Is pelvic floor loading in female runners associated with post‐run changes in pelvic floor morphometry or function?

Abstract

Objectives

To investigate whether, among female runners, transient changes in pelvic floor morphometry, pelvic floor muscle (PFM) function, or pad weight gain observed after a standardised running protocol are associated with the exposure of the pelvic floor to loading during the run.

Methods

Adult female runners with (n = 19) and without (n = 19) running‐induced stress urinary incontinence (RI‐SUI) completed a 37‐min treadmill‐based running protocol with a pressure sensor placed in the posterior fornix of the vagina and a triaxial accelerometer adhered to the pelvis, and an incontinence pad adhered to their undergarment. Pelvic morphometry and PFM function were assessed before and after the run using transperineal ultrasonography and intra‐vaginal dynamometry. Urine leakage volume was estimated based on incontinence pad weight gain. Separate linear regression models were used to evaluate the associations between variables representative of pelvic floor load exposure (posterior fornix sensor pressure [PFSP] and pelvic accelerations) and changes in pelvic morphometry, PFM function, and incontinence pad weight observed after the run.

Results

After the run, the levator hiatus was larger and the bladder neck sat lower in the pelvis, but there were no significant differences in PFM active or passive forces measured using dynamometry. These changes were not different between those with and those without RI‐SUI. Higher pelvic accelerations were associated with greater reductions in passive PFM stiffness after the run (R ² = 20%–27%), but not with changes in pelvic morphometry. No associations were found between any measures of pelvic floor load exposure and changes in PFM force‐generating capacity. Among runners with RI‐SUI, greater pad weight gain occurred among those who ran with slower vector accelerations (R ² = 0.27).

Conclusion

The magnitude of pelvic floor loading experienced during running does not appear to influence the transient loss in static pelvic organ support observed after running nor urine leakage volume.

Abbreviations

2D

two‐dimensional

3D

three‐dimensional

AUC

area under the curve

BN

bladder neck

IAP

intra‐abdominal pressure

LH

levator hiatus

LP

levator plate

MVC

maximum effort voluntary contraction

PFM

pelvic floor muscle

PFSP

posterior fornix sensor pressure

RI‐SUI

running‐induced stress urinary incontinence

SUI

stress urinary incontinence

Introduction

Running is a popular activity and a component of many sports. However, a significant number of female runners report discontinuing this form of exercise due to prevalent lower urinary tract symptoms [1], mainly stress urinary incontinence (SUI) [2, 3, 4, 5]. Exposure to high‐impact forces during activities such as running has been proposed as a contributor to incontinence experienced during athletic activities [6, 7]. High intra‐abdominal pressure (IAP) has also been hypothesised to lead to symptoms of SUI during athletic activities [8, 9, 10, 11, 12, 13, 14, 15].

Few studies on exercise‐induced SUI have focused on runners. A recent study found that a 37‐min run induced a loss of passive support without concurrent pelvic floor muscle (PFM) fatigue [14]. Another study found that after performing a half‐marathon, female runners experienced fatigue‐related changes in PFM function including strength, endurance, and ability to complete repeated contractions [16]. de Melo et al. [15] investigated biomechanical correlates of urinary incontinence among runners, finding no differences in running mechanics between runners with and without urinary incontinence. They also found no association between the severity of leakage experienced during running and these running mechanic parameters [15]. Similar results were found in a sample of postpartum runners where, during a 7‐min treadmill run, there was no evidence of an association between symptoms of pelvic floor dysfunction and both running kinetics and pelvic accelerations [17].

Pressure measured by a sensor positioned within the posterior fornix of the vagina has been theorised to reflect pelvic floor loading [18, 19, 20, 21, 22]. Such sensors have been used to infer pelvic floor loading in groups of females performing different physical activities [18, 19, 20, 21, 22]. Pressure sensor data have, however, been highly variable among individuals who perform different exercise activities. The sensor measures higher pressure with tasks that increase IAP (abdominal wall muscle contraction, higher respiration volume and greater trunk flexion) and those associated with inertial loading (higher pelvic acceleration) [23], but it remains unclear how exposure of the pelvic floor to these forces affects pelvic organ support or PFM function. The magnitude of accelerations experienced by the pelvis during running likely reflects the magnitude of mechanical stresses experienced by the PFMs and connective tissues related to inertial loads and ground reaction forces being transmitted up the kinetic chain [18]. Such mechanical stress may lead to tissue strain, and, consequently, a loss of pelvic organ support. Currently, it is not clear how exposure to loading during running ultimately contributes to urine leakage among females. Understanding these relationships will allow us to explore biomechanical interventions to prevent or alleviate symptoms of SUI experienced by females during physical activities such as running.

The aim of this study was to determine, among experienced female runners, if the magnitude of loading experienced by the pelvic floor tissues (inferred both through pressure recorded by a sensor placed in the posterior fornix and through pelvic accelerations) over the duration of a standardised treadmill running protocol is associated with the extent of observed changes in pelvic floor morphometry or PFM function observed after the run. A second aim was to assess if the extent of urine leakage measured among runners with running‐induced stress urinary incontinence (RI‐SUI) is associated with the magnitude of pelvic floor loading experienced during running. Our primary hypothesis was that greater pelvic floor loading would be associated with greater reductions in pelvic organ support and PFM stiffness measured after the run. Our secondary hypothesis was that, among female runners with RI‐SUI, higher pelvic floor loading would be associated with more urine leakage during the run.

