Describing Normative Foot Temperatures in Patients With Diabetes-Related Peripheral Neuropathy

Abstract

Background:

Prior research shows increased foot temperatures are predictive of diabetes-related foot complications. Our aim was to describe normative skin foot temperatures for individuals with diabetic peripheral neuropathy to better inform new technologies. We also explored for potential risk factors which correlate with changes in foot temperatures.

Methods:

We conducted a retrospective chart review of adult patients >18 years of age with diabetes mellitus and clinically diagnosed diabetic peripheral neuropathy with pedal digital thermometry performed between 2009 and 2018. A total of 58 patients met these criteria. Univariate modeling was based on covariates that may affect foot temperature including age, peripheral arterial disease, toe pressure, seasonality of measurement, smoking pack-years, caffeine use, insulin use, and calcium channel blocker use.

Results:

In patients with diabetic peripheral neuropathy, mean toe temperatures of 27.67°C (6.300°C), forefoot of 28.58°C (5.36°C), midfoot of 29.21°C (3.81°C), and rearfoot of 29.88°C(3.83°C) were demonstrated. A modest negative correlation between seasonality and toe and metatarsal temperatures (r = −0.38, P < .05; r = −0.43 P < .01, respectively) was demonstrated. Midfoot temperatures were modestly and positively correlated to the presence of small fiber symptoms (r = 0.33, P = .03). Positive modest correlation with rearfoot temperatures and amount of pack-year history (r = 0.30, P = .03) was seen.

Conclusion:

Normative foot temperatures in neuropathic patients were found to be inversely associated with seasonality at the toe and metatarsal level. Smoking and pack-year history demonstrate modest correlation previously unseen in temperature analyses and warrant further exploration. Normative temperatures in neuropathic patients can better inform new technologies for the prevention of diabetic foot ulcer and Charcot neuroarthropathy.

Introduction

Relative temperature changes using dermal thermometry may represent antecedent inflammatory changes occurring prior to frank diabetic foot ulceration (DFU), Charcot neuroarthropathy (CN), and cellulitis of the foot in patients with diabetes mellitus (DM).^1-8 Monitoring of dermal foot temperatures may represent an important clinical tool for diagnosing and provide an early intervention in patients with diabetes-related foot complications and diabetic peripheral neuropathy (DPN) where the loss of protective sensation leads to potential diagnostic delays.^9,10 When dermal foot temperatures are monitored in patients with previous DFU, there can be a four to tenfold prevention of recurrence with prompt recognition and subsequent offloading of pressure.^2,4-6,11 These studies have used the contralateral foot as a control to determine whether or not the temperature of the foot of interest is elevated.

Although proven to be significantly successful in reducing future foot complications unilaterally, using the contralateral foot as a control may not be helpful in patients with future acute bilateral foot complications. For example, bilateral CN has been reported to occur up to 9%–33% of the time.¹² In addition, due to the higher incidence of lower extremity amputations in those with diabetes, using the contralateral foot as a control may not be an option. Therefore, elucidating a better understanding of normative skin temperatures in DPN patients would be helpful in designing clinical tools to provide earlier detection of diabetes-related foot complications and better refine their use in bilateral cases.

The aim of this study was to determine normative foot temperature ranges in DPN patients. We also examined potential factors for correlation that may affect foot temperature in those with DPN, such as seasonality of measurement, smoking, peripheral arterial disease (PAD), age, caffeine, insulin, calcium channel blockers, and small fiber symptoms.

Methods

Overview and Setting

The University of Michigan Podiatry Clinic performed clinical assessments of 5000 distinct individuals with DM from 2009 to 2018. We conducted a consecutive retrospective study of 58 DPN patients who were seen in outpatient podiatry clinics at the University of Michigan between 2009 and 2018 to describe average foot temperatures. Chart review was performed by the authors and consecutive patients were assessed. The protocol was approved by the Institutional Review Board.

