Relationship Between Gestational Diabetes Mellitus and Postpartum Diastasis Recti Abdominis in Women in the First Year Postdelivery

Abstract

Objective

Postpartum diastasis recti abdominis (DRA) influences women’s appearance and health. Gestational diabetes mellitus (GDM) can affect the structure of the rectus abdominis muscles. However, the relationship between GDM and postpartum DRA is unknown. The objective of this study was to investigate the relationship between GDM and postpartum DRA.

Methods

This retrospective cohort study included 241 women in the first year postdelivery. Women with GDM were matched with those without GDM using propensity score matching. They underwent an oral glucose tolerance test during pregnancy and a random blood glucose test before delivery. At follow-up, DRA was diagnosed by palpation, and interrectus distance was measured using ultrasound to evaluate the severity of DRA. The strength of the rectus abdominis was evaluated using the manual muscle testing method.

Results

Among the 241 participants, 174 (72.2%) had postpartum DRA, and 46 women with GDM were matched with 46 women without GDM on the basis of propensity scores. Women with GDM had higher odds of experiencing postpartum DRA (adjusted odds ratio = 4.792; 95% CI = 1.672 to 13.736) and larger interrectus distance values at the upper part of the rectus abdominis than those without GDM. There was a weak and positive correlation between the fasting oral glucose tolerance test level and the interrectus distance values (0.267 ≤ r ≤ 0.367).

Conclusion

GDM was associated with postpartum DRA in women in the first year of delivery. Women with GDM had larger interrectus distance values at the upper part of the rectus abdominis than those without GDM. The fasting oral glucose tolerance test level showed a positive and weak correlation with the severity of postpartum DRA.

Impact

Women with GDM have higher odds of experiencing postpartum DRA than those without GDM. The upper part of the rectus abdominis deserves increased focus during and after rehabilitation. Controlling the fasting oral glucose tolerance test level may help reduce the severity of postpartum DRA.

Introduction

Diastasis recti abdominis (DRA) refers to the separation of 2 rectus abdominis muscles along the linea alba. Women usually develop DRA during pregnancy, and some of them recover spontaneously postdelivery. However, 33–74% of women still experience DRA and achieve disappointing recovery during the postpartum period.^1,2 Women with DRA may present an abdominal protrusion, which increases their anxieties about self-image.^1,3 Moreover, women with DRA have been found to be more susceptible to lumbopelvic pain and pelvic floor diseases, such as urinary incontinence, fecal incontinence, and pelvic organ prolapse, than women without DRA.⁴ DRA is defined by a width of greater than the breadth of 2 fingers between the rectus abdominis bellies along the linea alba as measured by palpation, which is frequently used in clinical practice and has good intrarater reliability (weighted κ > 0.70) according to literature.^5,6 Recently, the diagnosis and evaluation of DRA using interrectus distance (IRD), as measured using ultrasound, have been increasing owing to the high accuracy and validity of this method.^2,7,8 Although a standard cutoff value of IRD for diagnosing DRA is currently unknown,⁷ numerous studies have investigated the effect of DRA on women’s health through IRD. The parameter of IRD was found to be associated with the severity of symptoms related to DRA, such as abdominal pain and impaired trunk motor function.^2,4,9

So far, several risk factors, such as parity, maternal age, weight gain, mode of delivery, and birth weight, have been reported to be related to postpartum DRA.^7,10 Wu et al¹¹ found that the incidence of DRA among women aged 18–90 years (sample size, 644) with diabetes was 1.956 times higher than that of those without diabetes. Their findings indicated the correlation between diabetes and DRA, but they did not analyze whether gestational diabetes mellitus (GDM), a hyperglycemic state that occurs during pregnancy, had an effect on DRA. Vesentini et al¹² evaluated the structural and biochemical changes in the rectus abdominis in 92 pregnant women with and without GDM and reported an increase in slow fibers, decrease in fast fibers, and alterations in the ultrastructure of the rectus abdominis muscle of those with GDM; however, they did not assess DRA in their study, and thus, the relevance of these changes to DRA remains unknown. These findings suggest that it is worth exploring whether GDM, which is a form of diabetes diagnosed during pregnancy, would affect postpartum DRA.

GDM is defined as the occurrence of glucose intolerance during pregnancy. It is estimated that 7% of pregnancies are complicated by some kind of diabetes, 86% of which account for GDM.¹³ In addition to its recognized influence on neonatal complications, such as hypoglycemia, hyperinsulinemia, and birth weight greater than the 90th percentile, and maternal outcomes, such as increased risk of type 2 diabetes mellitus in later life and high rates of requiring cesarean section,^14,15 GDM has been confirmed to have adverse effects on women’s pelvic floor functions because hyperglycemic metabolism is known to impair pelvic floor muscle structure.^12,16–18 Although DRA is also an important muscle problem in the field of postpartum rehabilitation, the association between GDM and DRA remains unclear. Therefore, the purpose of the current study was to investigate the relationship of GDM with postpartum DRA and the strength of rectus abdominis in a retrospective cohort of women in the first year postdelivery. We also studied whether GDM was associated with ultrasound-measured IRD.

