Criteria for considering studies for this review
Types of studies
Randomized controlled studies.
Types of participants
Children and adults with CF (according to standard definitions, clinical features of CF plus a positive sweat test or the presence of two genes known to be associated with CF (Rosenstein 1998).
Types of interventions
We included studies comparing patient outcomes following therapies guided by the results of bronchoscopy (with BAL or PBB sampling) with outcomes following therapies guided by the results of any other type of sampling (including but not limited to cultures from sputum, throat swab and cough swab).
Types of outcome measures
per cent predicted forced expiratory volume at one second (FEV1) change and absolute values in litres
per cent predicted forced vital capacity (FVC) change and absolute values in litres
infant lung function-squeeze
per cent predicted FEV0.5, forced expiratory flow from 25% to 75% of vital capacity (FEF25-75) and FEF75 changes and absolute values in litres
per cent predicted FVC change and absolute values in litres
lung clearance index (LCI)
High-resolution computed tomography (CT) appearances using a recognised scoring system (e.g. Brody 2004; Loeve 2009)
weight (in kg or percentile)
body mass index (BMI) percentile
Number of positive isolates per child per year of follow-up (for BAL defined as a positive culture of the growths from 1000 CFU/ml)
Burkholderia cepacia complex
Stenotrophomonous maltophilia (S. maltophilia)
Achromobacter xylosoxidans (A. xylosoxidans)
any other organism
Clearance of the organism from the cultures
Time to first infection with P. aeruginosa*
Time to chronic infection using any recognized definition e.g. (Lee 2003)*
with P. aeruginosa
Complications and adverse effects related to bronchoscopy (e.g. fever, hypoxaemia and increased cough, unplanned admissions and other serious adverse events etc.)
Quality of life (QoL) measured using a validated tool such as Cystic Fibrosis Questionnaire-Revised version (CFQ-R) (Quittner 2009) and Cystic Fibrosis Quality of Life Questionnaire (CFQoL) (Gee 2000)
number of hospitalisations per patient per year
days as inpatient per patient per year
cost of care
Number of courses of antibiotics prescribed per patient per year (not including prophylactic antibiotics)
Number of pulmonary exacerbations (requiring oral or intravenous antibiotics) per patient per year (diagnosis based on clinical judgement or any approved or published definition e.g. Rabin 2004, Rosenfeld 2001)
* In the studies where the follow up was started just after the birth, we considered the age of first acquisition of infection and age of establishing chronic infection. In studies in older patients where follow up is started later in life, we considered time to acquire the first infection and time to acquire chronic infection.
Search methods for identification of studies
We identified relevant studies from the Cochrane Cystic Fibrosis and Genetic Disorders Group's Cystic Fibrosis Trials Register using the term: sampling techniques.
The Cystic Fibrosis Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of The Cochrane Library), quarterly searches of MEDLINE, a search of Embase to 1995 and the prospective handsearching of two journals - Pediatric Pulmonology and the Journal of Cystic Fibrosis. Unpublished work is identified by searching the abstract books of three major cystic fibrosis conferences: the International Cystic Fibrosis Conference; the European Cystic Fibrosis Conference and the North American Cystic Fibrosis Conference.
For full details of all searching activities for the register, please see the relevant sections of the Cochrane Cystic Fibrosis and Genetic Disorders Group Module.
Date of the latest search of the Group's Cystic Fibrosis Trials Register: 28 November 2013.
We also searched the ongoing study registers: clinicaltrials.gov and International Standard Randomised Controlled Trial Number (ISRCTN) Register (www.isrctn.org) using the search terms: cystic fibrosis AND bronchoscopy, cystic fibrosis AND bronchoalveolar lavage, cystic fibrosis AND sampling technique.
Date of the latest search of the ongoing study registers: 16 July 2012.
Searching other resources
We checked the reference lists of primary studies and review articles for additional references.
Data collection and analysis
We conducted the review according to the recommendations from theCochrane Handbook of Systematic Reviews of Interventions (Higgins 2011a).
Selection of studies
Two review authors (KJ and CW) independently screened the titles and abstracts of all the studies identified by the primary search and then obtained the full text of relevant studies. Both authors independently went through the studies while considering inclusion criteria to decide whether to include or exclude the studies. The authors planned to contact the investigators for more information for any study where the criteria for inclusion were unclear. They planned to resolve any disagreement by discussion, referral to the third review author and consensus. They have described the reasons for exclusion of the studies, initially considered eligible for inclusion according to title and abstract.
Data extraction and management
One review author (KJ ) extracted the data from the only included study; the second review author (CW) was familiar with the data (being the author of the study). They completed a pre-decided data extraction form. They resolved any disagreements by discussion, referral to the third review author and consensus. Extracted information included:
administrative details including authors, year of publication, country of study;
patient characteristics including number of patients in each group, age, gender, weight, BMI, concomitant morbidities and other baseline characteristics mentioned in the studies;
study characteristics including design of study, inclusion and exclusion criteria, duration of follow up, co-morbidities, primary and secondary outcome measures;
details of intervention techniques used for the collection of specimens and the description of any adverse effects;
data for primary and secondary outcome measures.
The review authors created a study flow diagram following the template described in the PRISMA statement (Moher 2009). They also completed a table of 'Characteristics of included studies' including information about study design, relevant information on the demographics and health of participants and a list of interventions and outcome measures.
If suitable data were available, they would have undertaken a meta-analysis for each outcome. Where possible, they have presented the results in a graph, or otherwise in a narrative way.
The authors planned to report data at two weeks (after finishing the antibiotic course), three months, six months, one year and annually thereafter. If there were studies reporting data at other time points, they would have considered reporting those as well. In the current version of the review, the authors have presented data at the five-year time-point.
