Effect of thermometry on the prevention of diabetic foot ulcers: a systematic review with meta-analysis*

Abstract Objective: to analyze the effect of cutaneous foot thermometry in people with Diabetes Mellitus, compared with the standard prevention of foot ulcers adopted in these patients. Method: a systematic review with meta-analysis. Protocol registered with PROSPERO (CRD42020202686). The recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) were followed. The search was performed in the following data sources: SCOPUS, Web of Science, MEDLINE via PubMed, MEDLINE via EBSCO, MEDLINE via Biblioteca Virtual em Saúde, Embase, CINAHL, Cochrane Library, LILACS via Biblioteca Virtual em Saúde, Google Scholar, Biblioteca Digital Brasileira de Teses e Dissertações, Catálogo de Teses & Dissertações-Capes, Open Grey and ProQuest Dissertations and Theses. The risk of bias was assessed by the Cochrane Collaboration Risk of Bias Tool (RoB 2), the meta-analysis was performed in the Review Manager 5.4 software and the Certainty of evidence in the Grading of Recommendations Assessment, Development and Evaluation system. Results: of the 670 records, five articles were eligible. The meta-analysis was calculated for the prevention of the incidence of diabetic foot ulcers outcome, with effect summarization (RR 0.53; 95%CI 0.29-0.96; p=0.02), with certainty of moderate evidence. Conclusion: thermometry showed a protective effect on the incidence of diabetic foot ulcers when compared to standard foot care.


Introduction
The current approaches adopted by health services in the prevention and early treatment of Diabetic Foot Ulcers (DFUs) are multiple and varied. Screening and education in health for the patient, family and health professionals are relevant pillars (1) . However, efforts to prevent DFU remain a challenge and demand high costs for global public health (2)(3) , raising the need for new preventive approaches.
DFUs are preceded by an increase in local dermal temperature by inflammation and enzymatic tissue autolysis resulting from pressure-activity imbalance, added to repetitive stress, neuropathic sensory loss and biomechanical abnormalities (4) . A temperature difference of 2.2ºC between the same site on both feet implies a risk for imminent ulcer/inflammation (5)(6) . However, clinical signs of inflammation are subtle for detection by patients or even by trained health professionals (7) .
Although many signs of inflammation are difficult to be objectively assessed, temperature can be easily measured.
The traditional method of assessing foot temperature is palpation with the back of the hand. However, with this method, humans are only able to discriminate temperature differences greater than 2ºC. Thus, skin thermometry emerges as a promising tool for identifying inflammation, providing early signs to prevent DFU incidence and reduce serious complications, such as high morbidity, frequent hospitalizations, lower limb amputation and deaths (8)(9) .
It is the method most used by the scientific community and patients, as it involves the use of a thermometer, a low-cost and easy to apply instrument (10) . As a result, the patients can modify their activity, measuring skin temperature, as well as dosing their insulin and checking their blood glucose (11) .
A study reported an association between increased local temperature and localized pressure, causing tissue damage (12) . A number of researchers used thermometry as a tool to diagnose occult neuropathic fractures in patients with diabetes (13) . Temperature assessment is a useful technique to identify patients at risk for ulceration (14) .
Similar findings were identified with a handheld infrared thermometer in patients with asymptomatic sensory neuropathy, neuropathic foot ulcers and patients with neuropathic fractures (Charcot arthropathy) (15) .
Foot temperature varies with the patient's activity level and environment. The reference is a corresponding area on the contralateral foot. It is noteworthy that the temperatures of these areas do not differ by more than 1ºC (5)(6) and a number of studies suggest that differences ≥ 2.2ºC in temperature can be considered a risk for ulceration (11,(16)(17)(18) .  (11,(16)(17) .
Previous systematic reviews have analyzed the effect of several interventions to prevent diabetic foot ulcers, including the use of thermal foot monitoring (19)(20)(21) .
Although reviews on the subject matter were identified, failures to provide reliable numerical summaries of effects were observed due to limitations in the quality of the individual studies. In addition to that, new evidence has emerged since then. Thus, a comprehensive evaluation of randomized controlled trials was necessary to allow us to make the best use of the currently available evidence.
This systematic review with meta-analysis aimed at analyzing the effect of cutaneous foot thermometry in people with Diabetes Mellitus (DM), when compared to the standard prevention of foot ulcers adopted in these patients.

