Interventions for the management of malignant pleural effusions: a network meta‐analysis

Amelia O Clive; Hayley E Jones; Rahul Bhatnagar; Nancy J Preston; Nick Maskell

doi:10.1002/14651858.CD010529.pub2

. 2016 May 8;2016(5):CD010529. doi: 10.1002/14651858.CD010529.pub2

Interventions for the management of malignant pleural effusions: a network meta‐analysis

Amelia O Clive ^1,^✉, Hayley E Jones ², Rahul Bhatnagar ¹, Nancy J Preston ³, Nick Maskell ¹

Editor: Cochrane Pain, Palliative and Supportive Care Group

PMCID: PMC6450218 PMID: 27155783

Abstract

Background

Malignant pleural effusion (MPE) is a common problem for people with cancer as a result of malignant infiltration of the pleura. It is usually associated with considerable breathlessness. A number of treatment options are available to manage the uncontrolled accumulation of pleural fluid including administration of a pleurodesis agent (either via a chest tube or at thoracoscopy) or indwelling pleural catheter insertion.

Objectives

To ascertain the optimal management strategy for adults with malignant pleural effusion in terms of pleurodesis success. Additionally, to quantify differences in patient‐reported outcomes and adverse effects between management strategies.

Search methods

We searched The Cochrane Central Register of Controlled Trials (CENTRAL), Ovid MEDLINE, Ovid EMBASE; EBSCO CINAHL; SCI‐EXPANDED and SSCI (ISI Web of Science) to April 2015.

Selection criteria

We included randomised controlled trials of intrapleural interventions for adults with symptomatic MPE in the review.

Data collection and analysis

Two review authors independently extracted data on study design, study characteristics, outcome measures, potential effect modifiers and risk of bias.

The primary outcome measure was pleurodesis failure rate. Secondary outcome measures were adverse effects and complications, patient‐reported control of breathlessness, quality of life, cost, mortality, duration of inpatient stay and patient acceptability.

We performed network meta‐analysis with random effects to analyse the primary outcome data and those secondary outcomes with enough data. We also performed pair‐wise random‐effects meta‐analyses of direct comparison data. If interventions were not deemed jointly randomisable, or insufficient data were available, we reported the results by narrative synthesis. We performed sensitivity analyses to explore heterogeneity and to evaluate only those pleurodesis agents administered via a chest tube at the bedside.

Main results

Of the 1888 records identified, 62 randomised trials, including a total of 3428 patients, were eligible for inclusion. All studies were at high or uncertain risk of bias for at least one domain.

Network meta‐analysis evaluating the rate of pleurodesis failure, suggested talc poudrage to be a highly effective method (ranked second of 16 (95% credible interval (Cr‐I) 1 to 5)) and provided evidence that it resulted in fewer pleurodesis failures than eight other methods. The estimated ranks of other commonly used agents were: talc slurry (fourth; 95% Cr‐I 2 to 8), mepacrine (fourth; 95% Cr‐I 1 to 10), iodine (fifth; 95% Cr‐I 1 to 12), bleomycin (eighth; 95% Cr‐I 5 to 11) and doxycyline (tenth; 95% Cr‐I 4 to 15). The estimates were imprecise as evidenced by the wide credible intervals and both high statistical and clinical heterogeneity.

Most of the secondary outcomes, including adverse events, were inconsistently reported by the included studies and the methods used to describe them varied widely. Hence the majority of the secondary outcomes were reported descriptively in this review. We obtained sufficient data to perform network meta‐analysis for the most commonly reported adverse events: pain, fever and mortality. The fever network was imprecise and showed substantial heterogeneity, but suggested placebo caused the least fever (ranked first of 11 (95% Cr‐I 1 to 7)) and mepacrine and Corynebacterium parvum (C. parvum) appeared to be associated with the most fever (ranked tenth (95% Cr‐I 6 to 11) and eleventh (95% Cr‐I 7 to 11) respectively). No differences between interventions were revealed by the network meta‐analysis of the pain data. The only potential difference in mortality identified in the mortality network was that those receiving tetracycline appeared to have a longer survival than those receiving mitoxantrone (OR 0.16 (95% Confidence Interval (CI) 0.03 to 0.72)). Indwelling pleural catheters were examined in two randomised studies, both of which reported improved breathlessness when compared to talc slurry pleurodesis, despite lower pleurodesis success rates.

The risk of bias in a number of the included studies was substantial, for example the vast majority of studies were unblinded, and the methods used for sequence generation and allocation concealment were often unclear. Overall, however, the risk of bias for all studies was moderate. We have not reported the GRADE quality of evidence for the outcomes, as the role of GRADE is not well established in the context of Network Meta‐analysis (NMA).

Authors' conclusions

Based on the available evidence, talc poudrage is a more effective pleurodesis method in MPE than a number of other frequently used methods, including tetracycline and bleomycin. However further data are required to definitively confirm whether it is more effective than certain other commonly used interventions such as talc slurry and doxycycline, particularly in view of the high statistical and clinical heterogeneity within the network and the high risk of bias of many of the included studies. Based on the strength of the evidence from both direct and indirect comparisons of randomised data of sclerosants administered at the bedside, there is no evidence to suggest large differences between the other highly effective methods (talc slurry, mepacrine, iodine and C. parvum). However, local availability, global experience of these agents and their adverse events, which may not be identified in randomised trials, must also be considered when selecting a sclerosant. Further research is required to delineate the roles of different treatments according to patient characteristics (e.g. according to their prognosis or presence of trapped lung) and to explore patient‐centred outcomes, such as breathlessness and quality of life, in more detail. Careful consideration to minimise the risk of bias and standardise outcome measures is essential for future trial design.

Plain language summary

Interventions for the management of fluid around the lungs (pleural fluid) caused by cancer

Review Question

We reviewed the evidence about the effectiveness of different methods to manage fluid around the lung in patients with a build up of this fluid caused by cancer.

Background

Malignant pleural effusion (MPE) is a condition whereby cancer of the lining of the lung results in fluid building up in the space between the lung and rib cage (pleural cavity), often resulting in breathlessness. Treatment options include removal of the fluid using either a temporary chest drain, a camera examination of the pleural cavity (thoracoscopy) or a semi‐permanent chest drain tunnelled under the skin (an indwelling pleural catheter). Introducing a chemical into the pleural cavity can also be used to prevent the fluid coming back (pleurodesis). We wanted to find out which method was the most effective in terms of preventing fluid build up and which was best in terms of side effects and patient‐reported outcomes such as pain, fever, breathlessness and quality of life.

Study Characteristics

We searched databases for trials comparing different interventions in adults with symptomatic MPE to April 2015, written in any language. Since we were only interested in rigorously conducted research, we restricted our search to randomised controlled trials (in which participants are randomly allocated to the methods being tested). We analysed the majority of the data using a technique called 'network meta‐analysis' which allows lots of different interventions to be compared in one analysis. This analysis ranks the interventions in order of their effectiveness.

Key Results

We found 62 studies involving 3428 patients.

In the network meta‐analysis, the use of thoracoscopy to remove the fluid and blow talc into the pleural cavity (talc poudrage) appeared to be more effective in preventing fluid build up than a number of other commonly used methods. However, we could not say definitely that it is better than some other methods such as giving talc or doxycycline through a chest drain.

Side effects, quality of life and patient satisfaction were reported inconsistently by the included studies, but are important factors to consider when selecting the best management strategy for a patient. There was enough data to perform network meta‐analysis for pain, fever and mortality. We found placebo caused the least fever and Corynebacterium parvum (C. parvum) and mepacrine were likely to cause the most. We found no differences in the pain caused by the interventions evaluated. Only one comparison showed a possible difference, revealing that those receiving tetracycline may live longer than those receiving mitoxantrone. As we only evaluated randomised controlled trials, it is possible some harms of treatments were not identified by this review.

Quality of the Evidence

Many of the studies were of low quality and the characteristics of the individual studies were quite different to each other. This high risk of bias makes it difficult to reach definite conclusions.

Conclusions

The available evidence shows that talc poudrage can stop fluid building up. However, we can not be sure that this is definitely the best method, and further research is needed. It is also important to consider global experience of these agents and knowledge of their safety and side effects when selecting the most appropriate pleurodesis method. Indwelling pleural catheters may help improve patient breathlessness, but may be less good at stopping the fluid coming back.

Further research is also required to look at particular patient groups and explore patient‐centred outcomes, such as breathlessness and quality of life in more detail. Ideally a fuller understanding of the potential harms of the treatments from the patients' perspective would also be beneficial.

Background

Malignant pleural effusion (MPE) is a common clinical problem, with an estimated annual incidence of at least 150,000 in the USA alone (American Thoracic Society 2000). Fifteen percent of people diagnosed with cancer will develop pleural effusion during the course of their disease as a result of malignant infiltration of the pleura. It often confers a poor prognosis (Rodrîguez‐Panadero 1989). Breathlessness results from compression of the underlying lung and impaired diaphragmatic and chest wall movement and is often relieved by pleural fluid aspiration.

Description of the condition

MPE is a condition whereby excess fluid accumulates in the pleural cavity, caused by direct pleural tumour invasion, resulting in increased permeability of the pleural microvessels and involvement of local lymph nodes causing reduced fluid reabsorption (Rodrîguez‐Panadero 2008). The most common primary sites which metastasise to the pleura are lung cancer in men and breast cancer in women, but other primary sites include lymphoma, genitourinary and gastrointestinal malignancy (DiBonito 1992; Sears 1987). In addition, the pleura may be the primary site of the malignancy, as is the case in mesothelioma. In the majority of cases, the diagnosis of pleural malignancy is made by cytological analysis of the pleural fluid or pleural biopsy. Depending on the clinical situation, confirmation of malignancy elsewhere and an otherwise unexplained (usually exudative) effusion may also be attributed to malignancy. Survival of these patients varies widely (Bielsa 2008; Burrows 2000) and estimation of an individual’s prognosis may help with the selection of the most appropriate management strategy (Clive 2014).

Trapped lung can occur when full lung expansion is limited by either a visceral pleural peel or endobronchial obstruction and in this situation, even once the fluid is drained, visceral and parietal pleural apposition does not occur. This results in pleurodesis attempts being less effective and often limits the treatment options to either an indwelling pleural catheter or surgery.

Description of the intervention

A number of different approaches may be used to manage MPE and the chosen method is likely to depend on clinical factors, patient preferences and local availability of the various techniques. Instillation of a sclerosant into the pleural cavity through an intercostal chest drain after complete fluid drainage has been the mainstay of treatment for many years (known as ‘bedside’ or ‘slurry’ pleurodesis). This technique aims to fuse the pleural layers together by means of local inflammation induced by the pleurodesis agent, thereby preventing pleural fluid re‐accumulation. The optimal management strategy to maximise pleurodesis success in terms of the size of chest drain, patient positioning, use of analgesia and type of sclerosant is still the subject of debate (Roberts 2010). The role of intrapleural fibrinolytics to break down septations and loculations within the effusion prior to administration of the pleurodesis agent is also yet to be formally established (Davies 1999; Gilkeson 1999; Hsu 2006).

Thoracoscopy is an alternative method, which is used to drain the effusion and deliver a sclerosant into the pleural cavity. This can either be performed under conscious sedation (local anaesthetic thoracoscopy), or as a surgical procedure under general anaesthetic (Video Assisted Thoracoscopic Surgery (VATS)). In both techniques, the pleural fluid is drained and the pleural cavity is visualised using a fibre‐optic camera. Loculations can be broken down and biopsies may be taken to gain a histological diagnosis. A pleurodesis agent can then be delivered by way of insufflation (poudrage) prior to the insertion of a chest drain (Rahman 2010).

An alternative approach in the management of MPE is the use of indwelling pleural catheters (IPCs). These are chest tubes, which are tunnelled under the skin and allow long‐term, intermittent fluid drainage to be performed in the community, thereby minimising recurrent hospital attendances. They have an established role in the management of pleural effusions in patients with trapped lung, but are increasingly being used for the primary management of malignant effusions as an alternative to chemical pleurodesis (Davies 2012; Demmy 2012). In a proportion of patients with IPCs, spontaneous pleurodesis occurs, allowing the drain to be removed without recurrence of the effusion (Tremblay 2006).

In certain clinical scenarios, none of the above options may be suitable and simple pleural fluid aspiration or medical management of a patient's breathlessness (for example using opiates) may be deemed more appropriate. This may be the case for patients in the terminal phase of their illness where invasive techniques may be felt to confer unnecessary discomfort.

How the intervention might work

Pleurodesis aims to fibrose the pleural layers together in order to obliterate the pleural space and by so doing prevent fluid recurrence. For pleurodesis to be successful the visceral and parietal pleural surfaces must be opposed and hence if lung expansion is incomplete (for example if the effusion is very loculated or the patient has trapped lung), pleurodesis is more likely to fail. The sclerosant stimulates an inflammatory reaction within the pleural cavity, which results in fusion of the visceral and parietal pleura.

