We performed this systematic review in accordance with the Cochrane Handbook for Systematic Reviews of Interventions33 and followed the tutorial of the Brazilian Journal of Physical Therapy.34 The protocol of this systematic review was registered at PROSPERO (CRD42013005068), and it can be assessed online.35

We included randomized controlled trials (RCT) evaluating the effect of MIRE on plantar tactile sensitivity and/or neuropathic pain in the lower limbs in patients with DSPN. Trials should have at least one intervention group and one comparison group. MIRE was defined as if delivered through LED with wavelengths in the infrared or near-infrared range (750–1300nm).20 The comparison group should not have been exposed to MIRE. Our primary outcome was plantar tactile sensitivity assessed by Semmes–Weinstein monofilament (SWM). Secondary outcome was neuropathic pain in the lower limbs. The exclusion criteria were intervention performed in other regions than feet and follow-up intervention fewer than three days.

We searched in the electronic databases MEDLINE (accessed by PubMed), Cochrane CENTRAL, and Google Scholar and in EMBASE (up to September 30, 2016). The search terms included MeSH and entry terms related to diabetic neuropathy, infrared-ray, and a methodological filter for RCTs.36 Outcome terms were not included to enhance the search sensitivity. There were no restrictions regarding language or publication date. The terms were adjusted to fit the requirements of each electronic database. We screened the list of references of included studies in order to identify additional RCTs. The complete search strategy applied on PubMed database can be assessed online in the systematic review protocol.35

Two reviewers (CCR, PSK) separately and independently screened the titles and abstracts of studies identified in the initial search. A standard screening checklist based on the eligibility criteria was employed for each study. Studies not meeting eligibility criteria according to titles or abstracts were excluded. Full texts of the remaining studies were retrieved for a second review independently by the two reviewers. For studies without sufficient information to evaluate the eligibility criteria, we contacted the authors by email to obtain additional information.

We extracted the following data from the included studies: description of study participants, intervention in the experimental group and in the control group, and description and results of the outcomes measurements. Two review authors (CCR, PSK) independently extracted the data from the eligible studies. Discrepancies were solved through discussion and a third author (CS) was consulted when consensus was not reached. We contacted the authors by email to obtain additional information when studies presented incomplete descriptions. Procedures for estimation of missing data33 were performed when possible. If data was still insufficient after these processes, the outcome was included in descriptive analysis only.

Two review authors (CCR, PSK) independently assessed the risk of bias of the included studies using the Cochrane Risk of Bias Tool.33 The following items were evaluated for each study: selection bias (random sequence generation and allocation concealment), performance bias (blinding of patients and investigators), detection bias (blinding of outcome assessors), attrition bias (description of losses and exclusions), and reporting bias (selective reporting). We used the grading of recommendation, assessment, development, and evaluation (GRADE) system to evaluate the quality for the body of evidence.37

In order to standardize units of measurement, the average insensitive plantar regions to the 5.07 SWM were converted to average sensitive plantar regions. Since the two groups in the study of Leonard et al.13 were composed of different individuals, they were included independently in the meta-analysis. Data imputed in the analysis of Swislocki et al.31 was the average effect between feet as originally the study assessed plantar sensitivity for the left and the right foot separately. Pain scale of one study29 was converted from the range 0–100 to 0–10.

For quantitative synthesis, results were pooled using mean difference (MD) when appropriate. As plantar tactile sensitivity measurement was inconsistent across trials, data was pooled using standardized mean difference (SMD). The change value, post minus pre intervention, was used for all the analysis. For meta-analysis, we pooled random-effect model, with DerSimonian and Laird as variance estimator.

Statistical heterogeneity of the effects across studies was assessed using Cochran's Q test and Inconsistency I2 test. Values above 25% and 50% were considered as indicatives of moderate and high heterogeneity. To explore the statistical heterogeneity, subgroup analyses were carried out considering the type of comparison (no-intervention or sham) and length of follow-up intervention: short-term follow-up (up to two weeks) or long-term follow-up (more than two weeks). A p value ≤0.05 was considered statistically significant. All analyses were conducted using Review Manager, version 5.3.

The search strategy yielded 2549 registers, 23 of which were considered as potentially relevant and were retrieved for detailed analysis. Fig. 1 shows the inclusion processes. Six studies, with a total of 304 patients with DSPN, met the eligibility criteria for the systematic review. Some studies randomized feet for plantar sensitivity outcome (total of 594 feet). Table 1 summarizes the characteristics of these studies.