Research question and eligibility criteria
The Population, Intervention, Control and Outcome (PICO) strategy was used for data research (24) . This systematic review with meta-analysis focused on participants (P) diagnosed with Diabetes Mellitus, with or without risk of developing diabetic foot ulcer. The use of cutaneous thermometry devices to assess foot temperature was arranged as an intervention (I). The use of standard foot care (therapeutic footwear, diabetic foot education, regular foot assessment by health Araújo AL, Negreiros FDS, Florêncio RS, Oliveira SKP, Silva ARV, Moreira TMM. professionals, and foot self-care) was provided as a comparison (C). And the outcomes (O) evaluated were studies that included the prevention of the incidence of diabetic foot ulcers outcome.
From this, the research problem was outlined: Which is the effect of cutaneous foot thermometry on people with DM, when compared to the standard prevention of foot ulcers adopted in these patients?
The studies were included when they met the We discarded publications with: study intervention with foot skin thermometry in people with and without DM; thermometry in people with simultaneous Diabetes Mellitus and active ulcer; study designs: cross-sectional, prospective and retrospective cohort, case-control, case reports or case series; types of publication: reviews, protocols, letters to the editor, congress abstracts, personal opinions, book chapters; unavailable in full. Additionally, another search for the references of the studies included was necessary. An additional search was also carried out in consultation with "experts/specialists" in the researched area via www.researchgate.net/. They were invited to suggest relevant articles on the chosen topic. However, no answers were obtained.

Search strategy
The search strategy was built using three controlled health vocabularies: Medical Subject Headings (MeSH), Descritores em Ciências da Saúde (DeCS) and EMTREE; together with natural language, in order to obtain a wide spectrum of results in different databases (25) . After the search terms have been defined, they were combined with the Boolean operators AND and OR.
The following high-sensitivity search strategy was used for all databases: ("diabetes mellitus" OR "diabetic patient" OR diabetes OR diabetic) AND ("temperature measurement" OR "temperature monitoring" OR "temperature recording" OR "thermal measurement" OR "thermal monitoring" OR "thermal recording" OR "thermomonitoring" OR "thermo-recording" OR "body temperature monitoring" OR thermometer OR thermometry OR thermogram OR thermomonitoring OR thermorecording OR thermomeasurement) AND (prevention OR prevent OR control OR prophylaxis OR "prevent ulceration" OR "ulcer prevention" OR "prophylactic treatment" OR "preventive therapy" OR "preventive measures" OR "disease prevention" OR "disease prophylaxis" OR "health protection" OR "preventive treatment" OR "prophylactic management" OR "prophylactic therapy" OR "prophylactic treatment") AND (foot OR ulcer OR ulceration OR feet OR "foot ulceration" OR "foot ulcer" OR "foot complication" OR "diabetic foot" OR "diabetic feet").

Selection of the studies and extraction of the information
The research results in each database were imported into the Rayyan ® reference manager developed by the Qatar Computing Research Institute (QCRI), for organization of the studies, removal of duplicates, and selection and screening of the studies (26) . Two authors of the review independently examined the titles and abstracts of all references. Subsequently, the full texts of potentially eligible studies were independently assessed by the two reviewers to determine whether all inclusion criteria were met. In case of disagreements, the third author of the review was consulted.
The data from the studies selected for the final sample were independently extracted by the two reviewers and then compared. A standardized clinical form created in Microsoft Excel ® was used and, finally, the data collected were compiled into a table. The data collected were the following: study characteristics (author, year, country, objective); population characteristics (total sample, Rev. Latino-Am. Enfermagem 2022;30:e3525. gender, type of diabetes and mean age), characteristics of the intervention and control groups (place of sample recruitment, duration in months, number of participants in each group, description of the intervention and of the control), characteristics of the result (outcomes evaluated, main outcomes) and conclusion. Afterwards, the results of collection were compared, discussing what was really relevant with the third reviewer. If there was information that was lacking, ambiguous, incomplete or not described in the primary studies, efforts were made to contact the authors.

