Impact of epithelial ovarian cancer screening on patient‐relevant outcomes in average‐risk postmenopausal women

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

The purpose of this systematic review is to identify and summarise the results of randomised controlled trials that assess the benefits and harms of transvaginal sonography or other screening measures (e.g. serum CA‐125 or HE4) and multimodal testing, for ovarian cancer screening in asymptomatic postmenopausal women with an average EOC risk by evaluating the screening‐related patient‐relevant outcomes (e.g. EOC‐specific mortality rates).

Background

Description of the condition

In postmenopausal women epithelial ovarian cancer (EOC) is one of the most frequent malignancies in developed countries. According to the definition of the European Union it has to be categorised as an 'orphan disease' (EU Parliament 1999), however with an age‐standardised incidence rate of 12.6 (per 100,000 population) and more than 44,000 cases per year, it is the sixth commonest cause of cancer among women in Europe (EUCAN 2013; Ferlay 2013), and the fifth most common in the UK (Cancer Research UK 2014). Furthermore, with a mortality rate of 7.4 (per 100,000 population) it is the most lethal gynaecological malignancy, ahead of breast or uterine cancer (EUCAN 2013; Ferlay 2013; Pavlik 2013). One reason for this high mortality is the lack of early or specific symptoms pointing to EOC and the concomitant late cancer diagnosis. Affected women complain of non‐specific symptoms, such as persistent abdominal distension, pelvic or back pain, loss of appetite or increased urinary frequency (Bankhead 2005; Goff 2012; Hippisley‐Cox 2012; NICE 2011). As these symptoms occur mostly when the ovarian cancer has spread from the ovaries into the abdomen (Aletti 2007), 70% to 80% of women are diagnosed at an advanced disease stage, with tumour spread into the abdominal organs or beyond the abdominal cavity (International Federation of Gynecology and Obstetrics (FIGO) stage III to IV) (Baldwin 2012; Hippisley‐Cox 2012; Jelovac 2011; Maringe 2012). Therefore, the prognosis for advanced‐stage disease is poor and the case‐fatality rate is high (five‐year survival rate: 20% to 35% or lower) (Baldwin 2012; Carlson 1994; Jacobs 2004; Jemal 2009; Slomski 2012; Whitehouse 2003), compared to early‐stage diagnoses where the cancer is confined to the ovaries (FIGO stage I) or to the pelvis (FIGO stage II) (five‐year survival rate: up to 95%) (Baldwin 2012; Bell 2006; Jemal 2009; Maringe 2012; van Nagell 2011; Whitehouse 2003). On average the five‐year survival rate in developed countries is greater than 40% (Baldwin 2012; Berrino 2007; Cancer Research UK 2014; Jemal 2009; Whitehouse 2003). This variation in the prognosis demonstrates the importance of early‐stage detection in order to improve the outcome and to reduce premature EOC‐specific mortality.

In order to find suitable and reliable screening methods, those women at increased risk of EOC due to genetic factors need to be distinguished from women who are at average risk. While the general population have a life‐time risk of developing an ovarian malignancy of 1.2% to 1.4% (US data) (Jelovac 2011; van Nagell 2012), the life‐time risk of women with known genetic mutations (e.g. BRCA 1/2, HNPCC) or a strong family history of EOC is 12% to 60% (Antoniou 2003; Bell 1998; Ford 1998; Jelovac 2011; King 2003). As the effect of any screening measure could vary between these average‐risk and high‐risk women, it may be necessary to design different screening strategies for each risk group and to evaluate these separately. In addition, the pathogenesis and prevalence of the most frequent EOC types could influence the choice of screening measures or screening intervals, particularly as screening aims to identify the tumours at an early stage, with confinement to the ovaries. For some time there has been broad consensus that this malignancy has its origin in the ovaries. However, new theories suggest that there are two subgroups of EOC, which differ in the site and natural history of their origin: low‐grade type 1 tumours and high‐grade type 2 tumours. The slow‐growing type 1 tumours evolve from intra‐ovarian lesions (i.e. borderline tumours) and are confined to the ovaries at detection, with a good prognosis. In contrast, recent data suggest that type 2 tumours may arise from intra‐epithelial lesions outside the ovary (e.g. in the fallopian tubes). This type of tumour is characterised by rapid growth with immediate access to the intra‐abdominal cavity, therefore the prognosis of affected women differs vastly from prognosis of women with type 1 tumours (Hong 2013; Kurman 2008b; Shih 2004). It is possible that the high percentage of type 2 tumours (probably 75% of all EOC) has a negative impact on the effectiveness of available screening measures in decreasing EOC‐specific mortality. Future screening tools might include the assessment of the tumour volume by transvaginal sonography and novel target‐orientated genetic tests or biomarkers in order to improve early detection rates of both tumour types (Kurman 2008a; Kurman 2010; Shih 2004).

