Can Clinical Postpartum Muscle Assessment Help Predict the Severity of Postpartum Pelvic Girdle Pain? A Prospective Cohort Study

Abstract

Objective

The purpose of this study was to evaluate whether the clinical assessment of pelvic floor muscles and the diastasis recti abdominis could predict the severity of pelvic girdle pain during the first year postpartum.

Methods

Between 2018 and 2020, 504 women were recruited to this prospective longitudinal cohort study. At 2 to 3 months postpartum, their pelvic floor muscles and diastasis recti abdominis were assessed using vaginal palpation, observation, and caliper measurement. The participants completed the Pelvic Girdle Questionnaire (PGQ) at 2 to 3, 6, 9, and 12 months postpartum. Mixed-effect models were used to determine how the results of pelvic floor muscle and diastasis recti abdominis assessments predicted the PGQ score. A sub-analysis for middle to high PGQ scores was conducted.

Results

Maximal voluntary pelvic floor muscle contractions ≥3 (Modified Oxford Scale, scored from 0 to 5) predicted a decreased PGQ score (β = −3.13 [95% CI = −5.77 to −0.48]) at 2 to 3 months postpartum, with a higher prediction of a middle to high PGQ score (β = −6.39). Diastasis recti abdominis width did not have any significant correlation with the PGQ score. A sub-analysis showed that a diastasis recti abdominis width ≥35 mm predicted an increased PGQ score (β = 5.38 [95% CI = 1.21 to 9.55]) in women with pelvic girdle pain.

Conclusion

The distinction between weak and strong maximal voluntary pelvic floor muscle contractions is an important clinical assessment in women with postpartum pelvic girdle pain. The exact diastasis recti abdominis width, measured in millimeters, showed no clinical relevance. However, a diastasis recti abdominis width ≥35 mm was associated with a higher PGQ score, and further research about this cutoff point in relation to pain is needed.

Impact

This study highlights the importance of clinical assessment of pelvic floor muscles in patients with postpartum pelvic girdle pain. A better understanding of the role of this muscle group will enable more effective physical therapist treatment of pelvic girdle pain.

Introduction

Pain in the lower back including the pelvic region during pregnancy and in the postpartum period is common and often categorized into the group of lumbopelvic pain. Approximately 5% to 25% postpartum women have persistent lumbopelvic pain.¹ Many studies do not differentiate between low back and pelvic girdle pain, which is important to do because the symptoms and treatments of these pain conditions are different. The use of the term “pelvic girdle pain” for pregnancy-related musculoskeletal pain in the pelvis has been recommended.² It is important to identify and treat women with pelvic girdle pain postpartum because of the risk that they may otherwise develop long-term problems, affecting their ability to perform daily activities for as long as 11 years postpartum.³

Although hormonal and mechanical changes as well as nonoptimal muscular stabilization of the pelvic joints may partly explain pelvic girdle pain, its complete etiology, especially in the postpartum period when the hormonal levels have returned to normal, remains unknown.² Pregnancy and childbirth can cause muscular alterations in the pelvic floor and abdominal muscles, which could be important pieces in this knowledge gap. The pelvic floor, whose muscular system is an important part of lumbopelvic stability,^4,5 loses some of its stability in the form of an enlarged levator hiatus already in late pregnancy.⁶ A vaginal birth affects both pelvic floor strength⁷ and stability. The first 6 months postpartum, the levator hiatus is still enlarged, which can lead to less support in case of increased abdominal pressure.⁸ Hypothetically, a weakness or decreased function of the pelvic floor muscles can influence the load transfer in the pelvis.⁹ Another important muscle group securing lumbopelvic stability are the abdominal muscles.⁹ During pregnancy, the abdominal muscles and their connection, the linea alba, overstretch because of the growing uterus. This stretch can result in a diastasis recti abdominis, which is defined as an impairment of the linea alba and an increased distance between the 2 parts of the musculus rectus abdominis.¹⁰ Postpartum women often have an increased interrectus distance.^11–13 Moreover, in surgical research, the bulging or protrusion of the abdominal wall is highlighted as an important component of the diastasis recti abdominis,^14,15 which recently has also been attracting more attention in physical therapy research.^16,17 The distended abdominal wall can result in a higher load on the pelvis during intraabdominal pressure and delay the recovery of pelvic girdle pain postpartum.¹⁸ A recent review showed that musculoskeletal alterations are among the main risk factors for persistent postpartum pelvic girdle pain.¹⁹

Clinically, we are observing that an increasing number of women seek care for a muscular checkup postpartum by primary care physical therapists. Clinically available and reliable methods for a postpartum assessment are observation, vaginal palpation, and caliper measurement.²⁰ However, there is no consensus yet as to how valid and clinically important these assessment methods are in relation to symptoms like pelvic girdle pain. Studies show contrary results about the increased or decreased function of the pelvic floor muscles in women with pelvic girdle pain.^21–24 The abdominal function is decreased in women with a diastasis recti abdominis.^12,25 However, 4 of 5 studies showed no associations between an increased diastasis recti abdominis and lumbopelvic pain,^13,26–28 whereas the last one showed the opposite.²⁹ Most studies used technical devices like vaginal balloon catheter, manometer, and ultrasound to assess muscle function. These devices are seldom available for primary care physical therapists who meet women with pelvic girdle pain. In this study, we aimed to evaluate whether the clinical assessments of pelvic floor muscles and diastasis recti abdominis with clinical available methods, like observation, vaginal palpation, and caliper measurement, could predict the severity of pelvic girdle pain during the first year after delivery.

