- Email: [email protected]
Skin temperature in the neuropathic diabetic foot
Journal of Diabetes and Its Complications 15 (2001) 260 – 264
Skin temperature in the neuropathic diabetic foot Edward J. Boykoa,b,*, Jessie H. Ahronic,d, Victoria L. Stenselc a
General Internal Medicine Section and Epidemiologic Research and Information Center (ERIC), Veterans Affairs Puget Sound Health Care System, 1660 South Columbian Way, Seattle, WA 98108, USA b Department of Medicine, University of Washington, Seattle, WA, USA c Research and Development Service, Veterans Affairs Puget Sound Health Care System, Seattle, WA, USA d Clinical Faculty, Department of Biobehavioral Nursing and Health Systems, School of Nursing, University of Washington, Seattle, WA, USA Received 25 January 2001; accepted 11 April 2001
Abstract Several authors have reported higher skin temperature in the feet of diabetic subjects with autonomic neuropathy. We reexamined this association in a cross-sectional study of 712 veterans with diabetes mellitus. Potential subjects included all diabetic patients enrolled in a general internal medicine clinic at a veterans affairs healthcare system. Sensory neuropathy was defined as any pedal insensitivity to the 5.07 monofilament. Autonomic neuropathy was determined using standard cardiovascular reflex tests. An infrared surface scanner was used to measure foot skin temperature at multiple sites. Subjects with sensory neuropathy had lower mean plantar foot skin temperature than those without (28.4C vs. 28.9C, P = .0101). Autonomic neuropathy as a dichotomous variable was unrelated to foot skin temperature. Foot skin temperature, though, negatively correlated with greater drop in systolic blood pressure with standing, which is an indicator of autonomic neuropathy (r = .08, P = .0385). Adjustment for potential confounding factors using multiple linear regression analysis resulted in diminution of the associations between foot skin temperature and sensory neuropathy or orthostatic blood pressure drop, but the latter association remained statistically significant in the right foot. Diabetic veterans with sensory or autonomic neuropathy do not have higher foot skin temperature. Our results suggest that skin temperature may be slightly lower with higher orthostatic blood pressure fall. Other causes exist for the frequently observed differences in skin temperature in the feet of diabetic subjects. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Diabetic foot; Peripheral neuropathy; Skin temperature; Autonomic neuropathy
1. Introduction Previous studies, as summarized by Flynn and Tooke (1995), investigated the effects of diabetic neuropathy on the microcirculation as an increase in blood flow through arteriovenous shunts, which dissipates heat produced as a result of an increased metabolic rate (Leslie et al., 1986). This increase in skin blood flow is thought to be due to dilation of arteriovenous shunts, which are normally under the control of sympathetic nerves. One expected consequence of this increase in skin blood flow is an increase in skin temperature, which has been reported by several authors. Archer, Roberts, and Watkins (1984) reported a * Corresponding author. Veterans Affairs Puget Sound Health Care System (S-111-GIMC), 1660 South Columbian Way, Seattle, WA 98108, USA. Tel.: +1-206-764-2830; fax: +1-206-764-2849. E-mail address: [email protected] (E.J. Boyko).
mean skin temperature on the plantar foot between 33.2C and 33.5C among diabetic subjects with either painful or sensory neuropathy compared to a mean of 27.8C in diabetic subjects without neuropathy. Similarly, Rayman, Hassan, and Tooke (1986) found a higher mean skin temperature on the plantar aspect of the left great toe among diabetic subjects with (32.2C) as compared to those without (28.7C) lower extremity sensory neuropathy. Not all authors, though, have reported statistically significant differences in skin temperature by neuropathy status. Flynn, Edmons, Tooke, and Watkins (1988) found that the median great toe nailfold temperature in 14 diabetic subjects with sensory or autonomic neuropathy (32.6C) differed by only 1C from diabetic subjects without neuropathy (31.5C). We presume that none of these three studies employed population-based sampling, since this was not mentioned in their methods. So, it is unclear how these results might apply to all diabetic subjects in a defined population.
1056-8727/01/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved. PII: S 1 0 5 6 - 8 7 2 7 ( 0 1 ) 0 0 1 5 6 - 8
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
Although the association between higher skin blood flow and skin temperature makes intuitive sense, researchers have reported that this association is not linear. Felder et al. (1954) found that relatively small increases in blood flow resulted in large increases in skin temperature at low skin temperatures, but this association diminished in magnitude above a 28C threshold. In subjects with warm limbs ( > 28C), one would therefore expect a smaller change in temperature with increasing blood flow. We incorporated skin temperature measurements as part of a large, prospective study of risk factors for diabetic foot ulceration. We report our findings from the baseline examination on the associations between measures of sensory and autonomic neuropathy in the lower extremities and directly measured foot skin temperature.
