Accuracy of MRI, ultrasound and vaginal assessment for the diagnosis of levator ani muscle avulsion in women

Risk factors for levator ani muscle avulsion

A variety of childbirth related risk factors for sustaining levator ani muscle avulsion are described in literature. Compared to a spontaneous vaginal delivery, forceps delivery increases the risk on levator ani muscle avulsion 3.8 to 14.7 times. The prevalence of levator ani muscle avulsion after forceps delivery is 35-71% (Kearney 2006a; Krofta 2009; Kearney 2010; Shek 2010a; Cassado 2011; Chan 2012; Chung 2014; Durnea 2014; van Delft 2014a), whilst ventouse delivery has not been shown to be a risk factor (Kearney 2006a; Valsky 2009; Shek 2010a; Durnea 2014; van Delft 2014a). Other risk factors are prolonged second stage of labour (Kearney 2006a; Valsky 2009; Kearney 2010; Shek 2010a; Durnea 2014; van Delft 2014a), and obstetric anal sphincter injuries (Kearney 2006a; van Delft 2014a). The literature is ambivalent regarding the risk factors: maternal age, fetal head circumference, fetal birth weight and episiotomy as some studies did, while others did not, find an association with levator ani muscle avulsion (Kearney 2006a; Dietz 2007b; Valsky 2009; Shek 2010b; Cassado 2011; Chan 2012; Cassado 2014; Durnea 2014; Low 2014; van Delft 2014a). Previous findings reported that epidural analgesia during delivery to be protective for sustaining levator ani muscle avulsion (Shek 2010a), whilst others did not find a positive or negative association (Kearney 2006a; van Delft 2014a).

Antepartum risk factors for sustaining levator ani muscle avulsion have been investigated, but the only significant risk factor was a lower body mass index, however the clinical significance is questionable as the body mass index was 28 versus 30 kg/m² (Shek 2010b). The anterior-to-posterior bony pelvis dimension is shorter in middle-aged women with complete bilateral levator ani muscle avulsion compared to women with normal muscles (Berger 2013). It is therefore plausible that a variation in bony pelvic dimensions may pose as a risk factor for delivery-induced levator ani muscle avulsion.

Effect of levator ani muscle avulsion on pelvic floor dysfunction

Levator ani muscle avulsion has a significant impact on symptoms and signs of pelvic floor dysfunction. Pelvic floor dysfunction refers to a wide range of symptoms that women can develop due to various pelvic floor disorders and include symptoms of pelvic organ prolapse, bladder, bowel and sexual dysfunction. Although pelvic floor dysfunction is not a life-threatening condition, it impacts on quality of life and can have a devastating impact on a woman's physical, social and psychological well-being (Boreham 2005; Oliveira 2013; Bezerra 2014; Doaee 2014).

Importance of diagnosing levator ani muscle avulsion

Apart from perineal, vaginal and anal sphincter injury, levator ani muscle avulsion is now believed to be a frequently unrecognised form of pelvic floor trauma following childbirth. Although current evidence shows that levator ani muscle avulsion cannot be effectively repaired surgically (Rostaminia 2013a), it is important to diagnose levator ani muscle avulsion to identify those women who are at higher risk of developing pelvic floor dysfunction and to assess women with symptoms of pelvic floor dysfunction in order to optimise treatment strategy.

Pelvic floor dysfunction is a major health care problem seen in 40% of women attending urogynaecology clinics (Dua 2014). As female life expectancy is increasing, a concomitant rise in the number of women suffering from pelvic floor dysfunction can be expected. It is estimated that over the next 30 years, the demand for treatment of pelvic floor dysfunction will increase by 45% (Luber 2001), and research is being focused on investigating preventive strategies for the development of pelvic floor dysfunction (DeLancey 2007a). To prevent the development of pelvic floor dysfunction, women with levator ani muscle avulsion could be counselled and advised on the benefits of pelvic floor muscle exercises, as this has been shown to be protective (Boyle 2012), although another study did not find this to have a positive effect (Bø 2014). There is a need for other preventive strategies for pelvic floor dysfunction to be explored and evaluated, including the assessment of peripartum interventions for the reduction of birth trauma, e.g. perineal protection, episiotomy, epidural analgesia, warm packs, perineal and pelvic floor massage, birth position, water birth.

