Electrical Stimulation and Time to Radiographic Healing of Acute Fractures: A Systematic Review and Meta-Analysis

Introduction

With approximately 18.3 million cases in the United States¹ per year, orthopedic fractures are extremely common and represent a large public health burden in terms of cost and socioeconomic impact secondary to work disruption and disability. The total economic impact of orthopedic fractures has been estimated at $265.4 billion dollars per year.¹ Further, delays in healing incur significant morbidity to the patient in the form of pain and functional limitations.² The time required for fractures to unite is important because of the resources used to care for patients until they are independent again as well as the amount of lost wages from being off work. Understanding and expediting the process of fracture healing would greatly improve patient care.

One method that has been described to augment bone healing is electrical stimulation (ES). Since Fukada and Yasuda first described the piezoelectric property of bone – that bone generates endogenous electrical fields by ionically-driven currents in response to mechanical stress – and its relationship to bone formation,³ there has been an abundance of evidence supporting the use of exogenous ES to augment bone formation in both the in vitro and in vivo animal model.^4-6 However, the clinical data in support of ES remain mixed. The lack of consensus in the literature may be due to differences in definition of radiographic union, varied fracture location and type, poor patient compliance with device utilization, or issues with delivery of ES to the fracture site.

Methods utilized to deliver ES to fractures include invasive (or semi-invasive) current, capacitive coupling, and inductive coupling. Invasive current is a method of delivery in which there is a cathode that is placed through the skin in or near the fracture site. Non-invasive ES delivers weaker, time-varied electromagnetic fields reaching the fracture site with capacitive or inductive coupling. Capacitive coupling devices use electrical fields transmitted transcutaneously though pads placed on the patient to generate current at the fracture site. Non-invasive inductive coupling ES includes pulsed electromagnetic field (PEMF) and combined magnetic field (CMF). PEMF is commonly used and described in the literature and typically involves coils placed on the patient to transcutaneously generate electromagnetic treatment fields.

There have been two recent meta-analyses studying the efficacy of ES for the purposes of bone healing in non-unions and spinal fusion,^7,8 which demonstrate that the relative risk of non-union at one year is reduced by 35% in the ES group.⁸ However, while these meta-analyses focused on studies comparing rate of non-union – a fracture that fails to heal within the period of time that is considered ‘normal’ for the particular fracture, none to date have compiled data exclusively on the effect of ES on time to healing of acute fractures. With the aim of answering the question if ES decreases the days to radiographic union of acute fractures, we performed a systematic review and meta-analysis of randomized-controlled trials with the outcome of interest being the days to radiographic union.

Methods and Materials

Utilizing the standard PRISMA protocol,⁹ a MEDLINE database search was conducted on September 25, 2020, using combinations of search terms of “electric stimulation AND bone healing”, “electric stimulation AND fracture,” “electric stimulation AND fracture healing,” “direct current AND bone healing”, “inductive coupling AND bone healing,” and “pulsed electromagnetic frequency AND bone healing.” The complete search strategy is included in Appendix 1. Additionally, all bibliographies of systematic reviews and meta-analyses on the topic were searched for articles that may have been missed in the MEDLINE search. There was no date of publication restrictions. Inclusion criteria were randomized-controlled trials (RCTs) concerning ES for the purpose of healing acute fractures of the appendicular skeleton with outcomes on radiographic union at regular time intervals. Studies involving skeletally immature patients or ES for non-unions, spinal fusions, and osteotomies were excluded. Two reviewers (P.J.N. and K.R.) agreed upon inclusion of studies for meta-analysis after full-text review. Data for article title, authors, year of publication, study type, number of patients studied, fracture type, ES type, fixation method, ES protocol, ES device specifications, radiographic union definition, radiographic interval, length of follow up for each group, number of patients lost to follow up for each group, non-unions in each group, and days to radiographic union was collected by each reviewer, and disagreements were handled through discussion or inclusion of a third reviewer (S.O.P) as a tiebreaker. For each study, the reviewers (P.J.N and K.R.) independently assessed the study quality per the Cochrane Collaboration tool for risk of bias assessment.¹⁰

