mented ketosis or requiring insulin within 1 month of per patient in 1987. In addition there is the cost of foot diagnosis’) and 80.3 % were classified as having Type ulceration which is a major cause of hospital a d m i ~ s i o n . ~ 2 diabetes. There is some evidence that diabetic foot disease is A non-diabetic comparison group of 751 individuals pre~entable,~,~ but data are not available on the precise aged 30 years or more (since no cases of foot disease frequency and characteristics of foot disease in the were found in diabetic patients under the age of 30 diabetic population. Such information is important to years) was also selected by obtaining the name of a facilitate adequate planning of health care provision and subject with the closest date of birth to the diabetic to identify high risk groups. The aim of this study was patient on the general practitioner’s agehex register in 7 to establish the prevalence, severity, and distribution out of the 10 practices. These 7 practices covered all of diabetic foot disease in a geographically defined geographical districts in the study area. All practices in population with reference to an age- and sex-matched the study area had agehex registers and all were in 1 non-diabetic group drawn from the same community. chronological birth order rather than alphabetical order. Records were checked for non-diabetic status and subjects Patients and Methods were asked to have a blood test 2 h after a meal to screen for diabetes. Subjects with an abnormal result During a surveillance programme between 1988 and were excluded. 1990, all the known diabetic patients registered with All individuals from the diabetic and non-diabetic groups were invited to attend for a structured interview Correspondence to: Dr D.P. Walters, Department of Ceratology, The and physical examination. Subjects were either seen at Radcliffe Infirmary, Woodstock Road, Oxford, OX2 6HE, UK.
Introduction
354
0742-3071/92/040354-05$05.00
@ 1992 by John Wiley & Sons, Ltd.
Accepted 13 December 1991 DIABETIC MEDICINE, 1992; 9: 354-358
Dm hospital research clinics or, if they could not travel, were visited at home by a single observer. A standard history was performed which included details regarding the duration and treatment of diabetes and history of intermittent claudication using the WHO Cardiovascular Questionnaire.a The examination included an assessment of foot ulceration, pulse palpation, and ank1e:brachial Doppler pressure ratios in the lower limb. Each patient was also examined neurologically for the presence or absence of light touch (using cotton wool), impairment of pain (using pinprick), presence or absence of limb jerks, and vibration thresholds using a biothesiometer (Bio-Medical Instruments, Newbury, OH, USA).9 The latter was recorded at the medial malleolus and great toe on each foot and the vibration perception threshold was the mean of four readings for each site. A neuropathic joint was defined as gross deformity of the foot associated with loss of sensation clinically. The X-ray findings (interpreted by a consultant radiologist aware of the clinical details) included subluxation, joint disorganization, and new bone formation. All feet with ulcers at or below the medial malleoli were recorded and graded by the Wagner System.'O The ulcer definitions were as follows: Grade 1 ulcer: a superficial ulcer; Grade 2 ulcer: extending down to the ligament, joint capsule or deep fascia; Grade 3 ulcer: the presence of radiologically confirmed osteomyel itis; Grade 4 and 5 ulcers: with at least one region of gangrene.
All feet with ulcers were X-rayed. An ulcer was defined as neuropathic if there was loss of light touch or impaired pain perception in at least one region of the foot with an ank1e:brachial pressure index 1.0 or more. An ischaemic ulcer was defined as an ulcer with a Doppler pressure of 0.9 or less with intact light touch and pain perception. Where both criteria coexisted the ulcer was defined as having a mixed aetiology. The site and duration of previous ulcers were also recorded checking with previous general practitioner, hospital, and chiropody records. Only ulcers diagnosed at or after the diagnosis of diabetes mellitus was made, were included in the analyses. All amputations were noted and the most proximal site was used for determining the prevalence if a previous amputation had been performed. Laboratory investigations included a HbA, measurement (Fluckiger Method"), plasma glucose taken 2 h after a main meal, and serum total-cholesterol (nonfasting enzymatic method, Bayer Diagnostic UK, Basingstoke, UK). Proteinuria was measured using Albustix (Ames, Slough, UK).
ORIGINAL ARTICLES Statistical Methods Results were analysed using the SPSS-X software package. l 2 To examine differences between subjects with and without foot ulcers univariate and multivariate analyses were performed. The chi-squared test (with Yates correction) was used for dichotomous variables where age ranges were similar. In instances where age differences existed between groups of subjects, multiple logistic regression analysis was used to compare variables. Multiple logistic regression analysis was employed to examine the relationship between the binary dependent variable (foot ulceration) and selected independent variables. All variables were initially analysed by backward stepping to identify all those significantly associated with neuropathy. Age and duration of diabetes were always forced into the regression (at every step) in view of their possible importance, and the impact of the other risk factors could thus be assessed independently of these variables. After identifying the significant variables, a condensed logistic regression model was computed to include age and duration of diabetes plus any significant variables.
