Study

First author, year

Appropriate and clearly focused question?¹

Yes/no/unclear

Comprehensive and systematic literature search?²

Yes/no/unclear

Description of included and excluded studies?³

Yes/no/unclear

Description of relevant characteristics of included studies?⁴

Yes/no/unclear

Appropriate adjustment for potential confounders in observational studies?⁵

Yes/no/unclear/notapplicable

Assessment of scientific quality of included studies?⁶

Yes/no/unclear

Enough similarities between studies to make combining them reasonable?⁷

Yes/no/unclear

Potential risk of publication bias taken into account?⁸

Yes/no/unclear

Potential conflicts of interest reported?⁹

Yes/no/unclear

Stein 2013

Yes

Yes

Unclear*1

Unclear*1

Not applicable

Yes*1

No*1

No*1

No*1

Davies 2015

Yes

Yes

Unclear*2

Yes

Not applicable

Yes

No*2

No*2

Unclear*2

Study reference

Describe method of randomisation¹

Bias due to inadequate concealment of allocation?²

(unlikely/likely/unclear)

Bias due to inadequate blinding of participants to treatment allocation?³

(unlikely/likely/unclear)

Bias due to inadequate blinding of care providers to treatment allocation?³

(unlikely/likely/unclear)

Bias due to inadequate blinding of outcome assessors to treatment allocation?³

(unlikely/likely/unclear)

Bias due to selective outcome reporting on basis of the results?⁴

(unlikely/likely/unclear)

Bias due to loss to follow-up?⁵

(unlikely/likely/unclear)

Bias due to violation of

intention to treat analysis?⁶

(unlikely/likely/unclear)

Stein 2013: included RCTs

TENS vs placebo

Cheing 2005

matched by age, ..and randomly assigned by drawing lots

Unclear

Unclear

Unclear

Unclear

Unlikely

Unlikely

Unlikely

Forst 2004

Generation of random sequence

Unlikely

Unclear

Unclear

Unclear

Unlikely

Unlikely

Unlikely

Gossrau 2011

patients were randomized

Unclear

Unlikely

Unclear

Unlikely

Unlikely

Unlikely

Unlikely

Kumar 1998

randomly assigned in a

single-blind fashion

Unclear

Unclear

Likely

Unclear

Unlikely

Unclear

Unclear

Kumar 1997

Participants were randomly assigned, in a

single-blind fashion

Unclear

Unclear

Likely

Unclear

Unlikely

Unclear

Unclear

TENS vs HF

Reichstein 2005

Participants randomly assigned using strata for non-painful and

painful neuropathic symptoms

Unclear

Likely

Likely

Likely

Unlikely

Unlikely

Unlikely

FREMS vs placebo

Bosi 2005

Randomisation was performed centrally

Unlikely

Unclear

Unclear

Unclear

Unlikely

Unlikely

Unlikely

PENS vs placebo

Hamza 2000

patients were randomly

Unclear

Likely

Likely

Likely

Unlikely

Unclear

Unclear

Electromagnetic fields vs placebo

Fezyioglu 2010

95 patients included, 50 patients were randomized; generation of random sequence

Unclear

Unclear

Unclear

Unclear

Unlikely

Unlikely

Unclear

Weintraub 2003

Randomized via computer assignment

Unlikely

Unlikely

Unlikely

Unlikely

Unlikely

Unclear

Unclear

Weintraub 2009

Randomized via computer assignment

Unlikely

Unlikely

Unlikely

Unlikely

Unlikely

Unlikely

Unlikely

Wróbel 2008

Patients were randomized

Unclear

Unlikely

Unlikely

Unlikely

Unlikely

Unclear

Unlikely

Update Stein 2013: recent relevant RCTs

FREMS vs placebo

Bosi 2013*1

Central computer random number generator

Unlikely

Unclear*1

Unlikely

Unclear*1

Unlikely

Unlikely

Unlikely

TENS vs PRF lumbar sympathectomy (minimally invasive procedure)

Naderi Nabi 2015*2

Randomly divided by triple blocks

Unclear*2

Unclear*2

Unlikely

Unclear*2

Unlikely

Unclear*2

Unclear*2

Internal neurostimulation (SCS)

Slangen 2014

van Beek 2015

computerized randomization

Unlikely

Unclear*3

Unlikely

Unclear*3

Unlikely

Unlikely

Unlikely

de Vos 2014

Duarte 2016

block stratified randomization

Unlikely

Unclear*4

Unlikely

Unclear*4

Unlikely

Unlikely

Unlikely

Surgical decompression

van Macaré van Maurik 2014, 2015

Web-based randomization system

Unlikely

Unclear*5

Unlikely

Unclear*5

Unlikely

Unlikely

Unlikely

Davies 2015: included RCTs

Psychological coping

Otis 2013

flip of a coin

Likely*6

Likely*6

Unlikely

Likely*6

Unlikely

Unclear*6

Unclear*6

Teixera 2010

random number assignment

Unclear*7

Unclear*7

Likely*7

Unclear*7

Unlikely

Unlikely

Unlikely

Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

External neurostimulation (and minimally invasive procedures)

Stein 2013

SR and meta-analysis of RCTs

Literature search up to April 2012

TENS vs placebo

A: Cheing 2005

B: Forst 2004

C: Gossrau 2011

D: Kumar 1998

E: Kumar 1997

TENS vs HF

F: Reichstein 2005

FREMS vs placebo

G: Bosi 2005

PENS vs placebo

H: Hamza 2000

[Not part of the research question: Electromagnetic fields vs placebo

I: Fezyioglu 2010

J: Weintraub 2003

K: Weintraub 2009

L: Wróbel 2008

Study design: RCT parallel / cross-over

Inclusion criteria SR: RCTs; electrical stimulation (TENS, HF, FREMS, or PENS) or electromagnetic field; treatment of PDN; outcome pain and sensitivity; placebo= no electric current or no exposure to magnetic fields; stimulation of lower or upper extremities

Exclusion criteria SR: inclusion subjects other than PDN patients; unreliable description of what was considered PDN; lack of control group data description

12 studies included (A-L)

Describe intervention:

See detailed description for individual studies (below)

Describe control:

See detailed description for individual studies (below)

End-point of follow-up:

See detailed description for individual studies (below)

For how many participants were no complete outcome data available?

See detailed description for individual studies (below)

Pain relief, sensitivity, length of treatment (group difference iin change from baseline; SMD or MD, [95%CI])

Meta-analysis only feasible for comparison with placebo, and only for TENS [and electromagnetic field methods, which are not part of the research question]

TENS vs placebo

RCTs A-E

Pain relief (various measures) at end of treatment (2-12 wks)

SMD (random effect model)

A: -1.08 [-2.05;-0.10]

B:-0.47 [-1.42; 0.48]

C: -0.05 [-0.67; 0.56]

D: -0.59 [-1.32; 0.14]

E: -0.47 [-1.32; 0.38]

Overall: -0.44 [-0.79; -0.09]

I2= 0% (p=1.00)

[below, for information only, results on electromagnetic field interventions]

Electromagnetic vs placebo

RCTs I-L

Pain relief

MD (random effect model)

I: -2.74 [-4.40; -1.08]

J: -0.20 [-1.13; 0.73]

K: -0.20 [-1.31; 0.91]

L: 0.50 [-2.18; 3.18]

Overall: -0.69 [-1.86; 0.48]

I2= 63% (p=0.04)

Authors conclude that TENS improved pain relief in patients with diabetic neuropathy; no improvement was observed with the use of electromagnetic field treatment

Note: only 1/5 studies on TENS shows a statistically significant treatment effect, and this study (study A) did not include patients with diabetic neuropathy but deals with hypersensitive hands (see Cheing 2005 in evidence table)

Note: SR uses a very broad strategy for pooling; different measures for pain are pooled, different intensities of treatment (TENS) are pooled, different magnetic field treatments (e.g. static and dynamic/pulsed) are pooled

Note: in meta-analyses pain relief with respect to baseline is studied, however, most publications only report baseline and endpoint, it is unclear how the SD values for pain relief were calculated in Stein 2013

Note: almost all RCTs have serious shortcomings and high or unknown riks of bias (except J and K)

Note: RoB determined but not reported at the level of individual studies [therefore the origiinal publications were analyzed and RoB determined; see below and separate RoB table]

Stein 2013: included RCTs on electric stimulation (A-H)

TENS vs placebo (A-E)

Cheing 2005

RCT (parallel)

Setting and Country: district general hospital, Hong Kong

Source of funding: not stated

Inclusion criteria:

clinical diagnosis of hypersensitive hands due to

peripheral nerve injuries

Exclusion criteria:

causalgia or shoulder–hand

syndrome; TENS or desensitization programme 1 month prior; cardiac

pacemaker or sensory loss in

hands prior to study

N total at baseline:

Intervention: 10

Control: 9

Important prognostic factors:

age ± SD:

I: 32±11

C: 38±13

Sex:

I: 8/10 M

C: 8/9 M

Pain (VAS)

I: 6.5 ± 1.6

C: 6.3 ± 1.8

Diabetes type

?

