Foot ulcers are an important cause of morbidity and high medical costs in people with diabetes, and have been called one of the easiest long-term complications of diabetes to detect in a primary care setting (6). Among the 17 million people with diabetes mellitus in the United States, up to 15% will develop a foot ulcer in their lifetime, and approximately 15% of those individuals will require amputation (18). Attributable medical costs for a 45- to 65-yr-old male with a new foot ulcer are nearly $28,000 over 2 yr (22). Foot ulcers have been identified in the causal pathway in 84% of lower-limb amputations (19).
The role of daily weight-bearing physical activity in the development of foot ulcers among people with diabetes mellitus and insensate feet is still poorly understood. A recent National Health Interview Study estimated that 36–40% of people with diabetes have insensate feet due to diabetic peripheral neuropathy (9). People with diabetes and insensate feet experience minor trauma and increased plantar pressures, putting them at increased risk for subsequent foot ulcer (5,12,20,21,26). In a UK-based cohort study of people with diabetes, Veves and colleagues (26) found that plantar foot ulcers developed after an average of 30 months of follow-up in only 35% of those with initially elevated plantar foot pressures (93% of whom had insensate feet). These findings led to the suggestion that an increased pressure load on the insensate foot may increase foot-ulcer risk even more in patients with an active lifestyle (25). The American Diabetes Association (ADA) recommends that people with diabetes mellitus and insensate feet should limit weight-bearing activity to decrease the risk of foot ulcers (4). The American College of Sports Medicine (ACSM) recommends for people with type 2 diabetes that “nonweight bearing activities should be performed by persons with peripheral neuropathy in order to mitigate irritation and/or trauma to the lower legs and feet” (3). Longitudinal observational studies and controlled trials have not been reported that definitely link weight-bearing activity to increased risk of foot ulcer.
An active lifestyle has been shown to be of great benefit to persons with diabetes. Mortality rates are lower in fit persons with diabetes: after adjustment for age, measures of glycemic control, cholesterol level, body mass index (BMI), blood pressure, and family history of cardiovascular disease, all-cause (14,27) and cardiovascular mortality (24) is lower in active type 2 diabetes patients. Even when activity does not enhance aerobic fitness, it can be beneficial: Glass and colleagues (10) found in a 13-yr observational study that all-cause mortality decreased among active elderly adults with and without diabetes, even when their participants’ activity did not enhance fitness. ACSM recommends that all persons with type 2 diabetes should participate in at least five 30-min sessions per week of low- to moderate-intensity physical activity (i.e., achieving 40–70% of maximal oxygen consumption) (3).
If health providers are to counsel patients with diabetes and insensate feet regarding their daily activity, it is vital that providers better understand the relationship between weight-bearing activity and the risk of foot ulcer. To investigate this issue, we conducted a 2-yr cohort study of 400 people with diabetes mellitus and a prior history of foot ulcer (23). This study had two aims: to identify characteristics associated with participants’ usual daily weight-bearing activity, and to determine whether weight-bearing activity increased the risk of foot ulcer. The study hypothesized that an increase in weight-bearing activity would not increase the risk of foot ulcer, even in participants with diabetes and insensate feet.
The study cohort included 186 men from the Veterans Affairs Puget Sound Health Care System (VA) and 123 men and 91 women from Group Health Cooperative (GHC), a large health maintenance organization in western Washington State. All took part in a randomized clinical trial investigating the protective effect of therapeutic footwear and insoles on foot re-ulceration among participants with diabetes and a history of prior foot ulcer (23). Data were also collected to assess the relationship between physical activity and foot ulceration. The Human Subjects Committees of the University of Washington and Group Health Cooperative approved this trial. Individual written informed consent was obtained before enrollment.
Study eligibility criteria included: diagnosed diabetes, ages 45–84, men from either the VA or GHC and women from GHC (due to limited numbers of eligible female veterans), history of a full thickness foot lesion or a foot infection requiring antibiotic treatment, no foot deformities requiring a custom shoe, and reported ability to walk one block and climb one flight of stairs per day. Exclusion criteria were: lower-extremity amputation of more than one digit; a lesion either unhealed or healed for less than 1 month; requirement of boots, custom shoes, or nontraditional footwear for daily activities; a history of or active Charcot foot deformity; or a terminal illness that would make 2-yr survival unlikely. For this analysis, we excluded one participant who became completely nonweight-bearing after enrollment and eight who missed multiple follow-up visits and had insufficient activity data, leaving 391 participants for analysis.