The outcome of interest was time to radiographic union. This was recorded as a continuous variable in days, with means and standard deviations calculated. Non-unions requiring secondary surgery were included in the analysis as a union at the longest measured time-point in each study. For example, if a scaphoid fracture went on to non-union and required open reduction, internal fixation (ORIF) with bone grafting at 3 months, this fracture would be counted as reaching radiographic union at 1 year, which is the longest time to union recorded in the study. This was done to capture the clinically relevant non-unions in the analysis. Heterogeneity between studies was tested using both the χ² test (significance defined as p < 0.05) and the I² tests (substantial heterogeneity defined as values >50 %).¹¹ The studies were also analyzed in subgroups for upper and lower extremity location of fracture and operative – ORIF, external fixator placement, or percutaneous pinning – and non-operative – closed reduction and casting – methods of fracture fixation. The meta-analysis was run on Metafor ® by the Department of Surgery biostatistician group.¹²

Results

Our query yielded 1652 references (dates ranged: 1964-2020). After title and abstract screen, 27 studies underwent full-text analysis. Five clinical trials and 301 patients were included in the meta-analysis (Figure 1). Of these 5 trials, two were studies pertaining to fractures of the lower extremity,^13,14 and three were for fractures of the upper extremity.^15-17 There were two studies concerning non-operative fractures,^15,16 and three studies for operative fractures.^13,14,17 There were four studies that used PEMF as their mode of stimulation,^13-16 and one study used continuous current.¹⁷ Martinez-Rondanelli et al.¹⁴ and Faldini et al.¹³ utilized a removable transcutaneous PEMF device – a device that the patient applied to the affected extremity daily. Faldini et al.¹³ used stimulators with 75Hz, 1.3μs impulse duration, and 2mT peak magnetic field, where as Martinez-Rondanelli et al.¹⁴ did not report specific values. Hannamen et al. 2012¹⁶ and Hanneman et al. 2014¹⁵ used cast-implanted transcutaneous PEMF devices – a device that was not removable and was implanted in the thumb spica cast. For both studies, the stimulators were programmed for 50mV pulse amplitude, 5μs pulse width, 5ms burst width, 62ms refractory period, and 15Hz repetition rate.^15,16 Finally, Itoh et al.¹⁷ administered continuous current through the pin of an external fixator directly into the fracture site of intra-articular distal radius fractures (2Hz, 30μA, 0-60kΩ). As noted above, this study was the only one we reviewed which utilized a semi-invasive means for stimulation – that is, ES delivered directly within the fracture site. The studies analyzed, including the specific methodology of ES, are summarized in Table 1.

• Download figure
• Open in new tab

Figure 1:

Flowchart of records included in the meta-analysis. A MEDLINE search was conducted per PRISMA protocol for relevant key terms to include randomized, sham-controlled trials for ES of acute fractures with time to radiographic union as a measured outcome. After applying all review criteria, our analysis included n=5 papers, with a total 301 patients. Four studies used non-invasive methods of stimulation, and 1 study used an invasive means of delivering current directly into the fracture site through the pin of an external fixator.

The Cochrane Collaborative risk of bias assessment demonstrated excellent inter-rater agreement (κ = 0.82). Only assessment of blinding of participants and selective reporting demonstrated discrepancy (κ = 0.54). The rest of the sources of bias, random sequence generation, allocation concealment, incomplete outcome data, and other sources of bias, were in perfect agreement (κ = 1.00) (Table 2). The agreed upon ‘high’ risk of bias found was in Martinez-Rondanelli et al.¹⁴ The study exhibited risk of selective reporting in that it did not report patient compliance with the removable PEMF device, which could affect the result of the study. The reviewers agreed upon an “unclear” risk of bias for blinding of participants in Itoh et al.¹⁷ They do not clearly state whether the study was blinded to the patients, but they do say that the physicians who made the determination of radiographic union were not part of the study. Knowing which patients were treated with ES could affect their decision making for determination of radiographic union, as reading radiographs is semi-objective.