Indwelling pleural catheters allow intermittent pleural fluid drainage, which relieves the pressure on the diaphragm and chest wall and promotes lung re‐expansion. By so doing, breathlessness is improved and spontaneous pleurodesis occurs in up to 50% of patients (Putnam 2000).

Why it is important to do this review

Due to wider availability of pleural interventions, such as thoracoscopy and indwelling pleural catheters, the management options available to patients with MPE are expanding. This review will help to delineate the specific roles of the different techniques and identify factors which may improve pleurodesis rates for those undergoing a bedside pleurodesis. This review includes an update of a Cochrane systematic review first published in 2004, 'Pleurodesis for malignant pleural effusions' (Shaw 2004) and will subsequently help to inform national guidelines in this area.

Given the availability of many pair‐wise comparisons for the method of pleurodesis administration and type of pleurodesis agent, this is a multiple interventions review. Network meta‐analysis has been performed to synthesise all the available evidence and investigate a treatment hierarchy.

Objectives

Methods

Criteria for considering studies for this review

Types of studies

We only included reports of randomised controlled trials (RCTs) in this review. This would have included randomised cross‐over trials and cluster randomised trials, although no studies of these types were identified. We included both single and multi‐centre studies. Studies, which were stated to be randomised but were found to be at high risk of bias for adequate sequence generation or allocation concealment, were excluded.

Types of participants

Inclusion

Adults over the age of 16.
Symptomatic pleural effusion resulting from an underlying malignant process (of any type and stage).

Exclusion

Studies recruiting both malignant and non‐malignant participants with no clear distinction between the two groups in the results section.
Studies evaluating the effect of a drug administered via any method other than the intra‐pleural route.
Studies including participants with effusions within a variety of body cavities (e.g. pleural, peritoneal, pericardial), where the effect of the treatments in the subgroup of patients with pleural effusions cannot be distinguished in the results section.

Types of interventions

We identified studies comparing the following.

Type of sclerosant.
Mode of administration of sclerosant (thoracoscopic pleurodesis and bedside pleurodesis).
Bedside or thoracoscopic pleurodesis and indwelling pleural catheter insertion.
Techniques used to optimise pleurodesis success rate, namely:
- chest drain size;
- type of analgesia given;
- duration of drainage after instillation of sclerosant;
- patient positioning after pleurodesis (for example, patient rotation);
- use of intrapleural fibrinolytics.

We generated a network of interventions, including comparisons between the types of sclerosant, mode of administration and IPC use. We assumed that any participant meeting the inclusion criteria could be, in principle, randomised to any of the eligible interventions. This is referred to as the interventions being ‘jointly randomisable’. However, if an intervention was not felt to be jointly randomisable, for example the treatment was specific to a certain tumour type, we reported the results separately from the network (Salanti 2012).

Interventions of direct interest

We included RCTs that evaluated one or more of the following intrapleural interventions: talc poudrage, talc slurry, bleomycin, tetracycline, doxycycline, iodine, C.parvum, IPC, mitoxantrone, mustine, mepacrine, interferon, triethylenethiophosphoramide and adriamycin, compared with another intervention or placebo. If we identified other sclerosants that we were not aware of, we considered them as eligible and we included them in the network after assessing their comparability with the pre‐specified set of competing interventions. We reported the findings for these interventions in the results and the conclusions of the review.

Types of outcome measures

Primary outcomes

The efficacy of pleurodesis was our primary outcome measure.

Definitions of pleurodesis failure varied between studies and although current practice would define this by a lack of recurrence of symptoms or need for a repeat pleural intervention to manage the effusion, in some older studies, less clinically relevant definitions were used (for example, re‐accumulation of effusion on imaging). We still included these studies in the review, and documented the method used to define pleurodesis for all studies in the assessment of the risk of bias.

For the purposes of the primary outcome, we used the following hierarchy of preferences to judge pleurodesis failure (if a study reported more than one definition of pleurodesis failure, the highest of these according to this hierarchy was used):

need for a repeat pleural procedure to manage recurrence of the effusion, or ongoing drainage of pleural fluid from an indwelling pleural catheter (if applicable);
evidence of significant pleural fluid re‐accumulation on radiological imaging (for example, chest X‐ray or ultrasound);
pleurodesis failure in the opinion of the trial investigators.

Similarly, we selected the time point used to define pleurodesis efficacy was selected using the following hierarchy of preferences:

2 ‐ 4 months;
> 4 ‐ 7 months;
> 7 ‐ 11 months;
> 11 ‐ 12 months;
< 2 months;
> 12 months.

Participants who died before the time point at which pleurodesis efficacy was assessed, were classified according to their last known pleurodesis outcome prior to their death (i.e. their last observation carried forward). If these data were not provided, we used the available reported data.

Secondary outcomes

Adverse effects and complications due to interventions, specifically the presence or absence of pain and fever after the intervention.
Patient‐reported control of breathlessness, as measured by a valid and reliable scale (for example, visual analogue scale (VAS), numeric rating scale or dyspnoea/breathlessness specific multidimensional scale)^*
The participants' quality of life and symptom control (including pain), as measured by a valid and reliable scale^*
Relative costs of the comparative techniques as reported by the individual trials. For ease of comparison, data reported in other currencies were converted to USD.^*
The overall mortality (we used the data for the reported outcomes closest to three months).
Median survival.
Duration of inpatient stay in days (both total length of stay and from time of intervention until discharge).^*
Patient acceptability of the interventions as judged by a valid scale (for example, visual analogue scale or numeric rating scale).^*

* if available

Search methods for identification of studies

Trials that compared at least two of the interventions (including placebo) were eligible. We included all possible comparisons formed by the interventions of interest.

Electronic searches

To identify studies for inclusion in this review, we searched the following databases:

The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library) Issue 3 of 12, 2015;
MEDLINE (Ovid) 1948 to 1/04/15;
EMBASE (Ovid) 1974 to 1/04/15;
CINAHL (EBSCO) 1980 to April 2015;
Web of Science Science Citation Index Expanded (SCI‐EXPANDED) and Social Sciences Citation Index (SSCI) searched to 2015.

The search strategies can be viewed in Appendix 1. There were no language restrictions. We included single and multi‐centre studies.

Searching other resources

We screened the reference lists from the included studies for additional publications. We also searched the reference lists from relevant chapters in key resources, such as the British Thoracic Society Pleural Disease Guidelines (Roberts 2010).

Data collection and analysis

Selection of studies

One author screened all titles and abstracts retrieved by the search for relevance (AOC). We identified potentially eligible studies and obtained the full papers. Two review authors (AOC and NAM) independently assessed each study for inclusion in the review and any disagreement was resolved through discussion or by a third author (NP).

Data extraction and management

Two of the review authors (AOC with NAM, NP or RB) extracted data from each included study. We resolved disagreements through discussion and referral to one of the other review authors. If an author was involved in one of the included studies, they did not perform the data extraction for that study. Data collected included the following.

Publication details including:
- title, author(s), date, country and other citation details;
- study aim and design;
- primary and secondary outcomes;
- number of participants randomised.
Details of the interventions and comparison group including type of intervention, duration, dose, mode of administration and number of doses.

Primary and secondary outcome measures (as detailed above) and data on adverse events and complications.

Assessment of the study’s risk of bias.

Data on potential effect modifiers including the following study and patient characteristics:

- how pleurodesis was defined (radiology only or including clinical need as well as radiology);
- whether patients with trapped lung were included or not;
- the size of the chest tube through which bedside pleurodesis was administered (defined as small (< 20 French), large (≥ 20 French) or unknown);
- the time point at which pleurodesis was defined;
- the tumour types included in the study.

We had planned to look at specific areas of study quality, which were incorporated into the assessment of the risk of bias. We requested additional data from the study authors as required. One author (AOC) entered data suitable for pooling into the Cochrane Collaboration's statistical software, Review Manager (RevMan) (RevMan 2014) . Where we performed network meta‐analysis, we transferred data to the WinBUGS software (Lunn 2000).

Assessment of risk of bias in included studies

We limited inclusion to studies that were randomised as a minimum. Two of the review authors (AOC with NP, RB or NM) independently assessed risk of bias for each study, using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), and adapted from those used by the Cochrane Pregnancy and Childbirth Group, with any disagreements resolved by discussion. In our original protocol, we had planned to include sample size in our risk of bias assessment. However, in view of Cochrane guidance stating imprecision should not be considered a risk of bias, we did not perform this assessment (Higgins 2011a). We assessed the following for each study.

Random sequence generation (checking for possible selection bias)

We assessed the method used to generate the allocation sequence as: low risk of bias (any truly random process, e.g. random number table; computer random‐number generator); unclear risk of bias (method used to generate sequence not clearly stated). We excluded studies using a non‐random process i.e. at high risk of bias (e.g. odd or even date of birth; hospital or clinic record number).

Allocation concealment (checking for possible selection bias)

The method used to conceal allocation to interventions prior to assignment determines whether intervention allocation could have been foreseen in advance of, or during, recruitment, or changed after assignment. We assessed the methods as: low risk of bias (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes); unclear risk of bias (method not clearly stated). We excluded studies at high risk of bias that did not conceal allocation (e.g. open list).

Blinding of participants and personnel (checking for possible performance bias)

We assessed the methods used to blind study participants and personnel from knowledge of which intervention a participant received. We assessed the methods as: low risk of bias (study stated there was blinding of participants and key study personnel and unlikely blinding could be broken, or no blinding or incomplete blinding but the outcome not likely to be influenced by lack of blinding); unclear risk of bias (insufficient information to permit judgement of low or high risk of bias); high risk of bias (no blinding or incomplete blinding, which is likely to influence the trial outcome or blinding attempted but likely it could have been broken and the outcome is likely to be influenced by lack of blinding).

Blinding of outcome assessment (checking for possible detection bias)

We assessed the methods used to blind outcome assessors from knowledge of which intervention a participant received. We assessed the methods as: low risk of bias (study stated that it was not blinded but the review authors judged that the outcome measurement is not likely to be influenced by lack of blinding or blinding of outcome assessment was ensured); unclear risk of bias (study provided an inadequate description to permit judgment of 'low risk' or 'high risk'); high risk of bias (no blinding of outcome assessment and outcome likely to be influenced by lack of blinding, or there was blinding of the outcome assessment but likely that the blinding could have been broken).

Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data)

We assessed the methods used to deal with loss to follow up for each of the given studies. Due to the challenges of inevitable missing outcome data given the predictable attrition of patients due to death in the palliative care population, we took into account whether missing data had been justified, whether the rate was similar in the different treatment arms, whether the treatment being evaluated was felt to have an impact on the degree of missing outcome data and whether an intention to treat analysis had been attempted. We assessed the methods used to deal with incomplete data as: low risk (rate of missing data were balanced between the treatment arms, seemed reasonable and had been justified; data had been analysed according to the patients' randomised treatment allocation; a suitable imputation method may have been used to account for missing data); unclear risk of bias (insufficient information given to allocate trial to 'high' or 'low' risk group); high risk of bias (imbalanced missing outcome data between the treatment arms or missing outcome data felt to be related to the true outcome; reasons for loss to follow up poorly justified; no attempt at ITT analysis; inappropriate imputation used).

Selective Outcome Reporting

We assessed the studies for selective outcome reporting using the following criteria: low risk of bias (all outcomes pre‐defined and reported, for example in a published protocol, or all clinically relevant and reasonably expected outcomes were reported); uncertain risk of bias (unclear whether all pre‐defined and clinically relevant outcomes were reported); high risk of bias (one or more clinically relevant and reasonably expected outcome was not reported and data on these outcomes were likely to have been recorded).

Other sources of bias

This section was used to report other biases, which were detected but did not fit into the above categories (for example, industry bias, academic bias or other methodological flaws that may have caused bias). We assessed the methods used to deal with other sources of bias as: low risk (the trial appeared to be free from other potential biases); unclear risk of bias; high risk of bias (other source of bias was identified).

Measures of treatment effect

Relative treatment effects

For proportions (dichotomous outcomes), such as pleurodesis efficacy and mortality, we calculated the Odds Ratio (OR) with 95% confidence intervals (CIs). For continuous data (such as length of hospital stay and cost) we planned to use the mean difference (MD) with 95% CIs and also the number needed to treat (NNT) to benefit for efficacy outcome, and the number needed to harm (NNH) for adverse events.

We planned to treat ordinal outcome measures (for example, breathlessness scales and quality of life data) as continuous so long as the scale was long enough. If different scales were used by the included studies, we planned to use the standardised mean difference in meta‐analyses.

We presented results from both pair‐wise standard meta‐analysis and network meta‐analysis (NMA) as summary relative effect sizes (OR, MD or SMD with 95% CIs) for each possible pair of treatments (Deeks 2011).

Relative treatment ranking

Based on the results of the network meta‐analysis, we estimated the rank of each competing intervention's effectiveness. We presented estimated ranks (medians) with 95% credible intervals (Cr‐Is) (representing uncertainty about the true rank) produced from the Bayesian analyses (Higgins 2011b).

