Description of the condition
A pneumothorax is defined as air in the pleural space, that is, air between the lung and the chest wall. A primary spontaneous pneumothorax (PSP) is a pneumothorax that occurs in a person without a precipitating event (including iatrogenic causes) or known underlying lung disease. Secondary spontaneous pneumothorax occurs as a result of known underlying lung disease, most commonly in persons suffering from chronic obstructive pulmonary disease (COPD) (Light 2007).
The most widely accepted mechanism for pneumothorax in cases of primary spontaneous pneumothorax is rupture of sub-plural bullae. Bulla formation is thought to be secondary to inflammatory change at a cellular level, most likely as a result of inhaled tobacco smoke (Sahn 2000). Other mechanisms include small airways inflammation and emphysematous changes. Spontaneous pneumothorax is a relatively common pathology, with studies suggesting 18 to 28 cases per 100,000 population in males and two to six per 100,000 population in females (Gupta 2000; Melton 1979).
Description of the intervention
Primary spontaneous pneumothorax can be treated either conservatively or by intervention. Interventional treatment options, which have been popular in the last 40 to 50 years, include simple fine needle aspiration, large or small bore intercostal tube drainage, thoracotomy and thoracoscopy. Conservative management, which involves observing the patient, giving appropriate analgesia and oxygen therapy, and only treating interventionally in rare cases, is the focus of this review.
How the intervention might work
The treatment principles for primary spontaneous pneumothorax are to remove the air from the pleural space, and to decrease the chance of recurrence (Light 2007). Many of the initial studies favouring interventions focused on the speed of normalization of x-ray findings or length of stay in hospital which was determined by normalization of x-ray findings, rather than clinical endpoints such as frequency of complication, analgesic requirements, and recurrence risk (Romanoff 1968). In 1966 Stradling and Poole, in a large observational study, reported efficacy and safety of conservative management. They expressed concern that interventional management would be practised on the basis of rapid radiological resolution (Stradling 1966).
Lung physiology dictates a lung will inflate with inspiration and that the air in the pleural space will be reabsorbed. The rate of reabsorption was estimated at 1.25% of the volume of the hemithorax each 24 hours (Kircher 1954).
Current guidelines on the management of primary spontaneous pneumothorax (Baumann 2001; De Leyn 2005; Henry 2003; MacDuff 2010) largely ignore the practice of conservative management, despite the earlier evidence of its efficacy and safety, and recommend intervention in all but the smallest PSPs. We believe that the evidence for this recommendation is limited, and clinical experience has suggested that conservative management can be effective and safe even in large PSPs. We plan to examine what evidence really exists for the superiority of an invasive approach.
Why it is important to do this review
Currently there are three major published guidelines outlining treatment options for PSP. The British Thoracic Society (Henry 2003; MacDuff 2010) and the Belgian Society of Pneumology (De Leyn 2005) both state simple aspiration to be the first line treatment for all PSP requiring intervention (Henry 2003; MacDuff 2010). In contrast to this, The American College of Chest Physicians consensus guidelines state there is no role for simple aspiration, and recommend the insertion of intercostal catheter as first line therapy in large pneumothoraces (Baumann 2001). This highlights the disagreement between current guidelines regarding interventional management as first line treatment for most PSPs, and there is no recommendation for even a trial of a conservative approach in anything but the smaller PSPs, defined by the British Thoracic Society as a rim of less than two centimetres of air in the pleural space, by the American College of Chest Physicians as three centimetres or less, and by the Belgian Society of Pneumology as "small and minimally symptomatic" (De Leyn 2005, p. 266). The major published guidelines cite the limited relevant evidence and low levels of evidence. They disagree on which pneumothoraces can be managed conservatively.
Literature currently exists comparing the interventional methods of management, including a Cochrane Review (Wakai 2007) comparing simple aspiration versus intercostal drainage, finding that there is no difference between the two interventions with regard to immediate successful resolution (Wakai 2007).