Data regarding diabetes, health, and functional status used in this study were collected at enrollment. Feet were classified as “insensate” when the 5.07 (10 g) Semmes-Weinstein monofilament response was absent at any point on either foot. We also noted the presence of dorsalis pedis and posterior tibial artery foot pulses. The presence of insensate feet and foot pulses was reassessed 1 yr into follow-up. Demographic and health status measures included age, gender, marital status, completion of high school, ethnicity, BMI, current history of smoking, history of any of six reported comorbid conditions (heart failure, respiratory disease, stroke, cancer, depression, or cardiovascular surgery), time since the diagnosis of diabetes, compliance with blood glucose self-monitoring, and standardized quality-of-life indicators shown to be responsive to the development of diabetic complications over time (the Medical Outcomes Study survey, or SF-36) (1). Participants were interviewed every 17 wk using a 24-h activity questionnaire to elicit information about the total minutes during the previous day spent sleeping, reclining, sitting, standing, walking, or performing structured exercise. When administered by trained interviewers in a previous study to subjects with diabetes and either insensate feet or prior amputation, this questionnaire showed strong criterion validity compared with a continuously worn step-activity monitor (intraclass correlation r = 0.47) and also showed substantial test-retest reliability for interviewer-collected diaries obtained 1 month apart (intraclass correlation coefficient = 0.52) (8). To our knowledge, this is the only physical activity questionnaire that has been validated for use in this population.
The study outcome was foot re-ulceration, defined as a break in the cutaneous barrier extending into or through the dermis to deeper tissue that did not heal within 30 d. For this analysis, six ulcers that were unrelated to daily activity were excluded: four decubitus ulcers that developed during short-term hospitalization, one foot ulcer that resulted from trauma when a medical worker stepped on a participant’s foot, and one foot ulcer that was due to acute vascular insufficiency. Foot ulcers that resulted from participants’ own minor traumatic events were included. A panel of three foot-care specialists blinded to the clinical trial treatment arm determined final ulcer classification. The ulcer onset was the date when the participant first noticed the foot ulcer. Minor lesions that did not become ulcers (i.e., abrasions and blisters or any erosion that healed within 30 d) were also recorded. Ascertainment of foot ulcers occurred at any point during the 17 wk after a follow-up visit (ascertainment intervals). Participants were followed for six consecutive ascertainment intervals.
Several definitions of activity duration were used in this study:
An active hour was 60 min that participants reported they accumulated in any weight-bearing activity (standing, walking or more active). Activity was reported in 15-min intervals.
Current activity was the reported number of the 24 h before a given follow-up visit that were active hours.
Long-term activity was the cumulative average number of active hours measured from enrollment through a given follow-up visit.
Short-term activity change for a participant was obtained by subtracting the number of active hours per day reported at a given follow-up visit from the number of active hours reported at the prior visit.
Activity duration was either analyzed continuously using the definitions above, or analyzed using activity levels. These levels were defined using tertiles of active hours reported by all participants at enrollment. “Least active” participants had fewer than 4.5 active hours per day, those “moderately active” had 4.6–7.4 active hours per day (middle tertile), and those “most active” had 7.5 or more active hours per day (upper tertile). Participants’ activity levels were updated at each follow-up visit.
Average activity intensity was the average metabolic equivalent task (MET) intensity above and beyond resting and was reported for weight-bearing activities at a given follow-up visit. Resting intensity was defined as 1.0 MET. All reported activities (including resting) were converted into METs using a standard activity task compendium (2). Average activity intensity was calculated using the following formula:MATH
STATA version 7.0 software was used (Statacorp, College Station, Texas). Statistical tests were two-sided with P < 0.05 considered statistically significant. We conducted a descriptive analysis to assess potentially confounding factors. Baseline health and demographic characteristics of participants were compared across the ordered activity levels (tertiles). We performed this analysis by calculating means and proportions and testing for differences using one-way analysis of variance and chi-square tests for trend, respectively. Differences in time spent at baseline in activities of different intensity were analyzed using nonparametric rank-sum tests, comparing moderately active and most active subjects with those least active.
Cross-sectional time series analysis with generalized estimating equations was used to describe increasing or decreasing trends in weight-bearing activity duration over the course of the study. It was also used to test whether foot sensitivity predicted any significant change in activity throughout the study after adjustment for potentially confounding factors.