Results Of the 11 50 diabetic patients identified, 1077 (93.7 %) were reviewed. From the 751 control subjects selected, 41 had moved and of the remainder 485 (68.1 %) attended for review of whom 5 proved to be diabetic. Home visits were performed on 130 (12.1 %) of the diabetic patients and 31 (6.5 %) of the non-diabetic comparison group. The prevalence of past or present foot ulceration was 7.4 (95 % CI 5.8-9.0) % in diabetic patients and 2.6 (95 % CI 1.1-3.9) % in the non-diabetic group. This yielded a prevalence odds of 2.94 (95 % CI 1.58-5.48; p < 0.001) for the occurrence of ulceration in diabetic patients. Table 1 shows the prevalence of foot ulceration in diabetic patients according to age and sex. The mean age of diabetic patients with past or present foot ulceration was 73.4 (SD 10.3) years. There was no significant Table 1. Prevalence of foot ulceration in diabetic patients according to age and gender Age (years)
Males (n = 526)
30-39 40-49 50-59 60-69 70-79 80+
0 1 2 7 19 12
Total
41 (7.8)
(0) (3.6) (2.6) (5.0) (11.5) (14.3)
Females (n = 478)
0 2 2 8 9 12
Al I ( n = 1004)
(0) (4.3) (3.4) (7.4) (5.9) (13.8)
33 (6.9)
0 3 4 15
(0) (4.0) (2.9) (6.0) 28 (8.8) 24 (14.0) 74 (7.4)'
Number (% of age group). "Prevalence in diabetic patients aged 30 years or more. DISTRIBUTION OF DIABETIC FOOT DISEASE IN A BRITISH COMMUNITY
355
ORIGINAL ARTICLES difference in prevalence between male and female patients. At review, 33 of the diabetic patients had ulceration present on examination, yielding a prevalence of 3.3 (95 % CI 2.2-4.4) % in diabetic patients aged 30 years or more. In Type 1 diabetes, the prevalence of past or present foot ulceration was 7.7 (95 % CI 4.1-1 1.3) %, yielding a prevalence odds of 11.29 (95 % CI 4.68-27.20; p < 0.001) for foot ulceration. The prevalence in Type 2 diabetes was 6.8 (95 % CI 5.1-8.5) % and the prevalence odds was 2.94 (95 % CI 1.27-5.48; p < 0.01) for foot ulceration. The median duration of ulceration present at review was 0.44 (interquartile range 0.14-1.66) months and for previous ulcers the median duration was 5 (interquartile range 2-7.5) months. Of the diabetic foot ulcers found at examination, 11 (33 %) were being treated in the hospital diabetic clinic, 19 (58 %) solely in the community, and 3 (9 %) were receiving no treatment in that they were only discovered at the time of examination. The majority of ulcers were Grade 1 (60.6 %) or 2 (33.3 %). There were no cases of Grade 4 or 5 ulcers, and 6.1 % were Grade 3. With respect to aetiology 39.4 % were neuropathic, 24.2 % vascular, and 36.2 % were mixed. In terms of the site of ulceration in diabetic patients with either a previous ulcer or an ulcer present at review, 24 were digital (including hallux valgus), 22 were on the heel, 18 under the 1st metatarsal head, 11 on the malleoli, and the remaining 8 were under the 2nd to 4th metatarsal heads. Of the ulcers in the nondiabetic group, 5 were digital, 4 were on the malleoli, and 4 were on the heel. In the non-diabetic group, only 3 individuals were found to have a foot ulcer at time of examination. Of these, 1 was traumatic in origin and the other 2 were found in subjects with hemiplegia. The prevalence of amputation was 1.3 (95 % CI 0.6-2.0) % in diabetic patients. Of the 14 patients with amputation, 3 had Type 1 diabetes and 11 had Type 2 diabetes. The site of amputation was above knee in 2 cases, below knee in 5 cases, part foot in 4 cases, and 3 were digital. In all but 2 cases, the amputation had been preceded by foot ulceration. No cases of amputation were found in the non-diabetic group. Neuropathic joints were found in 3 diabetic patients but in none of the non-diabetic subjects. Significant potential risk factors for foot ulceration are shown in Table 2.