Figure 1.

Flow diagram of the studies included.

(0.22MB).

Five studies13,27–29,31 compared MIRE to sham intervention, and one study30 compared MIRE to no intervention. Regarding the sham comparisons, two studies13,27 considered the foot as the unit of randomization, and three studies28,30,31 randomized the patients. MIRE wavelengths ranged from 870 to 890nm with an energy density ranging from 1.30 to 1.95Joulescm2min−1. Follow-up length ranged from 1.531 to 12 weeks29; median follow-up was four weeks. Session duration ranged from seven to 40min/day. Interventions occurred in outpatient environment.

All of the included studies showed adequate sequence generation, but none reported information about the allocation concealment. Out of six studies, one30 did not blind patients and investigators and 50% reported blinding of outcome assessors. Selective reporting was identified in one study that performed analysis per protocol,29 however 60% of studies did not report the type of analysis. Description of losses and exclusions were reported in 50% of the studies (Table 2). Quality of evidence, rated using GRADE, ended with low-quality evidence for both the outcomes. The evidence profile is presented in Table 3.

There was limited evidence that MIRE increased plantar tactile sensitivity on the first two weeks of intervention and that such effect was not sustained over time. This short-term effect is expected, given the natural history of DSPN featured by elevated thresholds of vibration and thermal perception that evolves to tactile sensory loss due to degeneration of all types of peripheral nerves.38 In addition, initial improvement of nerve fiber function by MIRE might have been responsible for enhancing the sensitivity of sensory pain nerve fibers, which in turn can be related to the lower decrease in pain relief in patients receiving MIRE compared to those received sham intervention. Management of pain is challenging in DSPN as decreasing of symptoms is associated with receptor degeneration and worsening of sensory function.39

Furthermore, a placebo effect cannot be ruled out. MIRE resulted in larger effects on pain relief and plantar tactile sensitivity in Al Enalzi,30 the only non-blinded study controlled by a non-intervention group, in comparison with all of the other sham-controlled blinded studies. It is important to clarify that sham intervention would avoid the placebo effect of MIRE and adequate blinding would prevent measurement bias, critical for subjective outcomes such as pain and sensitivity.40 There are no significant differences in patient characteristics or in MIRE treatment that could explain the larger effect found in Al Enalzi30 other than the non-blinded, non-intervention comparison. Thus, we assumed that this was the main bias related to the effect size found in this study.

Plantar sensitivity assessment by SWM is a good screening test for DSPN and it is useful for risk prediction for foot ulceration.38 On the other hand, this measure might not be quite as accurate to detect changes in sensory nerve fiber function over time. Nevertheless, the identified studies used this measure as a primary endpoint for assessment of sensitivity restoration. Accurate measures for nerve function, such as nerve conduction velocity and quantitative sensory testing, should be used as primary outcomes measurement in further studies.

The visual analog scale is one of the oldest, easiest, and best validated tools to assess pain but evaluation of neuropathic pain requires the understanding of the possible negative (sensory loss) and positive (pain and paresthesia) symptoms and signs.41 Only Swislocki et al.31 assessed quality of pain (by using the Neuropathic Pain Scale) and interestingly showed a decrease in tingling, cramping, and itchy symptoms in patients who received MIRE even without difference in pain intensity. In order to understand the effect on neuropathic pain better, further studies should detail its assessment by using specific scales.

From our knowledge, this is the first systematic review presenting pooled-effects of MIRE for patient-important outcomes. Pain and sensitivity are directly related to quality of life in patients with diabetes4; furthermore, loss of protective foot sensation is associated with increased risk of lower limb ulceration and amputation.5 A former qualitative review42 reported effects of MIRE in these outcomes but included other types of studies. Our systematic review included RCTs identified through an extensive database source without limitation of language and time of publication. In addition, we incorporated a comprehensive assessment of bias of individual studies and quality of evidence for each outcome using GRADE.

Although we conducted the systematic review following methodological standards, the presented results have some limitations. We did not use specific gray literature databases to search for studies beyond Google Scholar, which may have affected the inclusion of all eligible studies available. Regarding data analysis, although some of the studies had within-subject comparisons (feet were the unit of randomization, not patients), we used the change difference instead of post-intervention values to perform the meta-analysis. This conservative estimate was preferred in order to deal with the baseline differences found in some of the studies.