Methods

Study Design

This was an observational, cross‐sectional exploratory study. Our local institutional research ethics board approved the protocol prior to recruitment, which occurred between June 2019 and March 2021, with a break between March 2020 and January 2021 due to laboratory closures caused by the COVID‐19 pandemic. Prior to participating in any part of the protocol, all participants provided written informed consent.

Participants

Experienced runners were recruited through local running groups and physiotherapy clinics. The inclusion and exclusion criteria are presented in Table 1. All eligible participants provided demographic information including age, parity, obstetric history (i.e. vaginal vs Caesarean section, use of forceps or suction) and menopausal status. Participants were also asked to complete the International Consultation on Incontinence‐Female Lower Urinary Tract Symptoms Long Form module (ICIQ‐FLUTS) and the International Physical Activity Questionnaire (IPAQ) [24], and to provide general information about their running profile such as frequency, volume, and experience. Due to the novel nature of this study, there were no prior data available to inform a sample size estimate. Consequently, the sample size was determined based on practical considerations and the availability of resources, with assumptions that a sample size >25 would be adequate to detect moderate linear associations [25].

Table 1.

Inclusion and exclusion criteria.

Inclusion criteria

Exclusion criteria

Runners assigned female at birth Older than 18 years old Run at least 10 km/week for the past year Run at least twice per week for the past year Can run at least 5 km in under 50 min

Risk factors for physical activity according to Par‐Q Self‐reported obesity or obesity defined by a BMI > 35 kg/m2 History of urogenital surgery and/or use of incontinence medication Neurological impairments involving the central nervous system or the sacral nerves Known connective tissue disorders Pelvic organ prolapse that moves caudally past the hymen on Valsalva manoeuvre while standing Pregnant or had delivered a baby within the previous year Dyspareunia or pelvic pain Symptoms of low energy deficiency as measured by the Low Energy Availability in Females Questionnaire (LEAF‐Q) In situ devices that would not be suitable for ultrasonography, intra‐vaginal dynamometry, and/or intra‐vaginal pressure sensor placement Urinary urgency with associated urine leakage during running

BMI, body mass index.

Laboratory Assessment

The first author, a registered physiotherapist who has worked exclusively in pelvic health, and with over 5 years of experience at the time of data collection, conducted all laboratory assessment data. Additionally, she received over 50 h of training on ultrasonography and intra‐vaginal dynamometry to assess pelvic morphometry and PFM function prior to initiating this study and was trained on and deemed competent by the device developers on the correct use of the pressure sensor.

Further details on the procedures used in this study can be found in a previous publication [14]. In brief, using transabdominal ultrasonography, bladder volume was confirmed to be between 100 and 200 mL to ensure consistency among participants during the running protocol. This decision was made to assess urine leakage during exercise with a moderately full bladder since female runners tend to void their bladder prior to going for a run and even during the run as a mitigation strategy to prevent leakage [26]. Before running, the morphometry of the PFMs was assessed using two‐dimensional (2D) and three‐dimensional (3D) B‐mode transperineal ultrasonography through a GE Voluson S6 system (General Electric, Toronto, ON, Canada). While in a standardised standing position, static 3D volumes (to measure the area of the levator hiatus [LH]) and dynamic 2D clips (to measure levator plate [LP] length and bladder neck [BN] height) were obtained at rest and during maximum effort voluntary contractions (MVCs). Also, in standing position, an intravaginal dynamometer was used to measure MVC forces. The arms were opened to an anteroposterior diameter of 35 mm and the participant was instructed to perform an MVC. The participant then moved to a supine position and the dynamometer was used to record passive forces as the diameter of the dynamometer arms moved at a constant 50 mm/s from 15 to 40 mm diameter. The elongation was held for 7 s before the arms returned to their initial position. Three sequential repetitions of each task were performed using each device.

The dependent variables were selected as those previously shown to be impacted by an acute bout of running among experienced female runners both with and without RI‐SUI [14]. Imaging outcomes included the area of the LH at rest, acquired using 3D imaging in standing position, and BN height and LP length at rest, acquired using 2D imaging in standing position. The amount of change (Δ) in the LP length and BN height was acquired as the difference in these measures between rest and peak MVC [27, 28]. Dynamometry outcomes included peak force and rate of force development observed during active contraction in standing position and passive PFM elongation in supine position [29, 30]. The relevance of each outcome is presented in Box 1.