Methods

Design

This study is 1 part of the AfterBabyBodyStudy, a longitudinal prospective observational cohort study about women’s physical health during the first year after childbirth (clinicaltrials.gov: NCT03703804). All participants gave written informed consent prior to participation, and the study was approved by the local Ethical Review Authority. The STROBE guidelines were followed when reporting this study.

Participants

The participants were recruited through advertisements at antenatal and child care centers and social media in the Region Västra Götaland in Sweden between September 2018 and February 2020. Eligibility for participation included female sex, ≥18 years of age, and vaginal delivery or C-section within the last 3 months. The participants needed to understand verbal and written Swedish to complete the included assessments and questionnaires. Women with any history of pelvic girdle and/or lower back pain prior to pregnancy and/or obstetric anal sphincter injuries were excluded. The power calculation was based on another part of the AfterBabyBodyStudy, which looks at the association between different types of exercises on pelvic floor and abdominal function during the first year postpartum (manuscript in preparation).

Data Collection

Data were collected at 3 rehabilitation centers in West Sweden: a rural, a suburban, and a metropolitan rehabilitation center. The participants applied for participation by email and were telephoned by the first author to book a clinical assessment. In Sweden, women are offered follow-up appointments with a midwife at approximately 6 to 8 weeks postpartum. In case of pelvic health problems, like pelvic girdle pain or urinary incontinence, midwives have the possibility to refer the women to a physical therapist at 8 to 12 weeks postpartum. This period of time, in the text referred to as 2 to 3 months postpartum, was used to evaluate the clinical assessments of pelvic floor muscles and diastasis recti abdominis postpartum.

The participants could choose 1 of the 3 rehabilitation centers. Assessments were performed by 6 physical therapists (2 at each center) trained in the clinical assessment of pelvic floor muscle and diastasis recti abdominis based on a previous study.²⁰ One week prior to the clinical assessment as well as at 6, 9, and 12 months postpartum, the participants completed an online questionnaire in the cloud-based questionnaire tool Esmaker (Entergate AB, Halmstad, Sweden) to collect the below-described outcome measures and covariates. Two reminder emails were sent to the participants to increase adherence.

Outcome Measures

The outcome measures of this study were pelvic girdle pain and related disability at 2 to 3, 6, 9, and 12 months postpartum. In the above-mentioned cloud-based questionnaire, the participants were asked: “During the last 4 weeks, have you experienced pain in the lower back/pelvis (see the Suppl. Figure)? Do not report pain you had with fever.”³⁰ If the answer to the question was “yes,” the participant was given the Pelvic Girdle Questionnaire (PGQ). The PGQ is an instrument with good construct validity for assessment of pelvic girdle pain and related disability in a postpartum population. The questionnaire strongly correlates with more traditional questionnaires for lumbopelvic pain, such as the Oswestry Disability Index, but shows a higher sensitivity and specificity for pregnancy-related pain. The used questionnaire has been tested among Swedish women with a higher internal consistency than the original questionnaire (Cronbach α = 0.82–0.96).³¹ The PGQ contains 20 activity items and 5 symptom items. Each item can be rated on a scale from 0 to 3, which results in a sum that ranges from 0 to 75. The sum of the rated items is divided by 75 (the maximum possible score) and then multiplied by 100, resulting in a PGQ score from 0 to 100.³¹ We have chosen to use a subgroup definition of Stuge et al to define low (<28) and middle to high (≥28) PGQ scores.³²

Predictor Variables

There is no gold standard for the pelvic floor muscle assessment. Perineometer and dynamometer are the most accurate assessment methods; both have the weakness of measuring pressure instead of strength and the limitation of missing the lifting component of the contraction.³³ Vaginal palpation correlates strongly with ultrasound assessment regarding changes in the levator hiatus³⁴ and is the most common method in clinical practice to assess both the lifting and squeezing components of the pelvic floor muscle contraction. With the aim to evaluate clinically available assessment methods with a proven moderate to substantial interrater reliability,²⁰ vaginal palpation was chosen as assessment method for involuntary and maximal voluntary pelvic floor muscle contraction. The detailed protocol for the clinical muscle assessments is described elsewhere.²⁰