HbA1c

?

onset hypersens after injury 5 w (range, 3–7);

Groups comparable at baseline? Yes

TENS

N=10

Pulse= 200 μs; frequency= 100 Hz; the intensity was

adjusted to produce a tingling sensation that was strong but tolerable

10 20 min sessions for two consecutive weeks

in the hands

Suppl treatment (medication): not mentioned

Placebo

N=9

no active treatment

10 20 min sessions for two consecutive weeks

in the hands

Length of follow-up:

End of treatment= 2 weeks

Loss-to-follow-up:

No drop outs

VAS-pain in cm (SD)

VAS-pain

baseline (d1)

I: 6.2 ± 2.3

C: 6.5 ± 0.2

endpoint (d11)

I: 1.7 ± 1.7

C: 4.6 ± 2.0

Pain relief (d11-d1)

I: -4.5

C: -1.9

between group difference

statistically significant (p<0.05)

Authors conclude that TENS reduces pain and hypersensitivity but not grip strength; studies with larger sample size and longer followup are required

Note: study population appears not to include PDNP patients; treatment of hypersensitive hands after peripheral nerve injury

Forst 2004

[full text not available: details based on SR of Davies 2014, and Medline abstract]

RCT (parallel)

Setting and Country: Germany

Inclusion criteria:

mild-to-moderate

symptomatic diabetic neuropathy

N total at baseline:

Intervention: 12

Control: 7

Important prognostic factors:

age ± SD:

I: 57.6±11.5 /

C: 59.4±8.6

Sex:

I: 6/12 M

C: 4/7 M

Diabetes type

?

HbA1c

?

NTSS

Range 4-16

Groups comparable at baseline?

No [large difference in NTSS-6 score]

TENS

N=12

Pulse= 280 μs; frequency= 4 Hz; individual intensity

between 5-70 mA (Salutaris device)

60 min sessions daily for 12 weeks in the peroneal nerve

Suppl treatment (medication):

?

placebo

N=7

No active treatment (electrically inactive device)

60 min sessions daily for 12 weeks in the peroneal nerve

Length of follow-up:

End of treatment= 12 weeks

Loss-to-follow-up:

?

Incomplete outcome data:

?

total symptom score (NTSS-6) and visual analogue scale (VAS)

NTSS-6 (at 12 weeks)

baseline

I: 10.0 ± 3.3

C: 7.6 ± 3.1

endpoint

I: 6.8 ± 3.9

C: 6.5 ± 6.1

Pain (symptom) relief

I: -3.2

C: -1.1

VAS [pain or ‘symptoms’?]

At 6 weeks [!!]

I: 19.8 ± 5.0

C: ..

endpoint

I: 14.4 ± 9.6

C: ..

Pain (symptom) relief

I: -5.4

C: ‘unchanged’

‘significant improvement in

VAS rating was found after 6 wk of TENS therapy’

VAS

At 12 weeks

??

Authors conclude that TENS is a convenient non-pharmacological option for primary or adjuvant treatment of painful diabetic neuropathy

Note: NO ACCESS TO FULL TEXT; PAPER CANNOT BE JUDGED

Note: small sample size (N=19)

Note: the larger difference between baseline and endpoint in the TENS group (= treatment effect) is due to a higher TENS score at baseline! After 12 weeks of treatment the NTSS-6 score is identical in TENS and placebo group ..

Gossrau 2011

RCT (parallel)

Setting and Country: Univ Medical School; Germany

Source of funding: none? [statement in paper is unclear]

Inclusion criteria:

diabetes type I or II

for >1 year; PDN diagnosed by experienced

neurologist or diabetologist

Exclusion criteria:

pacemaker or heart defibrillator,

brain stimulator, history of alcohol abuse, malignancy

N total at baseline:

Intervention: 22

Control: 19

Important prognostic factors:

age ± SD:

I: 67.9 ± 12,1

C: 65.9 ± 7.0

Sex:

Not reported

Diabetes type

1 2 (% not reported)

HbA1C

I: 7.04 ± 0.71

C: 7.02 ± 0.84

Pain duration (w)

I: 46.3 1 ± 1.97

C: 58.1 ± 60.47

NPS (sum score)

I:43.2 ± 12.9

C:43.4 ± 13.3

Pain intensity (VAS)

I: 6.1 ± 2.0

C: 6.0 ± 2.0

Groups comparable at baseline? Yes

TENS: micro-TENS

N=22

Intensity= 30-40 μA; frequency= 2 Hz

30 min sessions for 4 weeks with three visits each week in the proximal dorsum pedis and on top of the caput fibulae on both legs

Suppl treatment (medication): not mentioned

placebo

N=19

No active treatment (electrodes were not

connected to TENS device)

30 min sessions for 4 weeks with three visits each week in the proximal dorsum pedis and on top of the caput fibulae on both legs

Length of follow-up:

End of treatment= 4 weeks

End of follow-up= 8 weeks (i.e. 4 weeks after end of treatment)

Loss-to-follow-up:

None? [not stated]

Incomplete outcome data:

Not reported

Neuropathic pain score (NPS): mean NPS [SD]; patients with >= 30% reduction in NPS defined as responders

Standardized questionnaires (Pain Disability Index [PDI],

], Center for Epidemiologic

Studies Depression Scale [CES-D]); safety (adverse events)

Neuropathic pain score (NPS)

baseline

I: 43.2 ±12.9

C: 43.4 ± 13.3

End of treatment

I: 36.2 ± 15.0

C: 32.7 ± 17.2

Pain relief: no statistically significant difference in pain relief (NPS decrease from baseline) between groups at end of treatment (p>0.18) or end of follow-up (p>0.5); idem for pain intensity (NRS item of NPS)

Responders end of treatment

I: 6/21= 29%

C: 10/19= 53%

p >0.09

Responders end of follow-up

I: 8/21= 38%

C: 6/19= 32%

p >0.74

QoL (PDI score)

baseline

I: 22.1 ±16.5

C: 21.8 ± 15.0

End of treatment

I: 17.7 ± 15.5

C: 18.0 ± 14.6

End of follow-up

I: 19.5 ± 15.6

C: 18.1 ± 13.5

Pain relief (with respect to baseline): no stat sign difference in PDI, at end of treamtment (p>0.8) or end of follow-up (p>0.5)

Depression (CESD)

baseline

I: 17.9 ±13.0

C: 15.4 ± 6.2

End of treatment

I: 15.1 ± 9.2

C: 12.5 ± 9.7

End of follow-up

I: 14.9 ± 9.4

C: 14.4 ± 9.7

Improvement (with respect to baseline): no stat sign diff at end of treatment (p >0.3) or end of follow-up (p >0.7)

Safety (AEs)

No AEs reported in either group

Authors conclude that pain reduction with TENS [micro-TENS] is not superior to a placebo treatment

Note: because micro-TENS uses very low current intensities, patients can be blinded to the intervention, however, because treatment intensity is reduced the efficacy of the treatment may be reduced as well

Note: authors criticize earlier RCTs in which patients can sense the group assignment (sensatinons due to TENS)

Note: authors state that the ambiguous conclusions of several studies suggest that the mode of TENS application, the location of the electrodes and the individuals disease stage can variously influence the therapeutic outcome of this procedure on neuropathic pain’

Kumar 1998

RCT (parallel)

Setting and Country: university medical center; USA

Source of funding: supported by manufacturer of H-wave machine (lended the machines); declaration of potential conflicts of interest is missing

Inclusion criteria: 31-70 years; type 2 diab; painful DNP >2 months; neurol deficit involving lower extremities

Exclusion criteria: vasc insufficiency, angina pectoris, cardiac arrhythmia, congestive heart failure, MI, untreated hypertension, cerebrovascular ischemia, psych disease or substance abuse, significant renal or liver disease; patients on corticosteroids, dilantin, or chemotherapeutic

agents

N total at baseline:

Intervention: 14

Control: 9

Important prognostic factors:

age ± SD:

I: 59±2

C: 58±4

Sex:

I: 4/14 M (29%)

C: 6/9 M (67%)

Diabetes type

type 2

HbA1C

Not reported

Duration NP sympt

I: 22 ± 6 months

C: 21 ± 5 months

Pain grade

I: 3.2 ± 0.2

C: 2.8 ± 0.3

Groups comparable at baseline? Yes except gender distribution

TENS (+ amitriptyline)