Description of the intervention

Various methods may currently be used to screen asymptomatic women, including physical examination, transvaginal sonography and serum tumour markers (e.g. cancer antigen 125 (CA‐125) and human epididymis protein 4 (HE 4)) (Aletti 2007; Gentry‐Maharaj 2012; Kyrgiou 2006).

Transvaginal sonography

Transvaginal sonography was first described as a screening method in 1989 by Higgins and colleagues (Higgins 1989). Prior to this, ultrasound scanning was mostly performed transabdominally (Higgins 1989). Transvaginal sonography uses a 5 MHz to 7.5 MHz ultrasound probe to examine the ovarian morphology and to measure the size of the ovaries in three dimensions, in order to compute the volume of the ovaries (van Nagell 2012). Due to the reduction in ovarian size in women after the age of 30 and the decrease in ovarian activity in women after their last menstruation, there are different upper limits for normal ovarian size in pre‐ (20 cm³) and postmenopausal women (10 cm³) (Pavlik 2000). Transvaginal sonography enables the detection of adnexal masses, as well as a first assessment of the individual risk of malignancy, by examining these suspicious masses in terms of the ovarian shape, internal structure and the morphological characteristics (e.g. visible solid or cystic components, existing ascites, non‐uniform ovarian echogenicity) (Manegold 2013; Menon 2009). If the ovary volume exceeds these thresholds, or if there is at least one morphological reference relating to a malignant lesion, further diagnostic evaluations (e.g. magnetic resonance tomography, laparoscopy) are indicated.

The ability of transvaginal sonography to distinguish accurately between EOC and non‐malignant ovarian abnormalities was demonstrated in a large prospective observational study in asymptomatic women aged 50 years and over, and in women above 25 years with a family history of EOC (van Nagell 2011). In this study, regardless of the cancer stage, a sensitivity of 86.4% and specificity of 98.8% was achieved by transvaginal sonography by using different threshold values of the ovarian volume depending on the menopausal status. These data may support the suitability of transvaginal sonography as a screening tool.

Cancer antigen CA‐125

The cancer antigen CA‐125, a high molecular glycoprotein, is the most used and described blood serum marker, which can rise to high concentrations in women with EOC. Bast 1983 showed that serum levels of CA‐125 higher than 35 U/ml can be found in more than 80% of women with a confirmed diagnosis of EOC. As serum CA‐125 levels above the threshold of 35 U/ml can also be the consequence of other conditions, such as inflammation of the peritoneum, endometriosis, benign cysts or non‐gynaecological malignancies, there is a significant risk of false‐positive screening results (Gentry‐Maharaj 2012; Medeiros 2009; Whitehouse 2003). In one quantitative systematic review (Medeiros 2009), with 17 included studies, the diagnostic accuracy of CA‐125 (threshold: 35 U/ml) was high in surgically confirmed EOC patients. The pooled sensitivity of 80% and the pooled specificity of 75% showed (despite high heterogeneity) an acceptable capability to discriminate between malignant or borderline lesions versus benign lesions of the ovaries.

The current clinical practice guideline from the National Institute for Health and Care Excellence (NICE) recommends that in cases with clinical symptoms, such as persistent abdominal bloating, pelvic/abdominal pain or other complaints, serum levels of CA‐125 are measured. If this marker exceeds the cut‐off value of 35 U/ml, a transvaginal sonographic examination is indicated (NICE 2011). While the NICE clinical practice guideline does not explicitly cover population‐based screening measures, the statement from the US Preventive Service Task Force in 2012 (Moyer 2012), as well as the clinical practice guideline from the German Cancer Society (DKG) in 2013 (DKG 2013), do not recommend EOC screening in asymptomatic women.