The pelvic floor muscles were assessed in a supine position with hips and knees flexed on a plinth. All assessing physical therapists had experience in assessing the pelvic floor muscles. The participants were asked if they experienced pain and/or numbness during palpation with the option to discontinue the assessment if this was the case. None of the participants were excluded for these reasons. The assessing physical therapists palpated the contraction of the levator ani muscle with the gloved index and middle finger inserted 2 to 3 cm into the vagina.³⁵ The participants were requested to cough forcefully. The involuntary contraction, defined as a squeeze around the pelvic openings and an inward lift,³⁶ was rated as absent or present. The participants were then encouraged to contract their pelvic floor muscles to the maximum for as long as possible. The participants had 3 attempts. The strongest of the 3 attempts was rated on the Modified Oxford Scale (ranging from 0 [no contraction] to 5 [maximal contraction] with lift, able to hold for >5 seconds).²⁰

The gold standard for the assessment of diastasis recti abdominis width is ultrasound; however, caliper measurement correlates highly with ultrasound (r = 0.85–0.99),³⁷ and the measurement at the umbilicus showed good interrater reliability in our previous study.²⁰ Caliper measurement is a low-cost method and easily available in clinical practice. There are no comparable methods for the assessment of diastasis recti abdominis bulging. The assessment by observation during increased abdominal pressure showed moderate interrater reliability.²⁰

The diastasis recti abdominis width was measured at the umbilicus with an electronic digital caliper (150 mm, carbon fiber, accuracy of ±0.2 mm; 24.se Sverige AB, Kalmar, Sweden) in a supine position on a plinth. The participants were asked to lift their head 2 to 3 cm to activate the musculus rectus abdominis, the shoulder blades still having contact with the plinth. The distance between the 2 parts of the musculus rectus abdominis were palpated and measured in millimeters with the outside jaws of the caliper. Diastasis recti abdominis bulging (see the Suppl. Figure) was assessed by observation of the linea alba during increased abdominal pressure.

Covariates

The first online questionnaire, sent out prior to the clinical assessment, included questions about maternal characteristics—such as age, body mass index (BMI), physical activity level (days with moderate [1 point] and/or vigorous [1.7 points] physical activity per week)—and self-reported obstetric events (including number of parities [primipara/multipara], mode of delivery [vaginal/C-section], and birth weight of the recent child).

Data Analysis

IBM SPSS Statistics for Windows, Version 25.0 (Armonk, NY, USA) was used for all data analysis. The participants’ characteristics are presented as means and SDs for ratio data and numbers and percentages for nominal and ordinal data. Differences between participants with no symptoms at all and pelvic girdle pain during the first year after childbirth were analyzed using t tests, χ-square tests, or Mann–Whitney U tests. All predictors and covariates were coded as presented in Table 1. A dropout analysis was conducted using the same tests. A P value of <.05 was considered statistically significant.

Table 1.

Maternal Characteristics and Obstetric Events of Participants With and Participants Without PGP^a

Characteristic	Participants With PGP Symptoms (n = 353 b )	Participants Without PGP Symptoms (n = 151)	P
Age, mean (SD), y	33.0 (3.7)	32.2 (3.3)	.02c
BMI, mean (SD), kg/m2	24.6 (3.4)	24.0 (3.1)	.06
Recent baby’s birth weight, mean (SD), kg	3.6 (0.5)	3.6 (0.5)	.88
Diastasis recti abdominis width, mean (SD), mm	25.7 (7.7)	24.7 (6.8)	.22
Educationd			.68
Primary school, high school, or other	31 (8.8)	15 (9.9)
University or college, <3 y	33 (9.3)	13 (8.6)
University or college, ≥3 y	287 (81.3)	123 (81.5)
Physical activity level ≥5 d/wke	163 (46.2)	61 (40.4)	.21
Delivery mode, last birth			.22
Caesarean section	49 (13.8)	15 (9.9)
Vaginal delivery	302 (85.6)	136 (90.1)
Vacuum extractionc	23 (6.5)	11 (7.3)
Parity			.15
Primipara	213 (60.3)	102 (67.5)
Multipara	138 (39.1)	49 (32.5)
Self-reported pelvic floor tear			.66
No tear	63 (17.8)	33 (21.9)
First- or second-degree perineal tear	227 (64.3)	98 (64.9)
Maximal voluntary contractions			.74
<3	182 (51.6)	74 (49.0)
≥3	166 (47.0)	72 (47.7)
Involuntary contractions			.20
Yes	56 (15.9)	31 (20.5)
No	296 (83.9)	119 (78.8)
Diastasis recti abdominis width			.10
≥35 mm	41 (11.6)	9 (6.0)
<35 mm	269 (76.2)	113 (74.8)
Diastasis recti abdominis bulging			.15
Yes	73 (20.7)	23 (15.2)
No	245 (69.4)	113 (74.8)

^a Data are reported as number (percentage) of participants unless otherwise indicated. BMI = body mass index; PGP = pelvic girdle pain.

^b Missing information regarding maternal characteristics and obstetric events in 2 cases caused by missing questionnaires at 2 to 3 months postpartum

^c Statistically significant (P < .05).

^d Not included as a covariate for further analysis.

^e Based on 2 physical activity questions, 1 about the number of days with at least 30 minutes of moderate-intensity aerobic physical activity (1 point) and 1 about the number of days with at least 20 minutes of vigorous-intensity aerobic physical activity (1.7 points).

