Family presence during resuscitation

Objectives

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

The primary aim of this review is to investigate whether offering relatives the option to be present during resuscitation of patients might lower the occurrence of PTSD‐related symptoms.

The secondary aim is to investigate what effect family presence compared to no family presence during resuscitation of patients has on other psychological outcomes of the relatives and on patient morbidity and mortality.

We also wish to investigate the effect of FPDR on medical treatment and care during resuscitation. Furthermore, we will investigate and report the personal stress seen in healthcare professionals and if possible describe their attitudes towards the FPDR initiative.

Background

The healthcare system has shifted from a paternalistic attitude towards a principle of personal autonomy, and family members often expect to be actively involved in care decisions involving the treatment of their relatives (Lederman 2014). Even during resuscitation, patients prefer to have their relatives nearby and many relatives want to be present if given the choice (Benjamin 2004; Bradley 2017; Chew 2014; De Stefano 2016; Dwyer 2015; Meyers 2000; Robinson 1998). This could be why family presence during resuscitation (FPDR) is a growing healthcare practice, even though the majority of the evidence is of low quality and lacking investigation of the relatives’ perceptions and psychological outcomes, and the impact on patient morbidity and mortality (Oczkowski 2015).

Description of the condition

Resuscitation can be described in general as a process of correcting the physiological disorders in a patient suffering from critical illness or reviving someone from unconsciousness or apparent death. The term 'resuscitation' in this review is therefore used as a description of cardiopulmonary resuscitation (CPR), trauma and critical care of acutely ill patients. In the USA, more than 200 thousand adults undergo CPR annually in hospital and in Germany more than 2000 in the emergency room (ER) (Mallikethi‐Reddy 2017; Merchant 2011; Zwingmann 2016). Furthermore, in the USA 2.2 million emergency department visits result in admission to critical care out of a total of 145.6 million visits (1.5%) (National Center for Health Statistics 2016).

Description of the intervention

When family members (e.g. siblings, parents, spouses, children or close friends) of the patient are present in the resuscitation room with the patient it is described as FPDR. The presence can be guided or unguided by a designated support person.

Whether family members should be offered the opportunity to be present during resuscitation or not is a question yet to be answered. This is because their presence during the resuscitation attempt may impact on their state of mind, the patient receiving care, and the performance of the healthcare professionals. There are concerns that FPDR can lead to post traumatic stress disorder (PTSD) in the relatives (Compton 2011); have a negative impact on the quality of critical care (Fernandez 2009); and that patient confidentiality can be violated as the patients' thoughts and preferences in the situation are unknown (Benjamin 2004; Lederman 2014). When FPDR is practised, the relatives are in direct contact with both the patient and the healthcare professionals during the resuscitation attempt and not placed outside (e.g. in a special waiting room). This means that they can interact with the patient (e.g. hold the patient's hand if they want to and talk to or with them) whenever needed.

How the intervention might work

Since the first study in 1987 (Doyle 1987), only three randomized controlled trials (RCT) and a quasi‐randomized trial have focused on relatives' perspectives concerning FPDR (Dudley 2009; Holzhauser 2006; Jabre 2013; Robinson 1998). Two studies found it to be psychologically beneficial for the relatives to be present in the ER during resuscitation in general, and not only during CPR (Holzhauser 2006; Robinson 1998). A cluster‐RCT and its one‐year follow‐up assessment, and a qualitative analysis of this RCT, showed positive effects of FPDR on relatives’ psychological outcomes including improved clinical indicators related to PTSD, better anxiety and depression scale scores, and less complicated grieving during prehospital CPR, and no negative effects on mortality (De Stefano 2016; Jabre 2013; Jabre 2014).

Much evidence concerning FPDR is qualitative and when investigating this literature, different themes emerge concerning relatives' perceptions. Some clinicians believe that being present during resuscitation may have more advantages than disadvantages as it allows the relatives to feel both needed and of use in the care and comfort of the patient (Holzhauser 2008). Especially in the event of death, it may help the relatives to acknowledge that all possible measures were taken to save the patient (De Stefano 2016; Holzhauser 2006). Watching the procedures and the communication—both verbal and nonverbal—among the healthcare professionals can play a role in supporting the understanding of the circumstances of the patient's death and help the relatives to cope with the patient's death in the context of the critical care performed (De Stefano 2016). A quasi‐experimental historical study and a large prehospital cluster‐RCT indicated that FPDR may have positive effects on relatives' psychological outcomes (Jabre 2013; Soleimanpour 2017). The communication between relatives and the emergency care team is considered very important and a qualitative analysis of the prehospital cluster‐RCT by Jabre and colleagues concluded that the emergency care team could facilitate the relatives' acceptance of the reality of death (De Stefano 2016). Therefore, a designated support person focusing on the relatives, as a part of the resuscitation team, is recommended (Bossaert 2015; Downar 2013; Dudley 2009; Meyers 2000).