The participants proceeded to insert a validated [31, 32, 33], reliable [34], sterilised, and calibrated (0–350 cm H₂O) pressure sensor into the posterior fornix of their vagina, which was zeroed to atmospheric pressure. Adequate positioning of the device was confirmed by the physiotherapist using digital palpation in the supine position by palpating the cervix and ensuring that the sensor was inserted posterior to it and as far as possible. The participants were then instructed to get dressed again in their underwear and in standardised Spandex running pants provided by the laboratory, which held the wires still while running. Posterior fornix sensor pressure (PFSP) data were acquired at a sampling rate of 40 Hz and stored on a Secure Digital card within an instrumented module that was secured to the participant's left iliac crest with adhesive tape. To ensure that the sensor was securely positioned in the upper third of the vagina, the participants were instructed to stand up and perform three maximal effort coughs, three maximal Valsalva manoeuvres, three vertical jumps, and three repetitions of a standardised lifting task with 6.8 kg in each hand. On visual inspection, if the sensor descended into the lower part of the vagina or came out, it was reinserted, and the process was repeated. If the sensor still descended, it was removed, and the participant continued the study without the sensor. A tri‐axial accelerometer (MetaMotionR; Mbientlab Inc.) with a range of ±16g was adhered using tape directly to the skin over the upper border of the left mid‐iliac crest and was used to record pelvis acceleration values in the vertical, anteroposterior and mediolateral directions at a sampling rate of 100 Hz.

With a pre‐weighed incontinence pad in their underwear, and a fitness tracker watch (Fitbit) on their wrist, participants proceeded to a NordicTrack Commercial 2450 treadmill. They gradually increased the belt speed to 7 km/h, jogged for 2 min, then increased the speed to 10 km/h, ran for 2 min, then increased the speed to 15 km/h, ran for 2 min then selected a running speed that they perceived as somewhat hard (15–16 on the Borg Rate of Perceived Exertion Scale) and maintained it for 30 min, after which they decreased the treadmill speed to 10 km/h and then 7 km/h, running for 30 s at each speed, at which point they reduced the speed to zero to end the protocol. Throughout the entire protocol, the instrumentation module recorded pressure data. If the participants experienced shortness of breath or discomfort during the running task, they had the option to reduce their speed or, if necessary, walk for 1 min before resuming their selected speed. Drinking water during the running protocol was permitted. Following completion of the running protocol, the physiotherapist examined the positioning of the pressure sensor, determining whether or not it remained within the posterior fornix of the vagina before removing it. The ultrasonography and intra‐vaginal dynamometry procedures were then repeated as described above, completed within 15 min of the end of the running protocol. Pad weight and total number of steps taken during the run were recorded. The researcher who conducted the laboratory assessment was not blinded to continence status as she could visually detect urine in the pad on removal for weighing.

Data Processing

The change in PFM outcomes was calculated as the difference between the post‐run measurements and the pre‐run measurements.

All PFSP data processing was performed using custom MATLAB programs and followed methods previously established by the developers of the device [35]. Following a power spectrum analysis, no filtering was deemed necessary to smooth the pressure data as there was no evidence of high frequency noise. Maxima and minima in the pressure data were identified using custom software, with successive peaks being at least 0.2 s apart. Confirmation via visual inspection ensured that pressure peaks were identified correctly. Four separate measures were derived from the PFSP data to quantify pelvic floor load exposure. Firstly, the difference between the mean of all maxima and of all minima was used to compute the ‘net mean peak PFSP’ during the 30‐min run at the self‐selected speed [35]. Secondly, the net mean peak PFSP calculated over the highest data point acquired during the three vertical jumps [36] was used to normalise the net mean peak PFSP during running, producing the third measure, the ‘relative net mean peak PFSP’. This measure, presented as a percentage, was used to allow meaningful comparisons among individuals considering variations in body mass and the mass of the abdominal/pelvic organs. Fourthly, the area under the PFSP curve (PFSP AUC) was averaged over 10 min of running at the self‐selected speed to represent cumulative loading [35]. While these measures were not independent (net mean PFSP vs relative net mean PFSP r = 0.777; net mean PFSP vs PFSP AUC r = 0.879, relative net mean PFSP r = 0.635), they were each selected as they provide different information about pelvic floor loading.

The raw pelvic acceleration time series were smoothed using a second order dual‐pass Butterworth filter with a 6‐Hz cut‐off. Vector accelerations (vertical [y], sagittal [x] and coronal [z]) were computed over 10 consecutive steps at the self‐selected speed within the 15‐min mark of the 30‐min run, using the peak acceleration values from each step (Equation 1), where a _x (sagittal), a _y (vertical), and a _z (coronal), represent acceleration amplitudes in the x, y, z directions, respectively, and $\left|\overrightarrow{a}\right|$ represents the vector acceleration. Again, while not independent (r = 0.693), both vector and vertical accelerations provide different information and were thus both included in the statistical analyses.

$$\left|\overrightarrow{a}\right|=\sqrt{a_x^2+a_y^2+a_z^2}$$ (1)

Statistical Analysis

All statistical testing was carried out using IBM SPSS version 28 and α = 0.05 was used for all tests. All outcomes were tested for normality using the Shapiro‐Wilks test. Descriptive statistics (mean, sd) were calculated for each outcome. Confidence intervals were calculated for the mean difference in PFM outcomes to describe the acute effects of the run. To test the primary hypothesis, separate linear regression analyses were conducted to evaluate the relationships between pelvic floor loading variables (net mean PFSP, relative net mean PFSP, PFSP AUC, vector acceleration and vertical acceleration) recorded during the run and changes in pelvic morphometry and PFM function after the run. The baseline value of each PFM morphometry or function outcome was included in its respective models as a covariate, since it was previously shown that baseline values of BN height and LP length can affect the extent of changes observed after a run [14]. To assess our secondary objective, among female runners with RI‐SUI, linear regression analyses were conducted to evaluate the relationship between each of the variables representing pelvic floor loading and pad weight gain after the running protocol. Residual plots were examined to ensure proper model fit.

