A search was undertaken to identify relevant studies in the electronic databases MEDLINE/Pubmed, Embase, CINAHL, Web of Science, PEDro, and Sport Discus. There was no time limit for publication date. Also, a manual search of reference lists and related studies was conducted. The following search was performed in PubMed on March 18, 2020; (“randomized controlled trial” OR “randomised controlled trial”) AND (“recti abdominis” OR “abdominal rectus diastasis” OR “diastasis recti”) AND (postpartum OR postnatal). Box 1 presents the inclusion criteria for eligible studies. Two independent reviewers screened the titles and abstracts and then evaluated articles available in full text for eligible studies. Any disagreement was solved through discussion until a consensus was reached. Other modalities, e.g. therapeutic taping technique or abdominal binding, could be included in one or more interventions or as a separate intervention.

We extracted data on participants' characteristics (age, parity, delivery mode), intervention with training dosage (mode of exercise, duration of the exercise period, frequency, training volume, and adherence), DRA cut-off value, measurement method, and primary and secondary outcome measures. In studies with insufficient information, authors were contacted for further details.

The PEDro scale was used to evaluate the risk of bias. The PEDro score ranges from 0 to 10 with higher scores indicating superior methodological quality. A total PEDro score equal to or less than three points are considered poor, score from four to five are considered fair, six to eight are considered good, and nine to 10 are considered excellent.22 The PEDro scale has been found to be a valid tool to evaluate methodological quality in clinical trials.23,24 Study selection and data extraction were evaluated independently by two reviewers. For risk of bias, when available we used the PEDro score available in the PEDro website, if not available two reviewers independently rated the trial.

To assess levels of evidence for the meta-analysis and the studies comparing abdominal training to a minimal intervention group, we used the Cochrane Collaboration Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.26 Two authors independently reviewed each study. The GRADEpro GDT27 was used to develop a summary of findings table. The quality of evidence for the meta-analysis was downgraded according to the presence of the following: risk of bias (downgraded by one level if more than 25% of the participants were from studies with poor or fair methodological quality), inconsistency of results (downgraded by one level if significant heterogeneity was present by visual inspection or if the I2 value was greater than 50%), and imprecision (downgraded by one level if fewer than 70 participants were included in the comparison or downgraded by two levels if participants from pilot studies were included in the meta-analysis). Single randomized trials were considered inconsistent and imprecise (that is, sparse data) and provided “low quality” evidence. This could be further downgraded to “very low” quality evidence if there was also high risk of bias.

Meta-analysis was considered appropriate only for those studies using similar outcome measures, measurement methods, and control groups. The Review Manager 5.4 software, from the Cochrane Collaboration, was used to conduct the meta-analysis. Mean, standard deviation, and sample size from each group were extracted and used to estimate effect sizes. Pooled effect sizes were calculated using fixed effect models and expressed as mean difference (MD) with 95% confidence intervals (CI) in the forest plot. The I squared value if lower than 50% was used to confirm homogeneity among included studies.

When trials were not sufficiently homogeneous, pooling of data via meta-analysis was not performed. Trials were grouped according to the type of intervention (i.e. TrA training, PFM training, and curl-up training). Outcome measures of the individual studies were extracted and difference between groups were expressed as MD and 95% CI.

The systematic literature search identified 31 potential records. In addition, two additional records were identified through personal knowledge. After removing duplicates (n = 15) and irrelevant studies (n = 6), 12 full-text articles were assessed for eligibility. A total of seven studies were included in this review. No relevant studies were identified through manual search of reference list. Supplemental Online Material shows the flow of studies in the review.

Studies were published between 2016 and 2020 and were conducted in six different countries. Detailed characteristics of included studies are presented in Table 1. Regarding the study design, two studies28,29 were pilot RCTs and five studies25,30–33 were RCTs. The sample size varied from nine30 to 175,25 and all women were between 18 and 45 years old. Time since birth for inclusion varied between a couple of days31 to three years.30 Parity and delivery mode were not reported in two of the included studies31,33 and the others contained a mix between primi and/or multiparous women and women with cesarean section and/or vaginal delivery.25,28,29,30,32 One study25 was a secondary analysis of a 2-arms RCT in which the primary aim was to evaluate the effect of PFM training on urinary incontinence.

Presence of DRA or IRD change was the primary outcome measure in all included studies. However, the studies used different measurement methods; ultrasound,28,29,32 palpation,25,31 and both ultrasound and caliper.30 The studies measured IRD at different places along linea alba and in different positions, i.e. rest, head lift, and modified curl-up. In addition, the included studies used different cut-off values for DRA, such as 2.0 cm, 2.5 cm, and 2 finger-widths. Secondary outcome measures varied among included studies. Secondary outcome investigated were symptoms of PFD measured with the Pelvic Floor Distress Index (PFDI),28,29,30 self-report low back disability measured with the Roland Morris Disability Questionnaire (RMQ)28 and the Oswestry Disability Index (ODI),30 and abdominal muscle strength measured with an isokinetic dynamometer (Biodex)32 and with a static trunk flexion endurance test.29 In addition, measures of self-reported physical function in the postpartum period29,33 and body image29 were assessed.