Subjects

Eligible patients were identified as age >18 years with the diagnosis of both DM and DPN using a cohort discovery tool, Data Direct, software developed by the University of Michigan Medical School Office of Research. Clinically, all patients were defined as having DPN by published criteria, which included conducting an annual diabetic foot examination using 10 g Semmes Weinstein Monofilament, ankle reflex exam, and 128 Hz vibratory tuning fork assessment.¹³ This resulted in a sample size of 2603 patients, thus, revealing a prevalence of DM with DPN of 52.1%. Of the aforementioned 2603 patients with DM and DPN, 452 patients had digital thermometry scans (9.04% of overall cohort). Patients receiving pedal temperature measurements were individuals who were considered at-risk for development of foot ulcers. This broadly included patients where the treating physician had concern for future skin breakdown on the affected foot, which was the foot with the clinical complaint. It included patients at risk of breakdown as determined by increased foot classification score,¹³ areas of bony prominence, local warmth, or localized pain in the presenting neuropathic patient.

Exclusions

After identifying patients with Data Direct, a manual review of charts was performed to exclude patients who did not have dermal foot temperatures during their podiatry visits. Patients who had active diabetes-related foot complications such as CN, infection, or DFU were excluded, as these pathologies have been shown to affect foot temperature and would obstruct our goal of obtaining normative temperatures. Upon exclusions, the sample size was n = 58 (1.16% of the overall cohort). “Infection” was defined as paronychia, cellulitis, tinea pedis, osteomyelitis, ulcers with suspicion of infection, and other suspected fungal infections. Patients with onychomycosis were not excluded.

Skin Temperature Measurement

An Exergen Dermatemp thermometer (Boston, MA, United States) was used for all measurements. A preprinted foot grid was used to standardize locations of all measurements taken. The thermometer was held at a standard distance, approximately 0.5 in. (1.27 cm) from the foot in the following 15 areas: first digit, third digit, fourth digit, fifth digit, the first metatarsal head, third metatarsal head, and fifth metatarsal head, the lateral/medial/central midfoot, the lateral/medial/central rear foot, and lateral/medial ankle (Supplemental Figure S1). Measurements were taken in a standardized fashion by the two practicing physicians in the study (JSW, BMS) from the plantar foot after patient was seated in the exam chair with shoes and socks removed for five minutes. We also collected data on variables that might affect skin temperature, such as insulin,¹⁴ calcium channel blocker,¹⁵ caffeine intake, tobacco history, age, toe pressure, PAD, and seasonality at the time of measurement, and small fiber symptoms.¹⁶

Seasonality

Seasonality was defined as the average temperature during the month in which the patient had his or her dermal temperatures measured.¹⁷ The interrogated data were time stamped according to the time the physician completed their note and not the arrival time of the patient, nor the exact time of day when the patient had their dermal temperatures taken. Given that our physical location is in a midlatitude climate, the breadth of temperatures can be broad and as a result the average temperature of the month when dermal temperatures were recorded was reported.

Tobacco Use

Smoking history was recorded as in number of pack-years, as reported by the patient during history. Historical smoking status was also considered and defined as “never smoker,” “current smoker,” or “former smoker.” Each patient was assigned a value based on pack-year calculation, which is one pack per day annually equates to one pack-year.

Caffeine Use

Caffeine use was recorded as present or absent based on self-reported regular caffeine use within a year before the encounter. Regular caffeine use was defined as any consumption of caffeinated drinks that was equal to or more than once per week. Caffeine amount was also recorded as the amount of caffeine consumed per day, which was estimated in milligrams based on self-reported regular consumption of tea, coffee, and/or soda reported within a year before the encounter. If caffeine use was not explicitly denied within the past year, no determination was made.

Toe Pressure and Peripheral Arterial Disease

Toe pressures taken within the immediate past 12 months of the dermal temperature reading were recorded. Toe pressures were extracted from arterial Doppler study reports that included segmental pressures, ankle-brachial index, toe-brachial index, and photoplethysmography. The presence of PAD was recorded if the patient had been diagnosed with PAD or peripheral vascular disease before the encounter in which dermal temperatures were taken or as a result of noninvasive vascular assessments.