Discussion It can be reasonably assumed that most cases of foot disease were detected in this survey, given the high review rate and the large number of home visits. The latter are necessary in view of the immobility that foot disease may cause. Using one observer eliminated interobserver variation, but the degree of intra-observer error was impossible to assess. The prevalence of foot ulceration in the non-diabetic group may be an
356
Dm underestimate since the population was smaller (because of time constraints) and the review rate was lower than with the diabetic patients. However, the 2-h postprandial blood glucose test may have failed to identify a small number of undiagnosed diabetic patients. The Oxford Community Study13 found that 5 % of the diabetic population had foot ulcers. This study was of small size in a predominantly elderly population with a low review rate. In a recent Swedish population study14 investigating foot problems in diabetic patients aged between 15 and 50 years, it was found that 3 % of the population had foot ulcers at the time of examination but a much higher prevalence (9.8 %) of previous ulceration was found. Questionnaire surveys, either sent to patients directly' or to general practitioners and allied health workers,16 found a prevalence of foot ulceration of 2.1 and 2.75 %, respectively (assumingthe known prevalence of diabetes was 1.2 %). The latter survey was community based and noted the prevalence of foot ulceration treated in the non-diabetic population to be 0.05 %. Although this is much lower than our study findings, it confirms the marked propensity to foot ulceration in diabetic patients. It is clear that ulcers tend to occur at high pressure points in diabetic and non-diabetic subjects." The high prevalence of heel ulcers found in the present survey i s at variance with previous clinic-based s t u d i e ~ . ~This ,'~ survey also suggests that most ulcers are superficial. Inevitably, ulcers seen in hospital clinics will often be more severe due to selection bias.18 Many of the significant, independent associations found with the presence of foot ulceration following multivariate analysis are predictable. More interesting was the lack of association between many variables, including age, smoking, diabetic control, vibration perception threshold, and Doppler pressure ratios, that would be expected to be associated with foot ulceration. However, the main aetiological determinants coexist, as was found in 36 % of cases in this survey. Previous clinic-based studies have reported a strong association between foot ulceration and retin~pathy,'~-~l which was confirmed in the present survey and was independent of duration of diabetes and measures of diabetic control. This would imply a propensity to both complications in a sub-group of diabetic patients, irrespective of the degree of exposure to diabetes. The lack of an association between Doppler pressure index and foot ulceration highlights the need for angiography in patients with foot ulceration if vascular insufficiency is suspected clinically, because of the higher prevalence of vascular calcification.18 It is clear that impaired pain perception and absent light touch are associated with ulceration. In contrast to the present survey, previous work has suggested that the vibration perception threshold is the most important predictor of foot ulceration.22 However, that study was not population based and investigated a relatively young group of diabetic patients (mean age of the ulcer group D.P.WALTERS €7 AL.
Dm
ORIGINAL ARTICLES Table 2. Significant associations between selected variables and foot ulceration following multiple regression analysis in diabetic patients Variable
Odds ratio
Duration of diabetes (years) Absent light touch Impaired pain perception Absent dorsalis pedis pulse Any retinopathy
1.05' 2.85 3.58 6.27 3.23
95 % CI 1.02-1.08 1.88-3.81 2.60-4.55 5.57-7.00 2.54-3.91
P 0,008 0.03 0.01
ORIGINAL ARTICLES
-
The Distribution and Severity of Diabetic Foot Disease: a Community Study with Comparison to a Non-diabetic Group D.P. Walters’, W. Gatling”, M.A. Mulleeb, R.D.