It is important to investigate how family presence might influence patient mortality. A patient who had survived cardiac arrest stated that he felt “aware of his wife’s presence”, and that this encouraged him to fight on (Belanger 1997). Three other patients who survived resuscitation reported that they felt supported by their relatives' presence and did not feel their confidentiality or dignity had been compromised (Robinson 1998). A qualitative study of nine surviving patients found several themes concerning the positive effects that family presence during invasive procedures, or CPR, had on patients: “it comforted them, provided help, and served, the patients believed, to remind providers of a patient’s “personhood” — he wasn’t just a patient; he was a person and had a family” (Eichhorn 2001).

A relevant concern is whether FPDR negatively affects the members of the resuscitation team, and if the relatives get in the way of the work of the healthcare professionals. Jabre and colleagues found no increase in the level of stress of the healthcare providers or difference in patient mortality and stated that "survival, the duration of advanced resuscitation, the type or dose of infused medications, and the number of shocks delivered were not affected by the presence or absence of the family member" (Jabre 2013). A study of physicians’ performance during a simulated cardiac arrest with relatives in different states of mind indicated that relatives may have a negative impact on the quality of acute medical care (Fernandez 2009). In contrast, a paediatric study of FPDR found no differences in the success rate of the critical interventions, and the healthcare providers surveyed believed there was a minimal effect of FPDR on resuscitation (Dudley 2009). Mangurten and colleagues believe a protocol for FPDR of paediatric patients is effective, and the providers in their study reported that the presence of parents did not negatively affect medical care (Mangurten 2006).

Why it is important to do this review

The European Resuscitation Council (ERC) recommends that relatives are offered the choice of being present during CPR (Bossaert 2015). The American Heart Association (AHA) does not give a recommendation; however they do encourage FPDR (Mancini 2015). This lack of a clear recommendation leads to disagreements among clinicians according FPDR (Bossaert 2016; Downar 2013; Lederman 2016; Meyers 2000; Porter 2013; Sak‐Dankosky 2014; Youngson 2016). These disagreements are perhaps led by fear, as seen in the reasons mentioned by the relatives for their choice of FPDR: "The anticipated fear of negative or positive reactions was often mentioned as a reason for their choice about being present for the resuscitation" (De Stefano 2016). However, the feeling of exclusion when relatives are not offered FPDR is a relevant factor to be aware of (De Stefano 2016).

As far as we are aware, no studies in the prehospital environment have been conducted covering FPDR in general, other than during CPR (Jabre 2013; Oczkowski 2015). Robinson and colleagues looked at FPDR including CPR, trauma and critical care in the ER but they had to terminate their trial prematurely, as the clinical team considered it unethical not to offer FPDR, even though the preliminary results were not convincing (Robinson 1998). Holzhauser and colleagues also looked at FPDR during both CPR and critical care in the ER, but excluded trauma patients (Holzhauser 2006). Dudley and colleagues investigated FPDR of paediatric patients (Dudley 2009).

We hypothesize that FPDR may be associated with overall positive results on relatives’ psychological outcomes, including decreased PTSD symptoms. Furthermore, we do not believe that it affects the healthcare professionals' performance negatively or that it increases morbidity and mortality of the patients. Considering the public health perspective, it may be possible that even some amount of overtreatment can be decreased. As a relative stated: "when I saw that they were being so aggressive [with the treatment], I felt it as an aggression, an assault on my husband" (De Stefano 2016).

These factors suggest the necessity of a high‐quality systematic review to evaluate the evidence.

Objectives

The primary aim of this review is to investigate whether offering relatives the option to be present during resuscitation of patients might lower the occurrence of PTSD‐related symptoms.

The secondary aim is to investigate what effect family presence compared to no family presence during resuscitation of patients has on other psychological outcomes of the relatives and on patient morbidity and mortality.