N=14 (2 w/o amitriptyline)

Pulse= 4ms, intensity= ≤35 mA; 35V; frequency= 2-70 Hz

30 min sessions daily for 12 weeks at home in both lower extremities

Suppl treatment (medication): treated with amitriptyline 50 mg/d

PS. Initially 26 patients treated with amitriptyline and after 4 weeks insufficient-/non-responders (n=23; pain grade >2) randomized to TENS and placebo; 3 patients (2 in TENS, 1 in placebo group) could not tolerate amitriptyline and did no longer take the drug

Placebo (amitriptyline only)

n=9 (1 w/o amitriptyline)

No active treatment (sham with non-functional machine)

30 min sessions daily for 12 weeks at home in both lower extremities

Suppl treatment (medication): treated with amitriptyline 50 mg/d

Length of follow-up:

End of treatment= 12 weeks

End of follow-up= 16 weeks (4 weeks after end of treatment)

Note: excluding 4 weeks of pretreatment with amitryptiline before randomization

Loss-to-follow-up:

none

Incomplete outcome data:

none reported

Graded pain (see Kumar 1997) on 6-point scale, mean (SD); overall change in symptoms (VAS); safety (AEs)

Pain grade

baseline

I: 3.2 ± 0.2

C: 2.8 ± 0.3

end of treatment

I: 1.4 ± 0.4

C: 1.9 ± 0.5

Pain relief

I: -1.8 ± 0.3

C: -0.9 ± 0.3

Diff= -0.9 ± 0.3 (p <0.03) in favour of TENS

Placebo: pain scores improved by 1 grade in 2 patients, by 2 grades in 3 patients; 4 patients (44%) had no change in pain grade; 1 (11%) experienced complete symptomatic relief >>5/9 (56%) improved with 1 or more grade

TENS: 3 patients improved by

3 pain grades, 8 by 2 grades, and 1 by 1 grade; 5 (36%) experienced complete symptomatic relief >>12/14 (86%) improved by 1 or more grade

overall change in symptoms visual analog scale (VAS)

I: -66 ± 10%

C: -55 ± 12%,

Safety (AEs)

No local or systemic side effects were noticed with electrotherapy

Authors conclude that TENS

was effective in reducing pain in patients who failed to respond to amitriptyline

or who had partial relief only; combining TENS with amitriptyline augmented symptomatic relief, suggesting that it may be a useful adjunctive modality

Note: to improve blinding of the intervention ‘investigator explained to each patient that electrical sensations at the electrodes might not be felt because of possible variation in sensory perception thresholds’ [..]

Note: Amitriptyline produced symptomatic relief in 15 (60%) of the 26 patients (after 4 weeks), 8 patients had no improvement

Note: authors state ‘After discontinuation of electrotherapy, there was a tendency for recurrence of

symptoms even though amitriptyline therapy was continued’ [details are not provided]

Kumar 1997

RCT (parallel)

Setting and Country: university medical center; USA

Source of funding: supported by manufacturer of H-wave machine (lended the machines); declaration of potential conflicts of interest is missing

Inclusion criteria:

31-70 years; type 2 diab; painful DNP >2 months; neurol deficit involving lower extremities

Exclusion criteria:

vasc insufficiency, angina pectoris, cardiac arrhythmia, congestive heart failure, MI, untreated hypertension, cerebrovascular ischemia, psych disease or substance abuse, significant renal or liver disease; patients on corticosteroids, dilantin, or chemotherapeutic

agents

N total at baseline:

Intervention: 18

Control: 13

Important prognostic factors:

age ± SD:

I: 53±4

C: 59±3

Sex:

I: 7/18 M (39%)

C: 5/13 M (38%)

Diabetes type

type 2

HbA1C

Not reported

Duration NP sympt

I: 16 ± 3 months

C: 22 ± 4 months

Pain grade

I: 3.17 ± 0.12

C: 2.92 ± 0.13

Groups comparable at baseline? Yes

TENS

N=18

Pulse= 400 μs, intensity= ≤35 mA; frequency= 2-70 Hz

30 min sessions daily for 4 weeks at home in both lower extremities

Suppl treatment (medication): All

patients were advised to discontinue analgesics,

including tricyclic antidepressants

Placebo

N=13

No active treatment

30 min sessions daily for 4 weeks at home in both lower extremities

Suppl treatment (medication): All

patients were advised to discontinue analgesics,

including tricyclic antidepressants

Length of follow-up:

End of treatment= 4 weeks

Loss-to-follow-up:

None? [not stated]

Incomplete outcome data:

Not reported

Graded pain (see Kumar 1997) on 6-point scale, mean (SD); overall change in symptoms (VAS); safety (AEs)

Pain (symptom) scale:

 0=no symptoms; 1= minimal burning pain, some discomfort, insign problem ADL; 2= mild burning, uncomformtable, some disturbance ADL, patient wants treatment; 3= burning, paresthesias disturbing sleep, difficulty ADL; 4= intense burning, paresthesias, disturbed sleep, unbearable, unable to function; 5= extremely intense burning, constant, excruciating, paresthesias, very disturbed sleep, patiient asking for strong analgesics

Pain grade

baseline

I: 3.17 ± 0.12

C: 2.92 ± 0.13

end of treatment

I: 1.44 ± 0.25 (p<0.01)

C: 2.38 ± 0.26 (p=0.04)

I/C sign different (p<0.03)

Pain relief

I: -1.73

C: -0.54 ± 0.21

Diff= -1.2 in favour of TENS

Placebo: pain scores improved by 1 grade in 3 patients, by 2 grades in 2 patients; 8 patients (62%) had no change in pain grade >>5/13 (38%) improved with 1 or more grade

TENS: 3 patients improved by

3 pain grades, 8 by 2 grades, and 4 by 1 grade; 3 (20%) experienced complete symptomatic relief >>15/18 (83%) improved by 1 or more grade

overall change in symptoms visual analog scale (VAS)

I: -52 ± 7%

C: -27 ± 10%

P <0.05

Safety (AEs)

1 patient in control group reported burning sensation

at the site of electrode placement; no other local or systemic side effects were noticed

Authors conclude that TENS ameliorated pain and discomfort associated with peripheral neuropathy. This novel modality offers a potential nonpharmacological

treatment option

Note: short-term (4 weeks of treatment), patients advised to stop taking analgesics

Note: 9 placebo treated patients received TENS at end of trial, and pain scores decreased significantly from 3.0 ± 0.62 to 1.56 ± 0.32 (P= 0.02)

Note: authors state that ‘a month after discontinuation of electrotherapy, there was a tendency for recurrence of symptoms, and therapeutic gains were being lost progressively’ [no details provided]

TENS vs HF (F)

Reichstein 2005

RCT (parallel)

Setting and Country: Diabetes center; Germany

Source of funding: non-commercial; declare ‘no conflicts of interest’

Inclusion criteria:

type 1 or 2 diabetes and polyneuropathy;

18–80 years; had HbA1c <11%

Exclusion criteria:

Patients taking medication that may influence neuropathic symptoms

(such as α-lipoic acid, tricyclic antidepressants or anticonvulsants);

patients with ulcers, amputations caused by ischaemia, or cancer

N total at baseline:

TENS: 21

HF: 20

60% painful DNP

40% non-painful

Important prognostic factors:

age ± SD:

T: 57.8±12.5

H: 64.2±12.7

Sex:

T: 10/21 M

H: 12/20 M

Diab type (T1/T2)

T: 3/18

H: 4/16

HbA1c (%)

T: 9.3±1.6

H: 9.3±2.4

Groups comparable at baseline? Yes

TENS (T)

N=21

Pulse= 4msec; frequency= 180 Hz, ≤35 V, ≤35 mA ;

Intensity adjusted according to patient and ranged from 20-30 mA

30 min sessions daily for three consecutive days in both lower extremities

Suppl treatment (medication): see excl criteria, medication for neuropathic pain excluded

Study makes use of strata for patients with non-painful (paraesthesiae, numbness) and painful neuropathic symptoms (in addition to non-painful

symptoms; burning, stabbing, shooting of an electrical nature)

HF (H)

N=20

high-frequency external

muscle stimulation (HF)

Pulse= ≤350 mA, voltage= ≤70 V, frequency= 4096 Hz,

increased to 32768 Hz within 3 s; max frequency

used for 3 s and then down-modulated from 32768 to 4096 Hz; Intensity adjusted to pleasant level that did not produce any pain or uncomfortable paresthesia

30 min sessions daily for three consecutive days in both lower extremities

Suppl treatment (medication): see excl criteria, medication for neuropathic pain excluded

Study makes use of strata

Length of follow-up:

End of treatment= 3 days

End of follow-up= 5 days (2d after end of treatment)

Loss-to-follow-up:

None

Incomplete outcome data:

None?