Similar to obstetric anal sphincter injuries, it has been suggested that levator ani muscle avulsion should be regarded as one of the key performance indicators of maternal birth trauma (Dietz 2015), and for this purpose all women would require postnatal imaging follow-up after vaginal childbirth. In order to be able to investigate causal factors and consequences of levator ani muscle avulsion and to find possible treatment options of levator ani muscle avulsion, the ability to diagnose levator ani muscle avulsion accurately is of crucial importance.

Ultrasound

Ultrasound is a non-invasive investigation that does not use ionising radiation. It is a fast and well-tolerated imaging technique with minimal patient discomfort, which could easily be carried out in an outpatient setting at the same time as the consultation and clinical examination. Compared to MRI, the costs of ultrasound are relatively low and include a fixed purchase price of the ultrasound machine and variable overheads for maintenance, ultrasound gel and probe covers. There are no risks or adverse events associated with the use of ultrasound. The only disadvantage of ultrasound is that the image acquisition and interpretation is operator dependent.

Vaginal assessment

Palpation of the levator ani muscle on vaginal assessment has also been suggested as a useful tool in the detection of levator ani muscle avulsion (Dietz 2006b; Kearney 2006b; Dietz 2008a; Dietz 2008c; Dietz 2008d; Dietz 2012b). The inter-observer agreement varies from moderate in women with pelvic floor dysfunction (Kearney 2006b; Dietz 2008c), to excellent in nulliparous women (van Delft 2013). Vaginal palpation is a simple, universally available and non-invasive test for no additional cost, however, it requires a substantial learning curve (Dietz 2012b).

Alternative test(s)

All available diagnostic tests for the diagnosis of levator ani muscle avulsion are under evaluation in this review: MRI, transperineal ultrasound, endovaginal ultrasound and vaginal assessment.

A number of tools could be used to define women who are at higher risk of having a levator ani muscle avulsion as described below.

Types of studies

We will consider cross-sectional (prospective or retrospective) and cohort studies that compare one or more index tests with the reference standard or studies that compare two or more index tests for the detection of levator ani muscle avulsion. We will only consider randomised controlled trials (RCTs) if patients are randomised to receive one or other index test and all patients receive the reference standard. We will exclude case control studies and case reports, as they are likely to overestimate sensitivity and specificity and may potentially cause bias (Whiting 2013).

Participants

We will include studies that recruited female patients. These women can either be asymptomatic (e.g. after recent delivery, routine gynaecological care) or have symptoms of pelvic floor dysfunction. We will exclude studies that used participants under the age of 16.

Index tests

Studies to be included for this review should compare one or more of the below index tests with the reference standard (MRI) or to each other.

For an overview of the different classifications for the diagnosis of levator ani muscle avulsion for each index test and reference standard please see Table 1.

Target conditions

The target condition is levator ani muscle avulsion. The different grading systems for levator ani muscle avulsion refer to ordinal outcome measurements (Table 1); no levator ani muscle avulsion, minor levator ani muscle avulsion and major levator ani muscle avulsion. The clinical significance of the minor levator ani muscle avulsion is under discussion as studies have proven that women with major levator ani muscle avulsions are more likely to develop symptoms of pelvic floor dysfunction than women with minor levator ani muscle avulsions (DeLancey 2007b; Dietz 2011; Rostaminia 2013b; Berger 2014; van Delft 2014d). In addition, minor levator ani muscle avulsion has the tendency to improve over time; 50-80% of minor levator ani muscle avulsions at postpartum assessment (six weeks to three months) were no longer evident on ultrasound or MRI at follow-up (six months to one year) (Branham 2007; van Delft 2014d), whereas the improvement of major levator ani muscle avulsion seems to be less common (17-62%) (Shek 2012; Chan 2014; van Delft 2014d). This indicates that the detection of major levator ani muscle avulsion has more clinical value than the detection of minor levator ani muscle avulsion. For the purpose of this review the outcome measurements will be dichotomised into levator ani muscle avulsion present (major) or absent (no or minor). Test positivity will be defined as the presence of major levator ani muscle avulsion.