PSP is a condition that should have a near 100% survival rate, occurring as it does in patients with no underlying lung disease. However, complications of intervention do occur, with rates of up to 30% with chest tube (Ball 2007; Vedam 2003), and interventions impact adversely on morbidity, mortality, and other outcomes such as pain and length of stay.
The somewhat contradictory published guidelines (Baumann 2001; De Leyn 2005; Henry 2003; MacDuff 2010), along with the lack of a firm evidence base has led to wide variations in the management of PSP, reflecting local preferences and expertise rather than an evidence-based approach. A systematic review of the literature, including literature which examines the use of a conservative approach to PSP, will help to create an evidence-based background for future treatment recommendations.
The objective of the review is to compare conservative and interventional treatments of adult primary spontaneous pneumothorax for outcomes of clinical efficacy, tolerability and safety.
Criteria for considering studies for this review
Types of studies
We will include randomized controlled trials (RCTs) and we will accept quasi-RCTs if a systematic method of allocation is used, such as alternation, assignment based on date of birth, case record number and date of presentation. We will include studies reported in abstract form (e.g. conference presentations) if data are available from the investigators.
Types of participants
We will include adults (aged 18 to 50 years) with a first episode of symptomatic primary spontaneous pneumothorax as identified by radiographic evidence, where there is no underlying lung disease. We will exclude: adults with lung disease such as chronic obstructive pulmonary disease, asthma, interstitial lung disease, cystic fibrosis, trauma, any other secondary causes or recurrent pneumothorax; and adults with an iatrogenic pneumothorax. We will exclude adults over 50 due to the increased likelihood of smokers over 50 having COPD changes.
Types of interventions
We will compare observational/conservative management with interventional management.
We define observational/conservative management as:
- Oxygen therapy
Interventional management strategies include:
- Intercostal or thoracic drainage
- Simple, manual needle aspiration
- Thoracotomy or thoracoscopy
- Chest tubes, suction, catheterization
Chest tube size will be included where available.
Types of outcome measures
- Immediate success - defined as clinical resolution with substantial radiological improvement of pneumothorax within 28 days without complications
- Failure of treatment - defined as need for a secondary intervention within 28 days
- Mortality - all causes within 28 days
- Duration of hospitalization - measured in nights in hospital from first presentation
- Immediate recurrence: 24 to 48 hours
- Early recurrence: one to two weeks
- Intermediate recurrence: two to four weeks
- Late recurrence: more than four weeks
- Complications: including of interventional and conservative management
Outcomes will not form part of the study eligibility assessment, so that studies that meet the participant, intervention and comparison criteria will be included in the review even if they report no relevant outcomes. Time-to-event data will be interpreted with care and in light of the duration of studies, in order to ensure that censoring is properly dealt with.
Search methods for identification of studies
We will conduct electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL), in the most recent issue of The Cochrane Library; MEDLINE via Ovid SP (1920 to date); CINAHL via EBSCO host (1980 to date); EMBASE via Ovid SP (1947 to date) and ISI Web of Science (1945 to date). We will not apply any language restrictions. The development of our final search strategy will be an iterative process, and we will employ strategies to maximize the sensitivity of the search as outlined in Chapter 6.4 of the Cochrane Handbook of Systematic reviews of Interventions (Cochrane Handbook). Given we are interested in pseudo-randomized as well as randomized trials, we will not limit the search to randomized controlled trials (RCTs). We will initially use the MEDLINE search strategy identified in Appendix 1, but will refine our terms by assessing retrieved studies for unidentified search terms. We will use both MeSH terms and free text as appropriate, given issues of precision and sensitivity.
The search strategy will be developed in conjunction with the Cochrane Anaesthesia Review Group's Trials Search Co-ordinator.
Searching other resources
In addition we will search relevant ongoing trials registers and relevant conference proceedings.