2. Methods Measurements were obtained from subjects enrolled in a prospective study of risk factors for diabetic foot ulceration at the General Internal Medicine Clinic of the Veterans Affairs Puget Sound Health Care System. Exclusion criteria included being nonambulatory, too ill to participate, or having a cognitive impairment that prevented informed consent. Diabetic subjects were identified by review of computerized pharmacy data for receipt of insulin, oral hypoglycemic medication, blood or urine glucose test strips, and review of medical record problem lists for the diagnosis of diabetes mellitus. Informed consent was obtained from all subjects. The University of Washington Human Subjects Committee approved the protocol used in this study. The initial interview employed a standard questionnaire to collect data on demographics, diabetes type, duration, and treatment, and past history of foot ulcer, amputation, and alcohol use. All subjects were asked the four standard CAGE (Ewing, 1984) questions and two or more positive responses were considered consistent with excessive alcohol use. A physical exam with emphasis on the lower extremities was performed by a research nurse practitioner. A random blood sample was drawn for measurement of glycosylated hemoglobin. Sensory neuropathy was defined as an inability to feel the 5.07 Semmes– Weinstein monofilament at any one of nine sites on each foot (Holewski, Stess, Graf, & Grunfeld, 1988). The 5.07 Semmes – Weinstein monofilament has a specificity of approximately 58% and sensitivity near 100% when the lack of perception at one or more sites is used as the cutoff for neuropathy (Armstrong, Lavery, Vela, Quebedeaux, & Fleischli, 1998). Assessment of autonomic neuropathy was done using two standard measures: R – R interval variation with six deep breaths and orthostatic systolic blood pressure decrease with standing (Ewing & Clarke, 1986; O’Brien, O’Hare, & Corrall, 1986). Autonomic neuropathy was
261
considered to be present if either of the following criteria were met: R – R variation 10 or orthostatic blood pressure decrease 30 mmHg. Measurements of lower extremity temperature were performed on the seated patient in a constant room temperature environment on the dorsal and plantar surfaces of both feet with an infrared surface temperature scanner (Omega Medical, Clearwater, FL). The manufacturer states that the instrument is accurate to ± 0.1C, which we confirmed in 200 subjects with repeated readings at the same site that were nearly identical. These skin temperatures were obtained immediately after the patient had walked 100 ft in a corridor adjacent to the neurovascular testing room. We attempted to maintain a constant room temperature during the testing. The forefoot summary measure is the mean of temperatures taken at five sites on the plantar surface of the foot: the great toe, the fifth toe, the first metatarsal head, the fourth metatarsal head, and the fifth metatarsal head. The plantar summary measure is the mean of these sites and the plantar heel; the all loci measure is the mean of these sites and a measurement on the dorsal forefoot. Brachial and lower extremity arterial blood pressures in both limbs were measured by standard segmental Doppler techniques (Cutajar, Marston, & Newcombe, 1973). The ankle –arm index (AAI) was calculated as the ratio of the highest ankle or dorsal foot pressure in each limb divided by the higher brachial pressure in either arm. Only measurements taken at entry into the study were considered in this analysis. Mean values were compared using the unpaired t test (Armitage & Berry, 1987). Nonparametric tests were performed (Mann –Whitney U) on data that were not normally distributed. Differences in proportions were analyzed using the c2 test. Nonparametric correlations were performed using the Spearman rank – order correlation coefficient (Armitage & Berry, 1987). Multiple linear regression analysis was performed using standard methods (Armitage & Berry, 1987).
3. Results Of 921 persons eligible for the study, 778 (85%) agreed to participate and were examined in the diabetic foot research clinic. Foot temperature measurements were successfully completed on 712 (92%) subjects. Characteristics of these subjects are displayed in Table 1 by neuropathy status. These diabetic subjects averaged 63 years of age and were 98% male, 79% white, and 92% had Type 2 diabetes mellitus. The average duration of diabetes was 11.4 years, the mean glycosylated hemoglobin was 11.3% (reference range 4.0 –6.8%), and the mean AAI of both limbs was 0.97. Insulin was the most frequently used treatment (47.3%). Fifty percent of diabetic subjects had sensory neuropathy and 65% had autonomic neuropathy. One-third of the subjects (33.4%) had a history of leg or foot ulceration, and 6.6% had a
262
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
Table 1 Baseline mean values for characteristics of study subjects by sensory and autonomic neuropathy status Sensory neuropathy (n = 712)
Autonomic neuropathy (n = 683)
Characteristic
Present (n = 367)
Absent (n = 345)
P-value
Present (n = 445)
Absent (n = 238)
P-value
Age (years, mean ± S.D.) Male [no. (%)] DM duration (years, mean ± S.D.) DM type Type 1 [no. (%)] Type 2 [no. (%)] Secondary DM [no. (%)] DM treatment Diet [no. (%)] OHA [no. (%)] Insulin [no. (%)] Glycosylated hemoglobin (%, mean ± S.D.) History of ulceration [no. (%)] History of LEA [no. (%)] Alcohol use Less than daily [no. (%)] Daily [no. (%)] Former [no. (%)] CAGE questionsa Treatment for alcohol use Right AAI (mean ± S.D.) Left AAI (mean ± S.D.)
64.7 ± 8.9 388 (51.0) 13.1 ± 9.9
61.7 ± 10.9 373 (49.0) 9.8 ± 9.0
< .0003 .0005 < .0001
63.7 ± 9.4 437 (65.1) 12.1 ± 9.7
62.6 ± 10.5 234 (34.9) 10.7 ± 10.0
.7064 >.9999 .0233
22 (5.6) 362 (93.1) 5 (1.3)
29 (7.5) 351 (90.7) 7 (1.8)
26 (5.8) 410 (92.1) 9 (2.0)
19 (8.0) 216 (90.8) 3 (1.3)
.4445
.4822
36 (9.3) 143 (36.8) 210 (54.0) 11.6 ± 3.4 162 (41.7) 38 (9.8)
53 (13.7) 176 (45.5) 158 (40.8) 11.0 ± 3.2 98 (25.3) 13 (3.4)
52 (11.7) 170 (38.2) 223 (50.1) 11.3 ± 3.2 165 (37.1) 36 (8.1)
31 (13.0) 111 (46.6) 96 (40.3) 11.3 ± 3.6 70 (29.4) 10 (4.2)
.0472
243 (63.4) 25 (6.6) 112 (29.5) 73 (19.3) 67 (17.7) 0.94 ± 0.28 0.94 ± 0.29
260 (70.1) 21 (5.6) 90 (24.0) 66 (17.6) 72 (19.3) 0.99 ± 0.23 1.01 ± 0.24
299 (67.7) 26 (5.9) 117 (26.5) 90 (20.4) 74 (16.8) 0.96 ± 0.26 0.97 ± 0.27
161 (67.9) 11 (4.6) 65 (27.4) 35 (14.8) 50 (21.1) 0.98 ± 0.25 0.98 ± 0.26
.0009
.0114 < .0001 .0003 .2008
.5564 .5780 .0251 .0033
.4697 .0445 .0534 .7825
.0709 .1698 .3387 .5990
Sensory neuropathy refers to presence of this condition on either foot. Abbreviations: DM = diabetes mellitus, OHA = oral hypoglycemic agents, LEA = lower extremity amputation, AAI = ankle arm index. a Two or more positive responses to the four standard CAGE questions about alcohol use.