Interventions, training dosage, and results for primary and secondary outcomes of included studies are presented in Table 2. Many treatment programs contained a plethora of different exercises, modalities, and combinations. Four studies compared the intervention to a minimal intervention group.25–29,31 The control groups included education,31 standard information after delivery,25 and instruction to maintain normal activity level.28 The interventions were performed as home exercise only in some studies28,30,31 and with individual supervision at the clinic.32 Two studies combined daily home training with either supervised weekly group exercise or individual treatment.25,29 The duration of the exercise period varied between six and 16 weeks and total number of repetitions varied from 4028 to 210 per week.29 Drop-out varied from no drop-out33 to 15.5% at 6 months post-test.29 Adherence to the exercise programs varied from 73%29 to 95%.28 No adverse effects were reported. Four studies28,32-34 reported a statistically significant difference between groups in reduction of numbers with DRA or decrease in IRD. Of these studies, two28,31 compared a physical therapy intervention to a minimal intervention group (i.e. education). Three studies25,29,30 did not find a statistically significant decrease in IRD after their training programs.

Supplemental Online Material shows the scores on the PEDro Rating scale. There were no disagreements between the assessors in the evaluation process. The PEDro score varied between four and eight points.

There were few reports on our selected secondary outcomes in the published RCTs (Table 2). One study reported a positive effect on body image and two studies measuring abdominal muscle strength reported positive effects.29,33 No statistically significant effects were found on low back pain or PFD.28,29,30 Two studies found contradictory results in self-report of physical function postpartum.29,33

PFM training was part of the exercise program30,33 or the primary intervention.25 Out of four RCTs reporting a positive effect on IRD or prevalence of DRA, only Thabet and Alshehri33 included PFM training. Out of three RCTs reporting no effect on IRD and prevalence on DRA, two studies included daily PFM training25 or PFM training as part of the exercise in both intervention groups.30 Therefore, it is reasonable to conclude that PFM training was not the exercise causing the effect found in four studies reporting effect on IRD or prevalence of DRA. Also, if PFM training has a positive effect, this should have been found in the study25 where the focus was on this muscle group. The latter is supported by the findings in several studies20,21,41,42 showing a significant widening, not narrowing, of the IRD during a single PFM contraction. PFM training is first-line treatment for urinary incontinence in women43 and has also shown to be effective in the early postpartum period.44 Although the immediate effect of contracting the PFM has shown a widening of the IRD, this widening is minimal (mm)21,42 and probably does not influence DRA. Women with DRA should therefore not be discouraged from doing PFM training in the postpartum period.

Of the four studies reporting a positive effect on IRD or prevalence of DRA, two studies included mainly exercises targeting TrA.28,33 One study33 did not include a minimal intervention group, and another28 had a very small sample size. In contradiction, two studies29,30 found no effect of TrA training but their results were not compared with a minimal intervention group and included women who may not be classified as having DRA,30 or had a very small sample size.29 Hence, there is very low quality evidence that TrA training is more effective than minimal intervention for treating DRA. Experimental studies have shown that TrA contractions widen the IRD,19,21,41 and the effect of training TrA over time may therefore also be questioned.

Two of the RCTs reported a positive effect on IRD or prevalence of DRA from curl-up exercises in women 0–3 days postpartum31 or two months postpartum.32 Kamel andYousif32 did not include a minimal intervention group. Due to the natural decrease in IRD postpartum2 and also the inclusion of other elements in the training protocol (e.g. neuromuscular electrical stimulation, prone lying) it is not possible to conclude whether curl-ups or twisted curl-ups are effective in the decrease of IRD or prevalence of DRA. Several experimental studies have shown that curl-up leads to an immediate decrease in IRD.41,42 A possible explanation for why curl-up might be more effective than PFM or TrA training is that the insertion and origin of TrA means that a contraction of the muscle pulls away from the midline. Because there is a co-contraction of TrA in a maximal voluntary contraction of the PFM,45 this may explain why contraction these muscle groups may increase the IRD. There is a need for more basic research to understand the influence of the abdominal muscles on the linea alba and IRD.

Regarding the secondary outcomes of this review, a lack of effectiveness was found on low back pain and PFD.28,29,30 No association between DRA and PFD in the postpartum period has been found in studies of other designs.2,45,46,47 Our results indicate that some of these exercise programs might improve body image, physical function, and abdominal muscle strength.29,32 However, whether these complaints are caused by or related to DRA is currently unclear.2,8

A limitation of our review is the inclusion of studies published in English, German, or Scandinavian languages only. Four of seven included studies did not involve a minimal intervention group. This is considered a limitation because of the natural remission of DRA until at least 12 months postpartum.2 Because intervention protocols often combined different exercises and modalities, it is not possible to conclude which specific exercises may have caused the effect in some of the RCTs.28,29,30–33 In addition, with the results of experimental studies in mind,20,41 inclusion of different types of exercises as part of the same intervention may have led to the effect of the exercises cancelling each other. Physical therapists should be cautious in promising effect of, or advocating specific exercises, in the treatment of DRA. There is an urgent need for larger, high-quality RCTs with designs to treat women with DRA, investigating the effect of single exercises on IRD and DRA in the post-partum period. As all the RCTs so far have included women with mild/moderate DRA only,25,28,29,30-33 there is also an urgent need to conduct RCTs in women with severe diastasis.