Multiple imputation was used to impute weight-bearing activity for follow-up visits that study participants missed (8.3% of all postenrollment visits) (15). The study data were stratified by participant characteristics associated with missed visits. These included marital status, ethnicity, footwear clinical trial treatment arm (intervention versus usual footwear), and follow-up visit. Missing observations were replaced randomly from a bootstrap sample of the complete observations drawn from each stratum. This produced multiple imputed data sets with identical numbers of observations.
Survival analysis was conducted using pooled logistic regression to adjust for temporal changes in weight-bearing activity and potentially confounding participant characteristics. This method produced odds ratios that approximated Cox regression hazard ratios in their interpretation (7). Regression models progressively controlled for: time in the study; foot-related characteristics, i.e., presence of insensate feet, presence of foot deformity, footwear clinical trial treatment arm, and presence of foot pulses; and measures of health status, i.e., the presence of comorbid conditions, age ≥ 65 yr, gender, marital status, education, ethnicity, duration of diabetes, frequency of self-monitoring of blood glucose, BMI, current history of smoking, and physical and mental health status (SF-36 composite standard physical and mental scores). Finally, pooled logistic regression controlling for all foot-related and health status characteristics was repeated using the imputed data sets. After averaging odds ratios across the imputed data sets, we calculated 95% confidence intervals using a weighted average of the variance within each imputed data set plus the variance across the data sets (13). Confidence intervals used robust standard errors to account for within-person correlation of observations.
Models that included continuous activity (e.g., current activity, long-term activity and short-term activity change) estimated linear trends in the relative risk of foot ulcer associated with a 1-h increase in weight-bearing activity. Analysis of the effect of activity levels on foot ulcer risk was performed to clarify the effect of a substantial increase in activity. Because the first ascertainment interval was only 4 wk long, a specific term was included to separately estimate foot ulcer risk during this period.
We assessed the relationship between minor foot lesions and subsequently measured activity in order to avoid “reverse-causation” bias. We used nonparametric rank-sum tests to evaluate differences in activity duration reported at the visit just before detection of a foot ulcer. We compared participants who developed a minor foot lesion within 1 month before that visit to participants having no minor lesion preceding a foot ulcer.
This study had 90% power to detect a ≥ 33% decrease or a ≥ 57% increase in the risk of foot ulcer, comparing moderately active to least active participants and 90% power to detect a ≥ 39% decrease or a ≥ 80% increase in the risk of foot ulcer, comparing most active with least active participants.
Table 1 describes the relationship between reported baseline weight-bearing activity duration and participants’ demographic and health characteristics and quality of life. At baseline, women reported significantly more activity. Activity was lower among male and female participants who were older or those who had not completed high school. Those with a history of stroke, congestive heart failure, respiratory disease, depression, or at least one comorbid condition reported significantly less activity at baseline. Least active participants had significantly lower physical, mental, and general health SF-36 scores. One participant did not provide activity data at baseline.
Table 2 shows the average time spent by subjects in activities of different intensities, according to subjects’ activity level (reflecting the total number of active hours per day reported at baseline). Moderately active subjects spent less time lying down than those least active, and more time standing or walking slowly. Similarly, subjects who were most active spent less time than least active subjects lying or sitting, and more time standing or walking slowly. There were no significant differences across activity levels in time spent walking briskly. No study subject reported participating in running or jogging activities at baseline. A few participants (N = 11) reported that at baseline they took part in structured exercises other than walking, including swimming, weight-lifting and stretching. Subjects who reported these activities participated an average of 57 min·d−1 (SD 31 min·d−1). Participation in structured exercise did not differ significantly across activity levels.
Figure 1 shows the temporal trend in weight-bearing activity duration for all reported activity during the study period and its association with foot sensation as estimated by the generalized estimating equation model. After controlling for the presence of insensate feet, footwear study treatment arm, the presence of foot deformity, current smoking status, gender, age, ethnicity, marital status, education, BMI, and a history of comorbid conditions, activity duration increased slightly for the first 5 months of follow-up and then decreased. By the end of 2-yr follow-up, a 5% significant decrease in activity duration was observed compared with activity at enrollment (95% CI 0.1–10% decreased). Those with insensate feet were slightly less active than those with intact sensation, but this remained nonsignificant throughout the study.
Overall, 62 participants from the study cohort developed a foot ulcer during 2 yr of follow-up. Of the 56 incident foot ulcers that may have been activity-related, 22 occurred in participants who were least active at baseline (cumulative incidence = 16.5%, 95% CI 10.4–25%), 18 occurred among those moderately active at baseline (cum. incidence = 13.4%, 95% CI 7.9–21.2%), and 16 among those most active at baseline (cum. incidence = 13%, 95% CI 7.4–21.1%) (data not shown in s). There was no significant difference in the cumulative incidence of foot ulcer, comparing either moderately or most active participants to those who were least active at baseline.