Unit of analysis issues

If repeated observations on the same participants occurred during the trial (for example, pleurodesis success rate at different time points), we analysed these separately. Only one measure per participant was used for the primary endpoint (according to the hierarchy of preferences detailed above Primary outcomes).

For the purpose of meta‐analysis, if a study had multiple doses for a certain substance, we combined and compared all relevant experimental intervention groups with the combination of all relevant control groups. We reported any evidence for effects of the different doses descriptively.

For cross‐over trials, we planned to analyse data using pair‐wise meta‐analysis, taking into account the cross‐over design. If meta‐analysis had been performed containing cluster randomised trials and the presented results had not accounted for clustering, then we planned to make an appropriate adjustment, as described in the Cochrane Handbook (Higgins 2011b).

We treated multi‐arm studies as multiple independent two‐arm studies in the standard pair‐wise meta‐analysis. In the network meta‐analysis, we accounted for the correlation between the effect sizes from multi‐arm studies.

Dealing with missing data

We attempted to contact the study authors of included studies to clarify any missing data. We would have imputed the missing standard deviations based on the average standard deviations from the other included studies if standard deviations for mean scores had not been reported and it had not been possible to obtain the information from the study authors. We only included data for those participants whose results were known if an intention‐to‐treat analysis was not reported by the study. However, we assessed the potential impact of these missing data in the 'Risk of bias' table.

Assessment of heterogeneity

Assessment of clinical and methodological heterogeneity within treatment comparisons

We extracted data from study reports regarding clinical heterogeneity such as details on the intervention and control treatments, participant characteristics and the outcomes evaluated.

We assessed the presence of clinical heterogeneity within each pair‐wise comparison by comparing the study population characteristics across all eligible trials. We only performed meta‐analysis when considered reasonable based on the degree of heterogeneity.

Assessment of transitivity across treatment comparisons

We assessed the assumption of transitivity by comparing the distribution of the potential effect modifiers across the different pair‐wise comparisons.

Assessment of reporting biases

We performed searches in multiple databases to ensure all potentially eligible studies were identified (Electronic searches). The review authors were alert to duplicated publication of results when analysing the studies to ensure each participant was only included once in the analysis.

If unpublished studies were identified, we tried to obtain sufficient information in order for them to be included in the analysis. The same applied for data published in abstract format.

In studies published in a language other than English, we made every effort to obtain a translation of at least the abstract. If sufficient information was available, we included the study in the analysis.

Data synthesis

Methods for direct treatment comparisons

Since we expected some clinical heterogeneity between studies (for example due to different definitions of pleurodesis success, different time points and doses used), we believed that the assumption of a single fixed intervention effect across included studies was unlikely to be valid. Our primary analyses therefore employed random‐effects models. Since pooled effect estimates from random‐effects models give relatively more weight to smaller studies, which is often considered undesirable, we performed sensitivity analyses using fixed‐effect meta‐analysis models. We performed standard pair‐wise meta‐analysis using a random‐effects model in Cochrane's statistical software, RevMan 2014 for every treatment comparison with two or more studies.

For binary outcome data, we meta‐analysed odds ratios (ORs). For continuous data we planned to use the mean difference (MD) or standardised mean difference (SMD) and perform a check to identify if continuous outcome data were skewed. If this was the case, we planned to analyse the data on a log scale.

If we assessed studies as unsuitable for meta‐analysis, or insufficient studies were identified for meta‐analysis to be performed, we planned to present data by means of a narrative synthesis.

If sufficient data were available, we used similar analysis methods to analyse the adverse effects data. Alternatively we summarised this qualitatively.

Methods for indirect and mixed comparisons

Wherever possible, we performed a multiple‐intervention, network meta‐analysis of primary and (separately) of each secondary outcome measure. We used a Bayesian random‐effects model, fitted using the WinBUGS software (Lunn 2000). We assumed a binomial likelihood and an uninformative normal prior distribution, with mean 0 and standard deviation of 100 for all baseline event rates and intervention effects on the logit scale. When network meta‐analyses were performed, we used the Stata software to generate a network plot (using the networkplot command) and inconsistency plot (using the ifplot command) (Chaimani 2013).

Subgroup analysis and investigation of heterogeneity

Assessment of statistical heterogeneity

In pair‐wise meta‐analyses we estimated the between‐study standard deviation (Tau²) separately for each intervention comparison. For the direct treatment comparisons, we quantified the heterogeneity across studies using the I² statistic, which we interpreted taking into account the magnitude and direction of effect as well as the confidence interval (Higgins 2003).

The assessment of statistical heterogeneity in the network meta‐analysis was based on the magnitude of and credible intervals for the between‐studies standard deviation (Tau) estimated from the NMA models. In network meta‐analysis we assumed a common Tau across all comparisons. We assumed a vague uniform(0,2) prior distribution for Tau.

As described below, reasons for heterogeneity were investigated using subgroup or sensitivity analyses.

Assessment of statistical inconsistency

Inconsistency in the network refers to differences between the direct and indirect effect estimates for the same comparison (Donegan 2013). We used both a loop‐specific approach and a global approach to evaluate these effects.

To evaluate the presence of inconsistency locally we used the loop‐specific approach. This assesses the consistency assumption in each closed loop of the network separately. We identified all the triangular loops (comprising three direct treatment comparisons, all compared with each other) and all the quadratic loops (involving four comparisons) in the network. We compared the differences between the direct and indirect estimates for these loops to generate inconsistency factors, with 95% CIs, calculated and displayed graphically using the 'ifplot' command in Stata (Chaimani 2013). We assumed the estimated between‐study standard deviation (Tau) from the Bayesian analysis of the full network for each loop. We used the magnitude of the inconsistency factors to infer the presence and degree of inconsistency in each loop.

In addition to this, we used a global approach, involving formally comparing the fit of the network meta‐analysis model (which assumes consistency) with that of an ‘inconsistency’ model (in which all consistency constraints are removed). The inconsistency model used is equivalent to fitting a random‐effects meta‐analysis model for all pair‐wise comparisons, with a shared between‐studies variance parameter but no assumptions about direct and indirect evidence forming coherent ‘loops’. We calculated the Deviance Information Criterion (DIC) for each model. If the DIC for the inconsistency model was more than five units higher than that of the consistency model, this was viewed as evidence of inconsistency (Dias 2013).

Assessment of statistical imprecision

We evaluated precision of results, and subsequent rankings, based on their 95% CIs (for pair‐wise analysis) or Cr‐Is (for Bayesian network meta‐analysis).

Sensitivity analysis

Sensitivity analysis and investigation of heterogeneity and inconsistency

We conducted subgroup or sensitivity network meta‐analyses by re‐running the model on restricted numbers of studies according to the following potential effect modifiers, which we felt could be sources of inconsistency and/or heterogeneity:

analysis only including studies which used a clinico‐radiological definition of pleurodesis failure;
analysis only including studies which analysed pleurodesis efficacy at one month after the intervention;
analysis only including studies which analysed pleurodesis efficacy at three months after the intervention;
analysis only including studies which analysed pleurodesis efficacy at more than six months after the intervention;
analysis only including studies which excluded patients with trapped lung;
analysis only including studies which administered bedside pleurodesis through a large‐bore chest tube (> 20 Fr)
analysis only including studies at a lower risk of bias (two or fewer domains at high risk of bias).

In the protocol, we had planned to investigate different tumour types, age of participants and baseline performance status, although there were insufficient data on this in the included studies to perform these subgroup analyses.

Sensitivity analysis

We performed a post‐hoc sensitivity network meta‐analysis evaluating only pleurodesis agents delivered via a chest tube (as opposed to being given at thoracoscopy). We removed the trials evaluating talc poudrage and IPC use from the main network and repeated the analysis.

Results

Description of studies

Results of the search

We performed the literature search in April 2015 (see Figure 1).

We identified 1888 records from database searches and 21 records from other sources before exclusion of duplicates. We screened 1650 abstracts, of which 207 full text articles were retrieved and assessed for eligibility by two independent researchers (AOC, NAM). Of these, 62 met the eligibility criteria (Characteristics of included studies) and 10 are pending classification as they are awaiting translation (Characteristics of studies awaiting classification). Six on‐going studies were also identified (Characteristics of ongoing studies).

Across the 62 included studies, a total of 3428 participants were randomised between 1977 and 2015. There was one foreign language study which was translated from the German (Schmidt 1997).

Included studies

The majority of studies (39/62) explored the efficacy of a variety of pleurodesis agents. Talc was evaluated in 23 trials, making it the most studied agent. The other most commonly examined agents were bleomycin and tetracycline. Two studies compared indwelling pleural catheters with talc slurry (Davies 2012; Demmy 2012).

Four studies evaluated the mode of administration of the pleurodesis agent (three studies comparing talc poudrage with talc slurry (Dresler 2005; Terra 2009; Yim 1996) and one comparing instillation of tetracycline thoracoscopically or through an intercostal cannula (Evans 1993)). A number of studies evaluated alternative methods to improve pleurodesis (one study examined catheter size (Clementsen 1998); three evaluated the duration of drainage after pleurodesis (Goodman 2006; Villanueva 1994; Yildirim 2005); one evaluated the duration of drainage prior to instillation of the sclerosant (Ozkul 2014); one assessed whether patient rotation improved pleurodesis rate (Mager 2002) and one evaluated the effect of talc particle size (Maskell 2004)). We identified one RCT which examined the role of intrapleural fibrinolytics (Okur 2011). One RCT evaluated administration of three different doses of silver nitrate through a chest tube (Terra 2015).

Two studies compared talc pleurodesis with surgical methods to treat malignant effusion (one comparing either talc pleurodesis with pleurectomy (Rintoul 2014) and one comparing talc slurry with thoracoscopic mechanical pleurodesis (Crnjac 2004)).

Additionally, we identified seven studies of agents specifically for the treatment of effusions due to lung cancer (Du 2013; Ishida 2006; Kasahara 2006; Luh 1992; Masuno 1991; Yoshida 2007; Zhao 2009).

There were a number of methodological differences between the included studies. Forty five of 62 studies included all tumour types; two included all except mesothelioma, one included only mesothelioma; one included only adenocarcinoma; six only breast cancer, and in seven studies only lung cancer patients were included.

The methods to define pleurodesis failure varied between studies. Eighteen of 62 studies used radiological criteria only to define a pleurodesis failure, 44 of 62 studies also incorporated symptomatic recurrence or need for a repeat pleural intervention into their definition. The time point at which pleurodesis was defined varied widely between studies, from 1 to 12 months.

The pleurodesis techniques were not standardised. Studies used a variety of chest drain sizes and durations of pleural fluid drainage after the sclerosant was administered. Additionally, patients with trapped lung were excluded from 25 of 62 studies, but not from the others.

Excluded studies

We placed 11 studies in the excluded studies section, having initially identified them as eligible for inclusion but with reasons for exclusion identified later (Characteristics of excluded studies). One study had insufficient data for extraction (Tattersall 1982). Three studies included data for patients with ascites, which could not be separated from those with pleural effusions even after attempting to contact the study authors (Kwasniewska‐Rokicinska 1979; Lissoni 1995; Nio 1999). As per the published protocol, seven studies were found to be high risk of bias for sequence generation and therefore excluded (Caglayan 2008; Dryzer 1993; Elayouty 2012; Engel 1981; Gust 1990; Maiche 1993; Manes 2000). Causes of the inadequate sequence generation included allocating patients to groups using alternation (Caglayan 2008); or according to certain clinical criteria (Maiche 1993), patient hospital number (Dryzer 1993), date of consent (Engel 1981) or date of diagnosis (Manes 2000). We excluded one study as the data contained both randomised and non‐randomised data, which was not distinguishable (Gust 1990) and we could not obtain contact details for the study authors. Another stated patients were ‘divided’ between groups, not mentioning if this process was random (Elayouty 2012) and there was no response from the study authors when contacted to clarify this further.

Risk of bias in included studies

A summary assessment of the risk of bias is presented in the Characteristics of included studies, Figure 2 and Figure 3. No studies were at low risk of bias for all domains.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Thirty three of 62 studies documented adequate sequence generation. The most commonly used methods were computer or telephone randomisation services, block randomisation, stratification, opaque sealed envelopes or a random number generator. Since studies with inadequate sequence generation were excluded as per the protocol, we assessed sequence generation as unclear in the remaining 29 studies. In all cases, the study was stated to have been randomised.

Regarding allocation concealment, we assessed 31 studies as low risk of bias for this domain. Since studies with inadequate allocation concealment were excluded as per the protocol, allocation concealment was ‘unclear’ for the remaining 31 studies.

Blinding

Blinding of participants and personnel (performance bias)

Due to the nature of many of the interventions evaluated in this review, blinding of the participants and clinicians was often not possible and therefore we assessed 40 of 62 studies as high risk of bias for this domain. Many of the pleurodesis agents have differing visual appearances and those studies randomising patients to different modes of administration of a pleurodesis agent, an indwelling pleural catheter or surgery, could not feasibly be blinded.