Mixed-effect models for repeated measures were used because of our interest in individual changes in the outcome measure, pelvic girdle pain during the first year postpartum, relative to predictors and covariates.³⁸ The mixed-effect models were used with the purpose of evaluating intra- and interindividual changes. Individual differences in levels and rates of change in pelvic girdle pain were examined using a 2-level mixed-effect model with repeated measures (ie, time) nested within individuals. The change component (ie, slope) was set to demonstrate longitudinal differences from 2 to 3 months to 12 months postpartum, that is, how pelvic girdle pain, on average, changed over time and which predictors and covariates were related to these changes. The average changes are defined as slope below.

Maximal voluntary pelvic floor muscle contractions were dichotomized into 2 groups—weak (<3) and strong (≥3)—because there were few participants on the outer edges of the scale: 13.5% with values of 0 to 1 and 11.4% with values of 4 to 5. In total, 166 participants (47.0%) showed maximal voluntary pelvic floor muscle contractions of ≥3, and 182 participants (51.6%) had contractions of <3. Five of the 353 participants with pelvic girdle pain were excluded because of incorrect pelvic floor muscle contractions (straining). The fixed effects of weak/strong maximal voluntary pelvic floor muscle contractions, absent/present involuntary contraction, and the covariates (age, BMI, mode of delivery, number of children, and birth weight of the recently delivered child) were tested in an initial model (model 1). Given that, in the first model, the average PGQ score was falling into the subgroup definition of a low level of pain (PGQ score = 14.61), the confounder middle to high levels of pain (PGQ score ≥ 28) was added to the model (model 2). In a last step, the fixed effect of the significant predictors on middle to high levels of pain was tested (model 3).

For the prediction of the diastasis recti abdominis assessment, the fixed effects diastasis recti abdominis width (measured in millimeters) and diastasis recti abdominis bulging (absent/present), as well as the covariates (age, BMI, mode of delivery, number of children, and birth weight of the recently delivered child), were added to a fourth model. Because of the need for more studies about more severe diastasis recti abdominis,³⁹ a sub-analysis with the dichotomized parameter diastasis recti abdominis width ≥35 mm was conducted (model 5). A diastasis recti abdominis width ≥35 mm is defined as moderate to severe according to Candido et al as cited in Gluppe et al.²⁸ In a last step, the confounder middle to high levels of pain was added, resulting in model 6. The most important results were graphically illustrated using lines and box plots.

Role of the Funding Source

The funders played no role in the design, conduct, or reporting of this study.

Results

Participation and Dropouts

A convenience sample of 504 postpartum participants was included in this study. All 504 participants underwent the clinical muscle assessment at 2 to 3 months postpartum, and 502 participants filled out the first web-based questionnaire prior to the clinical assessment. The response rates for the questionnaire declined over time: 93.7% at 6 months, 86.9% at 9 months, and 84.1% at 12 months (Fig. 1). One of the main reasons for loss to follow-up was a new pregnancy.

Figure 1.

Flowchart of study participation of the 504 participants. ^aReduced number due to incorrect pelvic floor contraction in 10 participants (straining). ^bReduced number due to exclusion of invalid measurements at start of the study. ^cTotal of 48 participants were rated as “cannot assess.” DRA = diastasis recti abdominis; MVC = maximal voluntary contraction; PGP = pelvic girdle pain.

Participant Characteristics

The study population had, on average, higher education levels, was slightly older, had lower BMI, and had fewer C-sections than women giving birth in the general Swedish population.⁴⁰ Of the 504 enrolled participants, 151 participants (30.0%) reported no pelvic girdle pain at all during the time period of 2 to 3 months to 12 months postpartum (Tab. 1). These participants were significantly younger than the participants with symptoms (mean age = 32.2 [SD = 3.3] years; P = .02). Regarding the clinically assessed muscle functions, there were no statistically significant differences between the 353 participants with pelvic girdle pain/disability and the 151 participants without pain/disability. A dropout analysis showed that the missing participants were significantly younger (mean age = 31.9 [SD = 3.8] years; P = .01) and had lower education levels (P < .01) (data not shown).

Outcomes

All predictors and covariates were tested for their correlation using the Spearman rank correlation test; only weak correlations were found between the different predictors and covariates. The median value of pelvic girdle pain/disability measured by the PGQ was 11.11 (interquartile range = 13.11) at 2 to 3 months postpartum. Maximal voluntary pelvic floor muscle contractions ≥3 predicted a decreased PGQ score (β = −3.13 [95% CI = −5.77 to −0.48]) at this point in time (Tab. 2). Involuntary pelvic floor muscle contractions and the added covariates showed no predictive effect on PGQ scores 2 to 3 months postpartum. The PGQ score declined, on average, by −0.40 (P = .04) point per measurement occasion during the study period (Tab. 2, model 1). The random effects analysis showed a significant variance of the slope (data not shown), which means that there were individual differences in how the participants’ PGQ scores changed during the first year postpartum.

Table 2.