prior lower extremity amputation. Increasing duration of diabetes and insulin use was significantly associated with both sensory and autonomic neuropathy. Diabetes type was not related to the presence of either type of neuropathy. Blood glucose control as reflected by a single glycosylated hemoglobin measure at entry to the study was significantly better in those without sensory neuropathy. Those with sensory neuropathy were significantly more likely to have had a previous ulcer or amputation. Previous ulceration and amputation were also related to the presence of autonomic neuropathy but not as strongly as sensory neuropathy. More frequent intake of alcoholic beverages, or two or more positive responses to CAGE questions, was not significantly associated with either
type of neuropathy. AAIs were significantly associated with sensory neuropathy bilaterally (right P-value = .0251, left P-value = .0033) but not with autonomic neuropathy. Skin temperatures ranged from 20.2C to 35.5C. Mean temperatures at all, plantar, and forefoot sites were similar on each foot (Table 2). As expected, the means of the dorsal sites were higher (31.0C right and 30.7C left) than the means of the plantar sites (28.6C right and 28.5C left). The plantar fifth toe had the lowest mean surface temperatures bilaterally (27.8C). The mean ± S.D. room temperature before testing was 24.2C ± 1.0C, and after testing, it was 24.5C ± 1.0C. Table 2 presents the mean skin temperatures by the presence of sensory and autonomic neuropathy. The presence
Table 2 Mean ± S.D. skin temperature (C) by presence of sensory and autonomic neuropathy Sensory neuropathy
Autonomic neuropathy
Characteristic
All subjects (n = 712)
Present (n = 366)
Absent (n = 344)
P-value
Present (n = 442)
Absent (n = 190)
P-value
All loci Plantar loci Forefoot loci
29.0 ± 2.3 28.6 ± 2.4 28.6 ± 2.5
28.7 ± 2.3 28.4 ± 2.4 28.3 ± 2.5
29.2 ± 2.2 28.9 ± 2.4 28.8 ± 2.5
.0052 .0101 .0051
29.0 ± 2.3 28.7 ± 2.4 28.6 ± 2.5
28.8 ± 2.2 28.5 ± 2.3 28.4 ± 2.4
.2525 .2340 .2425
Left foot
All loci Plantar loci Forefoot loci
All subjects (n = 712)
Present (n = 360)
Absent (n = 341)
P-value
Present (n = 435)
Absent (n = 191)
P-value
28.8 ± 2.4 28.5 ± 2.5 28.4 ± 2.7
28.6 ± 2.4 28.3 ± 2.5 28.1 ± 2.6
29.1 ± 2.4 28.7 ± 2.6 28.7 ± 2.7
.0085 .0198 .0071
28.8 ± 2.4 28.5 ± 2.5 28.4 ± 2.7
28.7 ± 2.3 28.3 ± 2.5 28.3 ± 2.6
.5785 .5635 .5705
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
of sensory neuropathy was associated with significantly lower skin temperatures in all regions, while subjects with and without autonomic neuropathy had similar skin temperatures. Table 3 examines whether skin temperature was correlated with the two tests of autonomic function expressed as continuous variables. No correlation was seen between all, plantar, or forefoot skin temperatures and mean heart rate variation with respirations on either foot. Weak negative correlations were seen between a fall in systolic blood pressure with standing and skin temperature, such that lower skin temperature was associated with a greater fall in systolic blood pressure. This correlation was statistically significant when comparing the mean right forefoot temperature to the fall in systolic blood pressure ( P = .0385), although the magnitude of the correlation was very small (r = .08). In Table 4, multiple linear regression analysis was used to determine whether differences in skin temperature by neuropathy status might be due to confounding by other variables, which differed by neuropathy status, as seen in Table 1. These analyses used the mean plantar skin temperature as the dependent variables, but similar results were obtained when either the mean all sites or mean forefoot temperatures was used. Adjustment for age, gender, diabetes duration, diabetes treatment (insulin vs. other), glycosylated hemoglobin, and ipsilateral AAI reduced the magnitude of the association between sensory neuropathy and plantar skin temperature on either foot. Skin temperature no longer remained statistically significantly associated with sensory neuropathy, although P-values for this association were near the .05 threshold value for rejection of the null hypothesis. Systolic blood pressure drop with standing and skin temperature in the right foot only remained related to a statistically significant degree, even after adjustment for age, diabetes duration, glycosylated hemoglobin, and diabetes treatment. Blood
Table 3 Spearman correlations between infrared foot skin surface temperature and continuous measures of autonomic function
Right foot Mean C all loci Mean C plantar loci Mean C forefoot loci Left foot Mean C all loci Mean C plantar loci Mean C forefoot loci
Heart rate variation with respirations (n = 530)
Systolic blood pressure decrease with standing (n = 669)
r
r
P-value
P-value
.01
.7958
.08
.0543
.02
.7005
.07
.0682
.01
.7694
.08
.0385
.00
.9220
.03
.4519
.01
.8168
.03
.4969
.00
.9543
.03
.4039
263
Table 4 Multiple regression analysis of skin temperature by measures of sensory and autonomic neuropathy while adjusting for factors related to neuropathy status Right foot mean plantar temperature Neuropathy measurement Sensory Sensory, adjusteda Blood pressure drop with standing Blood pressure drop with standing, adjustedb
Left foot mean plantar temperature
Coefficient ± S.E. P-value Coefficient ± S.E. P-value 0.422 ± 0.181 0.341 ± 0.196
.0203 .0826
0.490 ± 0.195 0.339 ± 0.209
.0122 .1062
0.017 ± 0.007
.0205
0.010 ± 0.008
.2198
0.018 ± 0.007
.0139
0.010 ± 0.008
.1879
Ipsilateral measurements of sensory neuropathy and AAI were inserted into these models. Sensory neuropathy was coded as 1 if the patient was unable to feel the 5.07 monofilament at one or more of nine locations on the foot and was coded as 2 if otherwise. Orthostatic systolic blood pressure drop was coded positively if a fall occurred and was coded negatively if otherwise. a Adjusted for age, gender, diabetes duration, insulin use, glycosylated hemoglobin, and ipsilateral AAI. b Adjusted for age, diabetes duration, glycosylated hemoglobin, and insulin use.
pressure decrease was similarly related to lower skin temperature in the left foot as well, but this association was not statistically significant.