For the outcome measure of time to radiographic union, the effect of ES did not reach significance between the control and ES groups (Mean difference (MD)) = −19.01, 95% CI = - 53.88 to 15.86, I² = 81%) (Figure 2). There was also an insignificant reduction in time to radiographic union in the upper (MD = −15.86, 95% CI = −78.41 to 46.64, I² =88%) and lower extremity (MD = −13.14, 95% CI = −27.64 to 1.37, I² = 20%) subgroup analysis (Figure 3). Additionally, there was no significant difference in either the operative (MD = −39.37, 95% CI = −87.06 to 8.32, I² = 87%) or non-operative (MD = 11.21, 95% CI = −25.18 to 47.59, I² = 20%) subgroup (Figure 4).

• Download figure
• Open in new tab

Figure 2:

Forest plot of effect of ES on time to radiographic union in days with aggregate mean difference at the bottom of the figure. Only 1 of 5 studies demonstrated a significant difference in the treatment group compared to controls (Itoh et al.). Taken together, the effect of ES was determined to be not statistically different from controls.

• Download figure
• Open in new tab

Figure 3:

Forest plots of the upper and lower extremity subgroups of the analysis with aggregate mean difference below each Forest plot. Neither subgroup analysis reached significance.

• Download figure
• Open in new tab

Figure 4:

Forest plots of the operative and non-operative subgroups of the analysis with aggregate mean difference below each Foster plot. Neither subgroup analysis reached significance.

Discussion

There were significant heterogeneities (I² > 50%) reported in the overall meta-analysis (I²= 81%) and the upper extremity (I² = 88%) and operative (I² = 87%) subgroups. The high heterogeneity is likely due to the fact that some fractures healed without issue within a period of time that was considered normal for that particular fracture. However, some fractures in both the control and treatment arms of the included studies exhibited delayed healing attributable to patient factors like fracture severity, nicotine use, and corticosteroid use. This is not to say that these factors, which caused some patients in each group to heal slower than normal, affected the results of the studies because Hanneman et al. 2012 and 2014^15,16 and Martinez-Rondanelli et al.¹⁴ have tables demonstrating that the rate of comorbidities known to affect fracture healing are similar between the treatment and control groups.

While Hanneman et al. 2012¹⁶ used a cast-implanted PEMF device study for treatment of scaphoid fractures and demonstrated an insignificant increase in time to healing in their treatment group (MD = 35.00, 95% CI = −19.12 to 89.12), the paper raised concerns about the reproducibility and inter-rater variability of radiographs for detecting union of the scaphoid.¹⁶ For this reason, this group repeated a similar study using computed tomography scans instead of standard radiographs for evidence of union. This repeated study demonstrated an insignificant decrease in time to radiographic healing (MD = −3.28, 95% CI = −42.78 to 36.22).¹⁵ These findings demonstrate the importance of consistent radiographic endpoints of union and sensitivity of imaging modalities.

The only study with statistically significant data for time to radiographic healing was Itoh et al.¹⁷ that used a 2Hz alternating current through a semi-invasive method of delivery (MD = - 69.30, 95% CI = −94.03 to −44.57). As stated previously, Itoh et al.¹⁷ used a spanning fixator for intra-articular distal radius fractures with a pin into the radial styloid, delivering alternating current directly into the fracture site.

Studies investigating removable PEMF devices like the Faldini et al.¹³ were limited by patient compliance issues with 53.3% of the treatment group reportedly using the transcutaneous PEMF device the prescribed 8 hours per day for 13 weeks. However, Hanneman et al.^15,16 implanted the PEMF device within the cast for the non-operatively managed scaphoid fractures, eliminating patient compliance issues, yet there was still no significant reduction in time to healing.