Results

Sample Demographic Information

Thirty‐nine female runners met the eligibility criteria. However, only 38 successfully completed the laboratory‐based assessment due to hardware issues with the dynamometer. Hardware difficulties due to faulty connections were encountered with the vaginal pressure sensor for another five participants. PFPS data were discarded for six participants due to the sensor having descended from the posterior fornix during the running task. Ultimately, PFSP datasets were available from 27 participants.

Demographic data are presented in Table 2. Baseline values, post‐run values and mean differences in all PFM outcomes are summarised in Table 3. The LH was significantly larger, and the BN was significantly lower within the pelvis in standing position at rest after the run. Runners were also able to shorten their LP further when performing an MVC after the run. The PFSP and pelvic acceleration data measured during the 30‐min self‐selected running speed are detailed in Table 4. The equations derived from the regression analyses are presented in Table S1. While most models were nonsignificant, as described below, the residual plots confirmed that all models met the necessary criteria for adequate model fit.

Table 2.

Demographic data for the whole sample (n = 38) and for participants in whom posterior fornix sensor pressure data were available for regression analysis (n = 27).

	All participants (n = 38)		PFPS subset (n = 27)
	Mean/n	sd or %	Mean/n	sd or %
Age, years	39.91	10.12	40.74	11.13
BMI (kg/m2)	23.03	3.99	23.25	3.68
Waist to hip ratio	0.79	0.07	0.79	0.08
Parous	17	45	11	41
Use of forceps (n, %)	4	11	3	11
Use of suction (n, %)	3	8	1	4
Menopause (n, %)	4	11	4	15
RI‐SUI (n, %)	19	50	15	56
ICIQ‐FLUTS total score (/48)	7.37	4.56	8.07	4.50
ICIQ‐FLUTS F (/16)	2.89	2.13	3.33	2.04
ICIQ‐FLUTS V (/12)	0.79	1.70	0.67	1.14
ICIQ‐FLUTS I (/20)	3.68	3.02	4.07	3.15
IPAQ total activity (MET‐min/week)	4389	5951	5531	6739
IPAQ subtotal vigorous activities (MET‐min/week)	1503	2268	1819	2616.
Self‐selected running speed (km/h)	10.30	1.64	10.16	1.77
Number of steps during running protocol	6489	384	6408	307
Weekly running distance (km)	36.2	23.6	36.1	23.0
Running experience (years)	15.70	10.54	15.49	11.17

Values are presented as mean ± sd or frequency and percentage (%).

BMI, body mass index; F, filling subscore; I, Incontinence subscore, IPAQ, International Physical Activity Questionnaire; ICIQ‐FLUTS, International Consultation on Incontinence‐Female Lower Urinary Tract Symptoms Long Form module; MET, metabolic equivalent units; PFSP, posterior fornix sensor pressure; RI‐SUI, running‐induced stress urinary incontinence; V, voiding subscore.

Table 3.

Baseline outcomes and post‐run changes measured by intra‐vaginal dynamometry and ultrasonography.

	Before the run		After the run		Difference (Δ)
	Mean	sd	Mean	sd	Mean	sd	95% CI	n
Dynamometry outcomes during MVC
Rate of force development (N/s)	9.45	6.08	9.45	6.72	0.00	2.70	−1.024 to 1.033	28
Relative peak force (N)	3.64	1.51	3.91	1.68	0.27	0.86	−0.054 to 0.602	28
Dynamometry outcomes during passive elongation
Rate of force development (N/s)	30.33	11.69	28.69	10.70	−1.64	6.58	−4.137 to 0.865	28
Relative peak force (N)	11.38	5.16	10.88	5.73	−0.50	2.05	−1.280 to 0.279	28
3D ultrasonography measurements
LH area at rest (cm2)	16.90	4.94	17.70	5.12	0.80	1.35	0.324 to 1.267	33
2D ultrasonography measurements at rest
BN height (mm)	16.75	4.78	14.59	4.13	−2.16	3.29	−3.273 to −1.049	35
LP length (mm)	55.59	8.24	56.92	9.06	1.33	4.55	−0.187 to 2.848	36
2D ultrasonography measurements change on MVC
BN height (mm)	3.36	3.56	4.01	2.94	0.66	2.55	−0.218 to 1.532	34
LP length (mm)	−10.38	4.64	−11.89	5.59	−1.51	4.07	−2.890 to 0.133	35

Confidence intervals significantly different from zero are shown in bold.