4. Discussion These findings argue against higher foot skin temperature in diabetic subjects with neuropathy. Although neuropathy has been associated with sympathetic dysfunction leading to increase in capillary flow, it is not clear whether this effect is present in our study subjects, since we observed that foot transcutaneous oxygen tension (also correlated with skin perfusion) was not generally related to autonomic or sensory neuropathy in our population (Boyko et al., 1996). Our data would argue that the opposite of what is generally believed concerning the association between neuropathy and skin temperature may be true. We found that autonomic neuropathy as reflected by orthostatic blood pressure drop was related to lower but not higher skin temperature. The asymmetry of this finding in our subjects will require further investigation. Similarly, on average, subjects with sensory neuropathy had lower mean foot skin temperature, although the statistical significance of this association disappeared after adjustment in multivariate models. Because of the low sensitivity and high specificity of using the inability to perceive the 5.07 monofilament at any one site as evidence of neuropathy, the presence of this condition may have been overestimated in this cohort.
264
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
Our study had more than adequate power to detect differences of far smaller magnitude than previously reported. Using the largest and therefore most conservative standard deviation, we observed for mean temperature (left forefoot, Table 2) of 2.7C, our comparison of 365 subjects with and 344 subjects without sensory neuropathy had 99% power to detect a difference of 0.9C (two-sided a = .05). The differences of 3.5C of Rayman et al. (1986) and 4.5C of Archer et al. (1984) therefore would have been easily detectable given our large sample size. Furthermore, none of the 95% confidence limits around the mean temperature differences by presence or absence of sensory neuropathy on either foot includes values greater than zero (data not shown), indicating a low likelihood for higher skin temperature associated with sensory neuropathy in this population. Several possibilities might explain the difference between our findings and previous research. Our population consisted of mostly male, elderly subjects with Type 2 diabetes mellitus. It is possible that different findings will be observed in younger, female patients with Type 1 diabetes. Archer et al. (1984) included mostly Type 1 diabetic subjects with a mean age of about 43 years in their report, of whom 50% were male. Rayman et al. (1986) did not specify diabetes type but included more men than women, with a mean age between 55.2 and 59.5 years. Our population consisted of unselected diabetic patients from a general internal medicine clinic population. It is not clear in previous research how patients were selected for study, therefore raising the possibility that selection bias may have been responsible for observed associations, as has been reported previously in diabetes research (Ballard & Melton, 1986; Melton, Ochi, Palumbo, & Chu, 1984). Despite the results of this study, differences in skin temperature in the feet of diabetic subjects may still be due to abnormal vasoregulation caused by dysfunction of peripheral nerves. We used cardiovascular reflexes as a correlate of autonomic dysfunction in the lower extremities. More direct methods of measuring autonomic function in the feet may be needed in order to demonstrate a relationship between skin temperature and autonomic neuropathy. In conclusion, we found no evidence that the presence of either peripheral sensory or autonomic neuropathy results in higher skin temperature in the feet of diabetic subjects. Although it is undoubtedly true that many diabetic subjects have warm feet, this finding should not be considered a correlate of either sensory or autonomic neuropathy. Skin temperature differences in the feet of diabetic patients may occur due to phenomena other than peripheral neuropathy.
Acknowledgments This research was supported by U.S. Department of Veterans Affairs Merit Review Rehabilitation Research and Development award A318-3RA.
References Archer, A. G., Roberts, V. C., & Watkins, P. J. (1984). Blood flow patterns in painful diabetic neuropathy. Diabetologia, 27, 563 – 567. Armitage, P. & Berry, G. (1987). Statistical methods in medical research, 2nd ed. (pp. 107 – 109, 127 – 129, 411 – 416), Chapter 10. Oxford: Blackwell. Armstrong, D. G., Lavery, L. A., Vela, S. A., Quebedeaux, T. L., & Fleischli, J. G. (1998). Choosing a practical screening instrument to identify patients at risk for diabetic foot ulceration. Archives of Internal Medicine, 158, 289 – 292. Ballard, D. J., & Melton, L. J. (1986). Sources of disparity in incidence and prevalence studies of diabetic retinopathy: influence of selective survival. Diabetes Care, 9, 313 – 316. Boyko, E. J., Ahroni, J. H., Stensel, V. L., Smith, D. C., Davignon, D. R., & Pecoraro, R. E. (1996). Predictors of transcutaneous oxygen tension in the lower limbs of diabetic subjects. Diabetic Medicine, 13, 549 – 554. Cutajar, C. L., Marston, A., & Newcombe, J. F. (1973). Value of occlusion pressures in assessment of peripheral vascular disease. British Medical Journal, 2, 392 – 395. Ewing, D. J., & Clarke, B. F. (1986). Autonomic neuropathy: its diagnosis and prognosis. Clinical and Endocrinological Metabolism, 15, 855 – 888. Ewing, J. A. (1984). Detecting alcoholism: the CAGE questionnaire. JAMA, the Journal of the American Medical Association, 252, 1905. Felder, D., Russ, E., Montgomery, H., & Horwitz, O. (1954). Relationship in the toe of skin surface temperature to mean blood flow measured with a plethsymograph. Clinical Science, 11, 251 – 256. Flynn, M. D., Edmonds, M. E., Tooke, J. E., & Watkins, P. J. (1988). Direct measurement of capillary blood flow in the diabetic neuropathic foot. Diabetologia, 31, 656. Flynn, M. D., & Tooke, J. E. (1995). Diabetic neuropathy and the microcirculation. Diabetic Medicine, 12, 298 – 301. Holewski, J. J., Stess, R. M., Graf, P. M., & Grunfeld, C. (1988). Aesthesiometry: quantification of cutaneous pressure sensation in diabetic peripheral neuropathy. Journal of Rehabilitation Research and Development, 25, 1 – 9. Leslie, P., Jung, R. T., Isles, T. E., Baty, J., Newton, R. W., & Illingworth, P. (1986). Effect of optimum glycemic control with continuos subcutaneous insulin infusion on energy expenditure in type 1 diabetes mellitus. British Medical Journal, 293, 1121 – 1126. Melton, L. J., Ochi, J. W., Palumbo, P. J., & Chu, C. P. (1984). Referral bias in diabetes research. Diabetes Care, 7, 12 – 18. O’Brien, I. A. D., O’Hare, P., & Corrall, R. J. M. (1986). Heart rate variability in healthy subjects: effects of age and the derivation of normal ranges for tests of autonomic function. British Heart Journal, 55, 348 – 354. Rayman, G., Hassan, A., & Tooke, J. E. (1986). Blood flow in the skin of the foot related to posture in diabetes mellitus. British Medical Journal, 292, 87 – 90.