Neuropathic symptoms on 1-10 VAS for pain (1= no symptom, 10= worst ever), numbness, numbness in painful areas, burning, paraesthesiae and dysaesthesia in the lower extremities; Total Symptom Score (TSS); safety (AEs)

improvement of symptoms defined as improvement of three points or more for at least one symptom

Neuropathic symptoms

Improvement (see defnition)

T: 7/21 (33%)

H: 16/20 (80%)

p<0.05

HF more effective than TENS

Strata

non-painful DNP

T: 4/9 44%

H: 7/7 100%

p<0.05

painful DNP

T: 3/12 25%

H: 9/13 69%

p<0.05

HF more effective than TENS in both strata

Total Symptom Score TSS

baseline

T: 6.6 ± 3.2

H: 7.0 ± 3.6

End of follow-up

T: 5.4 ± 3.8 (p<0.05)

H: 4.6 ± 3.4 (p<0.005)

Safety (AEs)

Both forms of electrical treatment were well tolerated, with no local or

systemic side effects; 1 patient reported some muscular

discomfort in his legs after first HF treatment, which ceased within several hours

Authors conclude that HF can ameliorate discomfort and pain associated with DSP, and suggests that HF is more effective than TENS

Note: short-term results (3 days of treatment + 2 days follow-up); this could e.g. reduce the efficacy of (TENS?) treatment

Note: patients with PDNP medication were excluded

Note: because of 2 active comparators placebo-effects are less important

Note: responder analysis shows that mean symptoms scores in responders decreased from 7.3±0.6 to 3.6±0.6 in the HF group and from 5.4±0.6 to 1.9±0.4 in the TENS group

Note: authors state that ‘reduction of symptoms was documented up to 2 days after

end of treatment. However, patients reported recurrence

of symptoms several days later, which was ameliorated

by further electrotherapy (data not shown)’

Note: power calculation is missing

FREMS vs placebo (G)

Bosi 2005

RCT (crossover)

Setting and Country: university med center; Italy

Source of funding: sponsored by manufacturer FREMS machine; declaration of potential conflicts of interest is missing

Inclusion criteria: type 1 or type 2 diabetes; 18-70 yrs; painful DNP with reduced sensory/MNCV lower limb; vibration perception at big toe >25 V

Exclusion criteria:

any other severe disease; pregnancy; renal disease; history of foot ulcers; lower limb vasculopathy (ABI<0.9 or TPP<50 mmHg)

N total at baseline:

Intervention: 31

Control: 31

Important prognostic factors:

age ± SD:

I: 63.1±3.1

C: 59.2±3.1

Sex:

Not reported

Diab type (T1/T2)

8/23

HbA1c

8.3±0.4

Pain (VAS 100-mm)

Daytime 32.3±6.8

Night 36.3±6.3

Pain duration

Not reported

Groups comparable at baseline? Not relevant (crossover)

Note: large difference in pain scores at baseline between patients first receiving placebo and those first receiving FREMS

FREMS

N=31

Pulse= 10 – 40 μs; frequency= 1 - 50 Hz, peak

amplitude variable from 0 – 255 V; patients modulated neurostimulation by increasing voltage up to max allowed ‘which corresponded to possible perception of burning at site of the electrode’

10 30 min sessions for three consecutive weeks

in the lower extremities

crossover: 3 weeks intervention, 1 week washout, 3 weeks of alternate intervention

Suppl treatment (medication): any analgesic or drug for chronic treatment of PDNP was discontinued >3 weeks before randomisation

placebo

N=31

No active treatment (no electric current)

10 30 min sessions for three consecutive weeks

in the lower extremities

Length of follow-up:

End of treatment= 3 weeks

End of followup= +4 months (4 months after completion of crossover study)

Loss-to-follow-up:

None

Incomplete outcome data:

None (for relevant outcome measures)

Daytime and night-time pain 0n 0–100 VAS; QoL using SF36; safety (AEs)

Pain (VAS 0-100)

End of treatment

daytime

pretreatment

I: 37.1±5.3

C: 31.2±3.9

end of treatment

I: 26.2±3.9 (p=0.0025)

C: 31.9±4.2 (NS)

Pain relief

I: -10.9 (29% reduction)

C: +0.7

Diff= -11.6 in favour FREMS

Night-time

pretreatment

I: 38.1±5.5

C: 33.3±3.8

end of treatment

I: 28.5±3.8 (p=0.0107)

C: 30.4±4.2 (ns)

Pain relief

I: -9.6 (25% reduction)

C: -2.9

Diff= -6.7 in favour of FREMS

QoL (SF36)

pretreatment

I: 103.7±1.5

C: 104.4±1.5

end of treatment

I: 105.6±1.3 (ns)

C: 105.9±1.5 (ns)

Change

I: +1.9

C: +1.5

Diff= +0.4 in favour of FREMS

End of follow-up

daytime pain

37.0±5.3 (baseline) to 25.1±4.2 (p<0.01; 32% reduc)

night-time pain

41.1±5.2 (baseline) to 26.5±3.9 (p<0.01; 36% reduc)

QoL (SF36)

103.6±1.5 (baseline) to 107.9±1.2 (p<0.001; 4% improvement)

Safety (AEs)

No systemic side effects recorded; patients reported very slight burning sensation with FREMS

Authors conclude that FREMS induced a significant reduction in daytime and night-time VAS pain score (all p<0.02), which persisted after 4 months of follow-up; FREMS is a safe and effective therapy for neuropathic pain in DPN

Note: analgesic medication was discontinued at least 3 weeks before randomisation

Note:authors state that no carryover effect was evident, however, the lower pretreatment pain scores in analyses of placebo indicate that there was a significant carryover effect. Also note the large difference in pain scores at baseline between patients first receiving placebo and those first receiving FREMS

Note: with respect to blinding >>‘preliminary study had shown that patients with a vibration perception threshold higher than 25V effectively had no perception of the electrical stimuli administered by the FREMS device (data not shown)’ [but note that patients modulated intensity ‘up to max allowed which corresponded to possible perception of burning at site of the electrode’ ..]

Note: power calculation is missing

PENS vs placebo (H)

Hamza 2000

RCT (crossover)

Setting and Country: university medical center; USA

Source of funding: non-commercial; declaration of potential conflicts of interest is missing

Inclusion criteria:

longstanding type 2 diabetes;

painful peripheral neuropathic

symptoms >6 months duration in both lower extremities; diagnosis PDNP confirmed by abnormal NCS

Exclusion criteria:

pregnancy, cardiac arrhythmias /

pacemakers, infection,

history of vascular insufficiency in legs, drug or alcohol abuse, psychiatric disease,

major organ disease, radicular pain, psychiatric disease

N total at baseline:

50 (cross-over)

Note: 1^st phase also analyzed as parallel RCT (see Results)

Important prognostic factors:

age ± SD: 55 ±8

Sex: 22/50 M

Pain intensity

6.3 ± 1.0

Pain duration

18 ± 7 months

Diab type

T2 only

Groups comparable at baseline? Not relevant (cross-over design)

Note: significant carry-over effects

PENS

N=50 (crossover)

Pulse= 500 μs; intensity ≤25 mA; frequency= 15 and

30 Hz; Intensity adjusted to highest tolerable level without producing muscle contractions

30 min sessions 3 times per week for 3 consecutive weeks in the leg and foot bilaterally

crossover: 3 weeks intervention, 1 week washout, 3 weeks of alternate intervention

Suppl treatment (medication): patients allowed to take oral nonopioid analgesic medication (= one of the outcome measures)

placebo

N=50 (crossover)

needles only (no electric current)

30 min sessions 3 times per week for 3 consecutive weeks in the leg and foot bilaterally

Length of follow-up:

End of treatment= 3 weeks

Loss-to-follow-up:

None

Incomplete outcome data:

4/50 (8%) for outcome=BDI

6/50 (12%) for POMS

Due to ‘questionnaire problems’

Complete data for all other outcome measures

Extremity pain, Physical activity, Quality of sleep (VAS (10cm; mean, SD); oral nonopioid analgesics (pills/day); QoL: SF-36 (physical component PCS; mental component MCS), depression (BDI; POMS); safety (AEs)

Pain (VAS-10cm)

Pretreatment

I: 6.2 ± 1.3

C: 5.2 ± 1.6

End of treatment

I: 2.6 ± 0.9

C: 4.8 ± 1.2

Pain relief

I: -3.6 (p<0.05)