Methods

Study Design and Setting

This retrospective, single-centered cohort study was conducted in the Seventh Affiliated Hospital of Sun Yat-sen University, Shenzhen, China. Women postdelivery less than 1 year and had their first postpartum examinations in the Pelvic Floor Disorders Center from June to December 2021 were invited to participate in our study. This study was approved by the Ethics Committee of the Seventh Affiliated Hospital of Sun Yat-sen University (2020SYS-USH-055). Written informed consent was obtained from all participants before participation.

Participants

Postpartum women, both primiparous and multiparous, were included in this study, and their data were analyzed. All of the participants had undergone a 75-g oral glucose tolerance test (OGTT) at 24 to 28 weeks of gestation. Furthermore, results of the random blood glucose (RBG) tests that were performed during hospitalization for delivery were collected to rapidly detect the control of blood glucose levels among the participants. The participants were divided into 2 different groups on the basis of their 2-h (fasting, 1 h, and 2 h) 75-g OGTT results, according to the recommendations of the International Association of the Diabetes and Pregnancy Study Groups.¹⁹ Women with 1 of the following blood glucose levels at different time points were allocated to the group with GDM: fasting level of ≥5.1 mmol/L; ≥10.0 mmol/L at 1 h; or ≥ 8.5 mmol/L at 2 h. The others with lower OGTT levels were allocated to the control group. The fasting blood glucose test results reveal basic blood glucose levels and insulin secretion. The 1- and 2-h blood glucose levels indicated glucose tolerance and regulation levels. The inclusion criteria were as follows: women with a postpartum period of <1 year; an age of 18 to 45 years; a singleton pregnancy with cephalic presentation; and the absence of any pelvic floor or rectus abdominis treatment (surgery or physical therapy) before or after delivery. The exclusion criteria were as follows: presence of previous history of diabetes mellitus (type 1, type 2, or overt diabetes at previous pregnancy); diabetes mellitus postdelivery (diagnosed at 6–12 weeks postpartum); smoking; complications affecting muscle structure or metabolism (such as severe digestive dysfunction, neurological diseases, and endocrine diseases) during pregnancy^20–22; and use of drugs that interfere with glucose and lipid metabolism, such as cortisone and phenytoin, during pregnancy. Baseline data on characteristics including age, occupation (manual/mental), education, body mass index, gestational weeks, weight gain, birth weight, postpartum day, gravidity, parity, number of vaginal/cesarean section births, mode of last delivery, lactation, OGTT, and RBG test levels were collected. Both groups were informed about the importance of glycemic control and advised to avoid exercises that may increase intraabdominal pressure, such as curling up and weightlifting. Women in the group with GDM received a standard treatment consisting of dietary modification and aerobic exercise, such as walking, swimming, and prenatal exercise classes.

Diagnosis of DRA

As there is no consensus on the cutoff value of ultrasound-measured IRD for the diagnosis of DRA, we used the traditional palpation approach^10,23 to diagnose DRA. A physical therapist with 5 years of experience in DRA diagnosis and rehabilitation carried out the palpating measurement. The participants were instructed to lie in a supine position with arms crossed over the chest and to perform an abdominal crunch until their shoulder blades left the examination bed. The physical therapist used their 2 fingers (index and middle) to measure the distance between the 2 rectus abdominis 4.5 cm above and below the umbilicus and at the umbilicus. A patient was diagnosed with DRA if a separation of ≥2 finger widths (2 cm) was observed in any of the 3 locations mentioned above, and no DRA was set, as all of the separations were < 2 finger widths.

Measurement of IRD

An ultrasound imaging unit (X5; SonoScape Medical Corp, Shenzhen, China) with a 7.5-MHz, 60- by 18-mm linear transducer was used to assess IRD in the B mode. The measurement was performed by a physical therapist with 4 years of experience in the assessment of abdominal muscles using ultrasound. Each participant was placed in a supine position on the examination bed, with their head resting on a thin pillow and their knees bent.⁹ The measurement site was along the midline of the abdomen at the following locations: at the superior border of the umbilicus; at 2, 3, and 4.5 cm above the superior border of the umbilicus; and at 2, 3, and 4.5 cm below the inferior border of the umbilicus.^10,24,25 To standardize the position of the transducer, each location was marked on the skin before measurement. The transducer was placed transversely across the markers on the abdomen with moderate pressure imposed to avoid a reflexive response. Images of the inner parts of the bilateral rectus abdominis were captured, and the distance between the 2 medial endpoints at their inside margins was measured as the IRD using an on-screen caliper (Fig. 1).

Figure 1.