We also wish to investigate the effect of FPDR on medical treatment and care during resuscitation. Furthermore, we will investigate and report the personal stress seen in healthcare professionals and if possible describe their attitudes towards the FPDR initiative.

Methods

Criteria for considering studies for this review

Types of studies

We will include randomized controlled trials (RCTs), including cluster‐RCTs and quasi‐RCTs.

Types of participants

The participants of this review will include relatives, patients and healthcare professionals.

We will include participants aged 18 years or older who have witnessed a resuscitation attempt of a patient (who is their relative) at the ER or prehospital.

We define the type of participants as being siblings, parents, spouses, children or close friends of the patient, or any other descriptions used by the authors.

We do not have any limitation on age or gender and define the type of patient as being a patient with cardiac arrest in need of CPR, a patient with a critical medical or traumatic life‐threatening condition, an unconscious patient or a patient in any other way at risk of apparent death.

We will include all types of healthcare professionals as described in the included studies.

Types of interventions

The intervention will be to offer the relatives the opportunity of FPDR at the ER or in the prehospital setting.

The control will be not to offer the relatives the opportunity of FPDR systematically in a standardized way but instead to follow standard practice; e.g. waiting in a special room with no opportunity of witnessing the resuscitation.

Regarding the patients and the healthcare professionals the intervention will be if there has been FPDR or not.

Types of outcome measures

The outcomes listed are not criteria for including studies. If we find a study that meets the inclusion criteria without the listed outcomes we will describe it.

Primary outcomes

• PTSD evaluation of the relatives measured at 0 to 3 months and more than 3 months. As there is a lack of consensus on time frames for PTSD evaluation, we may adapt our approach in the event that included studies employ alternative cut‐offs. Any changes will be outlined in the 'Differences between protocol and review' section of the review.

Secondary outcomes

• Depression, anxiety, or complicated grief in the relatives according to ICD‐11 measured at 0 to 3 months and more than 3 months. As there is a lack of consensus on time frames for measuring these outcomes, we may adapt our approach in the event that included studies employ alternative cut‐offs. Any changes will be outlined in the 'Differences between protocol and review section' of the review.

• Patient morbidity: readmissions within 30 days, number of ICU admissions within 30 days.

• Length of hospital stay.

• Patient mortality as measured in the included studies. If a study has several time points included, the one closest to 30 days will be used. 

• Adverse events in the patient population: number of adverse cardiac and pulmonary events or other adverse events described in the included studies and total number of adverse events.

Tertiary outcomes

• Duration of patient resuscitation as reported in included studies.

• Personal stress in healthcare professionals during resuscitation as measured by included studies.

• Healthcare professionals’ attitudes towards FPDR as measured in the included studies.

Search methods for identification of studies

Electronic searches

We will search for studies as described in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 4 (Lefebvre 2019).

There will be no language, publication year or publication status restrictions.

We will search the following databases for relevant trials.

• Cochrane Central Register of Controlled Trials (CENTRAL) (latest Issue)

• MEDLINE (Ovid, from 1946)

• Embase (Ovid, from 1974)

• Cinahl (EBSCO, from 1980)

• PsycINFO (EBSCO, from 1806)

We have developed a draft search strategy for MEDLINE. The search strategy can be found in Appendix 1, and we will modify it appropriately for the other databases.

Searching other resources

We will scan the reference lists and citations of included trials, and any relevant systematic reviews identified for further references to additional trials. We will scan the abstracts of conference proceedings of the American Heart Association, and the European Resuscitation Council. When necessary, we will contact trial authors for additional information.

We will also search www.ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch), and ISRCTN (www.isrctn.com), for unpublished and ongoing studies; Open Grey (opengrey.eu) for grey literature; and Google Scholar for additional trials.

Data collection and analysis

Selection of studies

Two authors (MAR and TS) will independently and in duplicate screen search results for eligible studies by scanning titles and abstracts. We will use Covidence for transparency. We will obtain full‐text records of all presumed eligible studies, and evaluate them before final inclusion. We will resolve disagreements by discussion, and consult with a third author (AMM) if needed. We will record the selection process in sufficient detail to complete a PRISMA flow diagram (Moher 2009).

Data extraction and management

One of the review authors (PJ) is the first author of an included primary study in this review (Jabre 2013). She will not extract data from any included studies.