Hill”
aDepartment of Diabetic Medicine, Poole General Hospital, Poole, Dorset and bDepartment of Medical Statistics and Computing, University of Southampton, UK
A surveillance programme was undertaken t o identify all diabetic patients with foot disease i n a defined population with the same age and sex structure as that of the UK. Of 1150 diabetic patients identified, 1077 were reviewed either at home or i n hospital. The presence of foot deformity, amputation, and foot ulceration was determined. The site, depth, and duration of ulcers were recorded and any previous ulceration noted. All feet with ulcers were X-rayed. A non-diabetic comparison group of 480 age- and sex-matched individuals were also examined by the same observer. The prevalence of past or present foot ulceration was 7.4 (95 % CI 5.8-9.0) % in diabetic patients and 2.5 (95 % CI 1.1-3.9) % in the non-diabetic group, yielding an odds ratio of 2.94 (95 % CI 1.58-5.48) ( p < 0.001) for the occurrence of foot ulceration in diabetic patients. O f the ulcers found on examination, 39.4 % were neuropathic, 24.2 % were vascular, and 36.4 % were mixed. Multiple logistic regression analysis of selected variables revealed that duration of diabetes, absent light touch, impaired pain perception, absent dorsalis pedis pulse, and the presence of any retinopathy were significant predictors of the presence of foot ulcers. The prevalence of amputation i n diabetic patients was 1.3 (95 % CI 0.6-2.0) %, but there were no amputations in the non-diabetic group. KEY WORDS
Diabetes mellitus Complications
Foot problems Amputation
10 general practices were identified using previously published methods.6 The study population has a similar Diabetic patients are prone to ulceration and gangrene age and sex structure to that of the UK and covered rural of the lower limb and are up to 70 times more likely to and urban areas around the Poole area of East Dorset. have an amputation than non-diabetic subjects.’ Apart From a general population of 97034, 1150 diabetic from being a serious cause of morbidity, diabetic foot patients were identified. Of these, 19.7 % were classified disease is economically of great importance. Connor2 as having Type 1 diabetes (defined as previous docuestimated the cost of a major amputation to be f8544 mented ketosis or requiring insulin within 1 month of per patient in 1987. In addition there is the cost of foot diagnosis’) and 80.3 % were classified as having Type ulceration which is a major cause of hospital a d m i ~ s i o n . ~ 2 diabetes. There is some evidence that diabetic foot disease is A non-diabetic comparison group of 751 individuals pre~entable,~,~ but data are not available on the precise aged 30 years or more (since no cases of foot disease frequency and characteristics of foot disease in the were found in diabetic patients under the age of 30 diabetic population. Such information is important to years) was also selected by obtaining the name of a facilitate adequate planning of health care provision and subject with the closest date of birth to the diabetic to identify high risk groups. The aim of this study was patient on the general practitioner’s agehex register in 7 to establish the prevalence, severity, and distribution out of the 10 practices. These 7 practices covered all of diabetic foot disease in a geographically defined geographical districts in the study area. All practices in population with reference to an age- and sex-matched the study area had agehex registers and all were in 1 non-diabetic group drawn from the same community. chronological birth order rather than alphabetical order. Records were checked for non-diabetic status and subjects Patients and Methods were asked to have a blood test 2 h after a meal to screen for diabetes. Subjects with an abnormal result During a surveillance programme between 1988 and were excluded. 1990, all the known diabetic patients registered with All individuals from the diabetic and non-diabetic groups were invited to attend for a structured interview Correspondence to: Dr D.P. Walters, Department of Ceratology, The and physical examination. Subjects were either seen at Radcliffe Infirmary, Woodstock Road, Oxford, OX2 6HE, UK.
Introduction
354
0742-3071/92/040354-05$05.00
@ 1992 by John Wiley & Sons, Ltd.
Accepted 13 December 1991 DIABETIC MEDICINE, 1992; 9: 354-358
Dm hospital research clinics or, if they could not travel, were visited at home by a single observer. A standard history was performed which included details regarding the duration and treatment of diabetes and history of intermittent claudication using the WHO Cardiovascular Questionnaire.a The examination included an assessment of foot ulceration, pulse palpation, and ank1e:brachial Doppler pressure ratios in the lower limb. Each patient was also examined neurologically for the presence or absence of light touch (using cotton wool), impairment of pain (using pinprick), presence or absence of limb jerks, and vibration thresholds using a biothesiometer (Bio-Medical Instruments, Newbury, OH, USA).9 The latter was recorded at the medial malleolus and great toe on each foot and the vibration perception threshold was the mean of four readings for each site. A neuropathic joint was defined as gross deformity of the foot associated with loss of sensation clinically. The X-ray findings (interpreted by a consultant radiologist aware of the clinical details) included subluxation, joint disorganization, and new bone formation. All feet with ulcers at or below the medial malleoli were recorded and graded by the Wagner System.'O The ulcer definitions were as follows: Grade 1 ulcer: a superficial ulcer; Grade 2 ulcer: extending down to the ligament, joint capsule or deep fascia; Grade 3 ulcer: the presence of radiologically confirmed osteomyel itis; Grade 4 and 5 ulcers: with at least one region of gangrene.