Sonographic images of the interrectus distance (IRD; indicated by a dotted line) measured at 7 locations in a 27-year-old woman at 44 d postpartum. (A)–(G) The IRD values were 32.69 mm (A), 20.83 mm (B), 25.56 mm (C), 27.32 mm (D), 8.44 mm (E), 20.82 mm (F), and 27.35 mm (G). SBU = superior border of the umbilicus.

Evaluation of the Strength of the Rectus Abdominis

The manual muscle testing method described by Hislop and Montgomery²⁶ was applied to evaluate the flexion strength of the rectus abdominis. The participants were placed in a supine position with their knees bent to 90 degrees and their feet on the plinth. The physical therapist who was in charge of the diagnosis of DRA instructed the participants to raise the trunk to palpate their rectus abdominis. The strength was graded on a scale from 0 to 5 (0 = no contraction, 1 = flicker contraction, 2 = only lifting the head with arms at sides, 3 = arms outstretched above the plane of the body, 4 = arms crossing over the chest, and 5 = hands clasped behind the head).

Statistical Analysis

As several covariates that have been recognized to impact on DRA presented significant differences between the group with GDM and the group without GDM (Tab. 1), we used a propensity score (PS)–matching approach to reduce the influence of confounding factors so as to avoid the underestimation of the effect of GDM.²⁷ The PS was estimated using a logistic regression model in which the status of GDM/no GDM was regressed on the covariates. The baseline variables with significant differences between the group with GDM and the group without GDM before matching were considered to enter the regression model for calculating the PS. To obtain a maximal number of matched pairs and to control for the covariates simultaneously, we included the vital covariates in the regression model for the PS calculation process. This model included age, gestational weeks, weight gain, and postpartum days, as well as the follow-up date in order to decrease the effect of the environment and evaluators on the measurement.

Table 1.

Characteristics of Participants With GDM and Those Without GDM Before and After 1:1 Greedy Matching^a

Characteristic	Before Matching b			After 1:1 Greedy Matching b
	Without GDM   (n = 190)	With GDM (n = 51)	P	Without GDM   (n = 46)	With GDM   (n = 46)	P
Age, y, median (IQR)	29.00 (27.00 to 32.00)	32.00 (29.00 to 35.00)	.001	30.76 (4.37)	31.52 (3.56)	.362
Occupation			.394			.613
Manual	38 (20.0)	13 (25.5)		9 (19.6)	11 (23.9)
Mental	152 (80.0)	38 (74.5)		37 (80.4)	35 (76.1)
Education			.799			.620
Primary school	1 (0.5)	0		1 (2.2)	0
High school	50 (26.3)	12 (23.5)		11 (23.9)	9 (19.6)
College/university	139 (73.2)	39 (76.5)		34 (73.9)	37 (80.4)
BMI, kg/m2
Before pregnancy, mean (SD)	20.73 (2.49)	22.03 (3.01)	.002	21.20 (2.68)	22.02 (3.03)	.175
Before delivery, median (IQR)	25.65 (24.07 to 27.77)	26.70 (24.61 to 28.25)	.158	25.76 (2.85)	26.78 (2.96)	.096
At follow-up, mean (SD)	22.60 (2.51)	22.95 (2.77)	.387	22.78 (2.45)	23.02 (2.63)	.649
Gestational weeks, median (IQR)	39.0 (38.5 to 40.0)	38.5 (38.0 to 39.5)	.005	39.0 (38.0 to 39.5)	39.0 (38.0 to 40.0)	.836
Weight gain, kg, median (IQR)	13.00 (10.00 to 16.00)	11.00 (9.50 to 13.50)	.005	11.40 (4.00)	11.87 (3.83)	.565
Birth weight, kg, mean (SD)	3.19 (0.387)	3.17 (0.447)	.782	3.20 (0.38)	3.23 (0.43)	.801
Postpartum days, median (IQR)	50.00 (46.00 to 58.00)	50.00 (46.00 to 59.00)	.359	54.00 (47.75 to 59.25)	50.00 (46.00 to 58.00)	.959
Gravidity			.526			.349
1	66 (34.7)	16 (31.4)		14 (30.4)	14 (30.4)
2	73 (38.4)	16 (31.4)		21 (45.7)	16 (34.8)
3	31 (16.3)	11 (21.6)		9 (19.6)	9 (19.6)
≥4	20 (10.5)	8 (28.6)		2 (4.3)	7 (15.2)
Parity			.447			.725
1	103 (54.2)	26 (51.0)		26 (56.5)	24 (52.2)
2	80 (42.1)	21 (41.2)		19 (41.3)	19 (41.3)
3	7 (3.7)	4 (7.8)		1 (2.2)	3 (6.5)
No. of vaginal births			.94			1.000
0	46 (24.2)	14 (27.5)		14 (30.4)	14 (30.4)
1	94 (49.5)	23 (45.1)		14 (30.4)	21 (45.7)
2	44 (23.2)	12 (23.5)		10 (21.7)	9 (19.6)
3	6 (3.2)	2 (3.9)		1 (2.2)	2 (4.3)
No. of cesarean sections			.325			.276
0	134 (70.5)	35 (68.6)		30 (65.2)	31 (67.4)
1	29 (15.3)	5 (9.8)		9 (19.6)	4 (8.7)
2	27 (14.2)	11 (21.6)		7 (15.2)	11 (23.9)
Mode of delivery			.889			1.000
Vaginal birth	136 (71.6)	36 (70.6)		32 (69.6)	32 (69.6)
Cesarean section	54 (28.4)	15 (29.4)		14 (30.4)	14 (30.4)
Lactation	177 (93.2)	46 (92.0)	.760	43 (93.5)	41 (91.1)	.488
OGTT, mmol/L, median (IQR)
Fasting	4.24 (4.09 to 4.43)	4.69 (4.47 to 5.03)	<.001	4.24 (4.06 to 4.38)	4.70 (4.48 to 5.03)	<.001
1 h	8.05 (7.04 to 8.65)	10.13 (9.32 to 11.03)	<.001	8.34 (6.54 to 8.65)	10.08 (9.20 to 10.97)	<.001
2 h	6.56 (5.67 to 7.23)	8.80 (7.67 to 9.40)	<.001	6.51 (5.64 to 7.24)	8.75 (7.61 to 9.16)	<.001
RBG before delivery, mmol/L, median (IQR)	4.69 (4.30 to 5.31)	4.94 (4.47 to 5.53)	.083	4.78 (4.35 to 5.34)	4.94 (4.47 to 5.56)	.487
DRA	133 (70.0)	41 (80.4)	.141	24 (52.2)	37 (80.4)	.004c
Flexion strengthd			.585			.458
3	61 (32.1)	18 (35.3)		21 (45.7)	17 (37.0)
4	35 (18.4)	10 (19.6)		5 (10.9)	9 (19.6)
5	94 (49.5)	23 (45.1)		20 (43.5)	20 (43.5)