Table 3 reports linear trends in the relative risk of foot ulcer associated with a 1-h increase in weight-bearing activity, updated at each follow-up visit. There was a significant decrease in the risk of foot ulcer with every additional hour of current activity or long-term activity updated at each visit, after adjustment for foot-related characteristics. An increase of 1 h of activity since the prior follow-up visit (short-term activity change) was associated with a borderline significant increase in foot ulcer risk. However, after control for health status characteristics, neither current activity nor short-term activity change remained associated with significant changes in the risk of foot ulcer. In contrast, after additional adjustment for health status characteristics, an increase of 1 h·d−1 in long-term activity updated at each visit remained associated with a significant 23% decrease in risk (95% CI 4–39%). After subsequent analysis of data sets with imputed observations, this finding remained significant.
Table 4 shows the effect of long-term activity levels on the risk of foot ulcer. Differences in the actual number of observations shown for each activity level reflect censoring after foot ulceration and changes in activity throughout the study. Persons who were “most active” initially became “moderately” active later on. After adjustment for all foot-related and health status characteristics, participants who were moderately active in the long term were at 50% less risk of foot ulcer compared with those least active (95% CI 78% less to 16% more risk), whereas the most active participants were at 80% less risk than the least active participants (95% CI 13–96%). Although this effect was smaller in the subsequent analysis of imputed data sets, it remained strong and significant.
An increase in average activity intensity was not associated with a significant change in the risk of foot ulcer. After controlling for foot-related and health status characteristics, with each additional 0.1 MET increase in average activity intensity over the resting level there was a nonsignificant 15% increase in subsequent foot ulcer risk (95% CI 13% decrease to 52% increase).
No potentially confounding foot-related or health-status characteristic independently increased the risk of foot ulcer except the presence of insensate feet. This increased the risk of foot ulcer strongly and significantly (OR 4.51, 95% CI 1.9–10.5), after controlling for activity level, average activity intensity, and all other foot-related and health-status characteristics.
The observed activity-related reduction in the risk of foot ulcer was marginally less for those with insensate feet. After adjusting for foot-related, demographic, and health status characteristics, foot ulcer risk for those with insensate feet decreased by 22% (95% CI 1–39%) for each additional hour of long-term activity reported. Those with sensate feet had a 33% decrease in risk (95% CI 10–49% decrease) for each additional hour of activity. However, the difference between these two estimates, using a test of interaction, was not statistically significant (P = 0.24) (data not shown in s). We therefore concluded that the effect of activity on foot ulcer risk was not dependent on the presence of sensate or insensate feet.
Eight (14%) of the 56 participants with foot ulcers developed minor foot lesions 1 month or less before the follow-up visit before developing a foot ulcer. At that follow-up visit, these participants reported a median of 5.75 active hours per day compared with 4.25 active hours per day reported by those with no minor lesion before their foot ulcer (Wilcoxon rank sum P = 0.50). Thus, we found no evidence that minor foot lesions depressed activity preceding a foot ulcer.
This prospective, longitudinal study documented a decreasing risk of foot ulcer with increasing duration of long-term weight-bearing activity. Contrary to the common assumption that prolonged weight-bearing activity leads to a greater risk of foot ulcer, we found that an increase in weight-bearing activity duration not only did not raise but rather appears to have reduced foot ulcer risk. Increasing activity intensity was likewise not associated with a significant increase in foot ulcer risk. Further, although foot insensitivity did increase the risk of foot ulcer, the presence of insensate feet did not modify the effect of activity.
It must be noted that weight-bearing activity reported in this study was accumulated over the course of an entire day, and that long-term increases in the duration of weight-bearing activity represented a change over a number of months. The slight increase in foot ulcer risk that we found after a short-term increase in activity did not retain significance after control for participants’ health status characteristics. This finding was in strong contrast to a dramatic decrease in foot ulcer risk associated with long-term increases in the duration of daily activity. Both these findings support Mueller and Maluf’s (17) “physical stress theory,” which states that gradually increasing physical stimulation leads to protective tissue hypertrophy and prevents plantar skin breakdown (11), whereas overstimulation without rest leads to tissue breakdown and ulceration (16).
