Levator ani muscle (LAM) motor control may play a role in the pathophysiology of stress urinary incontinence (SUI), as proprioception has a demonstrated impact on motor control and rehabilitation outcomes in other body regions. Yet, a recent study reported that women with SUI demonstrated more, not less, accurate target force reproduction than continent controls[1]. While research into LAM proprioception is in its early stages, it is important to develop valid and standardized approaches to facilitate comparisons among studies. The goal of this study was to propose novel and robust methods for assessing LAM proprioception during active contraction and passive elongation. The objectives were to: 
1)	Develop a standardized protocol to assess accuracy during a LAM force-matching task, through:
a.	Determining the optimal window length and location for the extraction of root mean square error (RMSE) from the target force to assess accuracy.
b.	Assessing the effect of visual feedback and target force level on accuracy.
c.	Determining if maximum voluntary contraction (MVC) force influences contractile accuracy. 
2)	Develop a standardized protocol to assess awareness of passive LAM elongation in women with and without SUI through:
a.	Describing women’s performance on the task and evaluating potential ceiling and floor effects in women with and without SUI.
b.	Determining if the reference intravaginal dynamometer (IVD) diameter influences accuracy.

This observational study received approval from the local institutional research ethics board. Adult women were recruited into SUI and control (no lower urinary tract symptoms) cohorts, being ineligible if they reported other symptoms of pelvic floor disorders. Participants provided informed consent. 

Testing was conducted with participants in supine using an automated IVD with previously reported specifications and reliability metrics[2,3]. The passive elongation task was performed first. A reference diameter of the IVD arms was set individually as the diameter that resulted in a stable passive force between 1.4-1.5N. Three test diameters were then assigned:  the reference diameter and ±5mm from it. During testing, the IVD arms were inserted intravaginally and opened to the reference diameter, held for 7s then closed, followed 30s later by opening to the test diameter, holding for 7s, and closing. The participant reported whether the arms had opened more, less or the same as the reference diameter. Arm opening speed was fixed at 20mm/s for all trials and each test diameter was presented 3x in a random order. 

For the force-matching task, MVC force for the LAMs was determined first. The IVD arms remained in situ and opened to a diameter of 35mm. After the stress relaxation response, the participant was asked to contract their PFMs as strongly as possible for 3s. The MVC force was recorded as the highest relative peak force across three trials. PFM force targets were then set at 25%, 50%, and 75% MVC. Up to three practice trials were provided with and without visual feedback at each target force. For the test, participants were instructed to contract their LAMs to reach but not to exceed the target force (with or without the target force displayed on the computer screen), to hold the contraction steady for 5s, then relax. Each target force was presented three times under each feedback condition, with the order fully randomized. One minute of rest was given between trials. 
RMSE from the target was calculated across each contraction from the first force peak to 2.5s after said peak using 3 different window lengths (0.25s, 0.50s, and 1.00s). A linear mixed model ANOVA was used to assess the impact of window length, feedback (visual or no visual), and target force on RMSE. This was followed by a series of RM-ANOVAs to determine the location along the force-time curve where the RMSE was stable. Spearman correlations were used to determine whether MVC force was associated with accuracy across conditions. Finally, a two-way RM ANOVA was used to assess the impact of feedback type and target force level on accuracy. Sphericity was assumed for all variables except for target force level which used the Greenhouse-Geisser correction.

Thirty women (44±13 years) completed the assessment; force-matching data from one participant were lost due to equipment failure. Sixteen participants reported SUI symptoms while 14 reported no lower urinary tract symptoms. Eighteen participants were parous (2±1 deliveries).

The average reference IVD diameter used for the passive elongation task was 36mm±3.25mm. While participants correctly identified the IVD opening diameter relative to the reference diameter 77.8% (25%) of the time, there was a large range in scores (22% – 100%). There were no ceiling or floor effects with only 4 individuals reaching 100% (controls n=3; SUI n=1) and none reaching zero, though 4 individuals were below 50% (controls n=2; SUI n=2). Spearman correlations indicated that there was no association between the reference diameter and accuracy on the passive elongation task (r = -.29, p = .13).

For the force-matching task, window length had no effect on measured accuracy (F(2, 2945)=1.84, p=.16). RMSE increased across the duration of the contraction under both feedback conditions (p=.00; Figure 1). With visual feedback, RMSE was stable between 1.0 and 1.5s after the first force peak when assessed using 0.25s and 0.50s window lengths and at force targets of 25% and 50% MVC. Without visual feedback, RMSE was stable between 1.0 and 2.0s after the first force peak for the same window lengths, but only at the 25% MVC target force (See Figure 1). 

Both feedback (F(1, 28)=7.58, p=.01, η2=.21) and target force (F(1.41, 39.34) = 35.16, p=.00, η2 =.56) significantly affected participants’ ability to match target forces; with no interaction (F(2, 56)=0.49, p=.62, η2=.02). Participant accuracy was significantly better with visual feedback (p=.01) and participants were more accurate at lower target forces (p=.00; Figure 2).
MVC force did not affect force accuracy except for the 25% target with visual feedback (r=-.45, p=.01), where participants who generated higher MVC forces were more accurate.

Accuracy during force-matching tasks should be assessed using a window length of 0.25s or 0.50s and accuracy values should be reported for a window between 1.00s and 1.50s after the initial force peak. Participants tended to be less accurate the longer they held a contraction, especially at the 75% MVC target. Both visual feedback and lower target force improved accuracy, which is consistent with findings in peripheral muscles. These results suggest that force matching should be assessed under visual feedback and no feedback conditions as well as at a range of target forces, because these factors influence outcomes.
The approach presented to evaluate elongation awareness resulted in a wide range of accuracy outcomes and was not influenced by the reference IVD diameter. There was no ceiling effect noted overall, nor in either cohort (SUI vs control).

This study represents a first and necessary step in systematically and objectively assessing the proprioception of the LAMs in women with and without SUI using any IVD capable of arm opening. Active force protocols should include different target forces and feedback conditions, and signal processing parameters should be standardized as they influence outcomes. Our proposed passive elongation task appears suitable for studying differences between women with and without SUI. Information gained from LAM proprioceptive assessment may improve our understanding of the role of the LAMs in pelvic floor disorders and may help to refine targeted treatment strategies.

Kharaji G, Nikjooy A, Amiri A, et al. Better Sense of Force Accuracy in Women with Stress Urinary Incontinence Compared with Women Without Stress Urinary Incontinence: A Surprising Result in a Case-Control Study. Middle East J Rehabil Heal Stud. 2022;10(1):1-9. doi:10.5812/mejrh-131059
Bérubé M-È, Czyrnyj CS, McLean L. An automated intravaginal dynamometer: Reliability metrics and the impact of testing protocol on active and passive forces measured from the pelvic floor muscles. Neurourol Urodyn. April 2018:1-14. doi:10.1002/nau.23575
Czyrnyj CS, Bérubé M, Lanteigne E, et al. Design and validation of an automated dual-arm instrumented intravaginal dynamometer. Neurourol Urodyn. 2021;40(2):604-615. doi:10.1002/nau.24600

Continence 7S1 (2023) 100998
DOI: 10.1016/j.cont.2023.100998