^a Data are presented as numbers (percentages) of participants unless otherwise indicated. Categorical variables are presented as numbers (percentages) of participants and compared using the χ² test. Continuous variables with a normal distribution are presented as mean (SD) and compared using the independent-samples t test. Continuous variables with a nonnormal distribution are presented as median (interquartile range [IQR]) and compared using the Mann–Whitney U test. BMI = body mass index; DRA = diastasis recti abdominis; GDM = gestational diabetes mellitus; OGTT = oral glucose tolerance test; RBG = random blood glucose test.

^b Age, BMI before pregnancy, gestational weeks, and weight gain were statistically different before matching (P < .05), and the differences were eliminated after 1:1 greedy matching. Differences in OGTT values existed both before and after matching (P < .05). Age, gestational weeks, weight gain, postpartum days, and the follow-up date were matched.

^c A significant difference in the prevalence of DRA was found between the group with GDM and the group without GDM after matching (37 [80.4%] vs 24 [52.2%] participants; P = .004).

^d The flexion strength of rectus abdominis was graded on a scale from 0 to 5 (0 = no contraction, 1 = flicker contraction, 2 = only lifting the head with arms at sides, 3 = arms outstretched above the plane of the body, 4 = arms crossing over the chest, and 5 = hands clasped behind the head).

Participants with GDM were matched with those without GDM using 1:1 greedy matching for the primary analysis. Additional sensitivity analyses were carried out using 1:1 optimal matching and 1:2 matching. The matching process was conducted with the MatchIt package in R studio (4.1.2) (https://cran.r-project.org/web/packages/MatchIt/vignettes/MatchIt.html).

In the greedy matching process, a participant in the group with GDM was assigned a participant in the group without GDM with the nearest PS as a match, and the distance between each matched pair was restricted to 0.1 times the SD of the PS by a caliper. Participants without matches were excluded. A standardized mean difference of <20% was suggested to represent a balanced sample.²⁸

In the optimal matching, 1:1 matches were made but restricted with a caliper of 0.25 that permitted the PS between each pair to differ by as much as 0.25 times the SD among all PS values. While a more relaxed caliper was used, the matched pairs were formed with the smallest sum of the absolute within-pair difference of the PS among any set of pairs.²⁷ The 1:2 matching was based on the greedy matching method with a caliper of 0.25, in which up to 2 participants in the group without GDM were matched to a participant in the group with GDM.

Statistical analyses were performed using SPSS 25.0 (IBM, Armonk, NY, USA). Several comparisons were conducted between the group with GDM and the group without GDM. Among these comparisons, continuous variables showing a normal distribution are presented as mean and SD and were compared using the independent-samples t test. Variables with a nonnormal distribution are presented as median (interquartile range) and were compared using the Mann–Whitney U test. The χ-square test was used to compare categorical variables, which are presented as numbers (percentages). The Spearman correlation coefficient was used to examine the correlation between the OGTT levels and the IRD at 7 locations. Correlations were categorized as very weak (0.00–0.19), weak (0.20–0.39), moderate (0.40–0.59), strong (0.60–0.79), and very strong (0.80–1.0).