C: -0.4 (ns)

Diff= -3.2 in favour of PENS (p<0.05)

Activity (VAS-10cm)

Pretreatment

I: 4.8 ± 1.2

C: 5.9 ± 1.3

End of treatment

I: 7.8 ± 1.1

C: 6.3 ± 1.2

Change

I: +3.0 (p<0.05)

C: +0.4 (ns)

Diff= +2.6 in favour of PENS (p<0.05)

Sleep (VAS-10cm)

Pretreatment

I: 5.7 ± 1.3

C: 6.8 ± 1.5

End of treatment

I: 8.6 ± 1.0

C: 7.1 ± 1.2

Change

I: +2.9 (p<0.05)

C: +0.3 (ns)

Diff= +2.6 in favour of PENS (p<0.05)

QoL (SF-36)

pretreatment

PCS 31.2 ± 7.3

MCS 41.0 ± 5.8

End of treatment

I: PCS 36.8 ± 6.7 (p<0.01)

I: MCS 43.9 ± 5.6 (p<0.01)

C: PCS 32.4 ± 7.5 (p<0.05)

C: MCS 42.0 ± 5.5 (p<0.05)

Improvement

I: +5.6 (PCS) and +2.9 (MCS)

C: +1.2 (PCS) and +1.0 (MCS)

Diff= +4.4 (PCS) and +1.9 (MCS) in favour of PENS (p<0.05)

Depression (BDI)

pretreatment

30.2 ± 11.6

End of treatment

I: 8.1 ± 4.6

C: 20.7 ± 8.2

Improvement

I: -22.1 (p<0.01)

C: -9.5 (p<0.05)

Diff= -12.6 in favour of PENS (p<0.01)

Depression (POMS)

Postactive and post-sham PENS treatments displayed

significant improvement on all POMS measures except for the vigor activity measure; PENS was associated with greater decreases on all POMS measures relative to the post-sham treatment (p<0.05)

Safety (AEs)

No side effects reported with either therapeutic modality

Extra:analysis of 1^st phase before crossover (i.e. as an parallel design 25/25) >>

Pain (VAS-10cm)

Diff= -3.6 in favour of PENS (p<0.05; 58% reduction)

Activity (VAS-10cm)

Diff= +2.0 in favour of PENS (p<0.05; 38% improvement)

Sleep (VAS-10cm)

Diff= +1.9 in favour of PENS (p<0.05; 32% improvement)

Oral analgesics (pills/day)

Diff= -1.8 in favour of PENS (p<0.05; 56% reduction)

Authors concluded that PENS produces short-term pain relief, improves mood, functionality, and quality of sleep, and decreases nonopioid analgesic requirements in PDNP; PENS shoud be viewed as supplementary therapy to conventional pharmacological therapy

Note: significant carryover effects (e.g. compare pretreatment pain scores between placebo and PENS), but results confirmed in separate analysis of 1^st phase as a parallel RCT (see results)

Note: short-term study (3 weeks) without additional follow-up (authors do state that ‘within 1 week of last PENS treatment session, the pain scores began to return to pretreatment levels’

Note: 92% patients preferred PENS over placebo; 88% reported improved sense of well-being after PENS; 92% expressed 'willingness to pay extra money for PENS therapy'

Note: power analysis with= 0.05 and= 0.10 (power

= 90%) determined that group size of 40 should be adequate to demonstrate a 25% change in the VAS pain scores between the two treatment modalities

Note: with respect to blinding, note that patients were told that the needle-only (sham) treatment represented an acupuncture -like therapy

Note: although clearly less

invasive than spinal cord stimulation, PENS is more complex than TENS

Stein 2013: UPDATE (recent relevant RCTs on electrostimulation)

FREMS vs placebo

Bosi 2013

RCT (parallel)

Setting and Country: university medical centers; Italy, UK, France, Germany

Source of funding: sponsored by manufacturer of FREMS machine; some authors received lecture fees from the sponsor

ClinicalTrials.gov NCT01628627

Inclusion criteria:

type 1 or 2 diabetes of >1 y duration and HbA1c <11.0%; 18-75 yrs; symptom DPN affecting lower extremities with at least 1 pos sens symptom (pain, burning, paraesthesia or prickling); abnormalities in amplitude, latency or conduction velocity in at least one motor nerve and/or sural nerve; measurable sensitive NCV and evocable potential in sural nerve; Michigan Diabetes Neuropathy Score (MDNS) >= 7 points; stable dose of (pain) medications for DPN, if any

Exclusion criteria:

previous treatment with TENS or other electrotherapy; implanted pacemaker, defibrillator or neurostimulator; active foot ulcer and/or previous major amputation of lower extremities; any concomitant severe disease limiting compliance or life expectancy

N total at baseline:

Intervention: 54

Control: 56

Important prognostic factors:

age ± SD:

I: 59.0±10.6

C: 61.3±8.3

Sex:

I: 72% M

C: 61% M

Medication PDNP

I: 26%

C: 16%

Diab type (T1/T2)

I: 10/40

C: 9/42

HbA1c (%)

I: 7.9±1.2

C: 7.6±1.2

Pain (VAS)

daytime

I: 31.6±26.3

C: 40.9±24.0

night-time

I: 41.3±29.7

C: 45.2±29.6

Pain duration

not reported

Groups comparable at baseline? Yes

FREMS

N=54

‘practical ITT’= 50

sequences of biphasic pulses composed of high negative voltage spike (max −300 V, 10–100 μs) followed by recharging phase of low voltage and long duration (0.9–999 ms); pulse frequency variable (mainly low range 1–50 Hz); to both lower limbs

3 series of 10 treatments at least 24 h apart, during 3 weeks, at 3-month intervals

Suppl treatment (medication): patients allowed to be on (stable) medication for PDNP (26% and 16% were on medication in FREMS and placebo group resp; none of the medications were discontinued)

Note:

T−1 (enrolment visit)

T0 (week 0, start 1^st cycle) T1 (week 3, end 1^st cycle) T2 (week 17: start 2^nd cycle) T3 (week 20, end 2^nd cycle)

T4 (week 34: start 3^rd cycle) T5 (week 37, end 3^rd cycle) T6 (week 51)

Placebo

N=56

‘practical ITT’= 51

Identical procedure but w/o electrical current (blinded switch)

3 series of 10 treatments at least 24 h apart, during 3 weeks, at 3-month intervals

Length of follow-up:

End of treatment (3^rd cycle)= 37 weeks

End of follow-up= 51 weeks (14 weeks after end of treatment)

Loss-to-follow-up:

With respect to ‘practical ITT’

Intervention: 11/50 (22%)

Control: 15/51 (29%)

Similar reasons in both groups: withdrew consent, unable to travel, moved

Incomplete outcome data:

ITT population includes all participants with at least 1 follow-up measurement; paper does not mention imputation

Daytime pain, 3^rd cycle:

N=60 (60/110=55%)

Night-time pain, 3^rd cycle:

N=57 (57/110=52%)

Missing data for pain varies between 16% (1^st cycle) and 48% (3^rd cycle)

Note:

36/56 (64%) in control, and

39/54 (72%) in FREMS group completed the study upto 51 weeks (= per protocol population)

Pain, daytime and night-time (VAS 100mm), mean (SD), % responders (>=30% reduction or >50% reduction)

Pain (VAS-100mm)

Effects analyzed per cycle, between start and end of cycle= 3 weeks

night-time pain and daytime pain were significantly reduced in FREMS compared with placebo group at T1 (both p<0.001), T3 (both p<0.001) and T5 (p<0.001 and p<0.05, resp); beneficial effect of FREMS was no longer detectable 3 months after the end of the last treatment series (T6); similar results observed in PP population (data not shown)

rate of responders with either ≥30% or ≥50% reduction in night-time or daytime pain score was significantly higher in patients treated with FREMS compared with placebo after second and third treatment cycle

Responders at 3^rd cycle (T4-5)

Daytime pain (VAS 100mm)

>=30% reduction

I: 66.7 %

C: 30%

Diff= 36.7% (p=0.004)

NNT=2.7

>=50% reduction

I: 50.0 %

C: 23.3%

Diff= 26.7% (p=0.032)

NNT=3.7

Night-time pain (VAS 100mm)

>=30% reduction

I: 64.3 %

C: 27.6%

Diff= 36.7% (p=0.008)

NNT=2.7

>=50% reduction

I: 53.6 %

C: 24.1%

Diff= 29.5% (p=0.022)

NNT=3.4

Safety (AEs)

No treatment-related severe or non-severe adverse

events were recorded during the study, either in FREMS

or placebo group; FREMS patients reported only a very slight burning sensation

Authors concluded that FREMS proved to be a safe treatment for symptomatic DPN, with immediate,

although transient, reduction in pain, and improvement

in cold sensation threshold, but with no effect on NCV

Note: 22/110 randomized patients had a VAS score of 0 at baseline for daytime pain or night-time pain >>the study population (also) contains patients with relatively mild PDNP: findings may not apply to more severe PDNP

Note: considerable % missing data, in particular at later cycles (but results are similar at 1^st cycle with 16% missing data)

Note: with respect to blinding, patients were iinstructed not to consider any perception at site of electrode placement during the treatment as a sign of active treatment; but note that patients were invited to

modulate the delivery of sham or true neurostimulation by

progressively increasing the voltage of the electrical stimulation from 0 V to 255 V (..)