Four studies were assessed as low risk of performance bias (Bjermer 1995; Maskell 2004; Terra 2015; Zaloznik 1983).

Blinding of outcome assessment (detection bias)

The assessment of pleurodesis success could often not be blinded as it was reliant on participants (who were not blinded) reporting symptoms, in association with the radiological findings of effusion recurrence. Very few studies reported whether the radiological assessments were performed in a blinded fashion. Twenty nine of 62 studies were at high risk of detection bias, and a further 25 of 62 studies had an unclear risk of bias for this domain. Eight studies were low risk of detection bias (Bjermer 1995; Maskell 2004; Masuno 1991; Ong 2000; Ozkul 2014; Terra 2015; Ukale 2004; Zaloznik 1983).

Incomplete outcome data

The majority of studies were low risk of bias because although some inevitable attrition due to death was reported, the rates were comparable for the treatment arms and were deemed reasonable for the size of the population. We classified 12 studies as high risk of bias (eight due to very high attrition rates (Kefford 1980; Kessinger 1987; Masuno 1991; Ostrowski 1989; Patz 1998; Ruckdeschel 1991; Sorensen 1984; Zaloznik 1983; Kefford 1980); one due to very imbalanced loss to follow up (LTFU) between the treatment arms (Fentiman 1986); one the number randomised was not stated (Zimmer 1997); one the numbers provided did not add up (Hillerdal 1986); one excluded patients from the analysis who discontinued treatment due to an allergic reaction (Gaafar 2014)). Three were unclear risk of bias (Kuzdzal 2003: number of randomised patients not stated, only number analysed; Alavi 2011: unable to access tables, and numbers only given as percentages, rather than absolute values; Ozkul 2014: numbers of patients lost to follow up not stated).

Selective reporting

The majority of studies were assessed to be at low risk of bias for selective outcome reporting. We classified two studies as unclear, one as minimal raw data were presented in the text and the tables could not be accessed (Alavi 2011) and the other because pleurodesis success data were not collected in an RCT of talc and tetracycline pleurodesis (although the study was not designed to evaluate this) (Maskell 2004). Five studies were high risk (four provided minimal or no data regarding side effects or survival, or both (Evans 1993; Kuzdzal 2003; Ozkul 2014; Salomaa 1995) and one did not report data on 15 of the randomised patients (Ruckdeschel 1991).

Other potential sources of bias

We classified nine of 62 studies as high risk of bias in the ‘other’ domain and three of 62 studies as unclear. This was for a variety of reasons (see Characteristics of included studies). The remaining studies had a low risk of bias for this domain.

Effects of interventions

PRIMARY OUTCOME

Selection of trials for inclusion in the network

All the interventions from the included studies were evaluated and assessed for inclusion in the network. A number of interventions were not felt to be jointly randomisable and hence were not included in the network. This was the case for specific surgical techniques (Crnjac 2004; Rintoul 2014), different talc particle sizes (Maskell 2004), interventions aimed to improve the efficacy of pleurodesis (Clementsen 1998; Evans 1993; Goodman 2006; Mager 2002; Okur 2011; Ozkul 2014; Villanueva 1994; Yildirim 2005), tumour‐specific intra‐pleural therapy (Du 2013; Ishida 2006; Kasahara 2006; Luh 1992; Masuno 1991; Yoshida 2007; Zhao 2009) and different doses of silver nitrate (Terra 2015).

Two interventions (silver nitrate and combined tetracycline and bleomycin), which we initially felt to be eligible for inclusion in the network had to be removed for the evaluation of pleurodesis efficacy. These agents were only evaluated in one trial each and no participants who received these agents had a pleurodesis failure, which led to computational problems such that a treatment effect could not be estimated (Emad 1996; Paschoalini 2005). One study was not included in the analysis of pleurodesis efficacy as there were no pleurodesis failures in either study arm (Yim 1996). Such studies cannot statistically contribute to the estimate of relative intervention effects (Higgins 2011b).

The majority of studies included all cell types and 36 of 62 trials (58%) did not exclude patients with trapped lung. Pleurodesis was defined using symptom recurrence and radiology in 44 of 62 studies (71%) and it was usually defined within four months of the intervention. It was very difficult to assess whether the distribution of potential effect modifiers was comparable for all the direct treatment comparisons because there were few studies per direct comparison (see Appendix 2).

The final network can be seen in Figure 4. Any studies in the systematic review which were not included in the network were reported descriptively.

Network plot of the pleurodesis efficacy network. The nodes are weighted according to the number of participants randomised to the intervention. The edges (line thicknesses) are weighted according to the number of studies included in each comparison.

Primary outcomes for the methods included in the network meta‐analysis

Direct meta‐analysis

Results of the direct, pair‐wise random‐effects meta‐analysis of the main pleurodesis agents are presented in Table 26. Given the small number of studies making the same direct comparisons, meta‐analysis was only possible for eight direct comparisons.

1. Direct meta‐analysis of pleurodesis efficacy using the random‐effects model showing the odds ratios (95% CI) of the rows compared to the columns.

Treatment	Talc slurry	Talc poudrage	Bleomycin	Tetracycline	C. parvum	Placebo	Mustine	Mitoxantrone	Mepacrine
Talc poudrage	0.76 (0.54, 1.09); n = 3; Tau² = 0; I² = 0%	NA	‐	‐	‐	‐	‐	‐	‐
Bleomycin	1.22 (0.55, 2.70); n = 5*; Tau² = 0.1; I² = 12%	9.70 (2.10, 44.78); n = 2; Tau² = 0; I²= 0%	NA	‐	‐	‐	‐	‐	‐
Tetracycline	0.78 (0.19, 3.13); n = 1*	12.10 (1.32, 111.30); n = 1	2.00 (1.07, 3.75); *n = 5; Tau² = 0; I² = 0%**	NA	‐	‐	‐	‐	‐
C. parvum	NA	NA	0.55 (0.01, 57.48); n = 2; Tau² = 11; I² = 94%	0.31 (0.05, 1.94); n = 1	NA	‐	‐	‐	‐
Interferon	NA	NA	3.25 (1.54, 6.89); n = 1	NA	NA	‐	‐	‐	‐
Iodine	0.47 (0.04, 5.71); n = 1	1.76 (0.26, 11.83); n = 1	1.25 (0.28, 5.59); n = 1	NA	NA	‐	‐	‐	‐
Indwelling pleural catheter	3.35 (1.64, 6.83); n = 2 Tau² = 0; I² = 0%	NA	NA	NA	NA	‐	‐	‐	‐
Placebo	13.93 (0.66, 293.99); n = 1	NA	NA	3.33 (0.51, 21.58); n = 1	NA	NA	‐	‐	‐
Mustine	NA	8.00 (1.40, 45.76); n = 1	NA	2.72 (0.74, 9.98) n = 2*; Tau²= 0; I²= 0%	3.00 (0.40, 22.71); n = 1	NA	NA	‐	‐
Mitoxantrone	NA	NA	3.18 (1.17, 8.65); n = 1	NA	NA	0.75 (0.32, 1.79); n = 1	NA	NA	‐
Mepacrine	2.08 (0.62, 6.96); n = 1	NA	0.16 (0.03, 0.89); n = 1	0.63 (0.05, 8.20); n = 1	NA	0.15 (0.03, 0.64); n = 1*	NA	7.61 (0.35, 163.82); n = 1	NA
Doxycycline	NA	42.69 (2.13, 856.61); n = 1	0.67 (0.24, 1.86); n = 2; Tau²= 0; I²= 0%	NA	1.91 (0.43, 8.48); n = 1	NA	NA	NA	NA
Triethylenethiophosphoramide	NA	NA	NA	NA	NA	2.06 (0.43, 9.80); n = 1*	NA	NA	4.95 (1.02, 24.10); n = 1*
Adriamycin	NA	NA	NA	1.11 (0.06, 20.49); n = 1*	NA	NA	0.37 (0.01, 10.18); n = 1*	NA	NA
n = the number of studies included in the pair‐wise comparison. * Indicates that the comparison included a three‐arm study. NA = no direct pair‐wise comparison available. Results that are statistically significant at the conventional level of P < 0.05 are shaded in grey. ‐ indicates the odds ratio is already expressed elsewhere in the table comparing the interventions the other way around.

Study	Reason study excluded from network	Intrapleural agent or intervention 1	Pleurodesis failure rate for agent 1	Intrapleural agent or intervention 2	Pleurodesis failure rate for agent 2	OR (95% CI) of agent 1 compared with agent 2***
Du 2013	Lung cancer specific therapy	Cisplatin and bevacizumab	6/36	Cisplatin	17/34	0.20 (0.07, 0.60)
Emad 1996*	No pleurodesis failures in the Combined group	Tetracycline**	3/19	Combined tetracycline and bleomycin	0/19	8.27 (0.40, 172.05)
Emad 1996*	No pleurodesis failures in the Combined group	Bleomycin**	2/19	Combined tetracycline and bleomycin	0/19	5.57 (0.25, 124.19)
Ishida 2006*	Lung cancer specific therapy	OK‐432	8/17	Cisplatin	11/17	0.48 (0.12, 1.92)
		OK‐432	8/17	OK‐432 and cisplatin	1/15	12.44 (1.32, 117.03)
		Cisplatin	11/17	OK‐432 and cisplatin	1/15	25.67 (2.68, 245.84)
Kasahara 2006	Lung cancer specific therapy	High dose OK‐432	5/19	Low dose OK‐432	3/19	1.90 (0.38, 9.44)
Luh 1992	Lung cancer specific therapy	OK‐432	3/26	Mitomycin C	9/27	0.26 (0.06, 1.11)
Maskell 2004	Two Talc slurry preparations	Mixed particle talc	3/14	Graded talc (particles >20µm)	2/14	1.64 (0.23, 11.70))
Masuno 1991	Lung cancer specific therapy	LC9018 and Adriamycin	10/38	Adriamycin	23/38	0.23 (0.09, 0.62)
Paschoalini 2005	No pleurodesis failures in Silver Nitrate group	Talc slurry	1/9	Silver nitrate	0/16	5.85 (0.21, 158.82)
Rintoul 2014	MPM specific surgical technique	Talc pleurodesis (slurry or poudrage)	25/62	VATS pleurectomy	24/60	0.88 (0.43, 1.82)
Terra 2015*	Comparison of different doses of Silver Nitrate	90 mg silver nitrate	0/20	150 mg silver nitrate	0/20	not estimable
		90 mg silver nitrate	0/20	180 mg silver nitrate	2/20	0.18 (0.01, 4.01)
		150 mg silver nitrate	0/20	180 mg silver nitrate	2/20	0.19 (0.01, 4.01)
Yoshida 2007*	Lung cancer specific therapy	OK‐432	8/33	Bleomycin	11/35	0.70 (0.24, 2.03)
		OK‐432	8/33	Cisplatin and etoposide	10/34	0.77 (0.26, 2.27)
		Bleomycin	11/35	Cisplatin and etoposide	10/34	1.10 (0.39, 3.07)
Zhao 2009	Lung cancer specific therapy	rAd‐p53 and cisplatin	3/17	Cisplatin	9/18	0.21 (0.05, 1.01)
Three‐arm study. The results for the pair‐wise comparison between tetracycline and bleomycin are included in the network meta‐analysis. **Results that are statistically significant at the conventional level of P < 0.05 are shaded in grey

Type of method to optimise pleurodesis	Study	Intervention 1	Pleurodesis failure rate for intervention 1	Intervention 2	Pleurodesis failure rate for intervention 2	OR (95% CI) of intervention 1 compared with intervention 2*
Mode of administration	Evans 1993	Tetracycline pleurodesis at the end of thoracoscopy	2/15	Tetracycline pleurodesis through an intercostal cannula	5/14	0.28 (0.04, 1.76)
Chest tube size	Clementsen 1998	Small‐bore chest drain	2/9	Large‐bore chest drain	3/9	0.57 (0.07, 4.64)
Patient rotation	Mager 2002	Rotation after instillation of talc	2/10	No rotation after instillation of talc	1/10	2.25 (0.17, 29.77)
Duration of drainage after administration of the sclerosant	Goodman 2006	Drain removed 24 hours after pleurodesis	2/16	Drain removed 72 hours after pleurodesis	4/19	0.54 (0.08, 3.40)
	Villanueva 1994	Drain removal the day after pleurodesis	2/9	Drain removal when < 150 ml/day output	3/15	1.14 (0.15, 8.59)
	Yildirim 2005	Fractionated dose oxytetracycline (4 divided doses at 6‐hourly intervals)	0/12	Single bedside instillation of oxytetracycline	2/8	0.10 (0.00, 2.50)
Duration of drainage prior to administration of the sclerosant	Ozkul 2014	Early instillation of talc slurry after drain insertion	5/40	Instillation of talc slurry when daily drainage from chest tube < 300 ml/day	6/39	0.79 (0.22, 2.82)
Intrapleural fibrinolytics	Okur 2011	Intrapleural streptokinase	14/19	No intrapleural streptokinase	9/16	2.18 (0.53, 9.02)
Pleural abrasion at thoracoscopy	Crnjac 2004	Talc slurry	11/42	Thoracoscopic mechanical pleurodesis	6/45	2.31 (0.77, 6.93)
* Results that are statistically significant at the conventional level of P < 0.05 are shaded in grey

Date	Event	Description
9 April 2019	Amended	Comma deleted in ongoing study reference (OPUS Trial).
11 January 2018	Review declared as stable	See Published notes.