The odds ratio of DRA among participants with GDM and those without GDM in samples created by matching were estimated in a logistic regression model that adjusted for variables associated with DRA in previous studies, including parity, mode of delivery, birth weight, and lactation. A 2-sided P value of <.05 was considered significant. The 2-stage linear step-up procedure described by Benjamin et al²⁹ was used to adjust the false discovery rate (FDR) for multiple tests (7 measured locations).

Results

Population and Characteristic

Overall, 258 postpartum women who met the inclusion criteria were included in the study, 17 of whom were excluded because baseline data or blood glucose test results were missing (Fig. 2). Among the remaining 241 women, 51 (21.2%) were diagnosed with GDM, and 174 (72.2%) had DRA postdelivery. There were significant differences in age, body mass index before pregnancy, gestational weeks, and weight gain between women with GDM and those without GDM. After 1:1 greedy matching, a sample of 46 women in each group was obtained, and these differences between the 2 groups were eliminated (Tab. 1; Suppl. Fig. 1). Their mean age was 31.14 (SD = 3.98) years, and the median weeks of gestation was 39.0 (interquartile range = 38.0 to 39.5. The median postpartum duration was 52.0 (interquartile range = 47.0 to 58.0) days. The standardized mean difference between the matched subjects was <20% (Suppl. Fig. 2). Supplementary Figures 3 and 4 show results for sensitivity analyses performed on samples created using optimal matching and 1:2 matching. There were 51 women in each group of the optimal matching sample, which manifested a more variant propensity score density. In the 1:2 matching sample, the women with GDM were matched with up to 2 of those without GDM; consequently, the GDM and control groups included 46 and 77 women, respectively. Both the samples had an standardized mean difference of <20%. The IRD values of women with DRA and those without DRA are shown in Supplementary Tables 1 and 2. There were significant differences in IRD at all measured locations between women with DRA, as diagnosed with the finger-palpation method, and those without DRA before and after varied matching processes (P < .001).

Figure 2.

Flowchart of the study. GDM = gestational diabetes mellitus.

DRA and the Flexion Strength of the Rectus Abdominis in the Groups With and Without GDM

Analysis after 1:1 greedy matching showed that the DRA rate was higher in the group with GDM than in the group without GDM (37 [80.4%] versus 24 [52.2%]; P < .01). There was no significant difference in the strength of the rectus abdominis between the group with GDM and the group without GDM (P > .05) (Tab. 1). No statistically significant difference in the flexion strength of the rectus abdominis was found between women with DRA and those without DRA (P > .05) (Suppl. Tab. 3). Sensitivity analysis with optimal matching and 1:2 matching yielded similar results (Suppl. Tabs. 3 and 4).

Difference in the IRD Values Between the Group With GDM and the Group Without GDM

The IRD values in the group with GDM were larger than those in the group without GDM at the superior border of the umbilicus as well as at 2, 3, and 4.5 cm above this border (P < .05; FDR-adjusted P < .05) (Tab. 2). There were no significant differences in the IRD values between the 2 groups at the locations below the umbilicus (P > .05; FDR-adjusted P > .05). Sensitivity analysis revealed that the results in the optimal matching sample were similar (Suppl. Tab. 5). The IRD value at 2 cm above the umbilicus was larger in the group with GDM than in the group without GDM in the 1:2 sample (P < .05), but this difference lost its significance following FDR adjustment (FDR-adjusted P > .05) (Suppl. Tab. 6).

Table 2.

Comparison of the IRD Values Between the Group With GDM and the Group Without GDM After 1:1 Greedy Matching^a

Location	IRD, in mm, in Group b :		Estimated Difference, in mm, Between Group Without GDM and Group With GDM (95% CI)	P c	FDR-Adjusted P d
	Without GDM (n = 46)	With GDM (n = 46)
Superior border of umbilicus	19.97 (17.73 to 32.47)	27.02 (20.35 to 33.98)	−4.10 (−8.32 to −0.36)	.027	.047
2 cm above	16.98 (12.37 to 23.94)	24.17 (18.11 to 28.14)	−5.41 (−8.97 to −1.51)	.005	.032
3 cm above	15.90 (11.10 to 21.04)	19.99 (14.33 to 28.13)	−4.42 (−8.22 to −0.91)	.017	.047
4.5 cm above	13.77 (10.39 to 21.71)	19.63 (13.76 to 25.60)	−4.11 (−7.75 to −0.44)	.030	.047
2 cm below	14.68 (9.17 to 22.52)	18.57 (12.75 to 24.04)	−2.68 (−6.32 to 1.18)	.168	.176
3 cm below	7.53 (4.68 to 17.51)	13.97 (6.67 to 18.20)	−2.28 (−6.21 to 0.81)	.151	.176
4.5 cm below	8.08 (3.76 to 18.55)	9.68 (4.55 to 17.95)	−0.10 (−2.79 to 3.15)	.960	.864

^a FDR = false discovery rate; GDM = gestational diabetes mellitus; IRD = interrectus distance.