Skin temperature in the neuropathic diabetic foot Edward J. Boykoa,b,*, Jessie H. Ahronic,d, Victoria L. Stenselc a
General Internal Medicine Section and Epidemiologic Research and Information Center (ERIC), Veterans Affairs Puget Sound Health Care System, 1660 South Columbian Way, Seattle, WA 98108, USA b Department of Medicine, University of Washington, Seattle, WA, USA c Research and Development Service, Veterans Affairs Puget Sound Health Care System, Seattle, WA, USA d Clinical Faculty, Department of Biobehavioral Nursing and Health Systems, School of Nursing, University of Washington, Seattle, WA, USA Received 25 January 2001; accepted 11 April 2001
Abstract Several authors have reported higher skin temperature in the feet of diabetic subjects with autonomic neuropathy. We reexamined this association in a cross-sectional study of 712 veterans with diabetes mellitus. Potential subjects included all diabetic patients enrolled in a general internal medicine clinic at a veterans affairs healthcare system. Sensory neuropathy was defined as any pedal insensitivity to the 5.07 monofilament. Autonomic neuropathy was determined using standard cardiovascular reflex tests. An infrared surface scanner was used to measure foot skin temperature at multiple sites. Subjects with sensory neuropathy had lower mean plantar foot skin temperature than those without (28.4C vs. 28.9C, P = .0101). Autonomic neuropathy as a dichotomous variable was unrelated to foot skin temperature. Foot skin temperature, though, negatively correlated with greater drop in systolic blood pressure with standing, which is an indicator of autonomic neuropathy (r = .08, P = .0385). Adjustment for potential confounding factors using multiple linear regression analysis resulted in diminution of the associations between foot skin temperature and sensory neuropathy or orthostatic blood pressure drop, but the latter association remained statistically significant in the right foot. Diabetic veterans with sensory or autonomic neuropathy do not have higher foot skin temperature. Our results suggest that skin temperature may be slightly lower with higher orthostatic blood pressure fall. Other causes exist for the frequently observed differences in skin temperature in the feet of diabetic subjects. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Diabetic foot; Peripheral neuropathy; Skin temperature; Autonomic neuropathy
1. Introduction Previous studies, as summarized by Flynn and Tooke (1995), investigated the effects of diabetic neuropathy on the microcirculation as an increase in blood flow through arteriovenous shunts, which dissipates heat produced as a result of an increased metabolic rate (Leslie et al., 1986). This increase in skin blood flow is thought to be due to dilation of arteriovenous shunts, which are normally under the control of sympathetic nerves. One expected consequence of this increase in skin blood flow is an increase in skin temperature, which has been reported by several authors. Archer, Roberts, and Watkins (1984) reported a * Corresponding author. Veterans Affairs Puget Sound Health Care System (S-111-GIMC), 1660 South Columbian Way, Seattle, WA 98108, USA. Tel.: +1-206-764-2830; fax: +1-206-764-2849. E-mail address: [email protected] (E.J. Boyko).
mean skin temperature on the plantar foot between 33.2C and 33.5C among diabetic subjects with either painful or sensory neuropathy compared to a mean of 27.8C in diabetic subjects without neuropathy. Similarly, Rayman, Hassan, and Tooke (1986) found a higher mean skin temperature on the plantar aspect of the left great toe among diabetic subjects with (32.2C) as compared to those without (28.7C) lower extremity sensory neuropathy. Not all authors, though, have reported statistically significant differences in skin temperature by neuropathy status. Flynn, Edmons, Tooke, and Watkins (1988) found that the median great toe nailfold temperature in 14 diabetic subjects with sensory or autonomic neuropathy (32.6C) differed by only 1C from diabetic subjects without neuropathy (31.5C). We presume that none of these three studies employed population-based sampling, since this was not mentioned in their methods. So, it is unclear how these results might apply to all diabetic subjects in a defined population.
1056-8727/01/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved. PII: S 1 0 5 6 - 8 7 2 7 ( 0 1 ) 0 0 1 5 6 - 8
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
Although the association between higher skin blood flow and skin temperature makes intuitive sense, researchers have reported that this association is not linear. Felder et al. (1954) found that relatively small increases in blood flow resulted in large increases in skin temperature at low skin temperatures, but this association diminished in magnitude above a 28C threshold. In subjects with warm limbs ( > 28C), one would therefore expect a smaller change in temperature with increasing blood flow. We incorporated skin temperature measurements as part of a large, prospective study of risk factors for diabetic foot ulceration. We report our findings from the baseline examination on the associations between measures of sensory and autonomic neuropathy in the lower extremities and directly measured foot skin temperature.