Mixed-effect Models Showing the Fixed Effect and Rates of Change of Pelvic Floor Muscle Function for PGP^a

Parameter	Model 1		Model 2		Model 3
	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P
Intercept	14.61 (12.46 to 16.77)	<.001	12.72 (11.15 to 14.30)	<.001	12.42 (10.80 to 14.03)	<.001
Slope	−0.40 (−0.77 to −0.03)	.04	−0.26 (−0.61 to 0.09)	.14	−0.29 (−0.65 to 0.08)	.12
Middle to high level of pain			27.72 (24.51 to 30.93)	<.001	30.49 (26.29 to 34.70)	<.001
Slope of middle to high level of pain			−1.86 (−2.55 to −1.18)	<.001	−1.70 (−2.54 to −0.85)	<.001
Age	0.12 (−0.24 to 0.48)	.50	0.08 (−0.18 to 0.33)	.56	0.08 (−0.18 to 0.34)	.54
Slope of age	0.03 (−0.04 to 0.09)	.45	0.03 (−0.03 to 0.09)	.27	0.03 (−0.03 to 0.10)	.26
BMI	0.37(−0.01 to 0.75)	.06	0.22 (−0.06 to 0.49)	.12	0.23 (−0.05 to 0.50)	.11
Slope of BMI	−0.01 (−0.08 to 0.06)	.76	<0.01 (−0.06 to 0.06)	.98	<0.01 (−0.06 to 0.06)	.97
No. of deliveries	0.99 (−1.73 to 3.71)	.47	−0.39 (−2.34 to 1.57)	.70	−0.29 (−2.24 to 1.66)	.77
Slope of no. of deliveries	0.32 (−0.15 to 0.80)	.18	0.40 (−0.04 to 0.84)	.08	0.42 (−0.02 to 0.87)	.06
Birth weight	0.08 (−2.44 to 2.59)	.95	−0.17 (−1.96 to 1.63)	.86	−0.20 (−1.99 to 1.60)	.83
Slope of birth weight	−0.35 (−0.80 to 0.09)	.12	−0.27 (−0.68 to 0.13)	.19	−0.29 (−0.70 to 0.12)	.17
Mode of deliveryb	<0.01 (−3.98 to 3.98)	1.00	−0.37 (−3.25 to 2.50)	.80	−0.13 (−3.01 to 2.75)	.93
Slope of mode of deliveryb	−0.14 (−0.85 to 0.57)	.70	−0.07 (−0.71 to 0.58)	.84	−0.02 (−0.68 to 0.64)	.96
Pelvic floor muscle functions
MVC ≥ 3	−3.13 (−5.77 to −0.48)	.02	−2.37 (−4.27 to −0.47)	.01	−1.87 (−3.85 to 0.10)	.06
Slope of MVC ≥ 3	−0.11 (−0.57 to 0.35)	.64	−0.17 (−0.60 to 0.25)	.42	−0.14 (−0.59 to 0.30)	.52
MVC ≥ 3 × middle to high level of pain					−6.39 (−12.84 to 0.07)	.052
Slope of MVC ≥ 3 × middle to high level of pain					−0.57 (−2.05 to 0.90)	.44
Involuntary contraction	−0.42 (−4.07 to 3.24)	.82	0.09 (−2.51 to 2.69)	.95	−0.03 (−2.62 to 2.57)	.98
Slope of involuntary contraction	0.30 (−0.32 to 0.92)	.35	0.27 (−0.30 to 0.84)	.35	0.25 (−0.32 to 0.82)	.38

^a As measured with the Pelvic Girdle Questionnaire. Bold type indicates statistical significance (P < .05). β = estimate; BMI = body mass index; MVC = maximal voluntary contraction; PGP = pelvic girdle pain.

^b Vaginal delivery = 0.

In model 2, the slope of low levels of pain (PGQ score < 28) was not significant. The median value for participants (n = 24) with middle to high levels of pain was 41.33 (interquartile range = 16.62) at 2 to 3 months postpartum. As shown in Table 2 and Figure 2, participants with middle to high levels of pain had a sevenfold greater improvement during the first year postpartum than participants with low levels of pain. Figure 2 shows the mean changes in the 2 groups, where mixed-effect models have calculated values for missing cases over time. There was an inverse interaction between a strong pelvic floor muscle contraction and middle to high levels of pain, as shown in model 3 in Table 2. Maximal voluntary pelvic floor muscle contractions ≥3 predicted a decreased PGQ score (β = −6.39 [95% CI = −12.84 to 0.07]), with borderline significance (P = .052) in participants with middle to high levels of pain; no significant prediction was seen in participants with low levels of pain. This interaction is also illustrated in Figure 3.

Figure 2.

Predicted mean slope in participants with a Pelvic Girdle Questionnaire (PGQ) score ≥28 (n = 24) and a PGQ score <28 (n = 329) during the first year postpartum.

Figure 3.

Comparison of the distribution of maximal voluntary pelvic floor muscle contraction (mod. Oxford scale 0–5) between participants with a Pelvic Girdle Questionnaire (PGQ) score <28 and those with a PGQ score ≥28. All values were measured at 2 to 3 months postpartum.