We will scan the references of primary sources and we will search any other unpublished sources known to the authors including 'grey literature'. We will seek information about, and raw data from, published trials where appropriate.
Data collection and analysis
Selection of studies
Two of three authors (PM, OJ, MA) will independently review all studies identified to assess whether they meet the inclusion criteria, having reviewed the abstract and retrieved the full text of the article if insufficient information is available from the abstract. We will resolve discrepancies by discussion and the involvement of a fourth author (KO) if they remain unresolved.
Data extraction and management
We will combine the results of all studies and delete duplicates. Two authors will independently review the titles and abstracts of identified papers for eligibility based on whether: it is a controlled trial; involved participants aged 18 and over; spontaneous pneumothorax was the condition under study; and one intervention involved conservative management as defined in the protocol (screening form Appendix 2). We will resolve any disagreements between authors by discussion with all authors present. If a study satisfies all of these criteria, or it is unclear from the title and abstract, we will retrieve the full text and inclusion of the study will be determined based on the criteria described in addition to identified inclusion and exclusion criteria. Once again, two reviewers will independently determine exclusion, and disagreements will be resolved by discussion with all authors.
We will extract data from the included studies using an appropriately modified version of the Cochrane Anaesthetic Review Group's Data Collection Form (Appendix 3). Two authors will independently extract the data. We will resolve disagreements by discussion with all authors. If further information is required from the authors of the trial, KO will contact the authors and request the information or data.
Assessment of risk of bias in included studies
Assessment of risk of bias is a domain-based critical evaluation of included studies. Three authors (PM, OJ, MA) will independently assess the risk of bias using The Cochrane Collaboration's tool for assessing risk of bias (Cochrane Handbook, Chapter 8). We will assess each study according to the following domains: random sequence generation; allocation concealment; blinding of participants and personnel; blinding of outcome assessment; incomplete outcome data; selective reporting; and any other bias. A copy of the risk assessment form that we will use is attached (Appendix 4). We will assess trials as having low risk of bias if all identified domains are assessed to be satisfactory and a high risk of bias if one or more domains are assessed as a high risk or unclear risk. All authors will jointly discuss and resolve disagreements in any classification.
We will include the risks of bias for each study in a table of included study characteristics. We will explore the impact of risk of bias by graphing results according to risk of bias and, if appropriate, we will perform sensitivity analyses which exclude studies assessed as having high or uncertain risk of bias to determine whether their inclusion alters the results of the meta-analyses.
Measures of treatment effect
We will make a descriptive or quantitative comparison of outcomes depending on the characteristics of the studies identified for inclusion (heterogeneity, comparability of outcomes, the presence of serious publication or reporting bias).
Analyses will contrast the intervention effect of the outcomes between the two groups with respect to the direction and size of the effect, consistency across studies and strength of evidence.
We will make pairwise comparisons (see: Data synthesis). We will calculate treatment effect for dichotomous outcomes (immediate success, early and late recurrence rates, complications, 28-day all-cause mortality, etc) using risk ratios with a 95% confidence interval.
For continuous variables (length of hospital stay) we will calculate difference in means as an estimate of effect size, using fixed-effect mean difference with 95% confidence intervals. We do not anticipate that there will be continuous variables measured on different scales, so are not expecting that standardized mean difference will be required in the analysis.
We do not anticipate that analysis of time-to-event data will be required.
Unit of analysis issues
Given the nature of the outcomes, we do not anticipate that there will be any unit of analysis issues. If any issues arise we will deal with them accordingly, with omission from the data if necessary.
Dealing with missing data
If not all intended outcomes have been reported we will endeavour to contact the authors of the study and gather the relevant outcome data. If summary data are not available we will endeavour to contact authors to obtain the relevant summary statistics. Where this is not possible we will employ imputation methods if possible (Cochrane Handbook section 16.1), or we will determine and discuss the implications of exclusion from the analysis.