2. Methods Measurements were obtained from subjects enrolled in a prospective study of risk factors for diabetic foot ulceration at the General Internal Medicine Clinic of the Veterans Affairs Puget Sound Health Care System. Exclusion criteria included being nonambulatory, too ill to participate, or having a cognitive impairment that prevented informed consent. Diabetic subjects were identified by review of computerized pharmacy data for receipt of insulin, oral hypoglycemic medication, blood or urine glucose test strips, and review of medical record problem lists for the diagnosis of diabetes mellitus. Informed consent was obtained from all subjects. The University of Washington Human Subjects Committee approved the protocol used in this study. The initial interview employed a standard questionnaire to collect data on demographics, diabetes type, duration, and treatment, and past history of foot ulcer, amputation, and alcohol use. All subjects were asked the four standard CAGE (Ewing, 1984) questions and two or more positive responses were considered consistent with excessive alcohol use. A physical exam with emphasis on the lower extremities was performed by a research nurse practitioner. A random blood sample was drawn for measurement of glycosylated hemoglobin. Sensory neuropathy was defined as an inability to feel the 5.07 Semmes– Weinstein monofilament at any one of nine sites on each foot (Holewski, Stess, Graf, & Grunfeld, 1988). The 5.07 Semmes – Weinstein monofilament has a specificity of approximately 58% and sensitivity near 100% when the lack of perception at one or more sites is used as the cutoff for neuropathy (Armstrong, Lavery, Vela, Quebedeaux, & Fleischli, 1998). Assessment of autonomic neuropathy was done using two standard measures: R – R interval variation with six deep breaths and orthostatic systolic blood pressure decrease with standing (Ewing & Clarke, 1986; O’Brien, O’Hare, & Corrall, 1986). Autonomic neuropathy was
261
considered to be present if either of the following criteria were met: R – R variation 10 or orthostatic blood pressure decrease 30 mmHg. Measurements of lower extremity temperature were performed on the seated patient in a constant room temperature environment on the dorsal and plantar surfaces of both feet with an infrared surface temperature scanner (Omega Medical, Clearwater, FL). The manufacturer states that the instrument is accurate to ± 0.1C, which we confirmed in 200 subjects with repeated readings at the same site that were nearly identical. These skin temperatures were obtained immediately after the patient had walked 100 ft in a corridor adjacent to the neurovascular testing room. We attempted to maintain a constant room temperature during the testing. The forefoot summary measure is the mean of temperatures taken at five sites on the plantar surface of the foot: the great toe, the fifth toe, the first metatarsal head, the fourth metatarsal head, and the fifth metatarsal head. The plantar summary measure is the mean of these sites and the plantar heel; the all loci measure is the mean of these sites and a measurement on the dorsal forefoot. Brachial and lower extremity arterial blood pressures in both limbs were measured by standard segmental Doppler techniques (Cutajar, Marston, & Newcombe, 1973). The ankle –arm index (AAI) was calculated as the ratio of the highest ankle or dorsal foot pressure in each limb divided by the higher brachial pressure in either arm. Only measurements taken at entry into the study were considered in this analysis. Mean values were compared using the unpaired t test (Armitage & Berry, 1987). Nonparametric tests were performed (Mann –Whitney U) on data that were not normally distributed. Differences in proportions were analyzed using the c2 test. Nonparametric correlations were performed using the Spearman rank – order correlation coefficient (Armitage & Berry, 1987). Multiple linear regression analysis was performed using standard methods (Armitage & Berry, 1987).
3. Results Of 921 persons eligible for the study, 778 (85%) agreed to participate and were examined in the diabetic foot research clinic. Foot temperature measurements were successfully completed on 712 (92%) subjects. Characteristics of these subjects are displayed in Table 1 by neuropathy status. These diabetic subjects averaged 63 years of age and were 98% male, 79% white, and 92% had Type 2 diabetes mellitus. The average duration of diabetes was 11.4 years, the mean glycosylated hemoglobin was 11.3% (reference range 4.0 –6.8%), and the mean AAI of both limbs was 0.97. Insulin was the most frequently used treatment (47.3%). Fifty percent of diabetic subjects had sensory neuropathy and 65% had autonomic neuropathy. One-third of the subjects (33.4%) had a history of leg or foot ulceration, and 6.6% had a
262
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
Table 1 Baseline mean values for characteristics of study subjects by sensory and autonomic neuropathy status Sensory neuropathy (n = 712)
Autonomic neuropathy (n = 683)
Characteristic
Present (n = 367)
Absent (n = 345)
P-value
Present (n = 445)
Absent (n = 238)
P-value
Age (years, mean ± S.D.) Male [no. (%)] DM duration (years, mean ± S.D.) DM type Type 1 [no. (%)] Type 2 [no. (%)] Secondary DM [no. (%)] DM treatment Diet [no. (%)] OHA [no. (%)] Insulin [no. (%)] Glycosylated hemoglobin (%, mean ± S.D.) History of ulceration [no. (%)] History of LEA [no. (%)] Alcohol use Less than daily [no. (%)] Daily [no. (%)] Former [no. (%)] CAGE questionsa Treatment for alcohol use Right AAI (mean ± S.D.) Left AAI (mean ± S.D.)
64.7 ± 8.9 388 (51.0) 13.1 ± 9.9
61.7 ± 10.9 373 (49.0) 9.8 ± 9.0
< .0003 .0005 < .0001
63.7 ± 9.4 437 (65.1) 12.1 ± 9.7
62.6 ± 10.5 234 (34.9) 10.7 ± 10.0
.7064 >.9999 .0233
22 (5.6) 362 (93.1) 5 (1.3)
29 (7.5) 351 (90.7) 7 (1.8)
26 (5.8) 410 (92.1) 9 (2.0)
19 (8.0) 216 (90.8) 3 (1.3)
.4445
.4822
36 (9.3) 143 (36.8) 210 (54.0) 11.6 ± 3.4 162 (41.7) 38 (9.8)
53 (13.7) 176 (45.5) 158 (40.8) 11.0 ± 3.2 98 (25.3) 13 (3.4)
52 (11.7) 170 (38.2) 223 (50.1) 11.3 ± 3.2 165 (37.1) 36 (8.1)
31 (13.0) 111 (46.6) 96 (40.3) 11.3 ± 3.6 70 (29.4) 10 (4.2)
.0472
243 (63.4) 25 (6.6) 112 (29.5) 73 (19.3) 67 (17.7) 0.94 ± 0.28 0.94 ± 0.29
260 (70.1) 21 (5.6) 90 (24.0) 66 (17.6) 72 (19.3) 0.99 ± 0.23 1.01 ± 0.24
299 (67.7) 26 (5.9) 117 (26.5) 90 (20.4) 74 (16.8) 0.96 ± 0.26 0.97 ± 0.27
161 (67.9) 11 (4.6) 65 (27.4) 35 (14.8) 50 (21.1) 0.98 ± 0.25 0.98 ± 0.26
.0009
.0114 < .0001 .0003 .2008
.5564 .5780 .0251 .0033
.4697 .0445 .0534 .7825
.0709 .1698 .3387 .5990
Sensory neuropathy refers to presence of this condition on either foot. Abbreviations: DM = diabetes mellitus, OHA = oral hypoglycemic agents, LEA = lower extremity amputation, AAI = ankle arm index. a Two or more positive responses to the four standard CAGE questions about alcohol use.