Note: costs of FREMS are similar to TENS

Note: pain was secondary outcome; other outcome measures were not part of the research question (primary endpoint was the change in nerve conduction velocity (NCV) of deep peroneal, tibial and sural nerves. Secondary endpoints included the effects of treatment on pain, tactile, thermal and vibration sensations)

Note: power calculation is missing

TENS vs PRF lumbar sympathectomy (minimally invasive procedure)

Naderi Nabi 2015

RCT (parallel)

Setting and Country: university medical center; Iran

Source of funding: non-commercial; declaration of potential conflicts of interest is missing

Inclusion criteria:

type I or II diabetes; PDPN resistant to conservative treatment for >6 months; Hba1c<8%; normal creatinine; pain intensity >= 4 on 10-point NRS

Exclusion criteria:

implanted pacemaker or heart defibrillator, brain stimulator; history of alcohol abuse, malignancy, coagulopathy, anti-coagulant drug use, infection or irritation at the probe or electrode sites, or certain anatomical anomalies

N total at baseline:

Intervention: 30

Control: 30

Important prognostic factors:

age ± SD:

T: 56.6 ± 5.8

P: 56.7 ± 6.9

Sex:

T: 53% M

P: 50% M

Diabetes type

% not reported

HbA1c (%)

T: 7.68 ± 0.27

P: 7.65 ± 0.33

Pain (NRS 0-10)

T: 6.10

P: 6.46

Pain duration

Not reported (but >6 months)

Groups comparable at baseline? Yes

TENS (T)

N=30

10 sessions performed every other day (80 Hz, 50 Amp, 0.2 ms square pulses, 2-3 times sensory threshold for 20 min)

Note: amplitude specified ‘50 Amp’ is probably not correct (: should be 50 mA?)

Suppl treatment (medication): patients were instructed to discontinue all analgesics, and take 300 - 600 mg pregabalin each day for duration of the study

PRF lumbar sympathect (P)

N=30

PRF at L4-L5, first diagnostic; if lidocaine/triamcinolone injection reduced NRS pain by >= 50%, RF probe was placed: pulsed RF energy

at each level for three 180 sec cycles at a temperature of 45°C; skin T recorded to test sympathetic blockade: positive change of at least 2°C (foot) indicated a successful block

Suppl treatment (medication): patients were instructed to discontinue all analgesics, and take 300 - 600 mg pregabalin each day for duration of the study

Length of follow-up:

1 week, 1 month and 3 months after treatment

Loss-to-follow-up:

T: 7 (~20%)

Chose not to complete survey (2), or lost to follow-up (5)

P: 3 (~10%)

Withdraw for personal reasons (3)

Incomplete outcome data:

Not reported

Note: numbers specified do not add up

Pain (0-10), mean (SD), p-value for change to baseline; safety (AEs)

Pain (NRS 0-10)

Baseline

T: 6.1

P: 6.46

1 week after treatment

T: 3.96 (Diff 2.13 ± 0.89)

P: 2.76 (Diff 3.70 ± 0.59)

1 month after treatment

T: 5.23 (Diff 0.86 ± 0.68)

P: 4.3 (Diff 2.16 ± 0.74)

3 months after treatment

T: 5.9 (Diff 0.20 ± 0.40)

P:5.13 (Diff 1.33 ± 0.88)

Changes from baseline (Diff) for both groups and all timepoints: ‘p-values <0.0001’ [but is this correct?]

Safety (AEs)

‘therapeutic methods well tolerated and not associated with any serious adverse

Effects; skin irritation reported in ‘a few’ TENS group subjects

Authors conclude that both TENS and PRF lumbar sympathectomy are promising pain relief treatments for painful DNP, however, PRF lumbar sympathectomy seems to have a superior efficacy (greater pain relief, and longer lasting pain relief)

Note: patients were instructed to discontinue all analgesics, and take 300 - 600 mg pregabalin each day for duration of the study. This may have (greatly?) affected pain intensity

Note: PRF lumbar sympathectomy is compared to a single cycle of TENS; TENS treatment may have been suboptimal

Note: paper appears to contain many inaccuracies, and power calculation is missing

Internal neurostimulation (SCS)

Slangen 2014

Suppl analysis:

van Beek 2015

RCT (parallel)

Setting and Country: outpatient pain clinics, university medical centers; Netherlands

Source of funding:supported by grant of manufacturer; authors declare ‘no other potential conflicts of interest’

ClinicalTrials.gov NCT01162993

Inclusion criteria:

DM; 18-80 yrs; moderate-severe PDPN in lower limbs (MDNS); insufficient pain relief and/or unacceptable side effects with drug treatment according to

the guidelines and the algorithm described for PDPN by Jensen 2006 (Diab Vasc Dis Res 3:108–119); pain present for >12 months with mean intensity daytime or nighttime NRS >= 5

Exclusion criteria:

neuropathic pain most prevalent in upper limbs; neuropathy or chronic pain of other origin than DM; recent neuromodulation therapy; peripheral vascular disease with no palpable foot pulses; active foot ulceration; life expectancy,1 year; pacemaker; severe cardiac or pulmonary failure; unstable blood

glucose control

N total at baseline:

Intervention: 22

Control: 14

Important prognostic factors:

age ± SD:

I: 57.1 ± 12.4

C: 56.5 ± 8.0

Sex:

I: 68% M

C: 64% M

MDNS score

0/1/2/3

I: 3/4/9/6 patients

C: 3/3/5/3 patients

Diab type (T1/T2)

I: 1/13

C: 3/19

HbA1c (%)

I: 8.3 ± 2.0

C: 8.4 ± 2.7

Pain intens (MDNS) score 0/1/2/3

I: 3/4/9/6

C: 3/3/5/3

Pain duration (yrs)

I: 6.0 ± 5.1

C: 4.9 ± 3.6

Groups comparable at baseline? Yes

SCS (+ BMT)

N=22

position of lead over thoracic level and settings of external stimulator were tailored for each patient

in order to reach optimal paraesthesia coverage; connected to an external

stimulator (Medtronic); 2-week trial stimulation: stimulator only implanted if

 NRS pain daytime or nighttime for last 4 days >= 50% lower than baseline

score, or PGIC-pain and sleep >= 6

Suppl treatment (medication): BMT, Best Medical Treatment (guidelines; algorithm Jensen 2006)

5/22 (23%) had negative trial stimulation (including 1 patient dying of complication)

No SCS (BMT only)

n=14

Suppl treatment (medication): BMT only

Length of follow-up:

End of follow-up= 6 months

Loss-to-follow-up:

Intervention: 3/22 (13.6%)

1 deceased, 1 withdraw consent, 1 withdraw for AE

Control: 0/14 (0%)

4/22 had negative trial stimulation: 1 withdrew consent, remaining 3 were classified as failures for SCS in the analysis

Suppl analysis (van Beek 2015):

End of follow-up= 24 months

Treatment success (pain severity NRS / PGIC), pain interference with daily life (mBPI-DPN), pain characteristics (NPS), HRQoL (EQ-5D; MOS-SF-36), pain interference with sleep / sleep quality and quantity (MOS-SS), mood (BDI), and registered medication Use; safety (complications, severe AEs)

Treatment success

Defined as >= 50% relief of pain on NRS for 4 days during daytime or nighttime, or score of >= 6 on 7-point PGIC scale for pain and sleep (6= much improved; 7= very much improved)

Treatment success

I vs C

13/22 (59%) 1/14 (7%)

[13/17 (76% of patients with a positive trial stimulation)]

P<0.01

Details (see definition) >>

Daytime pain - NRS

9/22 (41%) vs 0/14 (0%)

Night-time pain - NRS

8/22 (36%) 1/14 (7%)

PGIC for pain

12/22 (55%) 0/14 (0%)

PGIC for sleep

8/22 (36%) 0/14 (0%)