Δ, absolute difference; 2D, two‐dimensional; 3D, three‐dimensional; BN, bladder neck; LH, levator hiatus; LP, levator plate; MVC, maximal voluntary contraction; N/s, Newtons per second; N, Newtons.

Table 4.

Pelvic floor loading outcomes measured using intra‐vaginal posterior fornix sensor and pelvic accelerometers during running for 30‐min at a self‐selected speed.

	n	Mean	sd	Range	CoV (%)
Net mean peak PFSP (cm H2O)	27	43.84	13.21	17.66–77.42	30.1
Relative net mean peak PFSP (%)	27	57.72	22.00	27.80–99.80	38.1
AUC PFSP (cm H2O·s)	27	177.10	37.22	114.69–255.66	21.0
Vertical acceleration (g)	38	2.62	0.28	2.06–3.26	10.2
Vector acceleration (g)	38	3.52	0.40	2.75–4.24	11.7

AUC, area under the curve; CoV, coefficient of variation; g, gravitational unit; PFSP, posterior fornix sensor pressure.

Relationship Between Changes in Pelvic Floor Loading Exposure During Running and Changes in Pelvic Morphometry and PFM Function After the Run

The outcomes of the linear regression analyses are presented in Table 5. The only regression models that were significant were the amount of change in relative peak passive force observed on PFM elongation and pelvic accelerations (both vertical and vector) observed during running. Figures 1 and 2 present scatterplots showing the relationships between pelvic floor loading exposure (net mean PFSP and pelvic vertical accelerations) and both dynamometry and ultrasonography outcomes, respectively. The outcome of the regression analyses is described in more detail below.

Table 5.

Regression coefficients and CIs for the relationships between pelvic floor loading (posterior fornix sensor pressure and pelvic accelerations) and change in pelvic floor muscle function and morphometry outcomes.

	Net mean PFSP			Relative net mean PFSP			AUC PFSP			Vertical acceleration			Vector acceleration
	R 2	B	95% CI	R 2	B	95% CI	R 2	B	95% CI	R 2	B	95% CI	R 2	B	95% CI
Dynamometry outcomes during MVC
Rate of force development	0.01	0.026	−0.128 to 0.180	0.02	0.016	−0.067 to 0.099	0.008	0.002	−0.045 to 0.050	0.00	0.071	−4.083 to 4.225	0.081	1.985	−0.72 to 4.690
Relative peak force	0.10	−0.019	−0.064 to 0.026	0.07	0.005	−0.018 to 0.029	0.118	−0.007	−0.021 to 0.007	0.01	−0.028	−1.348 to 1.292	0.094	0.683	−0.205 to 1.571
Dynamometry outcomes during passive elongation
Rate of force development	0.17	−0.104	−0.498 to 0.291	0.19	−0.077	−0.268 to 0.115	0.175	0.035	−0.083 to 0.154	0.26	−7.141	−15.913 to 1.631	0.27	−5.187	−11.181 to 0.807
Relative peak force	0.20	−0.098	−0.217 to 0.020	0.18	−0.046	−0.107 to 0.015	0.048	0.004	−0.036 to 0.045	0.20	−3.399	−6.228 to −0.569	0.273	−2.747	−4.585 to −0.908
3D ultrasonography measurements
LH area at rest	0.12	−0.022	−0.066 to 0.022	0.12	−0.014	−0.042 to 0.013	0.082	−0.003	−0.019 to 0.013	0.01	−0.347	−2.233 to 1.538	0.037	−0.635	−1.825 to 0.555
2D ultrasonography measurements at rest
BN height	0.21	0.046	−0.035 to 0.128	0.48	0.020	−0.027 to 0.067	0.491	0.016	−0.014 to 0.045	0.30	−1.455	−5.151 to 2.241	0.281	−0.130	−2.750 to 2.490
LV length	0.03	0.096	−0.076 to 0.267	0.11	0.054	−0.043 to 0.151	0.088	0.026	−0.034 to 0.087	0.08	4.538	−1.105 to 10.181	0.012	−0.804	−4.749 to 3.141
2D ultrasonography measurements change on MVC
BN height	0.35	0.000	−0.082 to 0.082	0.35	−0.003	−0.050 to 0.043	0.35	−0.001	−0.03 to 0.028	0.34	0.867	−1.873 to 3.607	0.352	0.841	−1.054 to 2.736
LP length	0.16	−0.095	−0.253 to 0.063	0.16	−0.055	−0.143 to 0.033	0.132	−0.025	−0.08 to 0.031	0.06	−2.377	−7.889 to 3.136	0.058	1.797	−2.089 to 5.684

2D, two‐dimensional; 3D, three‐dimensional; AUC, area under the curve; B, beta coefficient of the regression for the PFSP or pelvic accelerations; BN, bladder neck; LH, levator hiatus; LP, levator plate; MVC, maximal voluntary contraction; PFSP, posterior fornix sensor pressure; R ², coefficient of determination.

Fig. 1.