prior lower extremity amputation. Increasing duration of diabetes and insulin use was significantly associated with both sensory and autonomic neuropathy. Diabetes type was not related to the presence of either type of neuropathy. Blood glucose control as reflected by a single glycosylated hemoglobin measure at entry to the study was significantly better in those without sensory neuropathy. Those with sensory neuropathy were significantly more likely to have had a previous ulcer or amputation. Previous ulceration and amputation were also related to the presence of autonomic neuropathy but not as strongly as sensory neuropathy. More frequent intake of alcoholic beverages, or two or more positive responses to CAGE questions, was not significantly associated with either
type of neuropathy. AAIs were significantly associated with sensory neuropathy bilaterally (right P-value = .0251, left P-value = .0033) but not with autonomic neuropathy. Skin temperatures ranged from 20.2C to 35.5C. Mean temperatures at all, plantar, and forefoot sites were similar on each foot (Table 2). As expected, the means of the dorsal sites were higher (31.0C right and 30.7C left) than the means of the plantar sites (28.6C right and 28.5C left). The plantar fifth toe had the lowest mean surface temperatures bilaterally (27.8C). The mean ± S.D. room temperature before testing was 24.2C ± 1.0C, and after testing, it was 24.5C ± 1.0C. Table 2 presents the mean skin temperatures by the presence of sensory and autonomic neuropathy. The presence
Table 2 Mean ± S.D. skin temperature (C) by presence of sensory and autonomic neuropathy Sensory neuropathy
Autonomic neuropathy
Characteristic
All subjects (n = 712)
Present (n = 366)
Absent (n = 344)
P-value
Present (n = 442)
Absent (n = 190)
P-value
All loci Plantar loci Forefoot loci
29.0 ± 2.3 28.6 ± 2.4 28.6 ± 2.5
28.7 ± 2.3 28.4 ± 2.4 28.3 ± 2.5
29.2 ± 2.2 28.9 ± 2.4 28.8 ± 2.5
.0052 .0101 .0051
29.0 ± 2.3 28.7 ± 2.4 28.6 ± 2.5
28.8 ± 2.2 28.5 ± 2.3 28.4 ± 2.4
.2525 .2340 .2425
Left foot
All loci Plantar loci Forefoot loci
All subjects (n = 712)
Present (n = 360)
Absent (n = 341)
P-value
Present (n = 435)
Absent (n = 191)
P-value
28.8 ± 2.4 28.5 ± 2.5 28.4 ± 2.7
28.6 ± 2.4 28.3 ± 2.5 28.1 ± 2.6
29.1 ± 2.4 28.7 ± 2.6 28.7 ± 2.7
.0085 .0198 .0071
28.8 ± 2.4 28.5 ± 2.5 28.4 ± 2.7
28.7 ± 2.3 28.3 ± 2.5 28.3 ± 2.6
.5785 .5635 .5705
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
of sensory neuropathy was associated with significantly lower skin temperatures in all regions, while subjects with and without autonomic neuropathy had similar skin temperatures. Table 3 examines whether skin temperature was correlated with the two tests of autonomic function expressed as continuous variables. No correlation was seen between all, plantar, or forefoot skin temperatures and mean heart rate variation with respirations on either foot. Weak negative correlations were seen between a fall in systolic blood pressure with standing and skin temperature, such that lower skin temperature was associated with a greater fall in systolic blood pressure. This correlation was statistically significant when comparing the mean right forefoot temperature to the fall in systolic blood pressure ( P = .0385), although the magnitude of the correlation was very small (r = .08). In Table 4, multiple linear regression analysis was used to determine whether differences in skin temperature by neuropathy status might be due to confounding by other variables, which differed by neuropathy status, as seen in Table 1. These analyses used the mean plantar skin temperature as the dependent variables, but similar results were obtained when either the mean all sites or mean forefoot temperatures was used. Adjustment for age, gender, diabetes duration, diabetes treatment (insulin vs. other), glycosylated hemoglobin, and ipsilateral AAI reduced the magnitude of the association between sensory neuropathy and plantar skin temperature on either foot. Skin temperature no longer remained statistically significantly associated with sensory neuropathy, although P-values for this association were near the .05 threshold value for rejection of the null hypothesis. Systolic blood pressure drop with standing and skin temperature in the right foot only remained related to a statistically significant degree, even after adjustment for age, diabetes duration, glycosylated hemoglobin, and diabetes treatment. Blood
Table 3 Spearman correlations between infrared foot skin surface temperature and continuous measures of autonomic function
Right foot Mean C all loci Mean C plantar loci Mean C forefoot loci Left foot Mean C all loci Mean C plantar loci Mean C forefoot loci
Heart rate variation with respirations (n = 530)
Systolic blood pressure decrease with standing (n = 669)
r
r
P-value
P-value
.01
.7958
.08
.0543
.02
.7005
.07
.0682
.01
.7694
.08
.0385
.00
.9220
.03
.4519
.01
.8168
.03
.4969
.00
.9543
.03
.4039
263
Table 4 Multiple regression analysis of skin temperature by measures of sensory and autonomic neuropathy while adjusting for factors related to neuropathy status Right foot mean plantar temperature Neuropathy measurement Sensory Sensory, adjusteda Blood pressure drop with standing Blood pressure drop with standing, adjustedb
Left foot mean plantar temperature
Coefficient ± S.E. P-value Coefficient ± S.E. P-value 0.422 ± 0.181 0.341 ± 0.196
.0203 .0826
0.490 ± 0.195 0.339 ± 0.209
.0122 .1062
0.017 ± 0.007
.0205
0.010 ± 0.008
.2198
0.018 ± 0.007
.0139
0.010 ± 0.008
.1879
Ipsilateral measurements of sensory neuropathy and AAI were inserted into these models. Sensory neuropathy was coded as 1 if the patient was unable to feel the 5.07 monofilament at one or more of nine locations on the foot and was coded as 2 if otherwise. Orthostatic systolic blood pressure drop was coded positively if a fall occurred and was coded negatively if otherwise. a Adjusted for age, gender, diabetes duration, insulin use, glycosylated hemoglobin, and ipsilateral AAI. b Adjusted for age, diabetes duration, glycosylated hemoglobin, and insulin use.
pressure decrease was similarly related to lower skin temperature in the left foot as well, but this association was not statistically significant.
4. Discussion These findings argue against higher foot skin temperature in diabetic subjects with neuropathy. Although neuropathy has been associated with sympathetic dysfunction leading to increase in capillary flow, it is not clear whether this effect is present in our study subjects, since we observed that foot transcutaneous oxygen tension (also correlated with skin perfusion) was not generally related to autonomic or sensory neuropathy in our population (Boyko et al., 1996). Our data would argue that the opposite of what is generally believed concerning the association between neuropathy and skin temperature may be true. We found that autonomic neuropathy as reflected by orthostatic blood pressure drop was related to lower but not higher skin temperature. The asymmetry of this finding in our subjects will require further investigation. Similarly, on average, subjects with sensory neuropathy had lower mean foot skin temperature, although the statistical significance of this association disappeared after adjustment in multivariate models. Because of the low sensitivity and high specificity of using the inability to perceive the 5.07 monofilament at any one site as evidence of neuropathy, the presence of this condition may have been overestimated in this cohort.