In light of the robust data in in vitro and animal studies demonstrating augmentation of bone formation with electrical stimulation and the clinical success of the Itoh et al. study where an invasive method of stimulation was used,^4-6,18 one explanation for the results of this meta-analysis could be issues with delivery of the stimulation to the fracture site. The majority of animal studies to date use an invasive means of delivering current, and the most common method of ES in human studies is PEMF.⁶ In our meta-analysis, the study utilizing invasive means of continuous current directly into the fracture site was the only study that demonstrated significant reduction in time to radiographic union (MD = −69.30, 95% CI = −94.03 to −44.57).¹⁷

One possible explanation for the inconsistent results in human clinical trials for ES compared to results in animals and cell cultures is a function of scale. In vitro and animal studies have demonstrated that proliferation of bone/osteoblast cells has a dose-dependent response to the application of a capacitively coupled electric field. Greater electrical field strengths cause a greater proliferation of cells.^19-22 However, the fall off of an electric field from the signal source is very steep, ∼1/r² where r is the distance from the electrodes.²³

Similarly, distance to target impacts the resolution of a pulse electromagnetic field into current within tissues. The difference in distance and coverage of the electric field at the fracture site would differ greatly between humans and small animal studies based on the width of the bone and thickness of soft tissue.²⁴ This may also explain why implanting an electrode did significantly decrease the time to radiographic union, which would not be subject to the same issues of scale.

Additionally, special care must be taken when calibrating the PEMF device to ensure that the fracture site is being exposed to the expected intensity of magnetic and electric field. Lunt²⁴ highlights the many variables present that may alter the current generated at the fracture site. In addition to the width of the bone, the difference in resistivity of air, soft tissue, and bone causes a significant fall off in current generated in the fracture site when compared to theoretical and laboratory models.²⁴ This raises another issue which may explain the negative results in the PEMF stimulator studies like the ones included. For example, another randomized, sham-controlled trial found that PEMF device had no effect on the rate of revision surgery for delayed union or non-union tibial shaft fractures.²⁵ In this study, the PEMF device was strapped onto the exterior of either a plaster or fiberglass splint by the patient. There was no mention of specific calibration for each patient in the device preparation protocol. According to Lunt’s calculations,²⁴ altering the geometry, thickness, and resistivity of the splint material, patients’ soft tissues, and bone would proportionally affect the current generated within the fracture site.

Therefore, the expected current within the fracture site is not as predictable as one would assume. While invasive methods of ES are not feasible in many clinical scenarios due to a significant infection risk,²⁶ more research at the basic science and clinical levels is necessary to develop an effective methodology and paradigm for stimulating the fracture site in vivo.

One potential limitation of our study is that while Itoh et al.¹⁷ described more robust callus formation and improved radial height in the ES group, they used time of external fixator removal as a surrogate for time to radiographic union. The timing of external fixator removal can be affected by surgeon availability and operating room scheduling and could therefore alter the results of the study. Additionally, the study kept the external fixators in place for 39 and 48 weeks in the control and treatment arms, respectively. These are well above the normal time of external fixation for these types of fractures at our institution, which would typically be on the order of 4 to 8 weeks. And so, there is some question of the external validity of these results when applied to practice patterns within the United States. Finally, it is difficult to compare the results of the included studies as they apply ES with drastically different ES protocols and specifications. For example, the Hanneman et al. studies^15,16 used PEMF stimulators that were kept on continuously while the cast was in place. In contrast, Martinez-Rondanelli et al.¹⁴ started ES by means of a PEMF stimulator 6 weeks after fixation for 1 hour per day for 8 weeks. This difference emphasizes that further studies are required to determine the optimal protocol for ES for the purpose of acute fracture healing.

No statistically significant effect was demonstrated by ES for the purposes of decreasing time to radiographic union in acute fractures. More robust RCTs with specific methodologies to address issues with patient compliance with clear and reproducible definitions of radiographic union are needed. Additionally, new developments in methods of safe and efficacious delivery of ES to fracture sites may show more robust results.