All feet with ulcers were X-rayed. An ulcer was defined as neuropathic if there was loss of light touch or impaired pain perception in at least one region of the foot with an ank1e:brachial pressure index 1.0 or more. An ischaemic ulcer was defined as an ulcer with a Doppler pressure of 0.9 or less with intact light touch and pain perception. Where both criteria coexisted the ulcer was defined as having a mixed aetiology. The site and duration of previous ulcers were also recorded checking with previous general practitioner, hospital, and chiropody records. Only ulcers diagnosed at or after the diagnosis of diabetes mellitus was made, were included in the analyses. All amputations were noted and the most proximal site was used for determining the prevalence if a previous amputation had been performed. Laboratory investigations included a HbA, measurement (Fluckiger Method"), plasma glucose taken 2 h after a main meal, and serum total-cholesterol (nonfasting enzymatic method, Bayer Diagnostic UK, Basingstoke, UK). Proteinuria was measured using Albustix (Ames, Slough, UK).
ORIGINAL ARTICLES Statistical Methods Results were analysed using the SPSS-X software package. l 2 To examine differences between subjects with and without foot ulcers univariate and multivariate analyses were performed. The chi-squared test (with Yates correction) was used for dichotomous variables where age ranges were similar. In instances where age differences existed between groups of subjects, multiple logistic regression analysis was used to compare variables. Multiple logistic regression analysis was employed to examine the relationship between the binary dependent variable (foot ulceration) and selected independent variables. All variables were initially analysed by backward stepping to identify all those significantly associated with neuropathy. Age and duration of diabetes were always forced into the regression (at every step) in view of their possible importance, and the impact of the other risk factors could thus be assessed independently of these variables. After identifying the significant variables, a condensed logistic regression model was computed to include age and duration of diabetes plus any significant variables.
Results Of the 11 50 diabetic patients identified, 1077 (93.7 %) were reviewed. From the 751 control subjects selected, 41 had moved and of the remainder 485 (68.1 %) attended for review of whom 5 proved to be diabetic. Home visits were performed on 130 (12.1 %) of the diabetic patients and 31 (6.5 %) of the non-diabetic comparison group. The prevalence of past or present foot ulceration was 7.4 (95 % CI 5.8-9.0) % in diabetic patients and 2.6 (95 % CI 1.1-3.9) % in the non-diabetic group. This yielded a prevalence odds of 2.94 (95 % CI 1.58-5.48; p < 0.001) for the occurrence of ulceration in diabetic patients. Table 1 shows the prevalence of foot ulceration in diabetic patients according to age and sex. The mean age of diabetic patients with past or present foot ulceration was 73.4 (SD 10.3) years. There was no significant Table 1. Prevalence of foot ulceration in diabetic patients according to age and gender Age (years)
Males (n = 526)
30-39 40-49 50-59 60-69 70-79 80+
0 1 2 7 19 12
Total
41 (7.8)
(0) (3.6) (2.6) (5.0) (11.5) (14.3)
Females (n = 478)
0 2 2 8 9 12
Al I ( n = 1004)
(0) (4.3) (3.4) (7.4) (5.9) (13.8)
33 (6.9)
0 3 4 15
(0) (4.0) (2.9) (6.0) 28 (8.8) 24 (14.0) 74 (7.4)'
Number (% of age group). "Prevalence in diabetic patients aged 30 years or more. DISTRIBUTION OF DIABETIC FOOT DISEASE IN A BRITISH COMMUNITY
355
ORIGINAL ARTICLES difference in prevalence between male and female patients. At review, 33 of the diabetic patients had ulceration present on examination, yielding a prevalence of 3.3 (95 % CI 2.2-4.4) % in diabetic patients aged 30 years or more. In Type 1 diabetes, the prevalence of past or present foot ulceration was 7.7 (95 % CI 4.1-1 1.3) %, yielding a prevalence odds of 11.29 (95 % CI 4.68-27.20; p < 0.001) for foot ulceration. The prevalence in Type 2 diabetes was 6.8 (95 % CI 5.1-8.5) % and the prevalence odds was 2.94 (95 % CI 1.27-5.48; p < 0.01) for foot ulceration. The median duration of ulceration present at review was 0.44 (interquartile range 0.14-1.66) months and for previous ulcers the median duration was 5 (interquartile range 2-7.5) months. Of the diabetic foot ulcers found at examination, 11 (33 %) were being treated in the hospital diabetic clinic, 19 (58 %) solely in the community, and 3 (9 %) were receiving no treatment in that they were only discovered at the time of examination. The majority of ulcers were Grade 1 (60.6 %) or 2 (33.3 %). There were no cases of Grade 4 or 5 ulcers, and 6.1 % were Grade 3. With respect to aetiology 39.4 % were neuropathic, 24.2 % vascular, and 36.2 % were mixed. In terms of the site of ulceration in diabetic patients with either a previous ulcer or an ulcer present at review, 24 were digital (including hallux valgus), 22 were on the heel, 18 under the 1st metatarsal head, 11 on the malleoli, and the remaining 8 were under the 2nd to 4th metatarsal heads. Of the ulcers in the nondiabetic group, 5 were digital, 4 were on the malleoli, and 4 were on the heel. In the non-diabetic group, only 3 individuals were found to have a foot ulcer at time of examination. Of these, 1 was traumatic in origin and the other 2 were found in subjects with hemiplegia. The prevalence of amputation was 1.3 (95 % CI 0.6-2.0) % in diabetic patients. Of the 14 patients with amputation, 3 had Type 1 diabetes and 11 had Type 2 diabetes. The site of amputation was above knee in 2 cases, below knee in 5 cases, part foot in 4 cases, and 3 were digital. In all but 2 cases, the amputation had been preceded by foot ulceration. No cases of amputation were found in the non-diabetic group. Neuropathic joints were found in 3 diabetic patients but in none of the non-diabetic subjects. Significant potential risk factors for foot ulceration are shown in Table 2.
