Effect of physical activity and exercise on endometriosis-associated symptoms: a systematic review

Studies of interventions involving any type of PA and exercise were eligible for inclusion. PA was defined as “any bodily movement produced by skeletal muscles that requires energy expenditure” [21] and exercise was defined as “PA that is planned, structured, and repetitive for the purpose of conditioning the body” [21], consisting of cardiovascular conditioning, strength and resistance training, and flexibility.

The study population consisted of women with any degree of endometriosis as diagnosed with an imaging or surgical modality, who presented with pain in the pelvic region (including dysmenorrhea, dyspareunia, or CPP). The primary outcome measure was the pain intensity, but all outcomes were accepted.

Exclusion criteria were data presented in short communications, reviews, letters to the editor, and congress abstracts, and the application of passive interventions such as manual therapy to patients. The literature search was completed with support from a trained medical librarian. The search included the Cochrane Central Register of Controlled Trials, Embase, PubMed, MEDLINE, PsycInfo, CINAHL, AMED, Scopus, Web of Science, PEDro, and SveMed + , without time limitation up to December 2020. Publications could be in English, Swedish, Norwegian, Danish, or German. Search terms were identified through a pilot search for relevant literature. The electronic search strategy for this systematic review is presented in Additional file 2. In addition, the reference lists of included articles and identified reviews on the topic were scanned, and manually searched for further studies.

In the first step, all obtained references were independently screened on the basis of the title and Abstract by M.K.T. and T.T. using the Rayyan web application [22] that allows blinded assessments. In the second step, all Abstracts with conflicting decisions were reviewed by both authors until consensus was reached. In the third step, the same authors independently assessed the methodological quality of the manuscripts that met the inclusion criteria, using quality assessment questionnaires appropriate for the design of each study as provided by the National Heart Lung and Blood Institute [23]. We applied the method of “quality assessment of controlled intervention studies” for randomized controlled trials (RCT), and the “quality assessment of before-after (pre-post) studies with no control group” for intervention studies with no control group while adding relevant questions to determine exposure, risk, and confounding variables. The assessment tools include several criteria rated “yes,” “no,” and “other: CD, cannot determine; NA, not applicable; NR, not reported.” The quality of the included studies was rated as good, fair, or poor. We also used the Consensus on Exercise Reporting Template (CERT) [24], which is a 19-item checklist that yields a detailed description of the minimum criteria that should be reported in an exercise intervention. The template provides individual scores for each included article (ranging from 0 to 19), in addition to a summary score for each item.

The full text of eligible articles was read by two reviewers (M.K.T. and T.T.), who independently extracted the following data: author(s), year of publication, study period, country of origin, study design, sample size, inclusion and exclusion criteria, intervention type, description of intervention, follow-up period, primary and secondary outcomes, and dropout rate.

This study identified 1879 citations (Fig. 1). After removing duplicates, the remaining 1045 citations were screened for eligibility based on the title and Abstract. Seventeen publications were assessed for further inclusion reading the full-text versions of the articles, and four publications were included for quality assessment [25‐28]. We identified four studies that described an intervention incorporating PA and/or exercise: two were RCTs [27, 28] and two were pre-post studies with no control group [25, 26] (Tables 1, 2).

One study was rated as being of fair quality [27], while three were rated as poor quality [25, 26, 28]. The detailed assessment including signaling questions are presented in Tables 1 and 2. The RCT of Carpenter et al. [27] was judged as being of fair quality (Table 1). The main limitation of that RCT for the purpose of this review was that the participants were treated with danazol, which is a potent drug for treating endometriosis. Though having a control group, the study was underpowered for determining whether exercise had an additional effect to danazol. However, since the study was designed to investigate if exercise could alleviate the side effects of danazol, it was not flawed per se. Moreover, the sample was too small to allow comparisons of individual side effects, important secondary outcomes (pelvic pain, dysmenorrhea, and dyspareunia) were not reported, and the methods of randomization and outcome assessment were not reported.

The RCT of Goncalves et al. [28] was judged as being of poor quality due to significant differences in the baseline characteristics between the intervention and control groups (Table 1). The intervention group had a higher level of education, a high percentage of homemakers, and a lower rate of employment, which is a confounder for quality-of-life assessments. Also, one of the inclusion criteria was the presence of therapy-resistant CPP, which is a possible confounder for endometriosis-associated symptoms. Furthermore, the control group also received physiotherapy. Finally, the dropout rate in the intervention group was very high, at 30% (vs 0% in the control group).