How the intervention might work

Screening aims to identify a disorder at the earliest possible opportunity and disease stage, where early treatment can improve outcome. EOC screening should diagnose ovarian cancer at an earlier stage, before clinical signs or noticeable symptoms are noticed (Figure 1), ideally when the tumour is confined to the ovaries and potentially curable (Cho 2009; Kurman 2010). However, screening may only identify disease at an earlier time point, without improving the stage or natural history of the disease, by finding disease in an asymptomatic or subclinical stage, thereby reducing the lag time from disease development to diagnosis, without making any impact on true survival.

Figure 1.

Schematic timing of screening vs. no screening (own compilation)

Transvaginal sonography and serum tumour markers, as well as other screening measures, have to fulfil important requirements. Apart from a high level of acceptability and diagnostic accuracy, every screening procedure should be easy, rapid and inexpensive to perform, and have minimal associated adverse effects and inter‐observer variation (Hulka 1988; van Nagell 2012). In addition, a screening procedure has to be able to discriminate clearly and reliably between (seemingly healthy) women who would benefit from early therapeutical intervention, and those who would not (Gates 2003). These benefits could be manifold, besides a possible increase in survival, due to the use of effective therapy an early stage, and could increase quality of life due to the use of less aggressive therapy. Healthy women could also benefit from an accurate (negative) test as their fears could be allayed (Adriaensen 2013).

All screening methods have an associated risk of false‐positive diagnosis of the condition. For EOC the rate of this 'screening over‐diagnosis' seems to be higher with transvaginal sonography (1.2% to 2.5%) than with serum CA‐125 testing (Nelson 2004; Schnell 2011). In turn, this could result in the use of unnecessary diagnostic and therapeutic procedures (e.g. surgery or chemotherapy (Morrison 2012)) and corresponding mental or physical harm, or both. Beyond the sensitivity and specificity of the screening measures, due to the relatively low prevalence of EOC there is basically a higher risk of low positive predictive values (Gates 2003) and of a higher rate of false‐positive EOC diagnoses in actually healthy women. Alternatively, there is also a risk of false‐negative screening results, which could falsely reassure women and delay the necessary early‐stage therapy by disregarding specific symptoms (Gates 2003). Both aspects threaten the utilisation of screening measures and should be evaluated in order to balance the risks, harms and benefits of each measure (Nelson 2004; Wegwarth 2013; Woolf 2012).

Why it is important to do this review

Due to the importance of early‐stage detection of ovarian malignancies there is an urgent need for a systematic and high‐level evaluation of the evidence for EOC screening measures. Owing to the poor prognosis of late‐diagnosed EOC, it is necessary to appraise these screening measures, beyond their technical properties and their diagnostic accuracy, with respect to patient‐relevant outcomes, such as cancer‐specific or all‐cause mortality. There are a small number of randomised controlled trials (RCTs), however, there is currently no systematic review of the evidence that evaluates the suitability and patient‐relevant features of screening methods such as transvaginal sonography.

In 2011, the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial, with more than 78,000 participants, presented their first results as one of the few RCTs in this area (Andriole 2012). In addition, the results of the ongoing United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), with a study population of more than 200,000 women, are expected in 2015 (Menon 2009). The findings of these two large studies and possibly of additional RCTs, which could result from the systematic literature search, will allow for a high‐level, evidence‐based assessment of the benefits and of possible harms of transvaginal sonography or other EOC screening measures.

Objectives

The purpose of this systematic review is to identify and summarise the results of randomised controlled trials that assess the benefits and harms of transvaginal sonography or other screening measures (e.g. serum CA‐125 or HE4) and multimodal testing, for ovarian cancer screening in asymptomatic postmenopausal women with an average EOC risk by evaluating the screening‐related patient‐relevant outcomes (e.g. EOC‐specific mortality rates).

Methods

Criteria for considering studies for this review

Types of studies

We will include randomised controlled trials (RCTs), as well as cluster‐randomised trials.

Types of participants

All postmenopausal women aged over 45 years with an average population risk of ovarian cancer, with no symptoms pointing to an ovarian malignancy and with no previously confirmed ovarian or breast cancer diagnosis. Menopause is defined as absence of menses for 12 months or more.

Types of interventions

We will include all studies assessing any EOC screening measure (e.g. transvaginal sonography, serum CA‐125 testing) or multimodal screening algorithm (e.g. transvaginal sonography + CA‐125) in comparison to:

1. any other EOC screening measure or multimodal screening algorithm; or

2. no EOC screening.