The Preferred Reporting Items for Systematic
Reviews and Meta-Analyses (PRISMA) flowchart was used to document selection of the studies.

Risk of bias analysis in the individual studies
The risk of bias assessment of the studies included was analyzed using the Cochrane Collaboration Risk of Bias Tool (RoB 2) for randomized studies (27) . RoB 2 includes judgments about random sequence generation, allocation concealment, participant and staff blinding, results evaluator blinding, incomplete results data, selective reporting, and other sources of bias. The studies were categorized into groups labeled as low risk of bias, uncertain risk of bias, or high risk of bias. RoB 2 has the novelty of considering that it is not always possible to blind the participants. We should only lower the quality of the evidence if lack of blinding affects the interventions in an unbalanced way between the groups and if it affects outcomes (27) . Risk of bias was performed by two independent reviewers. In case of disagreements in the judgment of the two reviewers, an evaluation by the third reviewer was requested.
To summarize the risk of bias analysis, the Cochrane Collaborations's Review Manager 5 (RevMan 5.4) tool was used and allowed for the elaboration of the risk of bias summary (28) .

Data synthesis
A descriptive synthesis of the characteristics of the studies was carried out. The quantitative synthesis of the data was performed in a meta-analysis of studies considered combinable and homogeneous in relation to interventions and results. Cochrane Collaboration's Review Manager ® 5 (RevMan 5.4) (28) was used to perform the meta-analysis, presented using a forest plot graph. The heterogeneity present in the studies was evaluated by the I2 statistical test of inconsistency indices, considering values of 0-30% for unimportant heterogeneity, 31-50% for moderate heterogeneity, 51-80% for substantial heterogeneity and 81-100% for considerable heterogeneity (29) . The results were presented with 95% Confidence Intervals (95%CI). The random effect model was chosen (29) to perform the meta-analysis. As a measure of effect, for binary outcomes (the result can be 1 of 2 possibilities) the Risk Ratio (RR, or relative risk) was used, considering in this study the occurrence or not of diabetic foot ulcer. As for the estimate of the effect of the individual studies, this was calculated using the Mantel-Haenszel statistical method, ideal for studies with small sample sizes.

Evidence certainty classification
This research used Cochrane's Grading of Recommendations Assessment, Development and Evaluation (GRADE) system to classify the certainty of the evidence (30) . The criteria evaluated were study design, risk of bias, inconsistency, indirect evidence, imprecision, and other considerations. The certainty of the evidence can be characterized as high, moderate, low or very low. The construction of a funnel chart to assess the presence of publication bias was waived because the few randomized clinical trials were less than ten.
A table called "GRADE Evidence Profile" was created to summarize the findings of this evaluation. In this process, the GRADEpro GDT (Guideline Development Tool) online tool was used, which is freely accessible from www. gradepro.org. After full-reading, 38 did not meet the eligibility criteria, leaving 5 studies (Figure 1).

Characteristics of the sample Intervention Group/Control Group Main results
Lavery, et al. (11) 2004. United States.

Risk of bias assessment
The data were analyzed in Review Manager 5.4 (28) . Figure 3, only one study has uncertain risk of bias for "random sequence generation" (11) . Regarding the "allocation sequence", one study presents a high risk of bias, as it does not mention the number of participants allocated to each group (16) . In turn, two of the studies have uncertain risk of bias, as they present insufficient information (11,32) . One study presented an uncertain risk of bias for the assessment of the "selective reporting" criteria (16) .  identified (11,(16)(17)(31)(32) . In a study (16) , the number of participants randomized to the intervention group (thermometry) and

As shown in
control group (standard health care) was not mentioned and, therefore, it was not included in the meta-analysis.
In this meta-analysis, four RCTs (n=547) were included, as shown in the forest plot graph (Figure 4), and the prevention of the incidence of diabetic foot ulcers outcome was analyzed, presented as a binary outcome.