Scatterplots of the change in relative peak force and rate of force development (RFD) during maximal voluntary contraction (four upper panels) and during passive tissue elongation (four lower panels) after the run in relationship with pelvic floor loading (left: net mean posterior fornix sensor pressure [PFSP]; right: vertical accelerations) measured during the run. The regression lines and their 95% CIs are depicted. The equations are in the form B ₀ + B ₁ x + B ₂ y, where B ₀ is the intercept, B ₁ is the coefficient associated with the measure of pelvic floor loading and B ₂ is the coefficient associated with the baseline value of the pelvic floor muscle outcome indicated. Δ, change; g, gravitational unit; N/s, Newtons per second; N, Newtons; R ², coefficient of determination.

Fig. 2.

Scatterplots of the change in levator hiatus (LH) area at rest, bladder neck (BN) at rest and levator plate (LP) length after the run in relationship with pelvic floor loading (left: net mean posterior fornix sensor pressure [PFSP]; right: vertical accelerations) measured during the run. The regression lines and their 95% CIs are depicted. The equations are in the form B ₀ + B ₁ x + B ₂ y, where B ₀ is the intercept, B ₁ is the coefficient associated with the measure of pelvic floor loading and B ₂ is the coefficient associated with the baseline value of the pelvic floor muscle outcome indicated. Δ, change.

Dynamometry Outcomes

Contractile forces

There was no difference in MVC force or rate of force development after the run compared to before the run. The extent of pelvic loading (PFSP, pelvic accelerations) exposure during running did not affect the magnitude or rate of force generated during PFM MVC. The regression models suggest that pelvic floor loading during the run explains little to none of the variance (1%–12%) in the change in contractile force observed after the run (Fig. 1).

Passive tissue elongation

As noted above, exposure to higher pelvic accelerations, both vertical and vector, was associated with reduced rate of force development during passive elongation of the PFMs after the run (i.e. stiffness) and with peak passive resistance (Fig. 1) after the run, with models explaining between 20% and 27% of the variance in the data. Pelvic floor loading exposure, when measured through PFSP (net mean PFSP, relative net mean peak PFSP, and AUC PFSP) was not associated with changes in pelvic floor tissue stiffness measured using vaginal dynamometry after the run, with models explaining between 5% and 20% of the variance (Fig. 1).

Ultrasonography Outcomes

Levator hiatus

While LH area was larger when measured in standing position at rest after the run, the magnitude of pelvic floor loading exposure (measured through both PFSP and pelvic accelerations) was not associated with the extent of observed changes after the run. The models explained between 8% and 12% of the variance in PFSP and between 1% and 4% of the variance in pelvic accelerations (Fig. 2).

Bladder neck height

The BN height was lower when measured in standing position at rest after the run compared to before the run. Pelvic accelerations explained between 28% and 30% of the variance in the change in BN height at rest after the run (Fig. 2). During the MVC, there was more elevation of the BN after the run. For the PFSP variables, the regression models explained between 21% and 49% of the variance in the data, however, the significant predictor in these models was BN height before the run, not the loading exposure during the run. Runners who initially had a more cranial BN position before the run were more likely to experience greater caudal shifts in BN position afterward, both at rest and on MVC.

Levator plate length

The LP was not significantly longer after the run. Neither pelvic accelerations nor PFSP had a meaningful impact on the resting length of the LP, nor its ability to shorten on MVC after the run compared to baseline. The regression models explained between 3% and 16% of the variance in the data (Fig. 2).

Relationship between Pelvic Floor Loading and Pad Weight Gain among Runners with RI‐SUI

Among the runners with RI‐SUI, PFSP (net mean PFSP, relative net mean PFSP, AUC PFSP) explained little to no variance in pad weight gain (n = 15, R ² = 0.039 [95% CI −1.415, 0.703], slope = −0.356, P = 0.481; R ² = 0.000 [95% CI −0.726, 0.677], slope = −0.025, P = 0.941; R ² = 0.048 [95% CI −0.597, 0.271], slope = −0.163, P = 0.432, respectively). While similar results were obtained for vertical accelerations (R ² = 0.015 [95% CI −92.952, 56.604], slope = −18.174, P = 0.615), slower vector accelerations were associated with greater pad weight gain after the run (R ² = 0.273 [95% CI −100.193, −9.053], slope = −54.623, P = 0.022; n = 19).

Sensor Dislodgement

Five out of the six participants in whom the sensor dislodged were parous. We compared pelvic floor morphometry and dynamometry between the parous runners in whom the sensor became dislodged and those in whom the sensor remained within the posterior fornix (Table S2). There were no significant between‐group differences but the LH and LP of the parous runners who lost the sensor tended to be larger and longer, there was less tissue stiffness during passive elongation of the paravaginal tissues and PFM force‐generating capacity was lower.