Types of outcome measures

Primary outcomes

• EOC‐specific mortality rates

• Serious adverse events (including serious adverse events as a result of false‐positive and false‐negative screening results)

Secondary outcomes

• All‐cause mortality

• Time to EOC‐specific mortality

• EOC incidence by stage, tumour size or tumour type at diagnosis

• False‐positive screening results (we will include screening‐identified borderline tumours as false‐positive results due to their low malignant potential)

• False‐negative screening results

• Number of laparoscopies/surgeries in relation to the number of identified EOC (ratio of operations per true‐positive)

• Other harms resulting from the screening procedure or from an intervention following a positive screening result, including psychological stress or anxiety

• Health‐related quality of life (HRQoL)

• Costs associated with screening programmes

Search methods for identification of studies

We will search without any time or language restrictions.

Electronic searches

In order to identify eligible studies we will search the following electronic databases:

• Cochrane Central Register of Controlled Trials (CENTRAL) (current issue)

• Cochrane Gynaecological Cancer Group Trial Register ('SR‐GYNAECA')

• MEDLINE (accessed through Ovid) (1946 to date)

• EMBASE (accessed through Ovid) (1980 to date)

The search strategy for MEDLINE can be found in Appendix 1. For databases other than MEDLINE we will adapt the search strategy accordingly.

Searching other resources

We will identify all relevant articles in PubMed using the 'related articles' feature and carry out a further search for newly published articles.

Unpublished and grey literature

Furthermore, we will look for ongoing and unpublished trials using the following databases:

• the database of clinical trials of the US National Institutes of Health: www.clinicaltrials.gov;

• metaRegister of Current Controlled Trials: www.controlled‐trials.com;

• National Cancer Institute: http://www.cancer.gov/clinicaltrials.

If ongoing or unpublished trials are identified through these searches we will contact the principal investigators and ask for further information and relevant data.

Handsearching

Additionally, we will check the reference list of all included studies, as well as previously published systematic reviews that are known to us.

We will conduct a handsearch of available abstracts (from 2008 to 2013) from reports of conferences in the following sources:

• Gynecological Oncology (Annual Meeting of the Society of Gynecological Oncologists);

• International Journal of Gynecological Cancer (Annual Meeting of the International Gynecological Cancer Society);

• Annual Meeting of the European Society of Medical Oncology (ESMO);

• Annual Meeting of the American Society of Clinical Oncology (ASCO).

If there is an abstract but no complete study available we will contact the authors for the full text or further information.

Data collection and analysis

Selection of studies

We will transfer all identified references into the reference management database 'EndNote'. After deleting all duplicates, two authors (Christoph G Mosch (CM), Thomas Jaschinski (TJ)) will independently review all titles and abstracts. We will exclude references that clearly contradict the previously defined inclusion criteria. We will obtain the remaining abstracts as full text and both authors (CM, TJ) will also independently reviewthese for inclusion. We will document reasons for exclusion.

We will discuss and resolve any possible disagreements between the two authors. In case of further existing discrepancies we will include the third author (Michaela AM Eikermann (ME)) in order to reach consensus.

Data extraction and management

Two authors (CM, TJ) will extract independently the data from every included study using a previously created extraction form (according to Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011)).

For all included studies we will extract the following data:

• Author, year of publication and journal citation (including language)

• Country

• Setting

• Inclusion and exclusion criteria

• Study design, methodology

• Study population (number of enrolled patients in each group, baseline patient characteristics (e.g. age, comorbidities))

• Intervention details (e.g. definition of abnormal screening results, cut‐off values, screening intervals)

• Comparison definition/details (e.g. definition of abnormal screening results, cut‐off values, screening intervals)

• Risk of bias in the study (Assessment of risk of bias in included studies)

• Duration of follow‐up (if necessary for each endpoint)

• Outcomes (for each outcome, we will extract the outcome definition and the unit of measurement (if relevant); for adjusted estimates, we will record variables adjusted for in analyses)

• Results (we will extract the number of participants allocated to each intervention group, the total number analysed for each outcome and the missing participants)

We will extract results as follows:

• For time‐to‐event data (e.g. time to EOC‐specific mortality), we will extract the log of the hazard ratio (log (HR)) and its standard error from trial reports. If these are not reported, we will attempt to estimate the log (HR) and its standard error using the methods of Parmar (Parmar 1998).