Evidence certainty classification
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system (30) was used to assess the certainty of the evidence and results presented in the GRADE evidence profile.   Rev. Latino-Am. Enfermagem 2022;30:e3525.
Previous systematic reviews on the subject matter have already been published; however, there were inconsistencies regarding the methodological quality of the study included in the meta-analyses (20)(21) . The RCT included in the meta-analyses does not specify how many individuals were allocated to the intervention group (use of thermometry) and to the control group (standard health care) (16) . Thus, failures to provide a reliable numerical summary of the intervention's effect are observed. In addition to that, searches in data sources and comprehensive evaluation of new evidence were necessary to provide reliable and robust information. to be at risk of ulceration due to inflammation at the measurement site. Thus, the patients were advised to contact the study coordinator and reduce their activity until temperature normalized. In addition to that, they were required to make diary entries about foot temperature observations.
When evaluating the outcome (incidence of foot ulcers), it was observed that in three studies there was an association of temperature monitoring with a reduction in the ulceration rate, suggesting that the thermometry used by the patients assists in the early identification of plantar inflammation before skin rupture and DFU formation (11,(16)(17) . On the other hand, in two studies no differences were detected between the control and intervention groups (31)(32) .
In the RCT conducted by researchers from the United States, DFU incidence during a 15-month evaluation in the usual health care group was 29.3% (17) . In contrast, in the group that monitored foot temperature at the same anatomical points daily, the incidence was 8.5% (OR 4.48; 95% CI 1.53-13.14; p<0.008).
In another study conducted by the same group of researchers, there were complications related to Charcot arthropathy (n=2) (11) . However, it is recognized that active Charcot arthropathy has an individual effect on foot skin temperature. Therefore, the data from the Charcot arthropathy cases were excluded from the analysis. Of Temperature measurements can be easily performed by the patients or family members and can represent an effective adjuvant in DFU prevention. In addition to that, they provide quantifiable information that shows inflammation formation in specific foot regions so that preventive measures can be taken before skin ruptures.
It was shown that 12.2% of the 115 participants in the standard care group had ulcers and only 4.7% of the 110 individuals in the thermometry group did so (16) .
Handheld thermometers can show positive results when used by high-risk patients to prevent ulcerative processes.
This may offer an additional advantage to conventional DFU prevention practices and therapies.
Half of the 10 people in the standard care group presented ulcerations and 39% of the 21 individuals in the temperature monitoring group presented the same outcome, with no significant intergroup differences (p=0.532) (32) . Although there were no intergroup differences in DFU recurrence, thermometry proved to be feasible for the patients.
In when the participants reduced their activity when an inflammation point was identified, the intervention proved to be effective over usual care (31) .
On the other hand, ensuring that the patients use the thermometer daily at their homes and refrain from all daily activities when the temperature of their feet is high is a potential challenge. Moreover, in the real world, adherence to the device for long periods of time can be lower than those recorded in this study.
The summary of the effect of the intervention points to a reduction in the number of ulcers in the group using foot skin thermometry, when compared to the standard care group (RR=0.53; 95%CI 0.29-0.96; I²=55%; p=0.08). Although statistical heterogeneity is substantial (I²=55%), it should be noted that it is not statistically significant.
The number of patients needed to be treated to prevent the appearance of a new skin ulcer was 8.
This metric has been valuable in the clinical practice, especially in the selection of therapeutic interventions.
It also has the potential to be used as a support tool in risk-benefit assessments and to help in health-related decision-making (33) . Thus, skin thermometry represents an intervention that is easy to apply and its preventive role can contribute to the reduction of the high health costs and serious complications, such as hospitalizations, amputations and deaths.
Estimating the efficacy of using thermometry in DFU prevention is supported by certainty of moderate evidence.
to lower the certainty of the evidence when the CI overlaps the null done line or the total number of events is less than 300. It should be noted that the estimates from the four studies included favor the intervention and that there is some overlap in the confidence intervals. In this case, it is not justified to lower the certainty of the evidence. However, the number of events is far from the recommended "optimal information size" (34) . New research studies on thermometry as a tool for the thermal assessment of the feet in people with diabetes are underway, and they can be followed-up via the https:// www.clinicaltrials.gov/platform.

Conclusion
It was evidenced that the use of infrared