Small Fiber Symptoms

The presence of small fiber symptoms was defined as any patient reported paresthesia including but not limited to numbness, tingling, or burning in the feet within a year before the dermal temperature encounter. If the presence or absence of small fiber symptoms was not explicitly clear, no determination was made.

Statistical Analysis

For each patient, skin temperatures were averaged at the following levels of the foot: toe, metatarsal, midfoot, and rearfoot. We used STATA/SE 10.1 to summarize all variables and examine for normality of distribution for continuous variables. We created a correlation matrix for the assessment of continuous variables and performed a one-way ANOVA test for dichotomous variables. We assessed for equal variances using Bartlett’s test. To assess for selection bias, we evaluated regional temperature differences for suspected acute presenting condition, such as infection, ulceration, and acute CN. The unaffected index foot, the foot without suspected pathology concerns clinically, for each patient was selected and temperatures from that foot were used for the analysis of mean foot temperatures. These unaffected feet did not have acute clinical signs of pathology documented in the paper chart at the time of temperature recording. Preulcerative lesions and other acute processes such as edema, fractures, contusions, and deformities have not been shown to influence foot temperature and, therefore, were not considered in the selection of the index foot. Toe pressures from the index foot, but not the acute foot, were used. Statistical significance was considered when the P-value was lower than an alpha of .05.

Results

A total of 58 patients met the inclusion criteria for this study. Patients had a mean age of 59.7 (11.8). All patients had DM and as previously aforementioned, DPN. A total of 84% identified as White, 13% Black, 2% Asian, and 1% other. The average body mass index was 34 kg/m².

Patient normative dermal temperatures were recorded from four sections of the participants’ foot: toes, forefoot (metatarsal heads), midfoot, and rearfoot (Table 1). The mean temperature of the toe level in patients with DPN was 27.6°C (6.30°C). At the level of the forefoot, the average temperature was 28.58°C (5.36°C). At the level of the midfoot the average temperature was 29.21°C (3.81°C). Finally, the pedal temperature of the foot at the level of the rearfoot was 29.88°C (3.83°C).

Table 1.

Characteristics of Study Population and Normative Dermal Foot Temperatures.

Variable	Number (n)	Mean	Minimum	Maximum
Age (years)	58	59.69 (11.78)	32	81
Toe pressure (mmHg)	12	128.42 (59.85)	20	233
Seasonality (°C)	58	10.69 (18.89)	−9.78	24.17
Pack-years	58	12.85 (24.42)	0	141
Caffeine amount (mg)	12	261.75 (253.86)	0	910
Metatarsal level (°C)	39	28.58 ± 5.36	21.87	33.17
Midfoot level (°C)	53	29.21 ± 3.81	22.91	33.67
Rearfoot level (°C)	50	29.88 ± 3.83	25.34	34.67
Toe level (°C)	39	27.67 ± 6.30	20.37	33.58

We explored several variables which may influence pedal temperatures. Variables evaluated included insulin, caffeine intake, current or former smoker, calcium channel blocker use, and small fiber neuropathy symptoms (Table 2). In order of population proportionality, 72.4% (n = 42) of patients who had DM and DPN attested to small fiber symptoms, while 72.4% (n = 42) used insulin to manage hyperglycemia, 43.1% of patients (n = 25) admitted to regular caffeine intake at an average caffeine intake of 261.75 mg (253.86 mg) per day. In addition, 42.1% of patients (n = 24) were either previous or current smokers with an average pack-year history of 12.85 (24.42) years. Finally, 25.9% (n = 15) of the cohort used calcium channel blockers to manage hypertension.

Table 2.

Insulin Use, Caffeine Use, Smoking Status, Calcium Channel Blocker Use, and Small Fiber Symptom Prevalence Within Study Population.