Two review authors (MAR and TS) will independently and in duplicate extract relevant data using pre‐designed data‐extraction forms (Appendix 2). We will undertake a pilot run of two studies, to evaluate the construction of the data extraction form. Afterwards we will refine and finalize the form. After extraction we will compare data and resolve any disagreements by discussion, aided by a third author (AMM) if necessary. We will contact study authors if we need to clarify any significant uncertainty that may occur during data extraction. We will then export the extracted data to the newest version of Review Manager 5 (RevMan 5) for further management and analysis (Review Manager 2014).

Assessment of risk of bias in included studies

One of the review authors (PJ) is the first author of an included primary study in this review (Jabre 2013). She will not be assessing the risk of bias from any included study.

For each study, two authors (MAR and TS) will independently use Cochrane’s assessment of risk of bias tool to evaluate risk of bias (Higgins 2017). We will use the tool to rate risk of bias as low, high or uncertain in the following domains: adequate sequence generation; allocation concealment; blinding of outcome assessment; incomplete outcome data; selective reporting; and other bias. We will resolve disagreements by discussion, aided by a third author (AMM) if needed. We will present the final results of the evaluation in a risk of bias summary table, produced using Review Manager 2014.

As part of the 'Risk of bias' assessment, we will seek out protocols within the online study databases to assess selective reporting.

Measures of treatment effect

Dichotomous outcomes

Concerning the presence of PTSD and other psychological outcomes, mortality, morbidity, and adverse events, we will convert the effect of treatment to risk ratios (RR) if presented, and use meta‐analysis to summarize treatment effects as such. If events are rare, we will use Peto odds ratios for summarizing treatment effects.

Continuous outcomes

We will present and analyse continuous outcomes—the severity of PTSD and other psychological outcomes, length of hospital stay and duration of resuscitation—as mean difference (MD) with 95% confidence intervals (CIs). If continuous outcomes are measured on different scales but we consider them to be measuring the same outcome, we will use standardized mean difference (SMD) for analysis and presentation. When using SMD we will present results as standardised mean with 95% CIs and will not back‐transform them. Furthermore, we will use Cohen’s rule‐of‐thumb to describe results in 'Summary of findings' (SOF) tables, defined as 0.2 to 0.5: “small effect”; 0.5 to 0.8: “medium effect”; more than 0.8: “large effect”.  

Unit of analysis issues

Analysis issues may be present as we wish to include cluster‐RCTs. If an included cluster‐RCT takes into account the clustering either by adjustment of sample size or standard error or by using a statistical method taking clustering into account (e.g. generalized estimating equation (GEE), mixed‐effect model), we will use the results as they are. If an included cluster‐RCT does not take into account the clustering of patients we will reduce the effective sample size by the design effect calculated as 1 + (M − 1)*ICC where M is the average cluster size and ICC is the intracluster correlation, as suggested in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 23 (Higgins 2019). We will obtain the ICC from similar studies; and if this information is not available we will set it to 0.04.

Dealing with missing data

To ensure completeness of data we will contact trial authors in order to obtain and incorporate any relevant missing data. We prefer to perform intention‐to‐treat analysis (ITT) where possible. If trials have more that 20% dropout we will explore the impact of missing data on the overall effect.

Where individual studies do not account appropriately for missing data, or do not report how these were handled, we will consider whether it was likely to be missing at random, or not affecting the assessment of risk of bias. Where outcome data are missing, and cannot be recovered, we plan to adopt the approach suggested in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 10 (Higgins 2019), and use available‐case analysis. We will include data only for those participants whose results are known, and address the potential impact of the missing data by using the ‘Risk of bias’ tool. Ultimately we will consider the potential impact of including such studies in the overall assessment of intervention effect.

Assessment of heterogeneity

We will consider clinical, methodological and statistical diversity (Higgins 2019).

We will address clinical diversity through detailed reporting of the diagnostic and clinical definitions and characteristics of the included studies. We will perform a meta‐analysis if we consider the studies to be sufficiently homogeneous concerning participants, interventions and outcomes. We will assess methodological diversity using the Cochrane ‘Risk of bias’ tool. We will assess the statistical diversity by considering the consistency of study results, and how this impacts on the meta‐analysis. We will assess statistical heterogeneity by statistical test (Chi² test) and by calculating the I² statistic (Higgins 2002). In addition, we will calculate and present Tau². To further investigate between‐study variance, we will present results with prediction intervals.

Assessment of reporting biases

If we can include at least 10 studies, we will produce a funnel plot using Review Manager 2014, and will analyse the funnel plot for signs of publication bias.