Date	Event	Description
2 April 2019	Amended	Published Notes text amended.
21 August 2014	Amended	Updated the authors' Declaration of Interest statements.

Potential effect modifiers		Total n (%)	Talc poudrage vs talc slurry. n (%)	Bleomycin vs talc slurry. n (%)	Bleomycin vs talc poudrage. n (%)	Tetracycline vs bleomycin. n (%)	C. parvum vs bleomycin. n (%)	IPC vs talc slurry. n (%)	Mustine vs tetracycline. n (%)	Doxycycline vs bleomycin. n (%)
Number of studies		59	3	5	2	5	2	2	2	2
Cell types included	All	42 (71)	3 (100)	5 (100)	1 (50)	5 (100)	1 (50)	2 (100)	2 (100)	2 (100)
	Only breast	6 (10)	0	0	1 (50)	0	0	0	0	0
	Only Lung	7 (12)	0	0	0	0	0	0	0	0
	Other	4 (7)	0	0	0	0	1 (50)	0	0	0
Trapped lung	Excluded	24 (41)	2 (67)	3 (60)	1 (50)	1 (20)	0	1 (50)	0	0
Trapped lung	Included	35 (59)	1 (33)	2 (40)	1 (50)	4 (80)	2 (100)	1 (50)	2 (100)	2 (100)
Drain size	Unknown	26 (44)	1 (33)	1 (20)	1 (50)	4 (80)	1 (50)	0	0	1 (50)
	Small < 20 Fr	13 (22)	0	1 (20)	1 (50)	0	1 (50)	1 (50	1 (50)	1 (50)
	Large ≥ 20 Fr	17 (29)	2 (67)	3 (60)	0	1 (20)	0	1(50)	1 (50)	0
	Study comparing large with small drains	3 (5)	0	0	0	0	0	0	0	0
How pleurodesis was defined	Recurrence of effusion and need for repeat intervention	41 (69)	1 (33)	4 (80)	1 (50)	4 (80)	2 (100)	2 (100)	2 (100)	1 (50)
How pleurodesis was defined	Radiological recurrence only	18 (31)	2 (66)	1 (20)	1 (50)	1 (20)	0	0	0	1 (50)
Time point pleurodesis defined*	2 ‐ 4 months	23 (39)	1 (33)	1 (20)	1 (50)	2 (40)	1 (50)	0	2 (100)	1 (5)
	> 4 ‐ 7 months	1 (2)	1 (33)	0	0	0	0	0	0	0
	> 11 ‐ 12 months	1 (2)	0	0	0	0	0	1 (50)	0	0
	< 2 months	29 (49)	1 (33)	2 (40)	0	3 (60)	1 (50)	1 (50)	0	1 (50)
	Not stated	5 (8)	0	2 (40)	1 (50)	0	0	0	0	0
* If the study reported multiple time points, the one referred to here was that used in our primary analysis (according to our hierarchy of preferences (see Primary outcomes)). IPC = Indwelling pleural catheter.

	Number of pleurodesis methods evaluated	Number of trials included in network	Tau² (95% CI)	Global inconsistency identified?	Loop‐specific inconsistency identified? If so, loop showing inconsistency with ROR (95% CI)
Whole network	16	41	0.88 (0.42, 1.482)	No	Yes TS‐TP‐BL 7.0 (1.1, 43.8) TS‐BL‐ME 10.2 (1.1, 96.9)
Only data collected at 1 month	13	24	0.80 (0.14, 1.66)	No	Yes BL‐MX‐ME 54.2 (2.0, 1469.5)
Only data collected at 3 months	9	10	0.55 (0.02, 1.84)	No	No
Only data collected at 6 months	7	10	1.09 (0.09, 1.95)	No	Yes TS‐TP‐BL 11.7 (1.1, 123.3)
Trials excluding patients with trapped lung	9	13	0.71 (0.04, 1.85)	No	No
Trials using a clinico‐radiological definition of pleurodesis	16	29	1.16 (0.59, 1.87)	No	Yes TS‐BL‐ME 12.9 (1.3, 126.3) TS‐TP‐BL 11.6 (1.02, 132.5)
Trials using large‐bore chest tubes	9	16	1.32 (0.51, 1.95)	No	Yes TS‐BL‐TC 15.2 (1.5, 151.4) TS‐BL‐ME (12.2 (1.2, 121.2)
Trials with a lower risk of bias (high risk of bias for < 2 domains)	16	36	0.46 (0.03, 1.09)	No	Yes TS‐BL‐ME 12.2 (1.2, 125.3) TS‐TP‐BL 8.5 (1.2, 60.3)
BL = Bleomycin; CP = C. parvum; DX = Doxycycline; IO = Iodine; ME = Mepacrine; MU = Mustine; MX = Mitoxantrone; PL = Placebo; TC = Tetracycline; TP = Talc poudrage; TS = Talc slurry

Pleurodesis method	Estimated rank (95% Cr‐I)
C. parvum	2 (1, 6)
Viscum	2 (1, 11)
Talc poudrage	3 (1, 6)
Doxycycline	4 (1, 13)
Talc slurry	5 (3, 8)
Mepacrine	5 (2, 9)
Iodine	7 (2, 12)
Adriamycin	8 (1, 16)
Bleomycin	9 (6, 12)
Tetracycline	10 (7, 13)
Indwelling pleural catheter	10 (6, 15)
Mustine	12 (7, 16)
Triethylenethoiphosphoramide	12 (5, 16)
Interferon	13 (8, 16)
Mitoxantrone	14 (11, 16)
Placebo	15 (12, 16)