Variable	Proportion of study population
Insulin use	42/58 (72.41%)
Caffeine use	25/58 (43.10%)
Current or former smoker	24/57 (42.11%)
Calcium channel blocker use	15/58 (25.86%)
Small fiber symptoms	42/58 (72.41%)

In the univariate analyses, several factors were correlated with temperature changes at various described levels. These include a modest inverse (negative) correlation between ambient outdoor temperature and pedal temperatures at the level of the toes and forefoot. Average seasonality was 10.69°C (18.89°C), but the yearly seasonality (all months) included a robust range from −9.78°C to 24.17°C. Both relationships were statistically significant at r = −0.38, P < .05 and r = −0.43, P < .01 at the level of the toe and metatarsal zones, respectively. However, seasonality did not influence midfoot and rearfoot temperatures (P > .05).

In addition, midfoot temperatures were modestly and positively correlated to the presence of small fiber symptoms and this finding was also statistically significant (r = 0.33, P = .03). Furthermore, there was a positive modest correlation between rearfoot pedal temperatures and amount of pack-year history; again this was significant (r = 0.30, P = .03). These findings are highlighted in Table 3.

Table 3.

Univariate Descriptions per Foot Region.

Foot region	Variable	Correlation	P-value
Toe level	Seasonality	−0.38	.02
Metatarsal head level	Seasonality	−0.43	.006
Midfoot level	Small fiber symptoms	0.33	.03
Rearfoot level	Pack-years	0.30	.03

There was an increase in the loss of protective sensation, as tested with a monofilament, located in the midfoot with increased pack-year history and was significant (P < .05) as compared to other locations on the foot. Finally, all regions of the foot demonstrated increased temperatures with calcium channel blocker use but it did not reach statistical significance (P > .05).

Discussion

We believe we are the first group to report normative pedal skin temperatures by foot region in patients with DM and DPN using in-office digital thermometry. Previous literature has relied on using the patient’s contralateral limb as an internal positive control, but this is not always an available option. For example, patients with CN have bilateral involvement approximately 9%–33% of the time. Therefore, determining subtle temperature changes in this patient population would not likely be feasible using the contralateral limb. We found the average skin temperatures to be approximately 28.92°C (84.06°F) for the 58 patients with DM and DPN included in this study. Temperatures vary according to location on the foot and are presented in Table 1 for review. In general, as would be expected logically and physiologically, the more distal the location on the limb, the cooler the temperatures. Our findings are similar to a recent report from Yavuz et al that reported average toe temperatures of 30.7°C and central metatarsal temperatures of 31.7°C in the DPN group using a thermal imaging device.¹⁸ In our analysis, we explored for potential confounding variables and stratified temperatures for the entire region of the foot.

Interestingly, we found seasonality to have a significant inverse association with toe and metatarsal temperatures. There could be several explanations for this finding. First, patients may be wearing open or cooler footwear and exposure to air conditioning could have stimulated vasoreactivity. Second, dermal temperatures were likely taken later during the visit after the provider evaluated their condition, thus, there was not a standard protocol followed for acclimation.^7,19 Autonomic neuropathy has been implicated in skin temperature changes via loss of sympathetic tone apart from sensory-motor nerve dysfunction.²⁰ Decreased sympathetic tone was observed in DPN patients²⁰ and could partially explain our findings following prolonged sitting during a clinic visit. Quayle also found that sudomotor activity was affected less than vasomotor tone in DPN patients.¹⁶ In our previous investigations, we found sudomotor function demonstrated stronger correlations with midfoot and heel temperatures when compared with the forefoot.¹⁹ There could be clinical implications for our observation of seasonality or local temperature association with toe and metatarsal temperatures. Toe ulcers have been described as the most frequent location for DFU,²¹ and the seasonality effect described in our data could partially explain the regional variation of DFU care expenditures which are higher in warmer climates of the United States.²²

