Winemiller MH, Billow RG, Laskowski ER, Harmsen WS. Effect of Magnetic vs Sham-Magnetic Insoles on Plantar Heel PainA Randomized Controlled Trial. JAMA. 2003;290(11):1474–1478. doi:10.1001/jama.290.11.1474
Author Affiliations: Department of Physical Medicine and Rehabilitation (Drs Winemiller and Laskowski) and Section of Biostatistics (Mr Harmsen), Mayo Clinic, Rochester, Minn; and Northwest Orthopaedic Surgeons, Mount Vernon, Wash (Dr Billow).
Context Despite anecdotal reports, rigorous scientific evidence of the effectiveness
of magnetic insoles for the pain of plantar fasciitis is lacking.
Objective To determine whether magnetic insoles provide greater subjective improvement
for treatment of plantar heel pain compared with identical nonmagnetized insoles.
Design, Setting, and Participants Randomized, double-blind, placebo-controlled trial conducted from February
12, 2001, to November 9, 2001, of a volunteer sample of 101 adults with diagnoses
of plantar heel pain for at least 30 days from a multispecialty group practice
clinic in Rochester, Minn. Daily pain diaries were kept for 8 weeks.
Interventions Cushioned insoles, with either active bipolar magnets or sham magnets,
which were worn daily by the participants for 8 weeks.
Main Outcome Measures Reported average daily foot pain (by metered visual analog scale [VAS]
and by categorical response of change from baseline) at 4 and 8 weeks, and
impact of insoles on employment performance and enjoyment.
Results No significant between-group differences were found on any outcome variables
studied when comparing active vs sham magnets. Both the nonmagnetic and magnetic
groups reported significant improvements in morning foot pain intensity, with
mean (SD) VAS scores improving from 6.9 (2.3) and 6.7 (2.0), respectively,
at baseline to 3.9 (2.6) for each group at 8 weeks (P =
.94). At 8 weeks, 33% of the nonmagnetic group and 35% of the magnetic group
reported being all or mostly better (P = .78). At
baseline, foot pain interfered moderately with participants' employment enjoyment
(mean VAS, 4.2) and improved in both groups by 8 weeks (1.3 and 1.5, respectively; P = .68).
Conclusion Static bipolar magnets embedded in cushioned shoe insoles do not provide
additional benefit for subjective plantar heel pain reduction when compared
with nonmagnetic insoles.
The use of magnetic fields for pain relief has increased dramatically
in the past decade. Despite a paucity of scientific evidence and lack of approval
by the US Food and Drug Administration, a large number of people use magnets
to relieve their pain. An estimated $5 billion has been spent worldwide on
magnetic devices purchased to treat pain,1,2 with
annual US sales estimated at $500 million.3 The
vast majority of devices available to and used by the general public are static
magnets, typically ranging from 200 to 2500 G. Such magnets are generally
considered safe when applied to the skin and have few adverse effects.4,5 The physiologic effects of static magnets
on pain, however, are largely unknown.6- 9
Although some studies have found that bipolar magnets can relieve various
sources of pain,10- 13 all
had significant methodological flaws. Several other investigations have failed
to show additional benefit from static magnets.14- 20
Plantar heel pain, commonly referred to as plantar fasciitis, is a common
condition among athletes as well as the general population.21- 23 The
characteristic complaints are knife-like pain at the calcaneal insertion of
the medial plantar fascia, typically worse on first arising in the morning,
and often lasting months to years.23 Many treatment
regimens exist but effectiveness is variable.24- 26 Insole
materials have generally been found to be effective in relieving signs and
We identified only 1 study in the medical literature evaluating the
effectiveness of magnets for plantar heel pain.16 This
study found no difference in the effectiveness of a firm insole compared with
an insole imbedded with a ferromagnetic foil, although there were several
methodological limitations.25 We report a randomized,
double-blind, placebo-controlled trial to assess the effectiveness of bipolar
static magnets in insoles for treatment of plantar heel pain.