^b Data are presented as median (interquartile range [IQR]) and compared using the Mann–Whitney U test.

^c The IRD values of the group with GDM at the superior border of the umbilicus and at 2, 3, and 4.5 cm above the border were significantly larger than those of the group without GDM at these 4 locations (P < .05).

^d The FDR was adjusted with the 2-stage linear step-up procedure described by Benjamin et al²⁹ for multiple comparisons (7 measured locations). The difference in the IRD values between the 2 groups was significant after FDR adjustment (FDR-adjusted P < .05).

Correlation of the Blood Glucose Levels With IRD and Strength of the Rectus Abdominis

The participants underwent OGTT and RBG tests before delivery. The Spearman correlation coefficient test revealed that the fasting OGTT level had a positive and weak correlation with IRD at the superior border of the umbilicus; at 2, 3, and 4.5 cm above the superior border of the umbilicus; and at 2 and 3 cm below the inferior border of the umbilicus (0.267 ≤ r ≤ 0.367; P < .05; FDR-adjusted P < .05). The 2-h OGTT level was positively correlated with the IRD at 2 and 4.5 cm above the superior border of the umbilicus and at 3 cm below the inferior border of the umbilicus, and the correlations were weaker and disappeared after FDR adjustment (0.214 ≤ r ≤ 0.235; P < .05; FDR-adjusted P > .05) (Tab. 3). In the optimal matching sample used for sensitivity analysis, the IRD at all of the 7 measured locations presented positive and weak correlations with the fasting OGTT level (0.208 ≤ r ≤ 0.341; P < .05; FDR-adjusted P < .05). The 1-h OGTT level was positively correlated with the IRD at 4.5 cm above the superior border of the umbilicus; this correlation was insignificant after FDR adjustment (r = 0.214; P < .05; FDR-adjusted P > .05). The 2-h OGTT level had a weak and positive correlation with the IRD at 2, 3, and 4.5 cm above the superior border of the umbilicus and at 3 cm below the inferior border of the umbilicus; this correlation did not survive the FDR adjustment (0.204 ≤ r ≤ 0.248; P < .05; FDR-adjusted P > .05) (Suppl. Tab. 7). In the 1:2 matching sample, the fasting OGTT level had a positive and weak correlation with the IRD at the superior border of the umbilicus; at 2 and 4.5 cm above the superior border of the umbilicus; and at 2 and 3 cm below the inferior border of the umbilicus. However, the correlations were insignificant after FDR adjustment (0.195 ≤ r ≤ 0.215; P < .05; FDR-adjusted P > .05) (Suppl. Tab. 8). There was no correlation between the OGTT levels and the strength of the rectus abdominis (P > .05; FDR-adjusted P > .05). The RBG level was not correlated with the IRD or the strength of the rectus abdominis (P > .05; FDR P > .05).

Table 3.

Correlations of OGTT and RBG Before Delivery With IRD and Flexion Strength of the Rectus Abdominis After 1:1 Greedy Matching^a

Parameter	r Value for:
IRD at:	OGTT Fastingb	OGTT at 1 h	OGTT at 2 hc	RBG Before Delivery
Superior border of umbilicus	0.322d,e	0.116	0.184	0.038
2 cm above	0.367d,e	0.169	0.235f	0.107
3 cm above	0.324d,e	0.168	0.200	0.041
4.5 cm above	0.363d,e	0.193	0.214f	−0.011
2 cm below	0.267f,g	0.07	0.137	0.006
3 cm below	0.302d,e	0.142	0.224f	0.014
4.5 cm below	0.192	0.022	0.094	0.029
Flexion strength	0.020	−0.025	0.100	0.076

^a The correlations between the OGTT levels and the IRD at 7 locations were examined with the Spearman correlation coefficient. IRD = interrectus distance; OGTT = oral glucose tolerance test; RBG = random blood glucose test.

^b The fasting OGTT value was positively correlated with the IRD at the superior border of the umbilicus; at 2, 3, and 4.5 cm above the superior border of the umbilicus; and at 2 and 3 cm below the inferior border of the umbilicus (0.267 ≤ r ≤ 0.367; P < .05; false discovery rate [FDR]–adjusted P < .05). The FDR was adjusted with the 2-stage linear step-up procedure described by Benjamin et al²⁹ for multiple correlation tests (7 measured locations).

^c The 2-h OGTT value was positively correlated with the IRD at 2 and 4.5 cm above the superior border of the umbilicus and at 3 cm below the inferior border of the umbilicus, but the correlations did not survive the FDR adjustment (0.214 ≤ r ≤ 0.235; P < .05; FDR-adjusted P > .05).

^d P < .01.

^e FDR-adjusted P < .01.