Although, as in prior prospective studies (5,12,20,21,26), the presence of insensate feet was a strong risk factor for foot ulcer, in this study the absence of foot sensation did not significantly depress activity. We further found that activity duration initially increased and then decreased slightly over time, possibly after participants became accustomed to the study and resumed their typical daily routines.
Participants in this study only rarely engaged in structured exercise other than walking. These findings are consistent with the 1990 National Health Interview Survey, which showed that people with diabetes preferred walking to more strenuous activities such as jogging, calisthenics, or weight-lifting (9). Because participation in structured exercise other than walking was so rare in this cohort, we did not specifically analyze the effects of structured exercise programs on foot ulcer risk.
We imputed activity data for missed follow-up visits to correct for bias due to small gaps in participant follow-up, and to provide consistent results and study power across regression models. Our regression method using imputed data accommodated the increased uncertainty in our risk estimates after imputation. This approach produces estimates that are less biased than regression after deletion of all cases with missing data (13).
There were several potential study limitations. First, we assumed that activity reported to interviewers was representative of participants’ typical daily activity, and that participants’ activity remained relatively stable between follow-up visits. Although Del Aguila’s validation of the 24-h diabetes physical activity questionnaire used in this study found substantial test-retest reliability over 1 month (intraclass correlation coefficient = 0.52 for interviewer-collected diaries) (8), further studies validating the stability of this activity questionnaire over longer periods would lend more credence to these results. Second, these findings may not apply to persons with severe foot deformities requiring custom shoes or those with Charcot feet, who were excluded from the study. Finally, although we controlled for a number of health status indicators, some residual confounding may have persisted due to incomplete control for unmeasured poor health status, self-care, or glycemic control. Elevated glycosylated hemoglobin may have affected the progression of minor foot lesions to full-thickness foot ulcers; however, it was not collected during the clinical trial. The self-reported frequency of glucose self-monitoring, reflecting treatment compliance, was collected and estimates of foot ulcer risk were adjusted for these effects.
The results of this prospective cohort study nonetheless call into question prior recommendations that those with diabetes and insensate feet should participate in predominantly nonweight-bearing physical activities. In this study, increased duration of daily weight-bearing activity did not increase the risk of foot ulcer, even for those with insensate feet. Clinical trials are now indicated to confirm these observational findings and further explore the type of regular activities that persons with diabetes and insensate feet may safely undertake. If subsequent studies continue to be supportive, agencies that advise people with diabetes and insensate feet will be free to encourage participation in physical activity without unnecessary weight-bearing restrictions.
The authors wish to acknowledge the efforts of the following: Katrina Sullivan, D.P.M., Christy Vath, B.A., Matthew Maciejewski, Ph.D., and Shane Hayes, C.Ped., assisted in data collection, gave helpful comments during the analysis and gave oral critiques to early drafts of the manuscript. Jeff Rodenbaugh was the study programmer.
This research was supported by Rehabilitation Research and Development, the Epidemiology Research and Information Center, Department of Veterans’ Affairs, the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Health, and the Centers for Disease Control and Prevention. The views expressed in this article are those of the authors and do not necessarily represent the views of the agencies providing support or the American College of Sports Medicine.
1. Ahroni, J. H., and E. J. Boyko. Responsiveness of the SF-36 among veterans with diabetes mellitus. J. Diabetes Complications 14: 31–39, 2000.
2. Ainsworth, B. E., W. L. Haskell, M. C. Whitt, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med. Sci. Sports Exerc. 32: 498–504, 2000.
3. Albright, A., M. Franz, G. Hornsby, et al. American College of Sports Medicine position stand: exercise and type 2 diabetes. Med. Sci. Sports Exerc. 32: 1345–1360, 2000.
4. American Diabetes Association. Diabetes mellitus and exercise. Diabetes Care 22: S49–S53, 1999.
5. Armstrong, D. G., E. J. Peters, K. A. Athanasiou, and L. A. Lavery. Is there a critical level of plantar foot pressure to identify patients at risk for neuropathic foot ulceration
? J. Foot Ankle Surg. 37: 303–307, 1998.
6. Boulton, A. J., and L. Vileikyte. The diabetic foot: the scope of the problem. J. Fam. Pract. 49: 3–8, 2000.
7. D’Agostino, R. B., M. L. L ee, A. J. B elanger, L. A. C upples, K. A nderson, and W. B. K annel. Relation of pooled logistic regression to time dependent Cox regression analysis: the Framingham Heart Study. Stat. Med. 9: 1501–1515, 1990.