Data synthesis

If we consider it appropriate after assessment of heterogeneity, we will perform a meta‐analysis using outcomes expressed as RRs and 95% CIs. For dichotomous outcomes we will use the Mantel‐Haenszel test; for continuous outcomes we will use the inverse variance; and for dichotomous outcomes with rare events we will use Peto's odds ratio.

We will also calculate and present prediction intervals (PI) of our summary measures. A 95% PI estimates where the intervention effect will be in 95% of future comparable studies. In the case of heterogeneity, the 95% PI will be wider than the 95% CI of the mean effect. In the case where the 95% CI for the mean treatment effect shows statistical significance but the PI is on both sides of zero, it can be concluded that the treatment effect can be in a different direction in future similar studies (IntHout 2016).

Due to the outcomes chosen in this review, we believe the studies to be diverse and inherently heterogeneous, and therefore we will perform a random‐effects meta‐analysis where possible, independent of measured heterogeneity. If we deem meta‐analysis to be feasible and appropriate, we will produce a forest plot showing a weighted estimate of the summarized effect.

We will perform all analyses using RevMan 5 (Review Manager 2014). If we deem the individual study designs to be too diverse, and therefore that statistical combination is inappropriate, we will present the findings in a narrative fashion.

A selection of extracted data describing included studies will be presented in a study characteristics table, as recommended by the Cochrane Handbook for Systematic Reviews of Interventions, with four distinct categories: methods; participants; interventions; and outcomes (Higgins 2019).

Subgroup analysis and investigation of heterogeneity

We will perform an adult/paediatric subgroup analysis to examine if the effects are modified by the age of patients if there are more than three RCTs concerning relatives to patients less than 18 years of age. Furthermore we will, if possible, conduct subgroup analyses concerning cardiac arrest/other critical situations and trauma/medical, and if possible a subgroup analysis to explore if location of resuscitation (i.e. in or out of hospital) has significance.

We will conduct subgroup analyses based on the criteria for the aforementioned variables used by the primary study authors.

Sensitivity analysis

In case of unavailable data, we will assess whether it is missing at random, or if the missing data might be a source of bias. We will omit randomly missing data from analysis; while non‐randomly missing data will be imputed as sensitivity analyses, as deemed appropriate by the authors, according to the cause of their omission. In addition, we will challenge analyses that include imputed data by sensitivity‐analyses to ensure robustness of its findings, when assumptions are changed.

For dichotomous outcomes we will impute missing values as experiencing or not experiencing the event. For continuous outcomes we will impute missing values using different fixed values for all missing values.

We will perform sensitivity analyses by excluding trials we judge to be at high risk of bias.

Summary of findings and assessment of the certainty of the evidence

The outcomes will be assessed in GRADE using GRADEpro online software to assess the certainty of evidence according to the guideline (GRADEpro GDT).

We will use the principles of the GRADE system to assess the certainty of the body of evidence associated with specific outcomes in our review, and construct a 'Summary of findings' (SOF) table using GRADE software (GRADEpro GDT; Guyatt 2008). We will include the outcomes for the comparison in three SOF tables: one including the PTSD evaluation of relatives and their depression, anxiety or complicated grief evaluation; one including patient mortality, length of hospital stay and adverse events; and one including the personal stress of healthcare professionals as measured by the included studies and duration of patient resuscitation. If we find it appropriate, we will construct two SOF tables concerning PTSD and depression, anxiety or complicated grief, covering relatives to adults/paediatrics subgroups.

We will downgrade the certainty of evidence according to the GRADE approach (Chapter 12, Cochrane Handbook for Systematic Reviews of Interventions) (Higgins 2019). Two authors (MAR and TS) will independently undertake the GRADE assessments. We will report agreement reached by consensus and all reasons for downgrading.

History

Protocol first published: Issue 5, 2020

Appendices

Appendix 1. Search strategy

MEDLINE

1) ((famil* or next of kin* or kinship or relativ* or significant other* or spouse* or husband* or wife* or partner* or parent* or sibling* or friend* or companion*) adj5 (presen* or attend* or observ* or witness* or perception* or participat*)).mp.

2) exp emergency medical services/ or emergency medicine/ or exp resuscitation/ or exp Emergency Treatment/

3) ((Emergenc* or acute) adj3 (servic* or medicine or treat*)).mp.

4) (resuscitat* or prehospital* or pre‐hospital* or trauma* or CPR).mp.

