cohort of individuals who are known to be free of foot ulcers at the onset of the study, and evaluate those individuals at a later time to determine the subsequent foot ulcer presence or absence, using a clear foot ulcer definition. 5.
While randomised controlled trial cohorts allow for careful ascertainment of foot ulcer incidence, they may be unsuitable to estimate incidence in the overall population of people with diabetes in the region where the study was conducted (even in controls, or in a study where the intervention was not successful and the investigators attempt to draw conclusions from the total cohort) because the sample is often highly selected, i.e. very unlike the population from which the sampled participants are drawn.
In one recent clinical trial sample that was used to estimate incidence of foot ulcers, participants must have been 18, 70 years old, men or non-pregnant women, had a vibration perception threshold (VPT) of ≥25 V on at least one foot and have had no prior foot ulceration or lower limb amputation, only diabetic causes of neuropathy, and no history of alcohol abuse, previous treatment with radiotherapy or cytotoxic agents, uncontrolled hypertension or any renal disease.25 While these criteria are suitable to enrolls participants into a controlled trial, they make it impossible to generalise results from such a study to estimate foot ulcer incidence in the general population of people with diabetes.
JWBK089-01 JWBK089-Boulton April 21, 2006 20:30 Char Count= 0
RISK FACTORS FOR DIABETIC FOOT ULCERS 5
Table 1.2 Anatomic site and outcome of diabetic foot lesions in two prospective studies
All lesionsa (%) Most severe lesionb (%) N = 314 N = 302
Lesion site (%) Toes (dorsal and plantar surface) 51 52 Metatarsal heads, midfoot and heel 28 37 Dorsum of the foot 14 11 Multiple ulcers 7 NA Total 100 100
a Ref. 46 (Apelqvist et al.); study included consecutive patients whose lesions were characterised according to Wagner criteria from superficial non-necrotic to major gangrene. b Ref. 47 (Reiber et al.); study patients were enrolled with a lesion through the dermis that could extend to deeper tissue.
ANATOMIC LOCATION OF FOOT ULCERS The anatomic site of foot ulcers varies according to the population from which the patients are drawn. Table 1.2 presents data from two prospective studies that reported foot ulcer site. Those studies using patients from general diabetes clinics found that the most common ulcer sites were the toes (dorsal and plantar surface), followed by the metatarsal heads.46,47
US COUNTRYWIDE ESTIMATES OF FOOT-ULCER-RELATED CONDITIONS The ability to identify foot ulcers in individuals in clinics, offices and outpatient settings is difficult, due to limited surveillance systems.
However, information on patients hospitalised with foot-ulcer-related conditions is available in many countries.
In the United States, for example, the Healthcare Cost and Utilization Project Nationwide Inpatient Sample (NIS) re- ports a 20% stratified sample of US community hospitals discharges (excluding discharges from Department of Veterans Affairs and military hospitals).
In Table 1.3 the data for 2001 and 2002 by International Classification of Diseases (ICD-9) codes, weighted to reflect the US civil- ian population, show that the leading reasons for hospitalisation among diabetic patients with foot-ulcer-related conditions are cellulitis and abscess, lower limb ulcers and osteomyelitis.48
RISK FACTORS FOR DIABETIC FOOT ULCERS Independent risk factors for diabetic foot ulcers were identified from analytic and experimental studies that used multivariable modelling techniques and included a defined foot ulcer outcome. Results shown in Table 1.4 demonstrate that the most consistent independent foot ulcer risk factors were long diabetes duration, measures of peripheral neuropathy and peripheral vascular disease, prior foot ulcer and prior amputation. Long duration of diabetes, even after controlling for age, was a statistically significant finding in three studies.16,23,24 The independent role of plantar foot pressure remains unclear.
JWBK089-01 JWBK089-Boulton April 21, 2006 20:30 Char Count= 0
6 EPIDEMIOLOGY AND ECONOMIC IMPACT OF FOOT ULCERS
Table 1.3 Frequency of foot-ulcer-related conditions in hospitalised individuals with diabetes, 2001, 2002
ICD-CM Estimated US Estimated US Type of ulcer code frequency, 2001 frequency, 2002
Cellulitis, abscess or infected ulcer 681.1 26 685 29 347 Other cellulitis and abscess, foot, except toes 682.7 81 367 83 954 Ulcer of lower limbs, except decubitus 707.1 209 088 216 785 Osteomyelitis 730.07 60 989 66 591 730.17 730.27 730.37 730.87 730.97 Chronic non-healing ulcers 707.0 129 466 134 274 707.9 Atherosclerosis of lower limb with ulcer or gangrene 440.23 83 546 78 983 440.24
Source: Nationwide Inpatient Sample, 2001, 2002.48
Assessment of diabetic peripheral neuropathy is performed using several semi-quantitative and quantitative measures and neurological summary scores.
Associations between peripheral neuropathy and foot ulcers are uniform across the studies reported in Table 1.4.
Boyko and colleagues found an increased risk of ulcers in patients who were insensate to the 5.07 (10-g) monofilament, a semi-quantitative measure of light touch.30,50 Kastenbauer and associates found that elevated VPT ≥ 25 V prospectively predicted foot ulcers in a cohort of type 2 pa- tients followed up on average for more than 3 years.37 In a randomised clinical trial that used VPT ≥ 25 V as an entry criteria, Abbott and colleagues found that VPT deficits and a combined neuropathy deficit score (NDS) that include both reflexes and muscle strength measures were significant predictors of incident ulcers.25 Several years later, the same investigators, in a cohort study of 6613 diabetes patients from six UK health care districts, found similarly that NDS score, increasingly abnormal ankle reflexes and 10-g monofilament insensitivity all indepen- dently predicted new foot ulcers.27 Carrington and associates found that peroneal motor nerve conduction velocity was strongly associated with foot ulcer risk even after controlling for sen- sory neuropathy.39 Together these studies suggest that aberrations in the various sensory modal- ities and the presence of motor neuropathy independently predict increased foot ulcer risk.
Peripheral vascular function can be measured as absent pulses, transcutaneous oxygen ten- sion (TcpO2 ) decrements and low ankle, arm index (AAI).
These variables predicted foot ulcers in several of the studies in Table 1.4.
Low TcpO2 , indicating diminished skin oxygenation, and low AAI, indicating abnormal large vessel perfusion, were independent predictors of foot ulcers in the study by Boyko et al.30 In Boyko and colleagues’ study, laser Doppler flowmetry did not predict foot ulcer.
Kumar et al. defined peripheral vascular involvement as the absence of two or more foot pulses or a history of previous peripheral revascularisation.24 They re- ported that this variable was a significant predictor of foot ulcers.
Walters et al. found that an absent dorsalis pedis pulse was associated with a 6.3-fold increased risk of foot ulcer (95% CI 5.57, 7.0).23 Abbott and colleagues found that having two or less palpable pedal pulses on
JWBK089-01
Table 1.4 Risk factors for foot ulcers in patients with diabetes mellitus from the final analysis models of selected studies
Neuropathy (monofilament, Low AAI, reflex, vibration or TcpO2 JWBK089-Boulton
Study (type Study design, Long DM neurological or absent High History of analysis) diabetes type duration deficit score) pulses HbA1c Deformity Smoking of ulcer Amputation
Moss et al.16 Cohort, patients with Borderline + Borderline (Logistic early- and older younger regression) late-onset April 21, 2006