Assessment of risk of bias in included studies
Two authors independently assessed the studies fulfilling the inclusion criteria for risk of bias as per guidelines from theCochrane Handbook of Systematic Reviews of Interventions (Higgins 2011b). They assessed studies for risk of bias according to a standardised set of questions covering the following domains:
random sequence generation;
blinding of outcome assessment for the primary outcomes for HRCT assessment and lung function parameters;
incomplete data outcome;
selective reporting bias;
other sources of bias.
They classified the responses from each category into one of three grades (low risk of bias, high risk of bias and unclear risk of bias) and generated a risk of bias graph.
Measures of treatment effect
The authors planned to analyse binary data using risk ratio (RR) and 95% confidence intervals (CIs). They used the mean difference (MD) with 95% CIs to analyse continuous data. The MD measures the absolute difference between the mean value in two groups when outcome measurements in all studies use the same scale; they would have used the standardized MD when the studies were assessing the same outcome but measuring it in a variety of ways. We used time-to-event analysis using hazard ratios (HRs) with 95% CIs, e.g. for time to acquisition of chronic infection with P. aeruginosa. With reference to count data, e.g. for the number of isolates of organisms, we treated the data as continuous data and measured the intervention effect using the MD and 95% CIs between the groups. By convention, the changes during the study were reported as the effect measured post-intervention minus that measured pre-intervention; and differences between the study arms were reported as treatment arm effects minus control arm effects.
Unit of analysis issues
If the review authors identify further studies in the future, they plan to include cluster-randomised studies if the clustering has been taken into account and the intra-cluster coefficient is included. For the analysis process they will calculate the design effect to allow calculation of an effective sample size or an inflated standard error. They will consider estimates from an appropriately analysed cluster-randomised controlled study, e.g. effects from a mixed model. They plan to use generic inverse variance output method in these contexts.
The review authors planned not to include cross-over studies in the review, since CF is a chronic disorder with progressive worsening of lung disease following repeated infections. Some of the infections like P. aeruginosa, once established, cannot be eradicated and are associated with worse outcome. In addition, for outcome measures such as change in lung function and structural lung damage it may not be possible to revert completely to the same level, even with treatment. In view of this, the authors felt it was justified not to include cross-over studies.
For events happening multiple times, such as pulmonary infections, the unit of analysis will be individual patients (except in some cases e.g. the number of isolates of organisms per year, where we will analyse the number of events).
Dealing with missing data
In the included study, there were no outcome data available for fewer than 10% of the recruited patients. In future, when the authors include more studies, in case of missing data, they will attempt to contact the primary author of any such studies and if they do not receive any responses, they will perform an intention-to-treat analysis where possible.
Assessment of heterogeneity
If the authors identify more studies in future, they will use the chi2 test in the forest plot and also the I2 test for assessing the heterogeneity of results (Higgins 2003).
According to chapter 9 of theCochrane Handbook of Systematic Reviews of Interventions (Deeks 2011), a rough guide to interpretation of thresholds for the I2 statistic is as follows:
0% to 40%: might not be important;
30% to 60%: may represent moderate heterogeneity*;
50% to 90%: may represent substantial heterogeneity*:
75% to 100%: considerable heterogeneity*.
* The importance of the observed value of I2depends on (i) magnitude and direction of effects and (ii) strength of evidence for heterogeneity (e.g. P value of chi2 test or a CI for I2).
Assessment of reporting biases
Two review authors (KJ and AS) examined the included study for any evidence of selective outcome reporting bias by comparing the outcomes described in the protocol with the results published. In future, when they include more studies, they will construct a matrix indicating which outcomes were recorded in each study to establish if any studies omitted any key outcomes. They will compare the protocols (if available) or the 'Methods' section of published articles fulfilling the inclusion criteria with their 'Results' sections to establish selective reporting of outcomes that were pre-specified. When there is suspicion of, or direct evidence of, selective outcome reporting, they will contact the study authors to provide the study protocol and full information for the outcomes reported inadequately.
If there are a sufficient number of studies included (at least 10), then the review authors will construct a funnel plot and assess this to help identify evidence of publication bias or any other type of bias.
The authors followed the guidelines from theCochrane Handbook of Systematic Reviews of Interventions to conduct the comparison between the bronchoscopy-guided and the standard treatment groups covering all the primary and secondary outcome measures (Higgins 2011a).
If they had combined more than a single data set and had identified no heterogeneity or only mild to moderate heterogeneity based on I2 test results (Higgins 2003), they would have used a fixed-effect analysis model; if the value of I2 had been between 50% and 75% they would have performed a random-effects analysis. If the I2 values had been greater than 75%, they would not have considered it suitable to pool the studies together.
For each outcome for which the authors were able to extract suitable data, they constructed a forest plot to display effect estimates and 95% CIs using a fixed-effect model of analysis.
They have summarized the evidence in a summary of findings table (Summary of findings for the main comparison).
Subgroup analysis and investigation of heterogeneity
In future, if the authors include sufficient studies (at least 10), they plan to undertake the following subgroup analyses:
age groups: 0 to 5 years, 6 to 16 years and 17 years and above;
chronic or intermittent isolation of P. aeruginosa versus no P. aeruginosa;
different methods of sampling, e.g. BAL versus PBB or sputum culture versus throat swab.
In future, if the authors include sufficient studies (at least 10), they plan to undertake the following sensitivity analyses:
to analyse the effect of bias including effects of sequence generation, allocation concealment, intention-to-treat analysis and reporting bias which are not resolved after contacting the authors;
to assess the differences between fixed-effect and random-effects analysis on the results.