5) 2 or 3 or 4

6) 1 and 5

7) FPDR.mp.

8) 6 or 7

9) ((randomized controlled trial or controlled clinical trial).pt. or random*.ab. or placebo.ab. or drug therapy.fs. or trial.ab. or groups.ab. or clinical trials as topic.sh. or random allocation.sh.) not (exp animals/ not humans.sh.)

10) 8 and 9

Appendix 2. Data extraction form

We will extract the following information.

• Bibliographic data, including date of completion/publication.

• Publication status.

• Study settings:

• • study design

• • Methods of randomization

  • Methods of blinding

• • • Study period

• • Continuous or interrupted

• • • Study length

    • Setting

• • Single‐ or multi‐centre study

  • Number of centres if multi‐centre

  • Emergency department or other

• Length of follow‐up

• Country of origin

• Source of funding

• Participants:

• • number

• • Number randomized to each group

  • Number of exclusions and reasons for exclusions

  • missing

• • • age

    • gender distribution

    • method for selection of subjects?

• • Inclusion criteria

  • Exclusion criteria

• Interventions

Participants being offered to observe

• • Resuscitation in the Emergency Department

• • Both cardiac arrest and

  • Treatment of critical conditions including trauma.

• • • Resuscitation prehospital

• • Both cardiac arrest and

  • Treatment of critical conditions including trauma

• Comparators:

• • participants not offered observation but following standard practice of placement in for instance a special room.

• Outcomes:

Primary:

PTSD evaluation of participants as reported by the authors.

Secondary:

• • Depression, anxiety, or complicated grief in participants diagnosed by any physician as measured and reported by the authors.

  • Patient morbidity: adverse cardiac events, adverse pulmonary events, total adverse events, readmissions, number of ICU admissions.

  • Length of hospital stay.

  • Patient mortality as reported by the authors: in‐hospital, out‐of‐hospital.

Tertiary:

• • Duration of patient resuscitation as measured by the authors.

  • Personal stress in healthcare professionals as measured by the authors.

• Duration of follow‐up.

• Number of withdrawals (by group) and number of withdrawals (by group) due to adverse events.

Contributions of authors

Monika Afzali Rubin (MAR), Tintin Svensson (TS), Suzanne Forsyth Herling (SFH), Tobias Wirenfeldt Klausen (TWK), Patricia Jabre (PJ), Ann Merete Møller (AMM).

AMM is the guarantor. AMM and MAR contributed to the conception of the review. MAR developed the search strategy together with Janne Vendt, the information specialist of Cochrane Anaesthesia and Cochrane Critical and Emergency Care Group. TWK is the guarantor of the statistics. AMM, PJ and SFH provide expertise on family presence during resuscitation (FPDR). All authors contributed to the selection criteria, quality of evidence assessment strategy and data extraction forms. MAR drafted, and all other authors have contributed to, and have approved, the final protocol manuscript.

Co‐ordinating the review: MAR

Undertaking manual searches: MAR

Screening search results: MAR, TS

Organizing retrieval of papers: MAR

Screening retrieved papers against inclusion criteria: MAR, TS

Appraising quality of papers: MAR, TS, AMM

Abstracting data from papers: MAR, TS

Writing to authors of papers for additional information: MAR

Providing additional data about papers: MAR

Obtaining and screening data on unpublished studies: MAR, TS

Data management for the review: MAR

Entering data into Review Manager 2014: MAR

RevMan 5 statistical data: TWK

Other statistical analysis not using RevMan 5: TWK

Interpretation of data: MAR, TS, PJ, TWK, SFH, AMM

Statistical inferences: TWK

Writing the review: MAR

Performing previous work that was the foundation of the present study: PJ

Guarantor for the review (one author): AMM

Person responsible for reading and checking review before submission: MAR

Sources of support

Internal sources

• No sources of funding, Other

External sources

• No sources of funding, Other

Declarations of interest

Monika Afzali Rubin: nothing to declare.

Tintin Svensson: nothing to declare.

Suzanne Forsyth Herling: nothing to declare.

Tobias Wirenfeldt Klausen: nothing to declare.

Patricia Jabre is the first author of an included primary study (which she has conducted) in the Cochrane Review (Jabre 2013). She will not extract data from her own study. Instead, MAR, TS and AMM will extract these data, and check the interpretation against the study report and any available study registration details or protocol.

Ann Merete Møller: nothing to declare.

New