The diastasis recti abdominis width (measured in millimeters) and bulging had no predictive value for the severity of postpartum pelvic girdle pain (model 4). However, a sub-analysis with the predictor diastasis recti abdominis width ≥35 mm (n = 41) showed that participants with a diastasis recti abdominis width ≥35 mm (n = 41) had an increased PGQ score (β = 5.38 [95% CI = 1.21–9.55]; P = .01) (Tab. 3, model 5). In participants with middle to high levels of pain, a diastasis recti abdominis width ≥35 mm was a significant predictor (β = 4.05 [95% CI = 1.05–7.06]; P = .01). The best model (ie, model 6) explained 72.1% of the variance in the intercept and 31% of the variance in the slope. Most of the variance in the model was explained by individuals with middle to high levels of pain, who for obvious reasons experienced higher levels of pain at 2 to 3 months postpartum and who also experienced a reduction in pain over the study period. A sub-analysis of an interaction between a diastasis recti abdominis width ≥35 mm and middle to high levels of pain was not possible because there were too few participants in this group (ie, only 4 participants with both middle to high levels of pain and a diastasis recti abdominis width ≥35 mm).

Table 3.

Mixed-effect Models Showing the Fixed Effect and Rates of Change of Diastasis Recti Abdominis for PGP^a

Parameter	Model 4		Model 5		Model 6
	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P
Intercept	12.94 (10.89 to 15.00)	<.001	12.44 (10.46 to 14.42)	<.001	11.60 (10.15 to 13.04)	<.001
Slope	−0.46 (−0.80 to −0.12)	.01	−0.42 (−0.74 to −0.09)	.01	−0.51 (−0.95 to −0.07)	.02
Middle to high level of pain					28.31 (24.70 to 31.92)	<.001
Slope of middle to high level of pain					−1.63 (−2.62 to −0.64)	<.001
Age	0.29 (−0.11 to 0.68)	.15	0.27 (−0.12 to 0.66)	.18	0.17 (−0.11 to 0.45)	.24
Slope of age	−0.01 (−0.07 to 0.05)	.76	−0.01 (−0.07 to 0.05)	.80	0.00 (−0.09 to 0.08)	.91
BMI	0.23 (−0.18 to 0.64)	.27	0.24 (−0.17 to 0.64)	.25	0.09 (−0.21 to 0.38)	.56
Slope of BMI	0.01 (−0.05 to 0.08)	.66	0.01 (−0.05 to 0.08)	.70	0.02 (−0.07 to 0.11)	.62
No. of deliveries	0.17 (−2.85 to 3.19)	.91	−0.16 (−3.11 to 2.80)	.92	−0.88 (−3.00 to 1.24)	.41
Slope of no. of deliveries	0.15 (−0.32 to 0.61)	.54	0.16 (−0.30 to 0.62)	.48	0.23 (−0.39 to 0.85)	.47
Birth weight	1.34 (−1.28 to 3.96)	.32	1.31 (−1.27 to 3.90)	.32	0.05 (−1.81 to 1.91)	.96
Slope of birth weight	−0.53 (−0.94 to −0.11)	.01	−0.53 (−0.94 to −0.12)	.01	−0.36 (−0.90 to 0.18)	.19
Mode of deliveryb	−2.17 (−6.56 to 2.21)	.33	−1.88 (−6.16 to 2.41)	.39	−2.16 (−5.29 to 0.97)	.18
Slope of mode of deliveryb	0.21 (−0.50 to 0.91)	.56	0.18 (−0.51 to 0.87)	.61	0.47 (−0.42 to 1.36)	.30
DRA assessment
DRA width, mm	0.08 (−0.11 to 0.27)	.41
Slope of DRA width	−0.01 (−0.04 to 0.02)	.61
DRA width ≥ 35 mm			5.38 (1.21 to 9.55)	.01	4.05 (1.05 to 7.06)	.01
Slope of DRA width ≥ 35 mm			−0.42 (−1.05 to 0.21)	.19	−0.27 (−1.10 to 0.56)	.52
DRA bulging	1.26 (−2.10 to 4.61)	.46	1.04 (−2.19 to 4.28)	.53	0.36 (−1.95 to 2.67)	.76
Slope of DRA bulging	0.22 (−0.30 to 0.74)	.41	0.22 (−0.27 to 0.72)	.38	0.39 (−0.28 to 1.06)	.25

^a As measured with the Pelvic Girdle Questionnaire. Bold type indicates statistical significance (P < .05). β = estimate; BMI = body mass index; DRA = diastasis recti abdominis; PGP = pelvic girdle pain.

^b Vaginal delivery = 0.

Discussion

This large prospective observational study with high response rate of the participants showed that the clinical assessment of maximal voluntary pelvic floor muscle contraction by vaginal palpation is a predictive factor for the severity of pelvic girdle pain after delivery. Maximal voluntary pelvic floor muscle contractions ≥3 predicted a decreased PGQ score, especially in the group of participants with middle to high PGQ scores. The clinical assessment of involuntary pelvic floor muscle contraction did not predict the severity of pelvic girdle pain 2 to 3 months postpartum and neither did the clinical assessment of diastasis recti abdominis width in millimeters nor the observation of diastasis recti abdominis bulging. However, a sub-analysis of participants with moderate to severe diastasis recti abdominis width (≥35 mm) indicated increased PGQ scores. No predictors for the slope of the PGQ score during the first year postpartum were found. More research, for example on how physical activity and exercise affect the increase or decrease of pain during the first year postpartum, is needed.