Reference standards

MRI is considered to be the reference standard. MRI of the pelvic floor usually uses proton density T2-weighted scans, two-dimensional (2D) fast-spin proton density MR with an echo time of 15 ms and a repetition time of 4000 ms, performed at 5-mm intervals in the axial, sagittal and coronal planes in the supine position using a 1.5 Tesla superconducting magnet. The settings most often used have a slice thickness of 4 mm with a gap of 1 mm (Schwertner-Tiepelmann 2012). With reconstruction of the 3D dataset angled axial images in the plane of the minimal hiatal dimensions are acquired. The reference point used is the accurate pubic ligament, above which slices 2-10 are assessed for avulsion of the muscle at the left and right side of the pubic bone. If an avulsion is observed in the axial plane it has to be confirmed in the coronal plane.

For the classification of levator ani muscle avulsion for MRI please see Table 1. A minor avulsion is diagnosed if < 50% (but > 0) of the expected muscle bulk is missing uni- or bilaterally or if ≥ 50% of the expected muscle bulk is missing unilaterally without complete detachment of the muscle. A major avulsion is diagnosed in the case of a complete unilateral avulsion or if at least ≥ 50% of the expected muscle bulk is missing bilaterally (DeLancey 2007b).

Electronic searches

We will perform a computer-assisted search in MEDLINE (Ovid)from 1946 to date and EMBASE (Ovid) from 1974 to date. We will design similarly structured highly sensitive search strategies using search terms appropriate for each database (Appendix 1; Appendix 2). We will identify grey literature through the Science Citation Index from 1900 to date and Conference Proceedings Citation Index from 1970 to date (Appendix 3). In addition, we will search the Cumulative Index of Nursing and Allied Health (CINAHL) via EBSCO from 1981 to date as this has an appropriate subject focus. In addition we will search the Cochrane Register of Diagnostic Test Accuracy Studies and other databases available via the Cochrane Library: Database of Abstracts of Reviews of Effects (DARE), Cochrane Database of Sytematic Reviews (CDSR), Health Technology Assessment Database (HTA) and the Cochrane Central Register of Controlled Trials (CENTRAL). One review author (IvG) will develop the search strategy and an information specialist from the Radboud University will approve the official search.

We will not use search filters to reduce the number needed to screen, as those published have proved not to be sensitive enough (Whiting 2011a; Beynon 2013). We will not apply any language or date restriction to the electronic searches.

Searching other resources

We will screen reference lists of all included studies to identify potential eligible studies missed by electronic searches. Through PubMed, we will use the included studies to search for additional studies using the 'Related Articles' feature. We will use Google Scholar to search for possible eligible studies that have cited one of the included studies. We will aim to access theses and dissertation abstracts from institutions known to be involved in research regarding levator ani muscle avulsion (e.g. Radboud University Nijmegen, University of Oslo, University of Sydney). We will search for ongoing trials on ClinicalTrials.gov (https://clinicaltrials.gov/) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) Search Portal (http://apps.who.int/trialsearch/), which has data from all approved registers worldwide. We will also attempt to contact researchers involved in studies with possibly relevant but unpublished data to reduce publication bias.

Selection of studies

Two independent review authors (IvG, KvD) will screen all titles and abstracts of identified studies from searches of electronic databases to determine whether they might meet the inclusion criteria. For each potentially eligible study, identified by the electronic search or by other methods, we will obtain the full-text version for further assessment. The two review authors will independently evaluate each study for inclusion or exclusion using a study eligibility screening form based on pre-specified inclusion criteria (Appendix 4). We will resolve any disagreement by consulting a third author (RT). If this proves to be inconclusive, our default decision will be to include the study.

We will select all studies that have been published more than once and with overlapping patient data. We will arrange for studies in a language other than English that meet the inclusion criteria to be translated by a designated translator into English. We will present the study selection process in a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

Data extraction and management

Two review authors (IvG, KvD) will independently extract data by using a standardised data extraction form (Appendix 5). We will resolve disagreements by consulting a third author (RT). One author (IvG) will enter data into Cochrane's statistical software, RevMan v5.3, and another author (KvD) will double check the entered data. Data to be extracted include study characteristics, test results, assessment of methodological quality and possible sources for heterogeneity.