We also found smoking status and pack-year history to be associated with increased rearfoot temperatures. We coded smoking status in this study as pack-years based on patient use. As the amount of pack-years increases, the temperatures in the rearfoot increase whereas no other significant temperature changes are appreciated. It is known that smoking causes abnormalities in endothelial dysfunction, lipoprotein metabolism, coagulation, and platelet function. We attribute our findings and this relationship to the development of vascular calcification in patients with DM. Aging is a major cause of vascular calcification with reported incidence increasing by 30% from age 20 to 90 years.²³ In addition, vascular calcifications are reportedly four times more frequent in patients with DM than healthy subjects²⁴ and calcifications are shown to correlate with diabetic microvascular complication (nephropathy).²⁵

While microvascular dysfunction and vascular calcifications could explain the lack of temperature variation distally in the toes, metatarsals, and midfoot, the vessels in the rearfoot are larger, often >1.5 mm in diameter,²⁶ and therefore, the effect of prolonged smoking may not be completely realized. While counterintuitive, the rearfoot may be the proverbial last domino to fall leading to more serious critical limb ischemia and tissue loss. However, our data do not explore the relationship between temperatures in DPN with respect to critical limb ischemia. Smoking also affects another one of our findings: the association between pack-year history and an increase in small fiber symptoms in the midfoot. While temperatures in the midfoot were not significantly altered by smoking, the relative ischemia created in the microvasculature by smoking likely causes nerve death and, therefore, may increase paresthesia symptoms and clinical findings. As neuropathy typically starts distally in a stocking-glove distribution in DM, small fiber symptoms may occur in relation to this accelerated dying of nerve fibers induced by prolonged cigarette use. A recent systematic review also suggested a significant positive association between smoking and neuropathy,²⁷ and our study is the first to quantify a relationship between pedal temperatures, smoking history, and small fiber neuropathy symptoms and loss of protective sensation via monofilament testing.

Our study has several limitations. First, it is a retrospective study and a standardized acclimation procedure was not used. We mitigated this by having patients rest for five minutes prior to testing. Also, to the extent of study design, sampling bias, experimenter bias, and observer bias could not be eliminated but were reduced by having broad inclusion and exclusion criteria. Second, as a result of manual chart review, not all data points were readily available and thus missing data are present. For example, there is missing data on 33% for toe and metatarsal data that may affect the precision around the normative data estimates for toe and metatarsal temperatures. Temperature measurement was excluded if not all points were obtained in accordance with the predetermined locations for a specified foot region. As a result, incomplete data were removed to not skew normative data, and our estimates compare favorably with Yavuz et al.¹⁸ We also had toe pressure and caffeine amount for only 21% of patients, thus affecting statistical power to detect a significant correlation.

In addition, the time interval between the most recent laboratory result (ie, HbA1c) and the digital thermometry scan was not within three months and not included because of incremental changes which could occur for an individual subject. However, data were reviewed consecutively and this increased data fidelity. HbA1c is the only parameter not accounted for by the chart review within the immediate past year. Furthermore, skin temperatures were taken based on a suspicion of a diabetes-related foot complication. We explored for this selection bias and found no difference across foot complications such as CN, DFU, and infection. We also used an index foot without clinical involvement to investigate this selection bias. Last, we may have been too underpowered to describe associations by limiting our model to foot regions and missing data for other covariates, such as toe pressures. For example, calcium channel blockers have been shown to increase skin temperatures in diabetes.¹⁵ However, we found increased temperatures across all regions of the foot with calcium channel blockers use that did not reach statistical significance possibly due to the regional analysis used and lack of statistical power (n = 15).

Conclusion

We described normative foot temperatures in DPN patients by region of the foot. We have found seasonality to be inversely associated with toe and metatarsal temperatures. This could explain some of the regional variation in DFU prevalence in warmer climates. Normative temperatures in DPN patients can better inform new technologies for detecting precursors to foot ulcer and CN such as personal devices calibrated to alert patients with DPN when their foot temperatures are abnormal. Future research should focus on better elucidation of confounding factors for skin temperatures.