Our results are contrary to the results of a study from Stuge et al, which found no difference in voluntary pelvic floor muscle function assessed via palpation, manometry, and ultrasound between women with and without postpartum pelvic girdle pain.²⁴ Also, no associations between pelvic floor muscle strength as assessed by vaginal palpation and pelvic girdle pain were found during pregnancy.⁴¹ A strength of our study is the use of mixed-effect models that take into account the individual variation in the PGQ scores. The above-mentioned studies compared participants with and without pain/disability. An interesting finding in both studies were the associations between a higher vaginal resting pressure and deep pelvic floor muscle tenderness in women with pelvic girdle pain/disability.^24,41 Tonus and relaxation of the pelvic floor muscle were not investigated as predictors in our study because of the low interrater reliability of the assessment of pelvic floor muscle relaxation by vaginal palpation shown in our previous study.²⁰ However, Dufour et al have hypothesized that pelvic floor muscle weakness could be a result of increased tension. They showed that the majority of women with lumbopelvic pain have both pelvic floor muscle tenderness and weakness.⁴² Similar considerations were made in another study, which showed increased activity in the pelvic floor muscles and decreased pelvic floor muscle endurance in women with pregnancy-related low back and pelvic pain.²¹ They also showed less pelvic floor muscle activity during coughing in women with pregnancy-related low back and pelvic pain, as measured using electromyography. We could not find any predictive value of the absence of involuntary contractions during coughing, which may be explained by the use of our less sensitive method. The hypothesis about weakness as a result of increased tension is interesting when interpreting our results. More research about these associations is needed to find the right rationale for future treatment studies and the clinical treatment of pelvic girdle pain, including addressing tensed pelvic floor muscles with a different exercise approach than weak muscles. For prevention and treatment of urinary incontinence, it is known that supervised pelvic floor muscle training is more effective than pelvic floor training alone.⁴³ Evidence of benefits of pelvic floor muscle training for women with pelvic girdle pain is low⁴⁴; hence a supervised or individualized approach could be important in the treatment of pelvic girdle pain as well. Furthermore, studies are needed to determine if tension and weakness are results of the pain or vice versa.

Our study indicated that the presence of moderate to severe diastasis recti abdominis width is a predictive factor for higher PGQ scores in women with pelvic girdle pain. This is contrary to the latest review about correlations between diastasis recti abdominis and musculoskeletal dysfunction, which concluded that there were no associations between the presence of a diastasis recti abdominis and lumbopelvic pain/pelvic girdle pain.⁴⁵ Comparing our results with the reviewed articles,⁴⁵ the different cutoff points for diastasis recti abdominis must be considered. In earlier studies, diastasis recti abdominis widths between 2 and 2.5 cm were defined as predictive factors. These are values falling into the normal range of diastasis recti abdominis width within the first 6 months postpartum.⁴⁶ A recently published study even claims that all values below 3 cm should be described as normal.⁴⁷ The latest surgical recommendations state that a diastasis recti abdominis width ≥50 mm, or abdominal bulging, should be considered for surgery.⁴⁸ To the best of our knowledge, there are no other studies besides ours that have evaluated the relationships between diastasis recti abdominis bulging and pain/dysfunction.

Regarding the chosen diastasis recti abdominis width, there is just 1 comparable study by Gluppe et al, who used the same updated cutoff values and questionnaire as our study.²⁸ They showed that there was no association between pelvic girdle pain and diastasis recti abdominis using ultrasound for the assessment of diastasis recti abdominis width. Ultrasound is a more valid method than measurement using a caliper.⁴⁹ The study of Gluppe et al had a different outcome; they compared the prevalence of pelvic girdle pain among women with and without diastasis recti abdominis. Comparing these outcomes, we also did not see any significant differences between the 2 groups of participants with and without postpartum pelvic girdle pain regarding diastasis recti abdominis width (Tab. 1). In both our study and the study of Gluppe et al, the number of women with moderate to severe diastasis recti abdominis width was low: 50 and 17, respectively. Neither of these studies conducted a prior power calculation.

Limitations

The inclusion of middle to high PGQ scores as a predictor to the third and fourth as well as fifth and sixth model must be discussed as both a strength and a limitation. A predictor, which is related to the outcome, will always explain the majority of the model, which can be problematic. Therefore, we present the models without this predictor first (models 1 and 4). The variable middle to high levels of pain is not part of the 4 time points of PGQ scores, which is the dependent variable, capturing within-person change in pain. This variable is a dichotomous variable of individuals with low levels of pain and those with high levels of pain, included in the model as a fixed level 2 covariate. The decision to add middle to high PGQ scores in further models was made in our special interest of how to treat women with middle to high PGQ scores, who are the ones who usually seek help from physical therapists postpartum.