^f P < .05.

^g FDR-adjusted P < .05.

Multivariable Analysis of Postpartum DRA

The multivariable logistic regression analysis revealed that GDM was associated with higher odds of postpartum DRA (Tab. 4). Women with GDM were found to have 4.792 times higher odds (95% CI = 1.672 to 13.736; P = .004) of experiencing DRA in the first year postdelivery than those without GDM, after adjustment for age, parity, gestational weeks, weight gain, mode of delivery, birth weight, postpartum days, and lactation. Moreover, cesarean section was also associated with higher odds of postpartum DRA (adjusted odds ratio = 8.111; 95% CI = 1.953 to 33.679; P = .004). Sensitivity analysis using optimal matching and 1:2 matching methods showed similar results (Suppl. Tabs. 9 and 10).

Table 4.

Multivariable Analysis of DRA After 1:1 Greedy Matching^a

Variable	AOR (95% CI)	P
GDM	4.792 (1.672 to 13.736)	.004b
Parity	2.414 (0.938 to 6.215)	.068
Mode of delivery (cesarean section)	8.111 (1.953 to 33.679)	.004c
Birth weight	0.525 (0.154 to 1.793)	.304
Lactation	0.219 (0.020 to 2.371)	.211

^a Adjusted for parity, mode of delivery, birth weight, and lactation. AOR = adjusted odds ratio; DRA = diastasis rectus abdominis; GDM = gestational diabetes mellitus.

^b GDM (AOR = 4.792; 95% CI = 1.672 to 13.736; P = .004) was associated with DRA.

^c Cesarean section (AOR = 8.111; 95% CI = 1.953 to 33.679; P = .004) was associated with DRA.

Discussion

Our study showed that GDM was associated with postpartum DRA in the first year postdelivery. Women with GDM showed larger IRD values at the superior border of the umbilicus and the locations above it. Moreover, the fasting OGTT level had a positive and weak correlation with the IRD values at most of the measured locations.

In this study, 72.2% women were diagnosed with DRA within 1 year, but mostly less than 6 months, postdelivery using a relatively recognized palpation method. This was similar to the previously reported prevalence.^2,5–8 Some of the characteristics including age, gestational weeks, weight gain and postpartum days may be potentially associated with GDM and DRA.^7,10,14,15 For example, women with GDM are at a higher risk of preterm birth with shorter than normal gestational duration, which may, however, result in a smaller IRD and covering up of the effect of hyperglycemia on IRD to some extent. Moreover, the measurement conditions on different follow-up dates such as the environment and the evaluator’s performance may affect the measurement results. Hence, we used PS matching to control such effects of the confounding factors. The balance of these covariates allowed us to improve our understanding of the association between GDM and DRA.²⁷

To the best of our knowledge, the current study is the first study to investigate the association between GDM and DRA. In this study, women with GDM were found to have 4.792 times (95% CI = 1.672 to 13.736; P = .004) higher odds of having postpartum DRA than those without GDM; this result suggests that a hyperglycemic status can influence postpartum DRA even though it persists only during pregnancy. Interestingly, despite the absence of a statistically significant difference in the RBG level before delivery between the 2 groups, which was regarded as good blood glucose control in the group with GDM, a difference in the prevalence of DRA still existed between the groups. Patients with GDM in the study of Vesentini et al¹² also showed pathological muscle changes after blood glucose control or treatment, despite the lack of blood glucose tests before delivery. Because the RBG test is not sufficiently sensitive,³⁰ further studies are needed to verify whether this result revealed that the blood glucose control could not prevent postpartum DRA.

In addition to palpation, we used ultrasound to measure IRD at 7 locations and explored the association between GDM and the IRD values. Measuring IRD with ultrasound has been reported to show excellent reliability and validity,³¹ and thus, it has been widely used to assess the severity and influence of DRA.^4,8,9 The measured locations varied in the previous literature; however, we synthesized and finally chose 7 common locations to measure IRD,^10,24,25 covering the upper and lower parts of the rectus abdominis. Although our results were not perfectly consistent in the 3 matching samples, they pointed out that the IRD values at the upper part of the rectus abdominis among women with GDM were larger than those among women without GDM, and the IRD values at the lower part presented a similar tendency (Tab. 2; Suppl. Tabs. 5 and 6) but without statistical difference (P > .05). The diverse results between the upper and lower parts of the rectus abdominis were unclear. One of the acknowledged distinctions between the 2 parts is that the upper part is wrapped in the rectus sheath, which consists of connective tissue, mainly collagen; the posterior rectus sheath gradually disappeared below the arcuate line from the lower part.^32,33 GDM was found to influence the content and localization of collagen.¹² Nonetheless, further studies are needed to explore whether GDM affects the upper IRD by influencing changes in collagen in the posterior rectus sheath. Moreover, there were weak correlations between the fasting OGTT level and the IRD values at both upper and lower parts of the rectus abdominis, which suggests a potential effect of basic blood glucose level and insulin secretion on the rectus abdominis³⁴; therefore, the fasting OGTT level should be controlled to reduce the severity of postpartum DRA.