8. Del Aguila, M. A. Assessment of physical activity in patients with diabetes (Lower extremities, dpaq, test reliablity). Dissertation Abstr. Int.
5–06, Section: B, 2708.
9. Ford, E. S., and W. H. Herman. Leisure-time physical activity patterns in the U.S. diabetic population: findings from the 1990 National Health Interview Survey: Health Promotion and Disease Prevention Supplement. Diabetes Care 18: 27–33, 1995.
10. Glass, T. A., C. M. de Leon, R. A. Marottoli, and L. F. Berkman. Population based study of social and productive activities as predictors of survival among elderly Americans. Br. Med. J. 319: 478–483, 1999.
11. Jagoda, A., H. Madden, and C. Hinson. A friction blister prevention study in a population of marines. Mil. Med. 146: 42–44, 1981.
12. Kastenbauer, T., S. Sauseng, G. Sokol, M. Auinger, and K. Irsigler. A prospective study of predictors for foot ulceration
in type 2 diabetes. J. Am. Podiatr. Med. Assoc. 91: 343–350, 2001.
13. King, G., J. Honaker, A. Joseph, and K. Scheve. Analyzing incomplete political science data: an alternative algorithm for multiple imputation. Am. Pol. Sci. Rev. 95: 49–69, 2001.
14. Kohl, H. W., N. F. Gordon, J. A. Villegas, and S. N. Blair. Cardiorespiratory fitness, glycemic status, and mortality risk in men. Diabetes Care 15: 184–192, 1992.
15. Mander, A., and D. Clayton. Hotdeck imputation. Stata Tech. Bull. 51: 32–54, 1999.
16. Manley, M. T., and T. Darby. Repetitive mechanical stress and denervation in plantar ulcer pathogenesis in rats. Arch. Phys. Med. Rehabil. 61: 171–177, 1980.
17. Mueller, M. J., and K. S. Maluf. Tissue adaptation to physical stress: a proposed “Physical Stress Theory” to guide physical therapist practice, education, and research. [Review] [151 refs]. Phys. Ther. 82: 383–403, 2002.
18. Palumbo, P. J., and L. Melton. Peripheral vascular disease and diabetes. In:Diabetes in America.
U.S. Government Printing Office, 1985, pp. 401–407.
19. Pecoraro, R. E., G. E. Reiber, and E. M. Burgess. Pathways to diabetic limb amputation: basis for prevention. Diabetes Care 13: 513–521, 1990.
20. Pham, H., D. G. Armstrong, C. Harvey, L. B. Harkless, J. M. Giurini, and A. Veves. Screening techniques to identify people at high risk for diabetic foot ulceration
: a prospective multicenter trial. Diabetes Care 23: 606–611, 2000.
21. Prabhu, K. G., K. M. Patil, and S. Srinivasan. Diabetic feet at risk: a new method of analysis of walking foot pressure images at different levels of neuropathy for early detection of plantar ulcers. Med. Biol Eng. Comput. 39: 288–293, 2001.
22. Ramsey, S. D., K. Newton, D. Blough, et al. Incidence, outcomes, and cost of foot ulcers in patients with diabetes. Diabetes Care 22: 382–387, 1999.
23. Reiber, G. E., D. G. Smith, C. Wallace, et al. Effect of therapeutic footwear on foot reulceration in patients with diabetes: a randomized controlled trial. JAMA 287: 2552–2558, 2002.
24. Rosengren, A., and L. Wilhelmsen. Physical activity protects against coronary death and deaths from all causes in middle-aged men: evidence from a 20-year follow-up of the primary prevention study in Goteborg. Ann. Epidemiol. 7: 69–75, 1997.
25. Uccioli, L. Footwear in the prevention of diabetic foot problems. In: Diabetic Foot: Medical and Surgical Management, A. Veves, J. M. Giurini, and F. W. LoGerfo (Eds.). Totowa, NJ: Humana Press, 2002, pp. 445–461.
26. Veves, A., H. J. Murray, M. J. Young, and A. J. Boulton. The risk of foot ulceration
in diabetic patients with high foot pressure: a prospective study. Diabetologia 35: 660–663, 1992.
27. Wei, M., L. W. Gibbons, J. B. Kampert, M. Z. Nichaman, and S. N. Blair. Low cardiorespiratory fitness and physical inactivity as predictors of mortality in men with type 2 diabetes. Ann. Intern. Med. 132: 605–611, 2000.