Our results on the predictive value of moderate to severe diastasis recti abdominis width on the severity of pelvic girdle pain must be interpreted with caution considering that 9 of the participants in the group of 151 participants with no pelvic girdle pain at all during the first year postpartum had a diastasis recti abdominis width ≥35 mm. Moreover, the group of participants with moderate to severe diastasis recti abdominis width and middle to high PGQ scores consisted of only 4 participants. For these reasons, we cannot make any conclusive statement about any interaction between diastasis recti abdominis width ≥35 mm and middle to high levels of pain. However, for the entire group of participants with pelvic girdle pain, moderate to severe diastasis recti abdominis width was found to affect the severity of pain, even in women with low levels of pain. This result is based on data from 41 participants and is an important finding, considering that participants with low levels of pain showed only small improvements during the first year postpartum. Even lower levels of pain can have a negative impact on the life of a new mother,⁵⁰ and the assessment and treatment of a diastasis recti abdominis width ≥35 mm could be important for their rehabilitation. Another limitation we must discuss is that we chose a practical time window of 8 to 12 weeks postpartum for the first clinical assessment. Considering that the pelvic floor muscles are still recovering during the time period of 8 to 12 weeks,^8,51 there is a risk that the participants assessed later had stronger pelvic floor muscles and less pain.

The clinical assessment methods used in this study  are both a strength and a limitation. They are inexpensive and easily available, which makes this study highly clinically relevant. However, the assessment by observation and palpation is always subjective and less accurate than measurements conducted using technical devices.⁵² The assessment by vaginal palpation gives results on a 6-step Likert scale, which is less precise than the assessment by technical devices, which often give results in millimeters of mercury. Although all assessments used in this study were tested for their interrater reliability in a previous study and only the assessments showing moderate or substantial interrater reliability were included,²⁰ there is a risk for over- or underestimation of the function because of the subjectivity of the method. The assessment of diastasis recti abdominis bulging and involuntary pelvic floor muscle contraction showed moderate interrater reliability in our previous study, but with concerningly wide CIs,²⁰ which means that their predictive value on the severity of pelvic girdle pain must be interpreted with caution and that more research about the assessment of these functions is needed.

Another limitation is the assessment of the outcome variable, pelvic girdle pain by questionnaire, without confirmation of this condition by clinical assessment. The European guidelines recommend evaluation of pelvic girdle pain using specific tests, such as the Posterior Pelvic Pain Provocation Test.² However, the PGQ is a questionnaire with a high sensitivity for pelvic girdle pain and changes in postpartum pelvic girdle pain.⁵³

Moreover, the homogeneous sample of participants must be discussed as a limitation. The researchers put effort into obtaining as representative sample population as possible by choosing different primary care rehabilitation centers in low- and high-income districts in West Sweden. However, the recruitment strategy was advertising, and the participants had to contact the researchers themselves, which led to a highly educated sample population with low to normal BMI. This can lower the overall validity of this study, and the predictive values of BMI on pelvic girdle pain may not be representative.

Conclusion

Our study indicates that the clinical assessment of a maximal voluntary pelvic floor muscle contraction ≥3 predicts less pelvic girdle pain postpartum. Therefore, vaginal palpation of maximal voluntary pelvic floor muscle contraction is an important part of the clinical assessment in women with pelvic girdle pain 2 to 3 months postpartum. The next step should be to gain more knowledge about the question of whether weakness is a result of tension or vice versa. This is important to ascertain in order to provide the correct physical therapist treatment of patients with pelvic girdle pain postpartum. Moreover, our study has shown that the exact distance of diastasis recti abdominis width, measured in millimeters, does not play a role in the assessment of women with pelvic pain. However, screening for women with a diastasis recti abdominis width ≥35 mm could be clinically relevant because of its prediction of increased pelvic girdle pain/disability. It is important to understand which physical and psychosocial factors a moderate to severe diastasis recti abdominis width will typically affect in women with pelvic girdle pain and how to treat them.

Author Contributions

Concept/idea/research design: S. Vesting, A. Gutke, M.F. Olsén, G. Rembeck, M.E.H. Larsson

Writing: S. Vesting, A. Gutke, M. Fagevik Olsén, M. Praetorius Björk, G. Rembeck, M.E.H. Larsson.

Data collection: S. Vesting

Data analysis: S. Vesting, A. Gutke, M. Fagevik Olsén, M. Praetorius Björk, G. Rembeck, M.E.H. Larsson

Project management: S. Vesting, M.E.H. Larsson

Fund procurement: S. Vesting, M.E.H. Larsson

Providing participants: S. Vesting

Consultation (including review of manuscript before submitting): A. Gutke, M. Fagevik Olsén, M. Praetorius Björk, G. Rembeck, M.E.H. Larsson

Ethics Approval

This study was approved by the Swedish Ethical Review Authority (Dnr 088–18).

Funding

This study was funded by the Research and Development Centre Gothenburg and Södra Bohuslän and the Healthcare Committee, Region Västra Götaland.

Clinical Trial Registration

This study was registered at ClinicalTrials.gov (NCT03703804).

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.