Flexion strength is an important function of the rectus abdominis. In the current study, we investigated the relationship between the flexion strength of the rectus abdominis and GDM. Several studies demonstrated an adverse impact of DRA and IRD on the strength of rectus abdominis.^8,35 However, the current study did not show a relationship between DRA and the strength of rectus abdominis. Although women with GDM had higher odds of getting DRA, and IRD was found to be correlated with GDM and the fasting OGTT level in our study, we did not find any associations between the flexion strength of the rectus abdominis and GDM or any blood glucose levels. Because the correlations between IRD and the strength of rectus abdominis in the studies above^8,35 were very weak (|r| < 0.2), further studies should be carried out to explore the effect of GDM on the abdominal muscle function.

Limitations

The current study has several limitations. First, this is a single-center, retrospective study with a small sample size. However, as this is the first study to investigate the relationship between GDM and DRA, more multi-center clinical trials with large sample sizes can be conducted on the basis of our findings. Second, multiparous women were included in this study, and it remains unknown whether they had DRA in previous pregnancies. The persistent DRA before this pregnancy may influence the results of the current study to some extent. Third, we used the palpation method to diagnose DRA, the accuracy of which may sometimes be influenced by the thickness of abdominal subcutaneous fat.⁸ However, palpation is the most common approach in the clinical practice^10,23 with a more recognized standardization. To be more precise, we also used ultrasound to measure IRD and analyze the relationship between GDM and IRD. Third, the results in this study should be interpreted with caution as the multiplicity of 7 measured locations may inflate the type I error. We used a relatively mild method to adjust the FDR; however, the results may vary if other adjusting methods are used. Moreover, the RBG test used just before delivery was not sufficiently sensitive to assess the participants’ blood glucose level, and this test should be replaced by OGTT or fasting blood glucose test to better evaluate the condition of blood glucose control in further studies. Finally, our study only examined the flexion strength of the rectus abdominis. Further investigations should evaluate additional functions of abdominal muscles, including rotation strength and endurance.

Conclusion

Our findings showed that GDM was associated with postpartum DRA in women in the first year postdelivery. Moreover, the IRD values at the upper part of the rectus abdominis measured with ultrasound were larger in women with GDM than in those without GDM. Women with GDM had higher odds of postpartum DRA. An increased focus on continued assessment of the upper part of rectus abdominis during and after rehabilitation should be carried out for these women. Moreover, there was a weak and positive correlation between the fasting OGTT level and the IRD values at the entire rectus abdominis, which suggested that the fasting OGTT level should be controlled to reduce the severity of postpartum DRA.

Author Contributions

Jingran Du (Conceptualization-Lead, Data curation-Equal, Funding acquisition-Equal, Investigation-Lead, Methodology-Equal, Resources-Equal, Validation-Equal, Writing – original draft-Equal), Juntong Ye (Conceptualization-Equal, Data curation-Lead, Formal analysis-Lead, Investigation-Equal, Methodology-Lead, Software-Lead, Visualization-Lead, Writing – original draft-Lead), Hui Fei (Conceptualization-Equal, Investigation-Lead, Methodology-Equal, Resources-Equal, Writing – original draft-Equal), Mengxiong Li (Data curation-Equal, Investigation-Equal, Resources-Equal, Validation-Lead, Writing – review & editing-Equal), Juan He (Investigation-Equal, Methodology-Equal, Resources-Equal, Writing – review & editing-Equal), Yun Liu (Investigation-Equal, Resources-Equal, Writing – review & editing-Equal), Lixiang Liu (Investigation-Equal, Supervision-Equal, Writing – review & editing-Equal), Yangliu Ye (Investigation-Equal, Resources-Equal, Writing – review & editing-Equal), Juanhua Li (Investigation-Equal, Supervision-Equal, Writing – review & editing-Equal), Lili Hou (Investigation-Equal, Resources-Equal, Writing – review & editing-Equal), Yang Xu (Investigation-Equal, Supervision-Equal, Writing – review & editing-Equal), Haixia Zhang (Investigation-Equal, Resources-Equal, Writing – review & editing-Equal), Changlin Zhang (Funding acquisition-Equal, Writing – review & editing-Equal), Tian Li (Conceptualization-Equal, Project administration-Lead, Resources-Lead, Writing – review & editing-Lead).

Ethics Approval

This study was approved by the Ethics Committee of the Seventh Affiliated Hospital of Sun Yat-sen University (2020SYS-USH-055).

Funding

This study was supported by funds from the Guangdong Basic and Applied Basic Research Foundation (2019A1515110085).

Data Availability

The data underlying this article cannot be shared publicly for the privacy of individuals that participated in the study. The data will be shared on reasonable request to the corresponding author.

Disclosures

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