• For dichotomous outcomes (e.g. EOC‐specific or all‐cause mortality, EOC incidence, harms of screening measures/number of adverse events, false‐positive screening results), we will extract the number of patients in each treatment arm who experienced the outcome of interest and the number of patients assessed until the end of follow‐up, in order to estimate a risk ratio.

• For continuous outcomes (e.g. HRQoL, costs of screening), we will extract the final value and standard deviation of the outcome of interest and the number of patients assessed in each treatment arm until the end of follow‐up, in order to estimate the mean difference between treatment arms and its standard error.

If reported, we will extract both unadjusted and adjusted statistics. If possible, all extracted data should be related to the intention‐to‐treat (ITT) population, in which participants were analysed according to the group to which they were initially allocated. We will also note the time points at which outcomes were collected and reported. We will present all measures of screening effect with 95% confidence intervals (CI).

We will discuss and resolve possible disagreements between the two authors (CM, TJ). In case of further existing discrepancies we will include the third author (ME) in order to reach consensus.

Assessment of risk of bias in included studies

We will assess the risk of bias in the included studies using The Cochrane Collaboration tool (Higgins 2011). This will include assessment of (Appendix 2):

1. selection bias: random sequence generation and allocation concealment;

2. detection bias: blinding of outcome assessment;

3. attrition bias: incomplete outcome data;

4. reporting bias: selective reporting of outcome;

5. other possible sources of bias.

Blinding of participants and (medical) personnel (performance bias) is not possible with screening measures. Thus we will not assess this item.

Two review authors (CM, TJ) will apply the 'Risk of bias' tool independently and resolve differences by discussion or by appeal to the third review author (ME). We will summarise the results in both a 'Risk of bias' graph and a 'Risk of bias' summary. Further, we will try to estimate the magnitude and the likely direction of the bias. We will interpret the results of meta‐analyses in light of the findings with respect to risk of bias. Additionally, we will assess the impact of bias by undertaking a sensitivity analysis (studies with low risk of bias versus studies with high risk of bias).

Measures of treatment effect

We will use the following measures of the effect of treatment:

• hazard ratio (HR) for time‐to‐event data (e.g. time to EOC‐specific mortality);

• risk ratio (RR) for dichotomous outcomes (e.g. EOC‐specific or all‐cause mortality, EOC incidence, harms of screening measures/number of adverse events, false‐positive screening results);

• mean difference between screening arms for continuous outcomes (e.g. HRQoL, costs of screening).

Unit of analysis issues

Two review authors (CM, TJ) will review unit of analysis issues according to Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) and resolve differences by discussion. These include reports where:

• groups of individuals were randomised together to the same intervention (i.e. cluster‐randomised trials).

Dealing with missing data

We will not impute missing outcome data for the primary or secondary outcomes. If data are missing or only imputed data are reported we will contact the trial authors to request data on the outcomes only among those participants who were assessed. In the case that missing data cannot be provided we will perform the analysis based on the available data. If necessary, and if respective data are available, we will conduct a sensitivity analysis with 'best case' and 'worst case' scenarios (Higgins 2011).

Assessment of heterogeneity

We will assess heterogeneity between studies by visual inspection of forest plots, by estimation of the percentage of heterogeneity between trials which cannot be ascribed to sampling variation (Higgins 2011), by a formal statistical test of the significance of the heterogeneity (Deeks 2001) and, if possible, by subgroup analyses. If there is evidence of substantial heterogeneity (i.e. P value < 0.1 or I² > 50%) (Higgins 2011), we will investigate and report the possible reasons for this heterogeneity.

Assessment of reporting biases

We will examine funnel plots corresponding to meta‐analysis of the primary outcomes to assess the potential for small study effects such as publication bias if a sufficient number of studies (i.e. more than 10 studies) are identified. If there are any other reasons for asymmetries the authors (CM, TJ, ME) will discussthese.

Data synthesis

If sufficient clinically similar studies are available, we will pool their results in meta‐analyses using the Cochrane Review Manager software (Review Manager 5.2).

• For time‐to‐event data, we will pool HRs.

• For any dichotomous outcomes, we will calculate the RR (including 95% confidence interval (CI)) for each study and then pool these.

• For continuous outcomes, we will pool the mean difference (including 95% CI) between the treatment arms at the end of follow‐up if all trials measure the outcome on the same scale; otherwise we will pool using the standardised mean difference (including 95% CI).