Discussion

To our knowledge, this is the first study to explore the direct relationship between pelvic floor loading during a single exercise session and transient changes in pelvic floor morphometry and function observed afterward. The results provide some evidence that pelvic floor loading, particularly when inferred through pelvic accelerations, may contribute to stress relaxation of the pelvic supporting tissues in female runners. The effects were, however, not consistent across all measures. The observations using intra‐vaginal dynamometry suggest that faster pelvic accelerations during running, thus higher ground reaction forces [37], cause the connective tissues within and/or anchoring the PFMs to become less stiff, without negatively influencing their contractile function. However, the area of the LH and the LP length at rest, as measured by ultrasonography, were not affected by the magnitude of the pelvic accelerations, suggesting that tissue changes were related to stress relaxation and not to creep. There was also no evidence that higher pressures measured within the posterior fornix during running were associated with greater changes in BN support or PFM function over the course of the run. While there is evidence of tissue strain after the run (lower BN height, larger LH area) the extent of tissue strain does not appear to be associated with the magnitude of the loading forces experienced during the run. As such, our primary hypothesis was not supported.

Furthermore, our secondary hypothesis was not supported as higher exposure of the pelvic floor to loading was also not associated with greater pad weight gain among those with RI‐SUI. Interestingly, there was an inverse relationship between pad weight gain during running and pelvic vector accelerations while running. The underlying mechanisms for this observation remain unclear; runners with RI‐SUI might have adopted running mechanics that reduce vector acceleration to mitigate leakage or running with a soaked incontinence pad may have affected their running mechanics [26]. While minimising pelvic accelerations could offer a strategy to mitigate the extent of pelvic floor stress relaxation experienced during running, this might not impact leakage symptom severity. A case study demonstrated that instructing a participant to reduce vertical loading by running ‘more softly’ led to a reduction in tibial acceleration, vertical oscillation of her centre of mass, and, ultimately, a reduction in symptoms of urine leakage during running [38]. Increasing cadence/decreasing stride length, improving shock absorption through stronger lower limb muscles or appropriate footwear, and/or reducing running speed, among other biomechanical interventions could reduce pelvic loading [39, 40, 41]. Further research is necessary to evaluate the effectiveness of biomechanical strategies for mitigating pelvic floor loading and associated reductions in urine leakage.

Bladder neck height at rest and its elevation on MVC prior to the run significantly influenced BN height in standing position and change of position in BN height attained on MVC after the run, while the extent of pelvic floor loading (peak PFSP and pelvic accelerations) did not. A higher BN position before running appears to allow for more descent of the BN during the run. A potential explanation could be that a lower BN at rest might indicate that the supportive structures of the BN are already in an elongated position prior to the run, thus the tissues may be closer to their maximum strain, leaving them less room for descent in these women compared to those whose BN is located higher in the pelvis before the run [14].

Middlekauf et al. [42] reported no differences in maximal vaginal descent, vaginal resting pressure and maximal and mean PFM strength between a group of nulliparous females who completed a strenuous exercise protocol and those who completed a nonstrenuous (walking) exercise protocol. Assuming that the participants who completed the strenuous protocol experienced higher loads on their pelvic floor compared to the nonstrenuous protocol, their finding is consistent with the results of the current study; both suggest that the magnitude of peak or cumulative loading during exercise is not strongly associated with changes in PFM structure or function that occur due to the exercise. Indeed, previous research has demonstrated that approximately 50% of the variability in urethral mobility among females can be attributed to genetic factors [43], thus it is quite possible that intrinsic factors such as genetics may predispose some runners to greater loss of BN support than others, regardless of pelvic loading exposure during running. Our results and those of Middlekauf et al. [42] suggest that intrinsic contributors to the loss of pelvic organ support during running should be investigated in future work.

Measurement of Pelvic Floor Loading

Both the vertical component of pelvic accelerations and the overall vector accelerations were associated with reduced stiffness in the PFMs observed after the run. While pelvic accelerations are highly correlated with ground reaction forces [37], it is not yet known how ground reaction forces are transmitted to the pelvic floor. Similar to the findings of this study, a recent study found no association between running kinetics or pelvic accelerations and symptoms of pelvic floor dysfunction in a sample of postpartum runners [17].

The pelvic floor is a multiplanar network of soft tissues that is subjected to loading through the vertical oscillations of the abdominal and pelvic contents. Intra‐vaginal accelerometry may offer a direct means of measuring pelvic floor kinematics to improve our understanding of pelvic floor loading [44]. It appears that a history of vaginal childbirth may impact the capacity of the pelvic floor tissues to dampen pelvic floor oscillations [44]. Further investigation is warranted to understand not just the magnitude of loads to which the pelvic floor is exposed, but the response of the pelvic floor to those loads.

While the pressure sensor used in this work has been validated as an indirect means of measuring IAP during activities [18, 19, 20, 21, 22], when worn during physical activities performed in an upright position, pressure measured by a sensor placed in the posterior fornix of the vagina measures the summed forces acting to compress the sensor. Gravitational loading from the abdominal/pelvic contents and inertial loading from reaction forces likely contribute the majority of the compressive forces acting on the sensor during high impact activities. A previous study found a strong positive correlation between net mean peak PFSP and pelvic accelerations during walking (r = 0.7220) [18]. In contrast, weaker correlations were observed during running (r = 0.347) [45], further demonstrating that inertial loading appears to influence the PFSP. The signal processing methods used here removed any offset related to static gravitational loading and reflected loads during running that were between 27.8% and 99.8% of the pressure recorded during a vertical jump. While previous studies have referred to this pressure as IAP, we chose the more accurate term of PFSP to encompass the multidirectional and summative forces acting on the sensor, which largely represents loads experienced by the pelvic floor. The results of this study suggest that pelvic floor loading measured through PFSP is not associated with the extent of strain observed in the pelvic organ support structures after the run and is not associated with the amount of urine leaked.