We will examine each study to determine how the authors dealt with missing data. If intention-to-treat (ITT) analyses were not performed and the data are available to perform them, we will do so (Cochrane Handbook section 16.2). In the case of large numbers of incomplete data we will perform sensitivity analyses using 'best-case' and 'worst-case' scenarios (Cochrane Handbook section 16.2.2). We will make explicit the assumptions of whatever method is used to cope with missing data.
Assessment of heterogeneity
We will assess heterogeneity for outcomes using the Chi² test in Review Manager 5 (RevMan 5.2), with the null hypothesis being no heterogeneity for treatment effect (Cochrane Handbook). The Chi² test measures the deviation of observed effect sizes from an underlying overall effect. This test has low power to detect true heterogeneity when studies have small sample sizes or are few in number, hence we will use a P value of 0.01 (Higgins 2002). The I² statistic assesses the impact of heterogeneity on the meta-analysis (Higgins 2002). The magnitude is roughly interpreted as:
- 0% to 40%: might be unimportant;
- 30% to 60%: may represent moderate heterogeneity;
- 50% to 90%: may represent substantial heterogeneity; and
- 75% to 100%: may represent considerable heterogeneity.
Assessment of reporting biases
We will attempt to avoid publication bias by ensuring that all potential sources of unpublished studies are closely examined. In particular we will examine the 'grey literature' (reports that are produced by all levels of government, academics, business and industry in print and electronic formats but that are not controlled by commercial publishers), trial registers, and conference proceedings.
We will use a funnel plot of the effect estimate of each study plotted against a measure of their size or precision, to assess publication bias where 10 or more studies are included and provide outcome data. We will visually examine the funnel plot for symmetry, with asymmetry alerting us that there may be a problem which needs consideration. Given the nature of the data it is unlikely that we will conduct further analyses of funnel plot asymmetry.
Depending on the studies identified for inclusion, we will produce a descriptive or a quantitative comparison of outcomes. The authors will determine whether it is appropriate to proceed with quantitative analyses (meta-analysis) based on:
- Heterogeneity of studies, in particular with respect to the interventions examined
- Comparability of outcomes measured
- The presence of serious publication or reporting bias
If heterogeneity between studies is identified, we will investigate possible causes, perform prespecified subgroup analyses and consider whether pooling is appropriate, or pool with random-effects meta analysis.
The descriptive or quantitative analyses will contrast the intervention effect of the outcomes between the two groups with respect to:
- The direction of the effect
- The size of the effect
- Consistency across studies
- The strength of evidence for the effect
When meta-analysis is considered appropriate, pairwise comparison will focus on the intervention of interest (conservative/non-interventional treatment) compared to one or more different interventional treatments, logically grouped where appropriate. In addition we will compare conservative/non-interventional treatment to all interventional treatments, grouped as one. We will perform meta-analyses in Review Manager 5 software (RevMan 5.2).
Subgroup analysis and investigation of heterogeneity
We plan to perform a subgroup analysis on smokers. We will examine differences in outcome effect size in smokers versus non-smokers in order to determine if the intervention works differently in these two populations.
Smoking has been implicated in the pathogenesis of PSP, and is associated with up to 22 times risk (moderate smoker up to 22 cigarettes per day) (Bense 1987) of developing a first PSP.
We will conduct sensitivity analyses for the following:
- Studies identified as having a high risk of bias will be excluded from analyses to determine if this alters the results
- Imputation of missing data to determine 'best-case' and 'worst-case' scenarios.
If significant heterogeneity is detected, we will perform additional sensitivity analyses using random-effects versus fixed-effect models.
Summary of findings
We will use the principles of the GRADE system (Guyatt 2008) to assess the quality of the body of evidence associated with specific outcomes, being immediate success, failure of treatment, mortality, early and late recurrence, complications and length of hospital stay, by constructing a summary of findings (SoF) table using the GRADE software. The GRADE approach appraises the quality of a body of evidence based on the extent to which one can be confident that an estimate of effect or association reflects the item being assessed. The quality of a body of evidence takes into consideration within-study risk of bias (methodological quality), the directness of the evidence, heterogeneity of the data, precision of effect estimates and risk of publication bias.