For all included studies, we will extract data on the number of true-positives, false-positives, false-negatives and true-negatives with respect to the diagnosis of levator ani muscle avulsion and the cut-off that are used to define test positivity. We will dichotomise the test results if necessary and cross-tabulate in 2 × 2 tables the index test results (positive or negative) against the reference standard results (positive or negative). We will contact the authors to supply missing information if these data are not available in the published trial reports. We will contact authors with more than one included study to avoid possible overlap in patient populations. In cases of overlapping patient data, in the meta-analysis we will use the data from each population only once.

Assessment of methodological quality

We will assess the methodological quality of each study by using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool (Whiting 2011b). This tool is made up of four domains: patient selection, index test, reference standard and patient flow (Appendix 6). We will assess each domain in terms of risk of bias, and the first three domains are also considered in terms of applicability. We will label a domain as 'low risk on bias' when all signalling questions are answered with 'yes', as 'unknown risk on bias' when one or more signalling questions are answered as 'unclear' and none with 'high', and as 'high risk on bias' when one or more signalling questions are answered with 'no'.

Two authors ( IvG, KvD) will independently assess the quality of each study. We will resolve any disagreements by consulting a third author (RT). We will present the outcome of the methodological quality assessment in a table summarising the number of studies with low, high or unclear risk of bias for each of the four domains. We will develop a similar table for concerns regarding applicability. We will explore the influence of risk of bias on accuracy in a sensitivity analysis (Sensitivity analyses).

Statistical analysis and data synthesis

We will describe the characteristics of the included studies and their accuracy results as follows:

• the number of studies retrieved;

• the number of diseased and non-diseased participants included; and

• the main characteristics (e.g. index test, cut-off value for test positivity).

We will present estimates of sensitivity and specificity in forest plots. We will use RevMan v5.3 to document the descriptive analyses and to produce summary forest plots and receiver operating characteristic (ROC) plots.

If studies report results with different cut-off values, we will use the cut-off for test positivity as defined in the section on the target condition. If the defined cut-off was not used, we will use the study results in additional analyses with the other pre-specified cut-off value, if possible.

Investigations of heterogeneity

Main sources of heterogeneity include the different cut-off values for presence of the target condition and difference in a priori probability of levator ani muscle avulsion caused by presence or absence of symptoms of pelvic floor dysfunction in the target population. First, we will investigate heterogeneity through visual examination of forest plots and ROC plots of the estimated sensitivities and specificities. If deemed necessary and if data are sufficient, we will include each of these factors as a covariate in the regression analyses.

If, after adjustment for the covariates mentioned above, there is still substantial heterogeneity, we will also consider the covariates mentioned below to determine possible causes.

We will consider the following covariates for analysis:

• types of machines (e.g. BK Ultrasound, GE Healthcare Voluson E8);

• types of ultrasound probes (e.g. curved, linear, 2D, 3D, 4D, differences in Hz);

• types of MRI acquisition (e.g. T1-3 sequences);

• operator characteristics (e.g. experience in performing and assessing index test or reference standard);

• prevalence of target condition in different settings (e.g. secondary, tertiary hospital);

• geographic location where the study is conducted (e.g. Asia, Europe, North-America).

Sensitivity analyses

We will perform sensitivity analyses to determine the effect of excluding studies that are deemed to be at high risk of bias, according to the QUADAS-2 checklist, by re-analysing the data without these studies. Our primary analysis will include all studies; sensitivity analysis will exclude studies of low quality (high likelihood of bias) to determine whether results have been influenced by inclusion of the lower-quality studies. Additionally, we will perform sensitivity analyses to determine the effect of excluding studies that were flagged as possibly less appropriate for inclusion (when disagreement between authors could not be resolved).

Assessment of reporting bias

We will not undertake assessment of reporting bias. Diagnostic test accuracy reviews show more heterogeneity in included study designs than intervention reviews (mainly RCTs). In general, RCTs are more likely to be published when results are positive or significant and publication bias could cause an important threat to the validity of these intervention reviews. The subject 'diagnostic tests for the detection of levator ani muscle avulsion' allows for publication of favourable and non-favourable results and it is less likely reports will remain unpublished. There is no evidence of reporting bias in test accuracy reviews and methods to detect this are not very reliable for diagnostic test data (Deeks 2005).

Inclusion Ccriteria

Exclusion criteria

Study characteristics