This study has some limitations. As our sample lacked diversity, the findings of this research are representative of White, experienced runners. They are also specific to treadmill running and may not be translatable to other types of high‐impact activities. Due to the exploratory nature of this study, we investigated several outcomes and did not use alpha adjustments to account for the multiple outcomes, which may be considered a limitation. Indeed, alpha correction is not required when outcomes are independent [46], however, the five measures used to infer pelvic floor loading were dependent. We chose to include all measures as they each represent different contexts of pelvic floor loading and thus had the capacity to inform on different aspects of the biomechanical influences on the tissues. There was large variance in the PFSP data, which may have influenced our ability to detect meaningful relationships between PFSP and pelvic morphometry and PFM function.

Ideally a multivariable regression analysis would be used, however, due to the small sample size, the regression models were limited to two independent variables. However, other factors such as continence status, body mass index, parity, age, running mechanics, and weekly running distance may also contribute to the influence of pelvic floor loading on pelvic morphometry and PFM function. Adding these potential confounders would require a sixfold increase in the sample size, which was not feasible for this exploratory study. Future studies should explore the influence of these characteristics on running‐induced changes in pelvic morphometry and PFM function.

Hardware issues with IAP sensor (n = 5) and sensor dislodgement (n = 6) contributed to us having substantive data loss for the PFSP measures. Changes in pelvic floor morphometry induced by vaginal birth was likely the primary reason for this. While this same sensor has been widely used during dynamic physical activity performed by parous females, in this study, we measured PFSP over a longer bout of exercise, which might explain why dislodging of the sensor has not previously been reported as a problem. In future work, use of a sensor that is better able to be retained within the posterior fornix, or perhaps a device located in the stomach or rectum, would provide more robust data.

Previous research has shown that compressive shorts can provide pelvic organ support and relieve SUI [47] and as such, we opted to standardise the pants that the participants wore during running. These pants did not provide substantive abdominopelvic compression, however, the vaginal sensor itself and the lycra pants might have interfered with study outcomes.

Based on the outcomes of this study, there remains limited evidence to support advice for women with athletic incontinence to refrain from running as a form of exercise, or to change their running mechanics to reduce pelvic floor loading. Such advice could lead to kinesiophobic behaviours, which may, in turn, result in adverse health consequences associated with inactivity. Urinary symptoms themselves are a significant barrier to regular, sustained, and intense exercise [26]. This can, in turn, contribute to decreased physical fitness and overall wellbeing [26]. Much work is still needed before biomechanical interventions should be recommended. Future studies should aim to incorporate robust biomechanical data with tools to assess pelvic floor morphometry and function to better understand how different movement patterns or loading conditions impact female pelvic health. Longitudinal studies are also needed to assess the long‐term impact of loading on the pelvic floor. Additionally, further research is still needed to explore how conservative management strategies could benefit individuals with athletic incontinence.

In conclusion, although no significant correlations were found between pelvic loading exposure during a run and changes in PFM contractile strength after the run, higher pelvic skeletal accelerations were associated with greater loss in passive stiffness of the paravaginal tissues after the run. Intrinsic factors may influence reductions in pelvic organ support induced by loading exposure and should be investigated.

Funding

M.‐È. Bérubé was supported by scholarships from the following funding agencies: Fonds de recherche du Québec – Santé, l'Ordre Professionnel de la Physiothérapie du Québec, the University of Ottawa/CHEO Research Institute. This research was also supported by an Alun Morgan Memorial Research Award in Orthopaedic Physiotherapy to Dr. McLean and Ms. Berube, administered by the Physiotherapy Foundation of Canada, and used infrastructure awarded to Dr. McLean through the Natural Sciences and Engineering Research Council of Canada.

Disclosure of Interests

The authors certify that they have no affiliations with or financial involvement in any organisation or entity with a direct financial interest in the subject matter or materials discussed in the article.

Ethics Statement

The study protocol was approved by the Health Sciences and Sciences Research Ethics Board of the University of Ottawa. The participants provided written informed consent prior to participating in the protocol.

Author Contributions

All authors were fully involved in the design of the study, and in the preparation of this manuscript. Under the mentorship of Dr. McLean, Ms. Berube was primarily responsible for collecting data and preparing the original draft. All authors contributed to the revision process and provided final approval prior to this submission.

Supporting information

Table S1. Linear regression equations for pelvic floor muscle morphometry and function outcomes based on pelvic floor loading (PFSP or pelvic accelerations) exposure during the run.

Table S2. Comparison of mean (standard deviation) pelvic floor outcomes and pelvic accelerations between parous runners with and without posterior fornix sensor displacement and corresponding one‐away P‐value.

Data Availability Statement

Data may be requested from the corresponding author.