Pleurodesis agent	Estimated rank (95% Cr‐I)
Viscum	1 (1, 10)
Mepacrine	3 (1, 9)
C. parvum	4 (1, 9)
Talc slurry	5 (2, 9)
Doxycycline	5 (1, 10)
Iodine	5 (1, 12)
Bleomycin	6 (1, 13)
Adriamycin	7 (4, 10)
Tetracycline	8 (4, 11)
Mitoxantrone	11 (5, 13)
Mustine	11 (6, 13)
Interferon	11 (3, 13)
Placebo	12 (6, 13)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pleurodesis failure	21		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Bleomycin vs iodine	1	39	Odds Ratio (M‐H, Random, 95% CI)	0.8 [0.18, 3.57]
1.2 Bleomycin vs talc slurry	5	199	Odds Ratio (M‐H, Random, 95% CI)	1.22 [0.55, 2.70]
1.3 Bleomycin vs tetracycline	5	220	Odds Ratio (M‐H, Random, 95% CI)	0.50 [0.27, 0.93]
1.4 Bleomycin vs talc poudrage	2	57	Odds Ratio (M‐H, Random, 95% CI)	9.70 [2.10, 44.78]
1.5 Bleomycin vs C. parvum	2	78	Odds Ratio (M‐H, Random, 95% CI)	1.81 [0.02, 189.25]
1.6 Bleomycin vs doxycycline	2	122	Odds Ratio (M‐H, Random, 95% CI)	1.50 [0.54, 4.20]
1.7 Bleomycin vs IFN	1	160	Odds Ratio (M‐H, Random, 95% CI)	0.31 [0.15, 0.65]
1.8 Bleomycin vs mitoxantrone	1	85	Odds Ratio (M‐H, Random, 95% CI)	0.31 [0.12, 0.86]
1.9 Bleomycin vs mepacrine	1	36	Odds Ratio (M‐H, Random, 95% CI)	6.40 [1.12, 36.44]
1.10 Bleomycin vs combined tetracycline and bleomycin	1	38	Odds Ratio (M‐H, Random, 95% CI)	5.57 [0.25, 124.19]
1.11 Bleomycin vs cisplatin and etoposide	1	69	Odds Ratio (M‐H, Random, 95% CI)	1.1 [0.39, 3.07]
1.12 Bleomycin vs OK‐432	1	68	Odds Ratio (M‐H, Random, 95% CI)	1.43 [0.49, 4.17]
1.13 Bleomycin vs viscum	1	17	Odds Ratio (M‐H, Random, 95% CI)	5.33 [0.62, 45.99]
2 Pain	14		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Bleomycin vs talc slurry	2	73	Odds Ratio (M‐H, Random, 95% CI)	1.66 [0.41, 6.80]
2.2 Bleomycin vs tetracycline	4	220	Odds Ratio (M‐H, Random, 95% CI)	0.61 [0.29, 1.27]
2.3 Bleomycin vs talc poudrage	1	32	Odds Ratio (M‐H, Random, 95% CI)	0.28 [0.01, 7.31]
2.4 Bleomycin vs C. parvum	2	71	Odds Ratio (M‐H, Random, 95% CI)	0.70 [0.27, 1.85]
2.5 Bleomycin vs IFN	1	160	Odds Ratio (M‐H, Random, 95% CI)	32.34 [1.89, 552.23]
2.6 Bleomycin vs mitoxantrone	1	96	Odds Ratio (M‐H, Random, 95% CI)	0.48 [0.15, 1.56]
2.7 Bleomycin vs mepacrine	1	40	Odds Ratio (M‐H, Random, 95% CI)	0.46 [0.11, 1.94]
2.8 Bleomycin vs doxycycline	2	148	Odds Ratio (M‐H, Random, 95% CI)	0.84 [0.26, 2.70]
2.9 Bleomycin vs OK‐432	1	67	Odds Ratio (M‐H, Random, 95% CI)	0.40 [0.14, 1.12]
2.10 Bleomycin vs cisplatin and etoposide	1	69	Odds Ratio (M‐H, Random, 95% CI)	0.83 [0.32, 2.16]
3 Mortality	11		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
3.1 Bleomycin vs combined tetracycline and bleomycin	1	40	Odds Ratio (M‐H, Random, 95% CI)	1.0 [0.06, 17.18]
3.2 Bleomycin vs talc slurry	2	116	Odds Ratio (M‐H, Random, 95% CI)	0.89 [0.29, 2.75]
3.3 Bleomycin vs tetracycline	2	125	Odds Ratio (M‐H, Random, 95% CI)	0.63 [0.27, 1.44]
3.4 Bleomycin vs talc poudrage	1	32	Odds Ratio (M‐H, Random, 95% CI)	0.82 [0.20, 3.43]
3.5 Bleomycin vs C. parvum	1	55	Odds Ratio (M‐H, Random, 95% CI)	0.60 [0.19, 1.94]
3.6 Bleomycin vs IFN	1	160	Odds Ratio (M‐H, Random, 95% CI)	0.46 [0.25, 0.87]
3.7 Bleomycin vs mitoxantrone	1	96	Odds Ratio (M‐H, Random, 95% CI)	2.15 [0.95, 4.86]
3.8 Bleomycin vs OK‐432	1	68	Odds Ratio (M‐H, Random, 95% CI)	2.66 [0.98, 7.23]
3.9 Bleomycin vs doxycycline	2	122	Odds Ratio (M‐H, Random, 95% CI)	1.44 [0.53, 3.90]
3.10 Bleomycin vs cisplatin and etoposide	1	69	Odds Ratio (M‐H, Random, 95% CI)	2.22 [0.82, 6.01]
4 Fever	16		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
4.1 Bleomycin vs talc Slurry	3	99	Odds Ratio (M‐H, Random, 95% CI)	0.90 [0.31, 2.56]
4.2 Bleomycin vs talc poudrage	1	32	Odds Ratio (M‐H, Random, 95% CI)	0.87 [0.11, 7.05]
4.3 Bleomycin vs tetracycline	5	250	Odds Ratio (M‐H, Random, 95% CI)	2.05 [0.67, 6.34]
4.4 Tetracycline vs C. parvum	2	80	Odds Ratio (M‐H, Random, 95% CI)	0.43 [0.17, 1.12]
4.5 Bleomycin vs IFN	1	160	Odds Ratio (M‐H, Random, 95% CI)	151.35 [9.08, 2522.62]
4.6 Bleomycin vs mitoxantrone	1	96	Odds Ratio (M‐H, Random, 95% CI)	1.11 [0.37, 3.36]
4.7 Bleomycin vs mepacrine	1	40	Odds Ratio (M‐H, Random, 95% CI)	0.52 [0.14, 1.92]
4.8 Bleomycin vs doxycycline	2	148	Odds Ratio (M‐H, Random, 95% CI)	2.69 [0.08, 89.51]
4.9 Bleomycin vs combined tetracycline and bleomycin	1	40	Odds Ratio (M‐H, Random, 95% CI)	0.47 [0.04, 5.69]
4.10 Bleomycin vs OK432	1	67	Odds Ratio (M‐H, Random, 95% CI)	0.7 [0.23, 2.13]
4.11 Bleomycin vs cisplatin and etoposide	1	69	Odds Ratio (M‐H, Random, 95% CI)	2.22 [0.82, 6.01]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pleurodesis failure	9		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 Talc poudrage vs talc slurry	3	599	Odds Ratio (M‐H, Fixed, 95% CI)	0.77 [0.54, 1.09]
1.2 Talc poudrage vs bleomycin	2	57	Odds Ratio (M‐H, Fixed, 95% CI)	0.10 [0.02, 0.48]
1.3 Talc poudrage vs tetracycline	1	33	Odds Ratio (M‐H, Fixed, 95% CI)	0.08 [0.01, 0.76]
1.4 Talc poudrage vs iodine	1	42	Odds Ratio (M‐H, Fixed, 95% CI)	0.57 [0.08, 3.80]
1.5 Talc poudrage vs mustine	1	37	Odds Ratio (M‐H, Fixed, 95% CI)	0.13 [0.02, 0.71]
1.6 Talc poudrage vs doxycycline	1	31	Odds Ratio (M‐H, Fixed, 95% CI)	0.02 [0.00, 0.47]
2 Mortality	6		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Talc poudrage vs talc slurry	2	397	Odds Ratio (M‐H, Random, 95% CI)	1.02 [0.35, 3.00]
2.2 Talc poudrage vs bleomycin	1	32	Odds Ratio (M‐H, Random, 95% CI)	1.22 [0.29, 5.13]
2.3 Talc poudrage vs tetracycline	1	41	Odds Ratio (M‐H, Random, 95% CI)	5.25 [0.91, 30.22]
2.4 Talc poudrage vs iodine	1	42	Odds Ratio (M‐H, Random, 95% CI)	2.64 [0.58, 12.09]
2.5 Talc poudrage vs mustine	1	46	Odds Ratio (M‐H, Random, 95% CI)	0.43 [0.09, 1.96]
3 Pain	3		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
3.1 Talc poudrage vs talc slurry	1	482	Odds Ratio (M‐H, Random, 95% CI)	0.47 [0.23, 0.96]
3.2 Talc poudrage vs bleomycin	1	32	Odds Ratio (M‐H, Random, 95% CI)	3.62 [0.14, 95.78]
3.3 Talc poudrage vs iodine	1	42	Odds Ratio (M‐H, Random, 95% CI)	9.97 [0.50, 198.04]
4 Fever	4		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
4.1 Talc poudrage vs talc slurry	2	479	Odds Ratio (M‐H, Random, 95% CI)	0.60 [0.15, 2.37]
4.2 Talc poudrage vs bleomycin	1	32	Odds Ratio (M‐H, Random, 95% CI)	1.15 [0.14, 9.38]
4.3 Talc poudrage vs iodine	1	42	Odds Ratio (M‐H, Random, 95% CI)	4.22 [0.43, 41.45]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pleurodesis failure	4	628	Odds Ratio (M‐H, Random, 95% CI)	0.74 [0.52, 1.04]
1.1 Talc	3	599	Odds Ratio (M‐H, Random, 95% CI)	0.76 [0.54, 1.09]
1.2 Tetracycline	1	29	Odds Ratio (M‐H, Random, 95% CI)	0.28 [0.04, 1.76]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pain	3	114	Odds Ratio (M‐H, Random, 95% CI)	4.56 [1.66, 12.52]
1.1 Mepacrine vs bleomycin	1	40	Odds Ratio (M‐H, Random, 95% CI)	2.15 [0.52, 9.00]
1.2 Mepacrine vs tetracycline	1	22	Odds Ratio (M‐H, Random, 95% CI)	5.6 [0.81, 38.51]
1.3 Mepacrine vs placebo	1	23	Odds Ratio (M‐H, Random, 95% CI)	14.53 [0.71, 298.21]
1.4 Mepacrine vs triethylenethiophosphoramide	1	29	Odds Ratio (M‐H, Random, 95% CI)	23.71 [1.19, 474.06]
2 Fever	3		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Mepacrine vs bleomycin	1	40	Odds Ratio (M‐H, Random, 95% CI)	1.91 [0.52, 7.01]
2.2 Mepacrine vs tetracycline	1	22	Odds Ratio (M‐H, Random, 95% CI)	8.00 [1.13, 56.79]
2.3 Mepacrine vs placebo	1	23	Odds Ratio (M‐H, Random, 95% CI)	62.43 [2.85, 1365.52]
2.4 Mepacrine vs triethylene...	1	29	Odds Ratio (M‐H, Random, 95% CI)	23.83 [3.35, 169.39]
3 Pleurodesis failure	5		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
3.1 Mepacrine vs talc slurry	1	89	Odds Ratio (M‐H, Random, 95% CI)	2.08 [0.62, 6.96]
3.2 Mepacrine vs bleomycin	1	36	Odds Ratio (M‐H, Random, 95% CI)	0.16 [0.03, 0.89]
3.3 Mepacrine vs tetracycline	1	21	Odds Ratio (M‐H, Random, 95% CI)	0.63 [0.05, 8.20]
3.4 Mepacrine vs placebo	1	23	Odds Ratio (M‐H, Random, 95% CI)	0.07 [0.01, 0.73]
3.5 Mepacrine vs mitoxantrone	1	26	Odds Ratio (M‐H, Random, 95% CI)	7.61 [0.35, 163.82]
3.6 Mepacrine vs triethylene...	1	29	Odds Ratio (M‐H, Random, 95% CI)	0.20 [0.04, 0.98]
4 Mortality	2		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
4.1 Mepacrine vs talc slurry	1	89	Odds Ratio (M‐H, Random, 95% CI)	0.53 [0.20, 1.43]
4.2 Mepacrine vs mitoxantrone	1	28	Odds Ratio (M‐H, Random, 95% CI)	1.64 [0.23, 11.70]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pleurodesis failure	1	18	Odds Ratio (M‐H, Random, 95% CI)	0.57 [0.07, 4.64]
2 Pain	1		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
3 Mortality	1		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pleurodesis failure	1	87	Odds Ratio (M‐H, Random, 95% CI)	0.43 [0.14, 1.30]
2 Mortality	1		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pleurodesis failure	4	205	Odds Ratio (M‐H, Random, 95% CI)	1.26 [0.70, 2.30]
1.1 Rotation vs no rotation	1	20	Odds Ratio (M‐H, Random, 95% CI)	2.25 [0.17, 29.77]
1.2 Streptokinase vs no streptokinase	1	35	Odds Ratio (M‐H, Random, 95% CI)	2.18 [0.53, 9.02]
1.3 Mixed particle talc vs graded talc	1	28	Odds Ratio (M‐H, Random, 95% CI)	1.64 [0.23, 11.70]
1.4 Talc pleurodesis vs VATS parietal pleurectomy	1	122	Odds Ratio (M‐H, Random, 95% CI)	1.01 [0.49, 2.09]
2 Pain	1		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Streptokinase vs control	1	47	Odds Ratio (M‐H, Random, 95% CI)	3.0 [0.12, 77.47]
3 Fever	1		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
3.1 Mixed particle talc vs graded talc	1	46	Odds Ratio (M‐H, Random, 95% CI)	15.92 [1.81, 140.16]
4 Mortality	2		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
4.1 Mixed particle talc vs graded talc	1	43	Odds Ratio (M‐H, Random, 95% CI)	0.88 [0.25, 3.07]
4.2 Talc pleurodesis vs VATS partial pleurectomy	1	175	Odds Ratio (M‐H, Random, 95% CI)	0.92 [0.45, 1.90]

Methods	Single centre RCT comparing the efficacy of cosmetic talc with iodopovidone for pleurodesis (India)
Participants	Inclusion: recurrent symptomatic pleural effusion with improvement of breathlessness with thoracentesis; or primary or secondary pneumothorax Exclusion: allergy to iodine; thyroid disorder; trapped lung; air leak; advanced malignancy with expected survival < 30 days 36 participants randomised
Interventions	28 Fr intercostal drain to completely drain effusion or treat pneumothorax. Pleurodesis agent given when < 150ml/day drainage and complete lung re‐expansion on chest x‐ray. All participants received intrapleural lignocaine (2 mg/kg) and IV tramadol prior to pleurodesis. Iodopovidone: 20 ml 10% iodopovidone in 80 ml saline Cosmetic talc: 5 g sterilised 'baby powder' After agent administered, chest tube clamped for four hours. Repeat administration of agent if > 250 ml/day drainage. Drain removed when < 100 ml/day output Followed up at 1 week, 1 month, 3 months and 6 months and then every 3 months thereafter
Outcomes	Pleurodesis success according to need for thoracentesis (complete success = relief of symptoms related to the effusion and no re‐accumulation on CXR at 30 days; partial success = reduced dyspnoea related to the effusion with only partial re‐accumulation of fluid on chest x‐ray and no requirement for therapeutic thoracentesis; failure = lack of success as defined above) Chest pain (measured by visual analogue scale score) Complications Time to pleurodesis
Notes	People with trapped lung excluded. Unpublished data obtained from authors relating to subgroup of participants in the study with malignant pleural effusion‐ only this data was included for the purposes of this review Included in network meta‐analysis for pleurodesis efficacy and fever.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation sequence
Allocation concealment (selection bias)	Low risk	Opaque sealed envelopes
Blinding of participants and personnel (performance bias)   All outcomes	High risk	"Blinding of the allocation to treatments was not possible". Agents have different appearances
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Symptom recurrence, visucal analogue scale scores and complications would all be biased by lack of patient blinding. Mortality would not be effected by lack of blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	No loss to follow up. Intention‐to‐treat analysis performed
Selective reporting (reporting bias)	Low risk	All reported
Other bias	Low risk	Cosmetic talc used rather than medicinal talc, but sterilised and comparable particle size by electron microscopy. No external funding for the study

Methods	Single centre RCT of povidone‐iodine and bleomycin pleurodesis for malignant pleural effusion (Iran)
Participants	Inclusion: biopsy or cytologically proven malignant pleural effusion (all tumour types); recurrent and symptomatic effusion; chest radiograph confirming lung expansion of 90% after thoracentesis; Karnofsky Performance Score > 70 Exclusion: co‐morbidities that preclude general anaesthesia; bleeding disorders; massive thoracic skin infiltration; active infectious disease 39 participants randomised
Interventions	All participants underwent a 28 Fr intercostal drain under local anaesthetic (+/‐ IV opiates if required). Study agent administered intrapleurally the next day with 5 ml 2% lidocaine Bleomycin group: 1 mg/kg bleomycin in 60 ml saline. 1 dose Povidone‐iodine group: 5% (volume unclear). 1 dose After administration of the study agent, the drain was clamped for one hour and removed when < 200ml fluid output/day. If the fluid output remained high after 10 days, they were discharged home with a Heimlich valve in place
Outcomes	Effusion recurrence on chest x‐ray at 30 days Pain (measured by numeric scale) at discharge and day 30 Dysponea (measured by numeric scale) at discharge and day 30
Notes	Minimal raw data in results section ‐ tables quoted in text but not available on line. Attempted to contact study authors by e mails ‐ no response People with trapped lung excluded from trial entry Pleurodesis success measured only using chest x‐ray criteria Included in network meta‐analysis for pleurodesis efficacy.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation
Allocation concealment (selection bias)	Low risk	Block randomisation
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Differing appearances of bleomycin and iodine make blinding not possible (although not stated explicitly in paper)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Pain and dyspnoea may be biased by lack of blinding. Not stated whether CXRs were evaluated by a blinded clinician. No response from study authors regarding this
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Unable to see the tables. Response rates only given as % (no actual numbers), so unclear whether there was LTFU
Selective reporting (reporting bias)	Unclear risk	Raw data not provided for many of the outcomes. Tables missing
Other bias	Low risk	No other biases identified