If any trials have multiple treatment groups, we will divide the 'shared' comparison group into the number of treatment groups and comparisons between each treatment group and treat the split comparison group as independent comparisons.

We will use the random‐effects‐model with inverse variance weighting or the Peto odds ratio (in case of (dichotomous) rare events) for all meta‐analyses (DerSimonian 1986).

Subgroup analysis and investigation of heterogeneity

We will perform a subgroup analysis by grouping the trials according to:

1. screening measure/algorithm used in the studies (including the screening interval);

2. EOC type (type 1/type 2);

3. cancer treatment conducted after secured EOC diagnosis.

Sensitivity analysis

We will perform a sensitivity analysis in order to assess the robustness of our results regarding the risk of bias of the included studies (low risk versus high risk).

Appendices

Appendix 1. MEDLINE search strategy

MEDLINE Ovid

1 exp Ovarian Neoplasms/ 2 exp Fallopian Tube Neoplasms/ 3 ((ovar* or fallopian tub* or adnex*) adj4 (cancer* or tumo* or malignan* or carcinoma* or adenocarcinoma* or neoplas* or mass*)).mp. 4 or/1‐3 5 exp Mass screening/ 6 exp “Early Detection of Cancer”/ 7 (screen* adj5 (ovar* or fallopian tub* or adnex*)).mp 8 or/5‐7 9 4 and 8 10 randomized controlled trial.pt. 11 controlled clinical trial.pt. 12 randomized.ab. 13 placebo.ab. 14 clinical trials as topic.sh. 15 randomly.ab. 16 trial.ti. 17 or/10‐16 18 9 and 17

Appendix 2. Explanation of the items in the 'Risk of bias' tool

1) Random sequence generation

• Low risk of bias, e.g. participants assigned to the screening groups on the basis of a computer‐generated random sequence or a table of random numbers

• High risk of bias, e.g. participants assigned to the screening groups on the basis of date of birth, clinic ID number or surname, or no attempt to randomise participants

• Unclear risk of bias, e.g. not reported, information not available

2) Allocation concealment

• Low risk of bias, e.g. where the allocation sequence could not be foretold

• High risk of bias, e.g. allocation sequence could be foretold by patients, investigators or screening providers

• Unclear risk of bias, e.g. not reported

3.1) Blinding of participants and personnel

• Blinding of participants and (medical) personnel is not possible for a screening trial (thus, we will not assess this item)

3.2) Blinding of outcomes assessors

• Low risk of bias, if outcome assessors were adequately blinded

• High risk of bias, if outcome assessors were not blinded to the intervention that the participant received

• Unclear risk of bias, if this was not reported or unclear

4) Incomplete outcome data

We will record the proportion of participants whose outcomes were not reported at the end of the study. We will code a satisfactory level of loss to follow‐up for each outcome as:

• Low risk of bias, if fewer than 20% of patients were lost to follow‐up and reasons for loss to follow‐up were similar in both screening arms

• High risk of bias, if more than 20% of patients were lost to follow‐up or reasons for loss to follow‐up differed between screening arms

• Unclear risk of bias, if loss to follow‐up was not reported

5) Selective reporting of outcomes

• Low risk of bias, e.g. review reports all outcomes specified in the protocol

• High risk of bias, e.g. it is suspected that outcomes have been selectively reported

• Unclear risk of bias, e.g. it is unclear whether outcomes had been selectively reported

6) Other bias

• Low risk of bias, if you do not suspect any other source of bias and the trial appears to be methodologically sound (e.g. no inconsistent data)

• High risk of bias, if you suspect that the trial was prone to an additional bias (e.g. high amount of inconsistent data)

• Unclear risk of bias, if you are uncertain whether an additional bias may have been present

What's new

Date	Event	Description
10 September 2018	Amended	The Editors are looking for new contributors to complete and maintain this Cochrane Review. Contact ruh‐tr.gnoc‐cochrane@nhs.net for further information.

Contributions of authors

All the authors have contributed to the conception, development and drafting of this protocol.

Sources of support

Internal sources

• None, Other.

External sources

• None, Other.

Declarations of interest

Christoph G Mosch ‐ none known.

Thomas Jaschinski ‐ none known.

Dr Michaela Eikermann ‐ none known.

Edited (no change to conclusions)