264
E.J. Boyko et al. / Journal of Diabetes and Its Complications 15 (2001) 260–264
Our study had more than adequate power to detect differences of far smaller magnitude than previously reported. Using the largest and therefore most conservative standard deviation, we observed for mean temperature (left forefoot, Table 2) of 2.7C, our comparison of 365 subjects with and 344 subjects without sensory neuropathy had 99% power to detect a difference of 0.9C (two-sided a = .05). The differences of 3.5C of Rayman et al. (1986) and 4.5C of Archer et al. (1984) therefore would have been easily detectable given our large sample size. Furthermore, none of the 95% confidence limits around the mean temperature differences by presence or absence of sensory neuropathy on either foot includes values greater than zero (data not shown), indicating a low likelihood for higher skin temperature associated with sensory neuropathy in this population. Several possibilities might explain the difference between our findings and previous research. Our population consisted of mostly male, elderly subjects with Type 2 diabetes mellitus. It is possible that different findings will be observed in younger, female patients with Type 1 diabetes. Archer et al. (1984) included mostly Type 1 diabetic subjects with a mean age of about 43 years in their report, of whom 50% were male. Rayman et al. (1986) did not specify diabetes type but included more men than women, with a mean age between 55.2 and 59.5 years. Our population consisted of unselected diabetic patients from a general internal medicine clinic population. It is not clear in previous research how patients were selected for study, therefore raising the possibility that selection bias may have been responsible for observed associations, as has been reported previously in diabetes research (Ballard & Melton, 1986; Melton, Ochi, Palumbo, & Chu, 1984). Despite the results of this study, differences in skin temperature in the feet of diabetic subjects may still be due to abnormal vasoregulation caused by dysfunction of peripheral nerves. We used cardiovascular reflexes as a correlate of autonomic dysfunction in the lower extremities. More direct methods of measuring autonomic function in the feet may be needed in order to demonstrate a relationship between skin temperature and autonomic neuropathy. In conclusion, we found no evidence that the presence of either peripheral sensory or autonomic neuropathy results in higher skin temperature in the feet of diabetic subjects. Although it is undoubtedly true that many diabetic subjects have warm feet, this finding should not be considered a correlate of either sensory or autonomic neuropathy. Skin temperature differences in the feet of diabetic patients may occur due to phenomena other than peripheral neuropathy.
Acknowledgments This research was supported by U.S. Department of Veterans Affairs Merit Review Rehabilitation Research and Development award A318-3RA.
References Archer, A. G., Roberts, V. C., & Watkins, P. J. (1984). Blood flow patterns in painful diabetic neuropathy. Diabetologia, 27, 563 – 567. Armitage, P. & Berry, G. (1987). Statistical methods in medical research, 2nd ed. (pp. 107 – 109, 127 – 129, 411 – 416), Chapter 10. Oxford: Blackwell. Armstrong, D. G., Lavery, L. A., Vela, S. A., Quebedeaux, T. L., & Fleischli, J. G. (1998). Choosing a practical screening instrument to identify patients at risk for diabetic foot ulceration. Archives of Internal Medicine, 158, 289 – 292. Ballard, D. J., & Melton, L. J. (1986). Sources of disparity in incidence and prevalence studies of diabetic retinopathy: influence of selective survival. Diabetes Care, 9, 313 – 316. Boyko, E. J., Ahroni, J. H., Stensel, V. L., Smith, D. C., Davignon, D. R., & Pecoraro, R. E. (1996). Predictors of transcutaneous oxygen tension in the lower limbs of diabetic subjects. Diabetic Medicine, 13, 549 – 554. Cutajar, C. L., Marston, A., & Newcombe, J. F. (1973). Value of occlusion pressures in assessment of peripheral vascular disease. British Medical Journal, 2, 392 – 395. Ewing, D. J., & Clarke, B. F. (1986). Autonomic neuropathy: its diagnosis and prognosis. Clinical and Endocrinological Metabolism, 15, 855 – 888. Ewing, J. A. (1984). Detecting alcoholism: the CAGE questionnaire. JAMA, the Journal of the American Medical Association, 252, 1905. Felder, D., Russ, E., Montgomery, H., & Horwitz, O. (1954). Relationship in the toe of skin surface temperature to mean blood flow measured with a plethsymograph. Clinical Science, 11, 251 – 256. Flynn, M. D., Edmonds, M. E., Tooke, J. E., & Watkins, P. J. (1988). Direct measurement of capillary blood flow in the diabetic neuropathic foot. Diabetologia, 31, 656. Flynn, M. D., & Tooke, J. E. (1995). Diabetic neuropathy and the microcirculation. Diabetic Medicine, 12, 298 – 301. Holewski, J. J., Stess, R. M., Graf, P. M., & Grunfeld, C. (1988). Aesthesiometry: quantification of cutaneous pressure sensation in diabetic peripheral neuropathy. Journal of Rehabilitation Research and Development, 25, 1 – 9. Leslie, P., Jung, R. T., Isles, T. E., Baty, J., Newton, R. W., & Illingworth, P. (1986). Effect of optimum glycemic control with continuos subcutaneous insulin infusion on energy expenditure in type 1 diabetes mellitus. British Medical Journal, 293, 1121 – 1126. Melton, L. J., Ochi, J. W., Palumbo, P. J., & Chu, C. P. (1984). Referral bias in diabetes research. Diabetes Care, 7, 12 – 18. O’Brien, I. A. D., O’Hare, P., & Corrall, R. J. M. (1986). Heart rate variability in healthy subjects: effects of age and the derivation of normal ranges for tests of autonomic function. British Heart Journal, 55, 348 – 354. Rayman, G., Hassan, A., & Tooke, J. E. (1986). Blood flow in the skin of the foot related to posture in diabetes mellitus. British Medical Journal, 292, 87 – 90.