We would like to thank Ronan O'Sullivan (content editor), Cathal Walsh (statistical editor), Steven A Sahn and Andrew MacDuff (peer reviewers) for their help and editorial advice during the preparation of this protocol for the systematic review. We would also like to thank Dr Julia Walters for her statistical expertise and input.
Appendix 1. MEDLINE (Ovid SP) search strategy
1. exp Pneumothorax/ or pneumothorax.af.
2. exp Thoracotomy/ or exp Thoracostomy/ or exp Chest Tubes/ or Drainage/ or Suction/ or (aspiration or thoraco?tom* or suction).ti,ab. or chest tube*.mp. or (drain* adj3 (chest or intercostal or thoracic)).mp.
3. 1 and 2
4. ((randomized controlled trial or controlled clinical trial).pt. or randomized.ab. or placebo.ab. or clinical trials as topic.sh. or randomly.ab. or trial.ti.) not (animals not (humans and animals)).sh.
5. 3 and 4
Appendix 2. Screening form
Appendix 3. Data Extraction Form
Data collection form
Notes on using a data extraction form:
- Be consistent in the order and style you use to describe the information for each report
- Record any missing information as unclear or not described, to make it clear that the information was not found in the study report(s), not that you forgot to extract it.
- Include any instructions and decision rules on the data collection form, or in an accompanying document. It is important to practice using the form and give training to any other authors using the form.
1. General Information
2. Study Eligibility
DO NOT PROCEED IF STUDY EXCLUDED FROM REVIEW
3. Population and setting
5. Risk of Bias assessment
See Chapter 8 of the Cochrane Handbook
Provide overall data and, if available, comparative data for each intervention or comparison group.
7. Intervention groups
Copy and paste table for each intervention and comparison group
Intervention Group 1
Copy and paste table for each outcome.
Repeat for late recurrence
Length of hospital stay
Pleurectomy or pleurodesis
Copy and paste the appropriate table for each outcome, including additional tables for each time point and subgroup as required.
Dichotomous outcome – Early recurrence
Repeat for late recurrence
Pleurectomy or pleurodesis
Length of hospital stay
11. Other information
Risk of Bias Table
Risk of Bias Table
Appendix 4. Risk of Bias Tool
Risk of Bias Table
Contributions of authors
Michael Ashby (MA), Pete Mulcahy (PM), Kathryn J Ogden (KO), Greg Haug (KH), Oliver Jensen (OJ).
Conceiving the review: GH
Co-ordinating the review: MA
Undertaking manual searches: MA, OJ, PM
Screening search results: MA, OJ, PM
Organizing retrieval of papers: KO
Screening retrieved papers against inclusion criteria: MA, OJ, PM
Appraising quality of papers: MA, OJ, PM
Abstracting data from papers: MA, OJ, PM
Writing to authors of papers for additional information: KO
Providing additional data about papers: KO
Obtaining and screening data on unpublished studies: KO, MA
Data management for the review: KO, MA
Entering data into Review Manager 5 (RevMan 5.2): MA, OJ, PM
RevMan statistical data: KO
Other statistical analysis not using RevMan: KO
Interpretation of data: KO, GH, MA, OJ, PM
Statistical inferences: KO
Writing the review: MA, OJ, PM, KO
Securing funding for the review: N/A
Guarantor for the review (one author): MA
Person responsible for reading and checking review before submission: KO
Declarations of interest
Michael Ashby: none known
Greg Haug: none known
Pete Mulcahy: none known
Kathryn J Ogden: none known
Oliver Jensen: none known
Sources of support
- Department of Medicine, Launceston General Hospital, Australia.Home Institute
- University of Tasmania, Australia.Launceston Clinical School
- No sources of support supplied