Skip to main content

Table 3 Summary of studies that compared EA to ER

From: The effects of error-augmentation versus error-reduction paradigms in robotic therapy to enhance upper extremity performance and recovery post-stroke: a systematic review

Article Study design Number of participants: experimental group (E) and control group (C) Participants characteristics Equipment Experimental Protocol Outcomes and assessment tools Main results and interpretation (means ± standard deviation) Effect size (Cohen’s d) Quality of study (PEDro score)
Bouchard et al. [61] Randomized comparative experiment.
Trial not registered.
34 in total
EA group: 17 haptic guidance (ER) group: 17
ER group: age (mean) = 67 ± 7, months since stroke (mean) = 63 ± 54 AMFMa = 63 ± 8
EA group: age (mean) = 67 ± 6, months since stroke (mean) = 78 ± 64 AMFM = 60 ± 10
TEO, a robotic device with 10-degree flexion/ extension of the left/ right wrist, actuated by Dynamixel MX-106 servomotor Experimental task: flex paretic wrist at the right time.
ER group:
the robotic device adjusted its activation time to reduce the timing errors (k value decreased by 90%).
EA group: the opposite, the timing errors were increased (k value increased by 90%)
Training parameters: Four phases of baseline assess-ment (140 trials) before the intervention. 75 trials during the inter-vention, and 40 trials at the reten-tion phase.
Kinematic:
Absolute timing errors (ms)
A significant decrease of 1.1 ± 5.1 ms in absolute timing errors in the ER group (p = 0.032), and a non-significant increase of 0.4 ± 6.0 in the EA group (p = 0.45).
A between group comparison revealed no significant difference between the two groups (95% confidence interval: − 1.2 to 4.3)
0.27 (small effect size) 8/10, high quality
Cesqui et al. [58] Crossover design.
Trial not registered
15 in total
EA group: 6
ER group: 9
(before cross-over)
Ages: 20–71 years (mean = 42 ± 17) 8 males and 7 females, all participants suffered from stroke (stages unknown) CMb: First EA group: mean = 5 ± 0.89
First ER group: mean = 4 ± 0.86
InMotion2 Experimental tasks: reaching targets in a plane.
EA group: received divergent field (negative elastic force)
ER group: received active assistance
Training parameters:
One hour per session, ten sessions per therapy cycle which lasted two weeks before subjects switched groups.
Kinematic:
Metric indexes (movement smoothness, movement accuracy, path length ratio, movement direction variability)
Clinical: MSSc, MASd, ROM Shoulder and Elbow
MAS: in the first cycle, ER showed more improvement than EA (3.5 ± 2.8 vs 1.8 ± 3.6), but level of significance not provided.
In the second cycle, ER still showed more improvement than EA (0.9 ± 3.5 vs 0.3 ± 2.7), but level of significance not provided.
MSS: in the first cycle, ER showed more improvement than EA (2.9 ± 7.1 vs 1.8 ± 5.2), but level of significance not provided.
In the second cycle, ER still showed more improvement than EA (1.0 ± 4.8 vs 0.6 ± 6.4), but level of significance not provided.
ROMs: no significant changes (numerical values not provided)
Metric indexes: no numerical values reported, so unable to calculate differences between groups. The authors reported final metric indexes differences were not significant in the group started with EA (F = 1.61, p = 0.194) but in the group started with ER, there was a significant improvement indexes (F = 9,46, p = 0.006). They did not mention the comparison of metric indexes between groups.
MAS: first cycle 0.53 (medium effect).
second cycle 0.19 (very small effect).
MSS: first cycle 0.18 (very small effect).
second cycle 0.09 (very small effect).
3/10, poor quality
Patton et al. [24] Quasi-experimental design.
Trial not registered
31 in total,
Stroke Group EA = 9
ER = 9
C = 9
Healthy Group
EA = 2
ER = 2
Ages = 30–72 years (EA: mean = 54.3 ± 8.8;ER: mean = 48.0 ± 8.4;Control: mean = 51.2 ± 6.1), besides 4 healthy subjects, all participants suffered from a chronic stroke (16–173 months, EA: mean = 69.1 ± 50.2; ER: mean = 109.3 ± 45.8; Control: mean = 70.8 ± 60.4), FM: EA group mean = 40.2 ± 13.7
ER group mean = 25.5 ± 10.9
Control group mean = 37.3 ± 16.2
Free-extremity robot with two degrees of freedom. The participant’s arm was supported by a low-friction, low-impedance mechanism Experimental tasks: reaching
EA group: both stroke and healthy EA groups received force field that magnified errors (EA)
ER group: both stroke and healthy ER groups received force field that reduced error. In the stroke control group, the 9 participants with stroke did not receive interfering forces.
Training parameters:
One single session of 834 movements.
Kinematic: initial direction error (degrees). Adaptation capacity The stroke EA group showed improvement at initial direction error (8.9 ± 10.9) while the stroke ER group showed deterioration (− 6.8 ± 9.6). The different between EA and ER groups was significant [F(1,13) = 4.29, p < 0.001].
Stroke subjects showed less adaptation capacity than healthy subjects (26% less)
Initial direction error: 1.53 (very large effect) 1/10, poor quality
Tropea et al. [54] Crossover randomized controlled trial.
Trial not registered
18 in total
EA = 9
ER = 9 (before cross-over)
Ages: 21–71 (EA: mean = 49.7±
18.7; ER: mean = 44.9 ± 15.9), 9 males and 9 females, all participants suffered from a chronic stroke (mean/SD unknown), CM: First EA group: mean = 4.9 ± 0.9
First ER group: mean = 4.2 ± 1.0
InMotion2 robotic system Experimental tasks: reaching targets in a plane.
EA group: received divergent force field
ER group: received active assist during practice
Training parameters:
Two weeks of training per cycle, and two cycles in total. After each cycle, subjects switch groups.
Kinematic: the trajectory of the end-effector
Clinical:
MAS, MSS
MAS: in the first cycle, ER group showed more improvement than EA (2.9 ± 3.2 vs 1.2 ± 3.2), but not significantly.
In the second cycle, ER group still showed more improvement than EA (1.4 ± 1.2 vs 0.7 ± 2.3), but not significantly.
MSS: in the first cycle, ER group showed more improvement than EA (2.2 ± 2.0 vs 0.8 ± 3.5), but not significantly.
In the second cycle, ER group still showed more improvement than EA (1.4 ± 1.3 vs 1.1 ± 1.1), but not significantly.
Trajectory of end-effector: no numerical values reported, but authors stated that EA group had significantly straighter (p = 0.028) as well as smoother (p = 0.031) trajectory than ER group
MAS: first cycle 0.53 (medium effect)
Second cycle 0.40 (small effect)
MSS: first cycle 0.51 (medium effect)
Second cycle 0.25 (small effect)
6/10, high quality
  1. aAMFM: Arm Motor Fugl-Meyer
  2. bCM: Chedoke-McMaster scale score
  3. cMSS: Motor Status Score
  4. dMAS: Modified Ashworth Scale