The study of Friggi Sebe Petrelluzzi et al. [25] was judged as being of fair quality (Table 2). As in Goncalves et al. [28] only women with endometriosis and therapy-resistant CPP were included, representing a confounder. Furthermore, a sample-size calculation was not reported, and there was no control group. The intervention consisted not only of PA and exercise, but also a range of modalities including behavioral cognitive therapy, which confounds the contribution of PA and exercise to symptom improvement.

The study of Awad et al. [26] was judged as being of poor quality (Table 2). It was ultimately excluded from the synthesis since its design was fatally flawed by initiating medroxyprogesterone acetate, an effective hormonal treatment for endometriosis, at the same time as the intervention but without including a control group. Furthermore, no sample-size calculation was provided, and the inclusion and exclusion criteria appeared to be random from a clinical perspective.

The individual scores for the articles based on the CERT checklist (Additional file 3) ranged from 7 to 14. None of the articles provided a description of exercise progression [25‐28], and only one included a description of individually tailored exercises [27]. Exercise adherence was measured adequately in one study [27], as were motivational strategies [25]. The level of exercise was only described for two studies [27, 28].

The total study sample consisted of 109 participants [25, 27, 28] (Table 3). Two studies included women with surgically confirmed endometriosis [25, 27], while it was not specified how endometriosis was diagnosed by Goncalves et al. [28]. The stage of endometriosis was not reported for any of the studies. The age of the included women was provided for two studies [25, 28]. All of the women in two studies [25, 28] also had CPP. Details of prior hormonal or surgical treatments were not provided for any of the studies.

The performed interventions are listed in Table 1. No study performed a follow-up after the intervention had finished. Confounding interventions to PA and exercise were identified in all studies, as explained above. Limitations in the reporting of exercise interventions (according to the CERT) are also explained above.

The primary and secondary outcomes for all studies are reported in Table 3. Only one study had “pain” as the primary outcome [25]. The outcome reports were incomplete for all studies.

Goncalves et al. [28] reported that the degree of daily pain was significantly lower in the intervention group than the control group, although the difference in the mean scores on a visual analogue scale (VAS) was not provided (p < 0.001). Furthermore, the scores in pain-related domains on Endometriosis Health Profile-30 (EHP-30) were significantly lower in the yoga group than the control group at postintervention (32.39 ± 21.95 versus 55.05 ± 21.49, p < 0.001). Notably, the control group also received physiotherapy following the intervention at their institution.

Friggi Sebe Petrelluzzi et al. [25] did not find a significant improvement in pain intensity (change in VAS score from pre- to posttreatment: 4.00 ± 0.56 to 3.30 ± 0.65, p > 0.05). Carpenter et al. [27] found that the pelvic pain decreased in both the intervention and control groups, with medical treatment using danazol providing no additional effect relative to that obtained by PA and exercise. However, the exact results and significance level were not reported, leaving it uncertain about whether a type II error was present due to the sample being too small.

The study of Friggi Sebe Petrelluzzi et al. [25] measured stress levels using the Perceived Stress Questionnaire (PSQ), salivary cortisol levels, and the 36-item Short-Form Health Survey (SF-36). The PSQ was developed as an outcome measure in psychosomatic research and was validated for use in Brazil [29]. Perceived stress was significantly lower at pretreatment (0.62 ± 0.02) than posttreatment (0.56 ± 0.02, p < 0.05). Significant improvements in the vitality and physical functioning domains of the SF-36 were also found (p < 0.05), but these effects were no longer significant after performing a multivariate analysis that included each variable in the SF-36. There was an overall decrease in salivary cortisol levels from pretreatment to posttreatment (p = 0.04), but this was not correlated with perceived stress as measured with PSQ. Goncalves et al. [28] found significant improvements in certain EHP-30 items (control and powerlessness, emotional well-being, and self-image) in the intervention group compared with the control group (p < 0.001).

The study of Carpenter et al. [27] assessed how exercise during danazol treatment could improve pelvic floor symptoms such as dyspareunia and dysmenorrhea. Those authors reported that the symptoms improved in both groups, but the values and significance levels were not provided. Goncalves et al. [28] found that the sexual-intercourse domain of EHP-30 was lower after 8 weeks of Hatha yoga in both the intervention and control groups, but the result did not reach between- or within-groups significance.

The strengths of this systematic review include its originality, rigorous search strategy, and methodological robustness. Its main limitation is the low grade of evidence that could be obtained from the previous studies. The small samples, confounding factors, heterogeneity of interventions, and poor reporting of details about the exercise intervention and outcome measures restricts our ability to draw overall conclusions about the effect of PA and exercise in treating endometriosis-associated symptoms.