Participants were recruited between February 12, 2001, and November
9, 2001, via referrals from clinics in the department of physical medicine
and rehabilitation at the Mayo Clinic, Rochester, Minn, and from verbal and
posted advertisements. The advertisements solicited adults with "foot pain
for at least 1 month that is present more days than not," aggravated by standing
or walking. Each respondent's eligibility was assessed via screening questions
and a physical examination completed by 1 of 2 authors (R.G.B., M.H.W.). Inclusion
criteria included 18 years or older; foot pain for at least 30 days occurring
more days than not; foot pain intensity 3 or higher on a 10-point visual analog
scale (VAS); maximal tenderness on palpation of the medial plantar fascia/medial
calcaneus; sharp, shooting, or localized pain in the plantar aspect of the
foot; pain exacerbated by standing, walking, or on first arising in the morning.
All potential participants underwent ankle examination and were excluded
if they had evidence of chronic instability, ligament tenderness, cuboid syndrome,
peroneus longus tendinitis, plantar nerve entrapment, or stress fracture.
Participants were excluded for any neurologic deficit involving the lower
extremities. Despite no documented adverse effects from static magnets, women
likely to be pregnant (self-report of absent menses in the previous 2 months
in premenopausal women) and individuals with electromagnetically activated
implants were excluded. Approval from the departmental research committee
and the institutional review board was obtained before initiating the study.
The insoles were the Spenco "Active Comfort" magnetic insole (Spenco
Medical Corp, Waco, Tex), which have a magnetic foil imbedded in foam under
the proximal arch of the foot. The magnets have a bipolar multiple circular
array, with internal magnetization to 2450 G. We verified surface magnetic
field strength in a random subset of study insoles by using a hand-held gaussmeter
(Lake Shore 410 Gaussmeter, Lake Shore Cryotronics Inc, Westerville, Ohio).
The sham-magnetic insoles were identical to the active insoles, but were specially
made by the manufacturer with the same metal foil, but in a nonmagnetized
All insoles (active and inactive) were provided by the manufacturer
to the investigators at no charge and in a blinded fashion by using a random
tracking code. The insole pairs were physically mixed in a box and were completely
indistinguishable by appearance, touch, and location within a large box containing
116 magnetized and 117 nonmagnetized insole pairs. The investigators conducting
enrollment, randomization, and insole distribution were blinded as to which
type of insole was used. Ongoing blinding was encouraged by requiring all
participants to sign an agreement to refrain from covert attempts to determine
whether the insoles they were given had active magnetic properties.
Written informed consent was obtained for all eligible participants.
Randomization occurred when the investigator stirred the unsorted insole pairs
within the box and randomly chose a pair. Insoles were trimmed and placed
in the participant's primary pair of shoes, with instructions to wear them
for at least 4 hours per day, 4 days per week for 8 weeks. If different shoes
were to be worn on a given day, participants were asked to transfer the insoles.
Data were collected by questionnaire at baseline and 4 and 8 weeks, including
pain intensity (10-cm metered VAS for morning, evening, and mean daily pain),
categorical response to treatment (5-point Likert scale), adverse effects,
and subjective pain-related interference with employment performance and enjoyment
(10-cm VAS). Finally, participants were asked if they thought magnets have
significant potential to relieve pain. Telephone reminders were used as necessary
to encourage the return of questionnaires.
Participants kept diaries for 8 weeks, recording the number of hours
insoles were worn and rating pain intensities daily. Participants were provided
the insoles to keep at no charge and were given a $10 remuneration at the
end of the study. No additional interventions were provided or suggested.
Participants were discouraged from changing any current treatment regimen
during the study period.
The primary outcome variables were the 4- and 8-week categorical responses
to treatment (all/mostly better vs somewhat better/unchanged/worse), as well
as VAS scores. Based on historical outcomes,31 we
enrolled 82 participants to achieve 80% power to detect a categorical outcome
difference of 30% between groups, with a significance level of .05. A 2-tailed
alternative hypothesis was conservatively used a priori given the possibility
that the magnets could have a poorer outcome than the controls, although no
such reports are in the literature. An additional 19 participants were enrolled
to maintain statistical power with anticipated participant dropout. All primary
outcome analyses were performed according to the intent-to-treat principle.