All p-values <0.05

mBPI-DPN

PSI 4.0±2.8 vs 6.5±2.1

PII 3.5±2.6 vs 5.5±1.5

All p-values <0.01

NPS

All dimensions in favour of SCS: Deep pain, Surface pain, Intensity, Unpleasantness, Coldness, Hotness, Dullness, Sharpness, Sensitivity, Itching

Stat significant (p<0.05) except for 'Deep pain'

EQ-5D

Utility 0.50±0.33 vs 0.33±0.29

Health 57.6±24.3 vs 56.5± 14.2

Stat not significant

MOS SF-36

MCS 49.3±11.5 vs 46.7±12.0

PCS 32.3±10.5 vs 30.5±7.4

Stat not significant

MOS-SS

SPI-9 40.3±19.6 vs 52.9±16.5

Sleep (h) 6.6±1.8 vs 5.4± 2.4

Opt sleep (%) 23% vs 21%

Stat not significant

BDI

13.0±9.8 vs 14.4±6.3

Stat not significant

Safety

Serious AEs in 2 patients (SCS): dural puncture causing postdural hematoma over left hemisphere and death ; infection of SCS system 6 weeks after implantation, with incomplete recovery after removal SCS and treatment

Suppl analysis (van Beek 2015)

24 months of follow-up

Treatment success

11/17 (65%)

Daytime pain - NRS

8/17 (47%)

Night-time pain - NRS

6/17 (35%)

PGIC for pain

9/17 (53%)

PGIC for sleep

9/17 (53%)

Authors conclude that treatment was successful in 13/22 (59%; 76% of patients with a positive SCS trial stimulation) of patients with PDPN who were treated with SCS over a 6-month period, although this treatment is not without risks

No significant difference was observed in treatment success rate at 24-month follow-up

(11/17=65% of patients with a postive SCS trial stimulation) compared with treatment success rate at 6-month follow-up (13/17=76%)

Note: only patients with failure of pharmacological treatment and with high pain levels (NRS >= 5 for >12 months) were screened and included; therefore, results cannot be generalized to patients with less pain

Note: authors state that SCS is a relatively expensive therapy (~12,000 Euros per patient)

Note: score of 6 or higher on PGIC indicates a clinically important difference

Note: power calculation assuming difference in success of 40% between intervention and control (50% in SCS; 10% in control), power of 80%, alpha 0.05, allowing

for 10% lost to follow-up >>sample of 40 patients

with PDPN required

Note:authors explain non-significant group difference on EQ-5D, MOS SF36, MOS-SS, and BDI, by referring to adaptiation phenomenon and the limited number of patients in the RCT

Note: 5/22 (23%) had negative trial stimulation (including 1 patient dying of complication)

de Vos 2014

Suppl analysis:

Duarte 2016

RCT (parallel)

Setting and Country: 7 pain clinics; Netherlands, Denmark, Belgium, Germany

Source of funding: sponsored by manufacturer (St Jude Medical)

Dutch Trial Register ISRCTN03269533

Inclusion criteria:

adults; refractory DNP in lower extremities for >1 year; refractory to conventional pain treatments; average pain score (VAS) of >= 50

Exclusion criteria:

pain due to atherosclerotic lesions; infection; neuropathic pain in upper extremities; anticoagulant medication; psychiatric problems requiring treatment; addiction to drugs or alcohol, or incapable of cooperation

N total at baseline:

Intervention: 40

Control: 20

Important prognostic factors:

age ± SD:

I: 58

C: 61

Sex:

I: 63% M

C: 65% M

Pain (VAS 100-mm)

I: 73 ± 16

C: 67 ± 18

Diab type (T1/T2)

I: 10/30

C: 5/15

HbA1c (%)

not reported

Pain duration (yrs)

I: 7 ± 6

C: 7 ± 6

Groups comparable at baseline? Yes

SCS (+ CMP)

N=40

lead (St Jude Medica) implanted and positioned

where the patient reported optimal overlap between paresthesia and painful area; trial stimulation period of 7 days maximum:

if successful (‘significant pain relief’), an implantable pulse generator (St Jude Medical) was implanted; patients can adjust stimulation amplitude to their own convenience

Suppl treatment (medication): conventional medical Practice (CMP); medication adjustments and other conventional pain treatments, such as physical therapy, were

allowed at any time during the study, if needed (were registered)

3/40 (7.5%) had negative trial stimulation (including 1 patient in which implantation failed)

No SCS (CMP only)

N= 20

Suppl treatment (medication): conventional medical Practice (CMP); medication adjustments and other conventional pain treatments, such as physical therapy, were

allowed at any time during the study, if needed (were registered)

Length of follow-up:

End of follow-up= 6 months

Loss-to-follow-up:

Intervention: 4/40 (10%)

Implantation failed (1), negative trial stimulation (2), joined another study (breach of protocol)

Control: 2/20 (10%)

Other (unrelated) diseases (withdraw consent; 2)

Treatment success (% patients with >50% VAS pain reduction); reduction in pain intensity; pain characteristics; QoL (SF-MPQ and EQ5D); medication Intake (MQS=Medication Quantification Scale III); patient global impression of change; patient satisfaction (11-point scale); safety (AEs)

Treatment success

I vs C

25/40 (60%) vs 1/20 (5%)

P<0.001 [25/37= 68% with positve trial stimulation]

Medication Intake (MQS)

mean (SD)

Baseline

I: 10.6 (9.7)

C: 9.2 (7.8)

End of follow-up

 I: 7.7 (8.7; p<0.001)

C: 10.1 (8.2; ns)

MPQ QoL

mean (SD)

baseline

I: 16 (5)

C: 15 (6)

End of follow-up

I: 8 (7)

C: 14 (6)

p<0.001

EQ5D self-reported health

mean (SD)

baseline

I: 50 (19)

C: 46 (17)

End of follow-up

I: 61 (22)

C: 41 (20)

p<0.01

PGIC pain reduction

I: 29 (73%)

C: 3 (17%)

p<0.001

Patient satisfaction

I: 8/10 [??]

C: 4/10 [??]

p<0.001

Safety (AEs)

incidences of infection and lead complications in

the SCS group were 3% and 8%, respectively; adverse events related to the implantation procedure were generally resolved by device repositioning

Suppl analysis (Duarte 2016)

6 months of follow-up

EQ-5D index

baseline

I: 0.27 (0.26)

C: 0.47 (0.31)

End of follow-up

I: 0.65 (0.28)

C: 0.44 (0.33)

Diff= -0.21 [-0.39;-0.04] in favour of SCS (p<0.05)

EQ VAS

baseline

I: 50 (19)

C: 48 (16)

End of follow-up

I: 61 (23)

C: 41 (20)

Diff= -20 [-34; -7] in favour of SCS (p<0.01)

QALYs

6 months of follow-up

Adj for baseline EQ-5D score

I: 0.258

C: 0.178

Diff= 0.080 [0.044; 0.114] in favour of SCS (p<0.001)

Authors conclude that In patients with refractory PDN, SCS significantly reduced pain and improved quality of life

Note:quantitation of some outcome measures is not accurately described, and statements in text are not always clearly supported by information in tables/figures >>:authors state that ‘In SCS group, 26 patients (65%) stated that they had much or

very much pain reduction at 6 months compared to baseline, and 3 patients (8%) indicated some pain reduction. Only 3 patients (15%) from control group indicated some pain reduction; despite receiving the best conventional pain treatment, 8 patients (40%)

perceived an increase in pain’

‘In the SCS group, of the 37 patients with an implanted SCS system, all but 2 patients (n= 35, 95%) might or would definitely recommend SCS treatment to other patients with PDN; in the control group, only 4 patients (20%)’

Note: 3/40 (7.5%) had negative trial stimulation (including 1 patient in which implantation failed)

Note: patient satisfaction is measured, but presentation of the results is unclear

Note: suppl study (Duarte 2016) finds QoL advantage for for SCS which is not found in Slangen (2014), possibly because of differences in statistical analyses (baseline correction)

Note: SCS results in gain in QUALYs in comparison to CMP, but gain only= 0.080 QUALYs

Note: power calculation assuming percentage of patients with more than 50% pain reduction to be 50% in

SCS group vs 10% in the control group, 90% power,

5% level of significance >>sample size of 60 for a 2:1 randomization between the SCS and control

Note: bias as a result of not being able to blind the intervention >>authors note that recently developed paraesthesia-free devices

may allow for a double-blind design, with both arms

receiving a device but only one of the arms receiving active treatment

Surgical decompression

van Macaré van Maurik 2014, 2015

RCT (parallel)

Setting and Country: university medical centre; Netherlands

Source of funding: non-commercial; authors declare ‘no conflicts of interest’