Discussion It can be reasonably assumed that most cases of foot disease were detected in this survey, given the high review rate and the large number of home visits. The latter are necessary in view of the immobility that foot disease may cause. Using one observer eliminated interobserver variation, but the degree of intra-observer error was impossible to assess. The prevalence of foot ulceration in the non-diabetic group may be an
356
Dm underestimate since the population was smaller (because of time constraints) and the review rate was lower than with the diabetic patients. However, the 2-h postprandial blood glucose test may have failed to identify a small number of undiagnosed diabetic patients. The Oxford Community Study13 found that 5 % of the diabetic population had foot ulcers. This study was of small size in a predominantly elderly population with a low review rate. In a recent Swedish population study14 investigating foot problems in diabetic patients aged between 15 and 50 years, it was found that 3 % of the population had foot ulcers at the time of examination but a much higher prevalence (9.8 %) of previous ulceration was found. Questionnaire surveys, either sent to patients directly' or to general practitioners and allied health workers,16 found a prevalence of foot ulceration of 2.1 and 2.75 %, respectively (assumingthe known prevalence of diabetes was 1.2 %). The latter survey was community based and noted the prevalence of foot ulceration treated in the non-diabetic population to be 0.05 %. Although this is much lower than our study findings, it confirms the marked propensity to foot ulceration in diabetic patients. It is clear that ulcers tend to occur at high pressure points in diabetic and non-diabetic subjects." The high prevalence of heel ulcers found in the present survey i s at variance with previous clinic-based s t u d i e ~ . ~This ,'~ survey also suggests that most ulcers are superficial. Inevitably, ulcers seen in hospital clinics will often be more severe due to selection bias.18 Many of the significant, independent associations found with the presence of foot ulceration following multivariate analysis are predictable. More interesting was the lack of association between many variables, including age, smoking, diabetic control, vibration perception threshold, and Doppler pressure ratios, that would be expected to be associated with foot ulceration. However, the main aetiological determinants coexist, as was found in 36 % of cases in this survey. Previous clinic-based studies have reported a strong association between foot ulceration and retin~pathy,'~-~l which was confirmed in the present survey and was independent of duration of diabetes and measures of diabetic control. This would imply a propensity to both complications in a sub-group of diabetic patients, irrespective of the degree of exposure to diabetes. The lack of an association between Doppler pressure index and foot ulceration highlights the need for angiography in patients with foot ulceration if vascular insufficiency is suspected clinically, because of the higher prevalence of vascular calcification.18 It is clear that impaired pain perception and absent light touch are associated with ulceration. In contrast to the present survey, previous work has suggested that the vibration perception threshold is the most important predictor of foot ulceration.22 However, that study was not population based and investigated a relatively young group of diabetic patients (mean age of the ulcer group D.P.WALTERS €7 AL.
Dm
ORIGINAL ARTICLES Table 2. Significant associations between selected variables and foot ulceration following multiple regression analysis in diabetic patients Variable
Odds ratio
Duration of diabetes (years) Absent light touch Impaired pain perception Absent dorsalis pedis pulse Any retinopathy
1.05' 2.85 3.58 6.27 3.23
95 % CI 1.02-1.08 1.88-3.81 2.60-4.55 5.57-7.00 2.54-3.91
P 0,008 0.03 0.01