Methods	Two‐centre RCT of intrapleural quinacrine (mepacrine) vs tetracycline via tube thoracostomy for malignant pleural effusion (USA)
Participants	Inclusion: (1) documented cancer with pleural effusion (2) pleural fluid cytology or pleural biopsy confirming malignancy or exudate effusion presumed to be malignant (3) symptomatic from the effusion or rapidly re‐accumulating effusion > 500 ml All cell types. No exclusion criteria 20 participants randomised.
Interventions	Both groups had a closed tube thoracostomy, drained overnight prior to the installation Quinacrine group: intrapleural quinacrine (100 mg in 30 ml normal saline) once daily for four days Tetracycline group: one dose of intrapleural tetracycline (500 mg in 30 ml N saline) The drains were clamped for six hours post installation with patient rotation. Drain removed when < 60 ml/24 hour drainage
Outcomes	Pleurodesis success (defined on chest x‐ray criteria only at 30 days as 'Complete response' (complete lack of re‐accumulation of pleural fluid); 'Partial response' (re‐accumulation of pleural fluid < 50% of the volume present before the sclerosis); 'Failure' (re‐accumulation of fluid to > 50% of the volume present before the attempted sclerosis)) Side effects of treatment (pain, fever)
Notes	People with trapped lung not excluded. CR and PR counted as a pleurodesis success for purposes of analysis One participant allocated to quinacrine arm having had treatment failure with tetracycline not included in the analysis Included in network meta‐analysis for pleurodesis efficacy and pain.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified and unable to contact study authors
Allocation concealment (selection bias)	Unclear risk	Not specified and unable to contact study authors
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	No comment on whether study was blinded
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not stated whether CXR evaluation was blinded. Pain and fever outcomes may have been affected if patients were unblinded to treatment allocation, however not stated in the paper whether this was the case
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Two of 14 randomised to tetracycline excluded from analysis (one died and one LTFU). No LTFU in mepacrine arm (overall LTFU 13%)
Selective reporting (reporting bias)	Low risk	All specified endpoints reported
Other bias	High risk	Eight of 22 participants included in the study did not have proven pleural malignancy

Methods	RCT of mitoxantrone versus mepacrine via an intercostal drain (Sweden ‐ number of centres not specified)
Participants	Cytologically proven, symptomatic MPE with an expected survival of greater than three months (Karnofsky Performance Score > 60). Excluded if cytotoxic chemotherapy in the preceding month All cell types included 30 participants randomised
Interventions	Both groups had a 12‐14 Fr chest tube inserted and effusion drained. Pleurodesis agent was given through the chest tube and patient's position changed for two hours after administration Group 1: 1 dose of intrapleural mitoxantrone 30 mg in 50 ml N saline was given; the drain was closed for 48 hours and removed after the 'pleural cavity was emptied' Group 2: 2 doses of intrapleural mepacrine chloride 200 mg in 20 ml N saline were given on consecutive days and the drain removed when < 150 ml fluid production/day
Outcomes	Pleural fluid re‐accumulation at 4 and 12 weeks (defined as 'Complete response' (CR), 'Partial response' (PR) (if recurrence of pleural fluid but thoracocentesis not considered to be indicated) or 'Progressive disease'. Side effects/toxicity (visual analogue scale pain and fever scores) Symptom questionnaires (participant grades symptom on a numeric scale for four key symptoms‐ pain, shortness of breath, nausea and tiredness) Pharmacokinetics of mitoxantrone
Notes	People with trapped lung not excluded from the study CR and PR counted as pleurodesis success for analysis Included in network meta‐analysis for pleurodesis efficacy and mortality.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified and unable to find contact details for study authors
Allocation concealment (selection bias)	Unclear risk	Not specified and unable to find contact details for study authors
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Study personnel not blinded as drugs are of different colours. However, participants were blinded to treatment allocation
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Participants blind to treatment allocation, therefore fever, pain and symptom scores unbiased. "Radiological evaluation was made by an independent radiologist'
Incomplete outcome data (attrition bias)   All outcomes	Low risk	One participant in each study arm did not receive treatment due to "unexpected medical emergencies", therefore deemed non‐evaluable. Follow‐up data clearly documented for the remaining patients
Selective reporting (reporting bias)	Low risk	All pre‐specified outcomes reported
Other bias	Low risk	Drain suction use was imbalanced between the treatment arms (10/14 received suction in mepacrine group vs 1/14 in mitoxantrone group)

Methods	Single centre RCT of tetracycline pleurodesis using a small percutaneous catheter (CH10), compared to a large‐bore chest tube (CH24) inserted after thoracoscopy (Denmark)
Participants	Symptomatic, recurrent MPE, proven on pleural fluid cytology. Expected survival of > 3 months (all tumour types included) 21 participants randomised
Interventions	Group 1: small percutaneous catheter (CH10 65 cm) inserted under local anaesthesia Group 2: medical thoracoscopy, followed by insertion of a large‐bore chest tube (CH24) Both groups received pleurodesis with 500 mg tetracycline and 100 mg bupivicaine intrapleurally. The drain was clamped for six hours after instillation after which suction was applied. Drain removed when fluid output < 200 ml in 24 hours
Outcomes	Treatment response at 3, 6 and 9 weeks defined by roentgenographic response ('Complete response' ‐ no recurrence of pleural fluid; 'Partial response' ‐ slight re‐accumulation with blunted costophrenic angle; 'No response' ‐ complete recurrence of pleural fluid) and clinical response (by the need for new thoracentesis) Questionnaire evaluating discomfort in connection with the tube and the pleurodesis
Notes	Trapped lung not accounted for in inclusion/exclusion criteria, but one patient excluded as they had hydropneumothorax at time of instillation CR and PR included as pleurodesis successes for analysis Not included in network meta‐analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Allocation by lot"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not possible to blind, as different drain sizes used (although not stated explicitly in the paper)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	"All data were evaluated by the same physician, who was without knowledge of the result of the randomisation". However, symptom‐based adverse events and symptomatic need for repeat pleural intervention may be biased by lack of patient blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	All data reported and justified. Missing outcome data balanced between the two treatment arms (two excluded from group 1 (one died of cancer soon after drain insertion and one developed hydropneumothorax necessitating large‐bore drain), one excluded from group 2 (patient withdrew consent for study participation)
Selective reporting (reporting bias)	Low risk	All outcomes reported
Other bias	Low risk	No other biases identified

PERMALINK

Interventions for the management of malignant pleural effusions: a network meta‐analysis

Amelia O Clive

Hayley E Jones

Rahul Bhatnagar

Nancy J Preston

Nick Maskell

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Inclusion

Exclusion

Types of interventions

Interventions of direct interest

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Random sequence generation (checking for possible selection bias)

Allocation concealment (checking for possible selection bias)

Blinding of participants and personnel (checking for possible performance bias)

Blinding of outcome assessment (checking for possible detection bias)

Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data)

Selective Outcome Reporting

Other sources of bias

Measures of treatment effect

Relative treatment effects

Relative treatment ranking

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of clinical and methodological heterogeneity within treatment comparisons

Assessment of transitivity across treatment comparisons

Assessment of reporting biases

Data synthesis

Methods for direct treatment comparisons

Methods for indirect and mixed comparisons

Subgroup analysis and investigation of heterogeneity

Assessment of statistical heterogeneity

Assessment of statistical inconsistency

Assessment of statistical imprecision

Sensitivity analysis

Sensitivity analysis and investigation of heterogeneity and inconsistency

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Excluded studies

Risk of bias in included studies

2.

3.

Allocation

Blinding

Blinding of participants and personnel (performance bias)

Blinding of outcome assessment (detection bias)

Incomplete outcome data

Methods	Single centre RCT comparing thoracoscopic mechanical pleurodesis (TMP) with talc slurry (Slovenia)
Participants	Inclusion: breast carcinoma and a resulting morphologically confirmed MPE Exclusion: unfit for general anaesthetic (GA) 87 participants randomised
Interventions	TMP arm: thoracoscopy (under GA) with adhesiolysis, pleural biopsy and scarification of the visceral and parietal pleura to induce bleeding. Chest tube inserted at the end of procedure Talc slurry arm: chest tube inserted under local anaesthetic. 5 g talc in 100 ml saline insufflated through chest tube Participants in both arms had the drain removed when < 100ml/24hour drainage
Outcomes	Recurrence of effusion on chest x‐ray (CXR) at 1 day, 1 week, 1 month, 3 months and 6 months Duration of chest tube drainage Duration of hospitalisation Complications Mortality (30 days and 6 months)
Notes	People with trapped lung not excluded. Pleurodesis success defined using CXR criteria alone Not included in network meta‐analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not stated
Allocation concealment (selection bias)	Low risk	Sealed envelopes
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not feasible to blind the study as comparing talc slurry with thoracoscopy
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Not stated whether radiological assessments were done in a blinded fashion. Complication reporting, time of tube drainage may be effected by lack of patient and personnel blinding. Mortality outcome not effected by lack of blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	ITT analysis performed. Minimal missing data. 6/45 patients died within six months in TMP group vs 8/42 in talc slurry arm
Selective reporting (reporting bias)	Low risk	All stated outcomes reported
Other bias	Low risk	No documentation of patient experience (e.g. QOL or degree of discomfort), relative costs or need for repeat pleural intervention Pleurodesis success defined using radiology only. Participants who did not have evidence of recurrence at death were classified as pleurodesis successes

Methods	Unblinded, multi‐centre RCT comparing indwelling pleural catheter (IPC) with talc slurry pleurodesis (UK)‐ TIME‐2 Trial.
Participants	Inclusion criteria: clinically confident diagnosis of MPE requiring pleurodesis Exclusion criteria: age < 18, expected survival of < 3 months, chylothorax, previous ipsilateral lobectomy or pneumonectomy, previous attempted pleurodesis, pleural infection, WCC < 1000/microlitre, hypercapnic ventilatory failure, pregnancy, lactating mothers, irreversible bleeding diathesis, irreversible visual impairment 106 participants randomised
Interventions	Group 1: IPC inserted with drainage three times a week (or as required to relieve dyspnoea) Group 2: 12 F Seldinger chest tube and 4 g talc slurry as an inpatient All patients had standard oncological management for the primary tumour
Outcomes	Primary outcome: mean daily dyspnoea visual analogue score (VAS) over the first 42 days Secondary outcomes: proportion achieving clinically significant decrease in mean VAS dyspnoea; mean VAS dyspnoea at 6 weeks, 3 months and 6 months; mean daily chest pain VAS over the first 42 days; mean VAS chest pain at 6 weeks, 3 months and 6 months; nights spent in hospital; self‐reported quality of life; frequency of adverse events
Notes	Participants with trapped lung in group 2 did not receive talc pleurodesis, but remained in trial follow‐up Pleurodesis in the IPC group was defined as removal of IPC following spontaneous cessation of drainage with no significant fluid recurrence on chest x‐ray (CXR) or ultrasound scan (USS) and no further ipsilateral pleural intervention. In the talc group, pleurodesis failure defined as the need for further ipsilateral pleural intervention If participants died during follow up, included as a pleurodesis success if no intervention prior to death Included in network meta‐analysis for pleurodesis efficacy and mortality.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Central telephone randomisation
Allocation concealment (selection bias)	Low risk	Central telephone randomisation
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not possible to blind participants or personnel due to nature of interventions (IPC vs talc slurry)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	VAS scores, QOL and symptom recurrence (which informs assessment of pleurodesis efficacy) could be biased by lack of blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	LTFU clearly documented with reasons given
Selective reporting (reporting bias)	Low risk	All predefined endpoints reported
Other bias	Low risk	No other biases identified

Methods	Multi‐centre RCT comparing bedside talc pleurodesis and daily tunnelled catheter drainage for management of malignant pleural effusion (USA)
Participants	Inclusion: symptomatic patients with histo/cytologically proven malignancy and a previously untreated, unilateral pleural effusion requiring management; ECOG performance score 0‐2 Exclusion: active pleural infection; talc allergy; contraindications to talc use; trapped lung; survival < 60 days; severe comorbid medical conditions 68 participants randomised
Interventions	Talc pleurodesis group: 4 g to 5 g sterile talc slurry in 100 ml saline infused into pleural space via > 24 Fr chest drain. Tube clamped for two hours. Drain removed when < 150 ml drainage/24hours Indwelling pleural catheter (IPC) group: PleurX catheter inserted and drained daily (output volumes recorded). Removed when < 30 ml output on three consecutive days
Outcomes	Primary: compare the proportion of maintained successful treatments 30 days after the intervention (success defined as being (1) alive (2) no effusion recurrence (3) > 90% lung re‐expansion after complete drainage (4) completion of the intervention by two weeks ie drain removed or IPC functioning normally) Secondary: Quality of life (QOL); dyspnoea; patient satisfaction and acceptability; lung expansion; pleurodesis success; fluid drainage volume; days device in place; removal of device before death; survival
Notes	Pleurodesis success measured at 30 days according to chest x‐ray (CXR) and need for repeat pleural intervention People with known trapped lung excluded from trial entry Included in network meta‐analysis for pleurodesis efficacy and mortality.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Permuted block randomisation via a web‐based randomisation service
Allocation concealment (selection bias)	Low risk	Permuted block randomisation via a web‐based randomisation service
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Due to nature of interventions, not possible to blind participants or personnel (IPC vs talc slurry)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	"Pleurodesis success was classified by an unblinded local investigator" (personal communication). QOL, symptom recurrence and patient satisfaction questionnaires may be biased by lack of patient blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Five excluded from analysis in each arm, but justifications given
Selective reporting (reporting bias)	Low risk	All reported
Other bias	Low risk	Target recruitment numbers not reached