Nederlands Trial

Register number NTR2344

Inclusion criteria:

diabetes type 1 or 2; 18-90 yrs; painful bilateral DPN diagnosed with DNS and DNE; positive Tinel sign at posterior tibial, superficial, or common peroneal nerve; ABI 0.8-1.15 with toe-brachial index >=0.7; palpable pulsations in posterior tibial and dorsal pedal artery

Exclusion criteria:

BMI>35; general condition unsuitable for surgery; history of ankle fractures or amputations proximal to Lisfranc joint; ulcer on foot; sufficient effect of pain medication (VAS score 0-1); other causes for neuropathy; history of nerve compression at additional sites

N total at baseline:

Intervention: 42

Control: 42

Important prognostic factors:

analytic sample: n=38

Age: 62.7 ± 10.2

Sex: 58% M

Diab type (T1/T2)

8/30

Diab dur

17.6 yrs

Pain (VAS 10-cm)

6.1

Pain duration

not reported

HbA1c

not reported

Groups comparable at baseline? Yes [within patient comparison]

Surgical decompression

N=42

Decompression procedure of lower extremity nerves according to Dellon in one limb (tibial nerve at ankle site): the common peroneal, deep peroneal,

or the superficial peroneal nerve; by the same surgeon; contralateral limb used as control (within-patient comparison)

Suppl treatment (medication): patients could use their medication following Dutch Polyneuropathy Guideline or guidelines of the European Federation of Neurological Societies

No surgical decompression

N=42 (contralateral leg)

Suppl treatment (medication): patients could use their medication following Dutch Polyneuropathy Guideline or guidelines of the European Federation of Neurological Societies

Length of follow-up:

End of follow-up= 12 months

Loss-to-follow-up / Incomplete outcome data:

4/42 (7%)

Unrelated death (1), lost to fup (1), missing data at fup (2) [all 4 excluded from analysis]

ANALYTIC SAMPLE: n=38

VAS-pain (10-cm), mean (SD) [95%CI], % with MCID; matched pair analysis; HRQoL (SF-36, EQ-5D); safety (AEs)

VAS-pain (10-cm)

baseline

I: 6.1

C:6.1

End of follow-up

I: 3.5 [2.5; 4.4]

C: 5.3 [4.4; 6.2]

p <0.001

Pain relief

I: -2.6 [-3.7; -1.6] p<0.001

C: -0.8 [-1.7; 0.0] p >0.05

Diff= -1.8 in favour of decompression (p= 0.002)

Overall, while 73.7 % of patients improved in their VAS score (of which 35.7 % by more than five points), 26.3 % had no effect or worsened

MCID= 2.9 on VAS-pain (anchor-based):

42.5% patients have reduction in VAS-pain of at least MCID at 12-month follow-up

HRQoL

baseline / end of follow-up

[per patient, i.e. no at the level of intervention vs control]

SF-36

Phys functioning 56.4 / 55.0

Social functioning 67.6 / 64.5

Bodily pain 46.4 / 50.5

Mental health 70.5 / 69.3

Physical composite 36.5 / 36.1

Mental composite 47.3 / 46.3

all p-values non-significant

EQ-5D

Usual activities (%)

1 39.5 / 21.1

2 52.6 / 71.1

3 7.9 / 7.9

Pain/discomfort (%)

1 5.3 / 7.9

2 63.2 / 73.7

3 31.6 / 18.4

EQ-5D index 0.60 / 0.62

all p-values non-significant

Safety (AEs)

3 complications: hematoma caused by use anticoagulants (1); infected wound at ankle site (2), both treated with antibiotics (1 admitted to hospital)

Authors conclude that surgical decompression can alleviate pain in a clinically meaningful way in 42.5% of people with painful DPN; can be added as a therapeutic option for patients with painful DPN who show signs of chronic nerve compression (positive Tinel or other diagnostic criteria), when pain medication fails to reduce pain to an acceptable standard

Note: unilateral surgery did not improve quality of life as measured with SF–36 and EQ–5D questionnaire >>but note that pain defines only a small part of HRQoL, and that only one leg was operated upon

Note: only patients included with painful DNP ‘refractory to pain medication’, but this is very loosely defined, only patients with VAS pain between 0-1 are not included (..)

Note: after one year, participants were asked if they

wanted to undergo the same surgical procedure on the

contralateral leg; to date, only eight participants (out of 38)

have taken that opportunity

Note: VAS score in control leg also decreased significantly during follow-up >>pain

perception in operated and nonoperated leg possibly influenced each other

Note: Sample size calculation based on VAS ( continuous), diff pre/post of 1.5; Beta 0.9, alpha 0.05; with 10 % fup >>

total calculated sample size was 42

Cognitive behavioral therapy / psychological coping

Davies 2015

SR (of RCTs)

Literature search up to July 2014 (10 databases)

Psychological coping strategy vs standard care

A: Otis 2013

B: Teixera 2010

Study design: RCT (parallel)

Inclusion criteria:

SR: clear diagnosis of painful DPN; adults; relevant intervention (physical activity or psychological coping strategy); pain outcome measures; controlled

2 studies included on psychological coping [and 2 studies on physical activity which is not part of the research question]

Participant characteristics

A: US veterans; 62(11) years,

all male, all type2 DM, DM and PDN duration not stated (at least 3 months of pain). US veterans medical centre

B: All participants (n= 20): 74 (10.8) years, 5 male, all type 2, DM duration 12.6 (9.4)

years, PDN duration 7.7 (6.6) years. Community

medical practice and retirement communities, USA; uses a convenience sample [limiting generalizability]

Groups comparable at baseline? Yes (A), Unclear (B)

Psychological coping

A: Otis 2013

Cognitive behavioral therapy (CBT)

N=11

Individual CBT, 1 hour session weekly, ×11 sessions; content: cycle of pain, pain mechanisms, relaxation, identification and challenge negative thoughts, pacing, sleep strategies

B: Teixera 2010

Mindfulness relaxation

N=10

Mindfulness relaxation; 1 hour session then audio CD

for homepractice

Control

A: Otis 2013

no active treatment

N=8

Treatment as usual (TAU) i.e. no weekly sessions with therapist

B: Teixera 2010

N=10

Attention-placebo group, i.e. similar contact/attention as in intervention group (nutritional advice 1hour;

asked to keep a food

diary for 4 weeks)

Length of follow-up:

A: end of follow-up= 4 months (3 months treatment + 1 month)

B: end of treatment= 4 weeks

Loss-to-follow-up / Incomplete outcome data:

A: 3/11 (27%) in intervention, none in control group Reasons: unknown (1), transportation (1), family issues (1)

Note: participants received money in order to reduce study attrition (130 dollars in total if completed)

B: 2/22 (9%); 1 in each group

Reasons: declining health (..), loss of interest

Note: participants received money in order to reduce study attrition (50 dollars gift card if completed)

Pain; QoL; safety (AEs)

A: WHYMPI (pain severity; pain interference); BDI

West Haven Yale Multidimensional Pain Inventory [not validated for neuropathic pain]; WHYMPI-PS subscale for pain severity (0-6); WHYMPI-I subscale for pain interference with day to day activities (0-6); Beck Depression Inventory (0-63; 0-10 normal, 41-63 extreme)

Group differences (baseline-to-end of follow-up):

in intervention group, pain severity decreased mean

1.08 (0.79), Control arm

unchanged, mean 0 (0.51),

P <0.01 [p-value corrected based on original publication]

in intervention group, pain interference declined

CBT mean 1.35 (SD 1.22),

control arm increased mean

0.22 (SD 0.73), P <0.05 [p-value corrected based on original publication]

in intervention and control group, scores on BDI did not significantly differ between baseline and 4-month follow-up

Safety (AEs): not reported

B: NPS; NeuroQoL

NPS: no significant difference in pain intensity or pain

unpleasantness

NeuroQoL: no significant difference in overall, symptom related or pain quality of life

Safety (AEs): not reported

Davies (2015) concludes that there is insufficient evidence to make recommendations on psychological therapies for PDN; while the results from

Otis (study-A) appear encour-

aging, the high dropout rate limits the quality and validity of the findings

Authors of study-A suggest that CBT for painful DPN may provide the skills to become more active and experience less pain

Authors of study-B state that no statistically significant differences were found between Mindfulness meditation and control groups; however, mean values of pain quality of life and symptom-related quality of life differed and further

studies may show efficacy (..)

Note: RoB analysed using Cochr RoB tool. For reasons of consistency RoB was determined separately (see separate RoB table)

Note: both studies are characterized as pilot studies and have a small sample size; power calculation is missing in study-A; study-B referes to power calculation but does not meet the target; both studies appear underpowered