The proportion of participants who reported being all or mostly better was
compared by using the χ2 test. We also compared outcomes between
participants who did vs did not believe in the potential of magnets to relieve
pain. Two-group comparisons were performed at baseline to determine group
comparability by using the nonparametric Wilcoxon rank sum test or Fisher
exact test. SAS version 8.2 (SAS Institute Inc, Cary, NC) was used to analyze
all data. All comparisons were considered significant at P<.05.
A total of 101 participants were enrolled in the study (80 women and
21 men). Six participants did not complete the study (Figure 1). All baseline characteristics and treatment regimens were
statistically comparable between groups, except sex, proportions with worsening
pain, and use of leg elevation for relief (Table 1). However, subsequent adjustment for these variables using
multiple logistic regressions did not alter any of the outcome results (data
not shown). Occupations included secretary/data entry (n = 21), nursing (n
= 20), technician (n = 16), clinical assistant (n = 9), administration (n
= 9), physician (n = 5), food service (n = 3), and miscellaneous (n = 18).
The 5 participants who were lost to follow-up did not differ from the others
in terms of age, sex, duration or intensity of pain, or proportion with worsening
pain at baseline.
A random sampling of 56 of the study insoles revealed surface readings
of 2.2 G (range, 1.2-3.1; SD, 0.45) for sham insoles and 192.1 G (range, 178-200;
SD, 8.53) for active magnetic insoles.
No significant differences were found between the magnetic and nonmagnetic
groups on any of the primary outcome variables (VAS pain or categorical response
to treatment) at baseline, 4 weeks, or 8 weeks. At baseline, most participants
reported unchanging pain, although more in the nonmagnetic group reported
actively worsening pain (32%) than in the magnetic group (13%, P = .02). At 4 weeks, 44% of the nonmagnetic group reported being all
or mostly better compared with 31% in the magnetic group (P = .19; an alternate analysis with exact permutation also showed P = .20). The nonmagnetic and magnetic groups did not differ
at 8 weeks with the proportions of 33% and 35%, respectively, of participants
saying their pain was all or mostly better (P = .78)
(Figure 2). These results were insensitive
to imputations assigning either the best or worst possible values to missing
observations in the 2 groups.
Participants in both groups had similar morning pain scores at baseline
(P = .64) and both groups reported nearly identical
improvement in morning pain at 4 weeks (P = .63),
with little change by 8 weeks (P = .94) (Table 2). Again, these results were essentially
unchanged after imputing either best or worst possible values to the missing
There were no significant differences in participants' reported ability
to do or enjoy their employment during the study period. At baseline, foot
pain interfered moderately with participants' employment enjoyment (mean VAS,
4.2) and improved comparably in both groups by 8 weeks (mean VAS improvements
of 1.3 and 1.5 in nonmagnetic and magnetic groups, respectively; P = .68). However, categorical improvement in pain at 8 weeks (all
or mostly better) did correlate with less interference in ability to enjoy
their employment when compared with their own baseline responses (P<.001).
Compliance was equivalent between the groups, with 98% (nonmagnetic)
and 92% (magnetic) of participants still wearing their insoles every or most
days at 4 weeks (P = .38), and 83% and 87%, respectively,
at 8 weeks (P = .59). Minor problems with the insoles
were reported by 27% of participants in the nonmagnetic group and 13% in the
magnetic group (P = .11) at 4 weeks, consisting primarily
of tightness of shoe with the insoles and cosmetic breakdown of the insole
surface material. No serious adverse effects were reported with any insoles,
although one participant in the magnetic group reported foot spasms and another
participant in the nonmagnetic group reported transient foot burning. Of the
95 participants with complete follow-up, 89 specifically reported beginning
no new treatments during follow-up. The other 6 were equally divided between
the 2 groups.