Methods	Prospective, single centre RCT of thoracoscopic talc poudrage versus bedside bleomycin pleurodesis via a small‐bore chest tube (Switzerland)
Participants	Inclusion criteria: documented MPE (all cell types); complete lung expansion on post drainage chest x‐ray (CXR); improvement in symptoms after drainage; expected survival of > 1 month; capable of undergoing medical thoracoscopy Exclusion criteria: loculated effusion; previous drainages or previous pleurodesis; adverse reaction to the study medication; severe coagulation disorder 36 participants randomised
Interventions	Group 1: bedside pleurodesis via small‐bore chest tube (OD = 2.7 mm) of 60 IU bleomycin. Tube unclamped after two hours and left on suction until removal at least 48 hours later Group 2: thoracoscopy with induced pneumothorax under sedation. 5 g talc sprayed into pleural cavity under direct vision after drainage of effusion and disruption of adhesions. Drain kept under suction for at least 48 hours
Outcomes	Recurrence of effusion (defined as a newly detected effusion needing drainage or occupying > 33% of the pleural space on CXR as compared with the first CXR after drain removal, or death from any cause) at 30, 90 and 180 days Medication use Volume of fluid drained Duration of hospital stay Cost Symptom VAS Scores (pain, shortness of breath, cough and general well‐being)
Notes	People with trapped lung excluded from study enrolment Included in network meta‐analysis for pleurodesis efficacy, mortality, fever and pain.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation sequence
Allocation concealment (selection bias)	Low risk	Sequential sealed envelopes
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not possible to blind participants or personnel due to nature of interventions (talc poudrage vs bleomycin via chest tube)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Not stated if radiology was interpreted by a blinded physician. However length of stay, VAS scores and symptom recurrence may be biased by lack of participant blinding. Mortality would not be affected by unblinded nature of the study
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Five withdrawals in total, but a similar number in each group
Selective reporting (reporting bias)	Low risk	All stated outcomes reported
Other bias	Low risk	No external funding source

Methods	Multi‐centre RCT comparing talc poudrage with talc slurry pleurodesis in MPE. Both groups received 4 g ‐ 5 g sterile talc intrapleurally (USA)
Participants	Inclusion criteria: history of malignancy (all tumour types), pleural effusion requiring sclerosis, ECOG performance status 0‐2, life expectancy > 2 months, ability to undergo general anaesthesia Exclusion criteria: pregnancy, previous intrapleural therapy or radiation therapy encompassing the entire hemithorax, changes in systemic therapy within two weeks, chylous or bilateral effusions requiring therapy 501 participants randomised
Interventions	TS Group: talc administered as a slurry in 100 ml saline through a chest tube at the bedside TTI Group: talc insufflated during thoracoscopy in the operating room
Outcomes	Primary endpoint: the percentage of patients whose lung initially re‐expanded > 90% and who had a successful pleurodesis at 30 days after treatment (defined according to cvhest x‐ray (CXR) criteria) Secondary endpoint: time to recurrence of effusion; frequency of complications and toxicities; ability to re‐expand the lung as assessed by CXR; oain; patient satisfaction; quality of life (QOL).
Notes	People with trapped lung excluded from analysis Included in network meta‐analysis for pleurodesis efficacy, mortality, pain and fever.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation lists
Allocation concealment (selection bias)	Low risk	Computer‐generated randomisation lists
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not possible to blind the study due to the nature of the interventions (talc poudrage vs talc slurry)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Not stated if radiological assessment was blinded. QOL and complications may be affected by lack of patient and personnel blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Missing data accounted for and balanced between the treatment arms (10 in slurry group and 9 in thoracoscopy group excluded as ineligible or participant withdrew consent; 33/163 slurry participants and 25/177 thoracoscopy participants died within 30 days of randomisation)
Selective reporting (reporting bias)	Low risk	All outcomes reported on
Other bias	Low risk	Trapped lung defined by different means in the two treatment arms, which may have effected their primary endpoint. However, this does not have an impact on the pleurodesis success rates

Methods	Single centre RCT of intrapleural cisplatin +/‐ bevacizumab in MPE due to non‐small cell lung cancer (NSCLC) (China)
Participants	Inclusion: Advanced NSCLC; large uni‐ or bilateral pleural effusion; positive pleural fluid cytology; no intrapleural therapy in previous month; Karnofsky performance score > 60%; age > 18; predicted survival > 3 months; no major organ disfunction; no previous chemotherapy in previous six weeks Exclusion: squamous cell carcinoma; allergy to biological agents; no detectable lesions; uncontrolled central nervous system metastasis; pregnancy or breastfeeding; infected wound; refractory psychiatric illness 72 participants randomised
Interventions	Participants underwent pleural fluid drainage by thoracentesis. Treatment given intrapleurally. Rest for two hours. Then rotate every 15 mins. Given every two weeks for 3 cycles Cisplatin: 30 mg cisplatin intrapleurally Cisplatin and bevacizumab: 30 mg cisplatin and 300 mg bevacizumab intrapleurally
Outcomes	Treatment response ('Complete remission (CR)' = accumulated fluid disappeared and stable for at least four weeks; 'Partial remission (PR)' = > 50% of the accumulated fluid had disappeared, symptoms had improved and the remaining fluid had not increased for at least four weeks; 'Remission not obvious (NC)' = < 50% of the accumulated fluid had disappeared; 'Progression (PD)' = accumulated fluid had increased). Treatment success defined as CR + PR Progression‐free survival Overall survival Adverse reactions Quality of life (QOL) Pleural fluid VEGF levels
Notes	People with trapped lung eligible for trial involvement Pleurodesis defined clinically and using radiology Not included in network meta‐analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Methods not stated and no response from study authors to clarify
Allocation concealment (selection bias)	Unclear risk	Methods not stated and no response from study authors to clarify
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Not stated if blinded and no response from study authors to clarify
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not stated if anyone was blinded. If not blinded, QOL, performance status, side effects and symptom recurrence could be biased by lack of blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Missing data accounted for. ITT analysis
Selective reporting (reporting bias)	Low risk	All outcomes reported on
Other bias	Low risk	No other biases identified

Methods	Three‐arm, single centre RCT comparing intrapleural bleomycin, tetracycline and combination treatment for pleurodesis of MPE (Iran)
Participants	Inclusion: histologically or cytologically proven, symptomatic MPE (all cell types) Exclusion: none 60 participants randomised
Interventions	All participants had 28 Fr intercostal drain inserted into 6th intercostal space. Complete drainage of the effusion was confirmed on chest x‐ray (CXR). All participants given 10‐15ml 1% lignocaine intrapleurally Tetracycline arm: 20 mg/kg tetracycline (max 2 g) in 50 ml saline given intrapleurally. 1 dose Bleomycin arm: 1 u/kg (max 60 units) in 50 ml saline given intrapleurally. 1 dose Combination arm: 20 mg/kg tetracycline in 40 ml saline and 1 u/kg bleomycin in 50 ml saline, given intrapleurally, one after the other (tube clamped for five mins between instillations) Drain clamped for two hours post instillation with patient rotation. Suction connected after 24 hours. Drain removed when < 50 ml/8 hours drainage and complete lung expansion on CXR
Outcomes	Pleurodesis success (defined as 'complete response' (no accumulation of effusion on CXR), 'partial response' (effusion recurred but did not require aspiration) or 'failure' (participant required repeat thoracentesis for re‐accumulation of the effusion) at 30 days (also at 60 days, 90 days and 6 months) Side effects
Notes	All participants in the study were receiving chemotherapy or tamoxifen, or both People with trapped lung not excluded from participation in the study Participants who died prior to the analysed time point were excluded from the analysis Combination of clinical need for repeat intervention and radiological re‐accumulation of effusion used to define pleurodesis failure Included in network meta‐analysis for pleurodesis efficacy, mortality and fever.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"'...simple randomised manner". No further details given
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not stated explicitly and unable to contact authors. However, different volumes and regimes were used for the two groups
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Not stated if radiology reported blindly. Complication‐reporting may have been affected by lack of participant blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Minimal data on baseline patient characteristics, but all outcome data reported and withdrawals justified. Six participants died within six months of randomisation (2 in tetracycline arm; 1 in bleomycin arm and 3 in combination arm)
Selective reporting (reporting bias)	Low risk	All stated outcomes reported
Other bias	Low risk	No other biases identified

Methods	Single centre RCT of medical vs surgical pleurodesis with tetracycline (UK)
Participants	Inclusion criteria: cytology‐proven MPE and histological or cytological evidence of metastatic breast cancer Exclusion criteria: unsuitable for general anaesthetic (GA); > 75 years old; severe non‐metastatic lung disease; evidence of life‐threatening metastatic disease at other sites 34 participants randomised
Interventions	Medical group: intercostal cannula inserted into mid‐axillary line 7th/8th intercostal space and fluid aspirated. When drainage complete, 500 mg tetracycline in 100 ml N saline inserted IP. Drain removed after 24 hours Surgical group: under GA, bronchoscopy then thoracoscopy performed. 500 ml tetracycline in 100 ml saline inserted after fluid removed. Drain removed at 24 hours
Outcomes	Fluid re‐accumulation on chest x‐ray (CXR) Need for repeat pleural aspirations Mortality
Notes	Pleurodesis failure defined as need for repeat aspiration. Trapped lung not accounted for Not included in network meta‐analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details given regarding randomisation
Allocation concealment (selection bias)	Unclear risk	No details given regarding randomisation
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Unable to blind due to nature of the interventions (surgery vs chest tube)
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Need for repeat aspirations and other treatments given for cancer after pleurodesis may have been biased by lack of blinding of personnel and participants. Not stated if CXRs were reported by a blinded person
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Reasons given for withdrawals (5/34 excluded (15%) ‐ 3 never received the treatment; 1 was randomised in error; 1 participant's records were lost)
Selective reporting (reporting bias)	High risk	No data on safety or side effects
Other bias	Low risk	No other biases identified

Methods	Single centre RCT of talc poudrage and mustine (via chest tube) in patients with breast cancer. All patients underwent VATS procedure under general anaesthetic. (UK)
Participants	Inclusion criteria: histologically confirmed breast cancer and radiologically verified pleural effusion Exclusion criteria: no previous local treatment; non‐malignant cause for the effusion 46 participants randomised
Interventions	Talc group: talc poudrage performed during VATS (dose of talc not stated), two chest drains in place for five days (with or without suction) Mustine group: after VATS and once lung fully re‐expanded on CXR, 15 mg mustine solution instilled via intercostal drain. Clamped for two hours. Drain removed when drainage stopped
Outcomes	Success of pleurodesis (defined by lack of re‐accumulation of effusion on CXR) at one month; complications
Notes	If died prior to one‐month follow up, excluded from analysis of pleurodesis success Participants with trapped lung eligible for enrolment Included in network meta‐analysis for pleurodesis efficacy and mortality.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Stratified for metastatic disease requiring treatment. "balanced randomisation"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not possible to blind patients or personnel due to the nature of the procedures
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not stated whether radiographic interpretation of CXRs were performed by a blinded person. Reporting of complications could be biased by lack of participant and personnel blinding
Incomplete outcome data (attrition bias)   All outcomes	Low risk	3/23 non‐evaluable in talc group; 6/23 non‐evaluable in mustine group. All non‐evaluable patients died prior to one‐month follow up
Selective reporting (reporting bias)	Low risk	All stated outcomes reported
Other bias	High risk	Different number of intercostal drains in the two groups. Different duration of drainage for two groups

Methods	Single centre RCT of intrapleural talc and tetracycline in MPE secondary to breast cancer (UK)
Participants	Inclusion criteria: histologically confirmed breast cancer and a symptomatic pleural effusion on radiology Exclusion criteria: previous treatment for effusion, other than simple needle aspiration; non‐malignant cause for effusion; unsuitable for general anaesthetic; history of sensitivity to tetracycline 41 participants randomised
Interventions	Talc group: thoracoscopy, talc insufflated (dose not stated). Intercostal drain remained in situ for five days Tetracycline group: thoracoscopy. Tetracycline 500 mg in 40 ml N saline inserted 16 ‐ 24 hours later via chest tube. Intercostal drain left in place for 3 ‐ 5 days
Outcomes	Pleurodesis success (defined by lack of re‐accumulation on CXR); complications; mortality
Notes	Pleurodesis success defined according to CXR only Participants with trapped lung eligible for trial entry Included in network meta‐analysis of pleurodesis efficacy and mortality.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomised with stratification for metastatic disease"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Not possible to blind patients or personnel due to the nature of the procedures
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not stated whether radiographic interpretation of CXRs were performed by a blinded person. Reporting of complications could be biased by lack of participant and personnel blinding
Incomplete outcome data (attrition bias)   All outcomes	High risk	Participants were excluded from the primary analysis if they died within the first month. Higher proportion of deaths in the talc group (6/18 = 33%) compared to the tetracycline group (2/23 = 9%)
Selective reporting (reporting bias)	Low risk	All stated outcomes reported
Other bias	Low risk	No other biases identified