Participants were asked whether they thought they had received active
or inactive magnets. Forty-two percent of participants in the nonmagnetic
group and 48% in the magnetic group correctly guessed their group assignment
at 4 weeks. At 8 weeks, 53% in each group guessed correctly, suggesting that
adequate participant blinding had occurred.
There was no baseline difference between the groups in terms of the
proportion of participants who felt magnets have a significant potential to
relieve pain (P = .28). Of those participants (irrespective
of group) believing in the potential of magnets, 15 (42%) of 36 had categorical
improvement at 4 weeks compared with 17 (35%) of 48 who did not feel magnets
have a significant potential (P = .56). At 8 weeks,
these categorical improvements were 40% and 29%, respectively (P = .30). Participants believing in magnets also tended to have less
morning pain at 4 weeks (mean VAS, 4.09 vs 4.88; P =
.23) and had significantly less pain at 8 weeks (3.18 vs 4.70; P = .04).
Both groups used cushioned insoles and both reported subjective improvement
in their symptoms; however, static magnets imbedded within these insoles did
not provide additional relief. Many people use magnetic devices empirically
and without specific diagnosis, and the results of this study may be generalized
for people with common forms of plantar heel pain.
Although previous literature has not found the effectiveness of static
magnets to be related to magnetic strength, it is important to note that the
strength of magnets used in this study is comparable with widely available
devices. We investigated several other brands of insoles and magnetic shoes
available at local retailers. Using the same gaussmeter, maximum surface magnetism
of these devices ranged from 35 to 236 G (data not shown) comparable with
the 192 G in the studied insoles. Our negative results thus are not likely
due to lack of magnetic strength.
The insoles in our study contained a metal foil in the arch area, magnetized
in a bipolar multiple circular array. Other marketed insoles have varied configurations
of magnets, which may or may not have different clinical effectiveness profiles.
Pulsed electromagnetic fields, however, are less comparable and no conclusions
should be drawn from this study on their effectiveness.
Although participants had the opportunity to test their insoles for
magnetic activity, we doubt that many of them did. First, all participants
signed an agreement and verbally committed not to test their insoles. While
this could have incited curiosity, we thought the honor system would be most
effective to dissuade participants from testing their insoles. Second, participants
in both groups correctly guessed their group assignment equally at a level
no better than chance, suggesting that adequate and equal blinding had actually
The randomization in this study was based on an unpredictable physical
device (drawing from an equal assortment of indistinguishable insoles from
a large box). The unequal sizes of the treatment groups was a result of having
more insole pairs in the box than we intended to use in this study, but in
itself this assortment was a random event which did not introduce any systematic
bias or alter the results.
Comparison of baseline characteristics revealed 3 statistically significant
differences between groups. Other than possible clustering of covariates,
there are no other identified causes for this number of differences, which
would be expected about 13% of the time given the 25 comparisons in Table 1. The control group did have relatively
more worsening pain at baseline, which could have represented more active
disease. Active disease would be expected to improve more during follow-up
and therefore mask true efficacy of magnets. However, baseline pain intensities
were nearly identical between active and control groups, and the control group
actually had a slightly longer duration of symptoms, which argues against
more active disease. Finally, multivariate adjustment for baseline worsening/static,
age, and sex (data not shown) all revealed no significant changes from the
univariate results presented in our study.
There may have been minor differences in participants' compliance with
insole use but with high rates of overall compliance and the even distribution
between groups of the 6 participants who began new treatments during the study,
it is unlikely that differences in compliance alone could explain our negative
Our results only marginally support the secondary hypothesis that participants
believing magnets have a significant potential to relieve pain would be more
likely to have a response to either intervention. Because participants were
blinded, the magnitude of this placebo effect could be understated compared
with devices purchased independently and known to be magnetic. Also, individuals
volunteering for a study using magnets may have believed in the potential
of magnets more than the general population but probably less than the typical
consumer who purchases magnetic devices based on belief or personal recommendation.
Although many claims have been made regarding the therapeutic use of magnets,
our outcomes showed static magnets to be ineffective in the treatment of plantar